FUNCTIONAL CONTROL OF SOLUBLE RHODOPSIN MIMICS USING HIGH RESOLUTION STRUCTURE-BASED DESIGN AND EVALUATION By Alireza Ghanbarpour A DISSERTATION Michigan State University in partial fulfillment of the requirements Submitted to for the degree of Chemistry – Doctor of Philosophy 2019 ABSTRACT FUNCTIONAL CONTROL OF SOLUBLE RHODOPSIN MIMICS USING HIGH RESOLUTION STRUCTURE-BASED DESIGN AND EVALUATION By Alireza Ghanbarpour Visualizing the microenvironments of protein/small molecule interactions is the missing link in evaluating the structure-function relationship in many “redesigned” protein systems requiring small molecule binding. The primary goal of this thesis is to manage protein/small molecule interactions to achieve new functions through rational protein engineering in a protein scaffold, which is “evolutionarily naive.” The snapshots of each engineering step are collected using high resolution protein crystallography, opening doors to the design strategies of future measures. Finally, the mechanism of the system is elucidated by connecting structural information and biochemical assays. The protein scaffolds used in our study are hCRBPII and CRABPII, belonging to the iLBP protein family. By reengineering their binding pockets, we generated a rhodopsin mimic ligating with small molecules with aldehyde functionalities through protonated Schiff base formation. In Chapter I, we employ the aforementioned strategy to create a new model system based on reengineered CRABPII, mimicking the critical steps of microbial rhodopsin isomerization in a single crystal. Using atomic resolution crystal structures, different mechanisms of retinal/protein interactions with light are demonstrated. Specially, a new photoswitchable protein is identified that does not require chromophore isomerization or a conformational change. In Chapter II, the effect of ligand binding on the conformational states of the domain-swapped dimer of hCRBPII is investigated. A new protein conformational switch is created through a designed disulfide bond that can be activated and adopt new conformations in response to retinal/fatty acid binding and/or reduction potential of the environment. A novel allosterically regulated zinc-binding site is engineered, whose binding affinity can be tuned by the conformational states of our protein. Additionally, using merocyanine, a synthetic fluorophore, a new “swap back” domain-swapped dimer is identified in hCRBPII at atomic resolution, leading to the largest conformational change in the protein. This demonstrates the power of our system to adopt new conformations with different small molecules. Through systematic mutational studies and high resolution crystal structures, the role of the hinge loop region in imposing new conformations/functions in the iLBP family is explored. In Chapter III, the discovery of the domain-swapped trimer as an unprecedented fold for the iLBP family is mentioned. Through a designed disulfide bond and metal- binding site, we are able to favor trimer formation. The mechanism of each step is examined using crystal structures and binding and stability assays. Finally, in Chapter IV, the mechanism of a new class of fluorescent protein tags using the hCRBPII rhodopsin mimic bound with synthetic fluorophores is inspected. By exploiting high resolution crystallography, the microenvironments of protein/ligand interactions is visualized in different fluorescent protein tags applications. ACKNOWLEDGMENTS I would like to thank my undergraduate and master’s supervisors, Professor Hamid Reza Khavasi, Professor Mostafa Pouramini, from Shahid Beheshti University, and Dr. Mojtaba Bagherzadeh from Sharif University of Technology, who gave me the chance to begin my scientific journey as a researcher in their labs. I am very thankful to my PhD supervisor, Professor James Geiger, who provided me with great knowledge and resources and helped me visualize protein structures from a unique perspective. Through our many conversations, he not only educated me on how to be a structural biologist, but also taught me what it means to approach science with humility. I am very grateful for all the support and invaluable advice I received from him in every step of my education. As a first year PhD student, I received helpful guidance from the senior graduate students, Dr. Nosrati, Dr. Assar, and Dr. Yapici. In particular, Dr. Nosrati played an essential role in my training. Many of my projects would have been hard to accomplish without the help of my talented collaborators, who gave me immense support from beginning to end. I would like to thank Dr. Nairat, who assisted me with laser experiments, Dr. Pinger, who assisted me with the radioactive zinc labelling assay, Dr. Santos, who generously helped me with the mutagenesis studies of many joined projects, Dr. Assar, who assisted me with the thermal stability assay, Professor Jin, who patiently provided support for the Isothermal calorimetry assay, and Mr. Adam Jenkins, who performed the time-resolved experiment on our system from Professor Larsen’s lab at UC Davis. iv I am very grateful to Dr. Chrysoula Vasileiou for providing invaluable ongoing guidance and feedback. I would like to express my sincere gratitude to Dr. Borhan, who educated me in various aspects of my projects. He has had a great influence on me as a researcher by helping me to think critically about my science and develop outstanding ideas during my projects. I would like to thank my other committee members, Professor Jian Hu, who generously gave me freedom to use the mosquito robot in his lab at any time, and Professor Daniel Jones, who kindly provided support and consultation during my PhD studies. Lastly, I am very thankful for my family, especially my wife and my parents, who have always been supportive, loving, understanding, and encouraging during my education. v TABLE OF CONTENTS LIST OF TABLES ........................................................................................................... ix LIST OF FIGURES ....................................................................................................... xii KEY TO ABBREVIATIONS ........................................................................................ xxvii CHAPTER I: MIMICKING MICROBIAL RHODOPSIN ISOMERIZATION IN A SINGLE CRYSTAL ................................................................................... 1 I-1. Rhodopsin and retinal isomerization ...................................................................... 1 I-2. Microbial rhodopsin features ................................................................................ 5 I-2-1. Bacteriorhodopsin photocycle ..................................................................... 8 I-3. Engineering hCRABPII to mimic microbial rhodopsin isomerization .................... 11 I-3-1. Choosing an initial protein template ........................................................ 11 I-3-2.Thermal stability of 13-cis retinal in the crystal ........................................... 23 I-3-3. Atomic resolution crystal structure of 13-cis retinal in the dark .................. 25 I-3-4. Mechanistic insight and prediction ............................................................. 27 I-3-5. Irradiation study of M1-L121E:A32Y .......................................................... 32 I-3-6. Other mutational studies on M1 construct ................................................. 34 I-4. 13-cis bound hCRBPII crystal structure .............................................................. 38 I-5. Time-resolved x-ray crystallography experiment with rhodopsin mimic .............. 39 I-6. Conclusion and future direction .......................................................................... 43 I-7. Experimental procedures .................................................................................... 44 I-7-1. Materials and methods ............................................................................ 44 I-7-2. List of primers .......................................................................................... 45 I-7-3. Transformation of PCR product ................................................................. 47 I-7-4. Expression of hCRABPII variants .............................................................. 47 I-7-5. Protein isolation and purification ................................................................ 48 I-7-6. UV-vis measurements ............................................................................... 48 I-7-7. pKa determination of mutants .................................................................... 49 I-7-8. Irradiation of protein/retinal complexes in solution ..................................... 49 I-7-9. Protein crystallization ................................................................................ 49 REFERENCES .............................................................................................................. 55 CHAPTER II: CREATING A NEW CLASS OF PROTEIN CONFORMATIONAL SWICTHES ............................................................. 63 II-1. Protein conformational switches ......................................................................... 63 II-2. Using human cellular retinol binding protein II as an engineering scaffold ......... 69 II-2-1. Crystallization of apo and holo retinal-bound variants of hCRBPII ......... 72 II-2-2. Elucidating the mechanism of protein switch .......................................... 76 II-3. Creating a disulfide bond as a trigger for allosteric behavior ............................ 80 II-4. Generating an allosterically regulated zinc binding site .................................... 85 II-5. Using Fatty acids as input signals ...................................................................... 91 II-6. The observation of a ligand-induced “swap back” at atomic resolution .............. 94 II-7. Hinge loop region: nature’s economic way to create conformationally distinct vi proteins? .................................................................................................................. 100 II-7-1. Choosing a suitable construct .............................................................. 102 II-7-2. “Phase problem” in the domain-swapped dimer .................................. 102 II-7-3. Amino acid insertion: odd vs. even number of insertions ..................... 104 II-7-4. The effect of the nature of amino acid insertion ..................................... 106 II-7-5. Investigating the effect of the insertion position on the conformational state .................................................................................................................................... 107 II-7-6. The effect of insertion on the monomer fold ........................................... 107 II-8. Using domain-swapped dimer as a functional regulator ................................... 109 II-9. Does the domain-swapped dimer of hCRBPII possess a novel functionality in humans? .................................................................................................................. 111 II-10. Conclusion and future direction ...................................................................... 112 II-11. Experimental Procedures ............................................................................... 114 II-11-1. PCR protocol ...................................................................................... 114 II-11-2. List of primers ..................................................................................... 115 II-11-3. Hinge loop insertion primers and PCR protocol .................................. 116 II-11-4. Transformation of PCR product .......................................................... 117 II-11-5. DNA purification ................................................................................. 118 II-11-6. Expression of hCRBPII mutants ......................................................... 118 II-11-7. Protein isolation and purification ......................................................... 118 II-11-8. Crystallization, data collection, and refinement .................................. 120 II-11-9. UV-vis measurements ........................................................................ 122 II-11-10. Fluorescence quenching assay ........................................................ 123 II-11-11. CD spectroscopy .............................................................................. 123 II-11-12. Radioactive 65ZnCl2 labelling assay .................................................. 123 II-11-13. ITC experiment ................................................................................. 125 REFERENCES ........................................................................................................... 137 CHAPTER III: DISCOVERY OF A DOMAIN-SWAPPED TRIMER AS A NOVEL FOLD FOR THE ILBP FAMILY .................................................. 142 III-1. Oligomerization in biology: an efficient regulator for many biological processes .................................................................................................................................... 142 III-2. Domain swapping ............................................................................................ 143 III-3. Reviewing several important examples of domain-swapped proteins ............. 145 III-4. Domain swapping in the iLBP family: a novel tool for protein engineering ...... 149 III-5. Discovery of domain-swapped trimer .............................................................. 151 III-6. Stability of dimer vs. trimer .............................................................................. 158 III-7. Effect of ligand binding on trimer structure ...................................................... 159 III-8. A cross-subunit disulfide bond drives trimerization .......................................... 162 III-9. Introduction of a metal-binding site in the domain-swapped trimer .................. 164 III-10. Using “designed disulfide bond” to favor the domain-swapped oligomeric forms of other iLBP family members .................................................................................. 167 III-11. Conclusion and future direction ..................................................................... 168 III-12. Experimental Procedures .............................................................................. 169 III-12-1. DNA and protein purification ................................................................ 169 III-12-2. PCR conditions for hCRBPII ................................................................ 169 vii III-12-3. List of primers .................................................................................... 169 III-12-4. Transformation of PCR product ......................................................... 170 III-12-5. DNA purification ................................................................................ 170 III-12-6. Expression of hCRBPII mutants ........................................................ 171 III-12-7. protein isolation and purification .......................................................... 171 III-12-8. Crystallization, data collection, and refinement ................................. 172 III-12-9. Radioactive 65ZnCl2 labelling assay ................................................... 173 REFERENCES ........................................................................................................... 179 CHAPTER IV: ELUCIDATING THE MECHANSIM OF HCRBPII-BASED FLUORESCENT PROTEIN TAGS USING HIGH-RESOLUTION CRYSTAL STRUCTURES.......................................................................................... 186 IV-1. Fluorescent protein tags ................................................................................... 186 IV-2. Elucidating the electronic effects on the photophysics of hCRBPII-based fluorescent proteins .................................................................................................. 188 IV-3. Using high-resolution protein crystallography to optimize a ratiometric pH sensor .................................................................................................................................... 200 IV-4. Elucidating the mechanism of excited-state proton transfer using high-resolution crystal structures....................................................................................................... 207 IV-5. Conclusion and future direction ........................................................................ 213 IV-6. Experimental procedures ................................................................................. 214 IV-6-1. ThioFluor analog synthesis ................................................................. 214 IV-6-2. FR1 synthesis ...................................................................................... 214 IV-6-3. protein expression and purification ...................................................... 214 IV-6-4. Crystallization conditions ...................................................................... 214 REFERENCES ........................................................................................................... 221 viii LIST OF TABLES Table I-1. The initial efforts at obtaining a 13-cis bound crystal structure. (Other than the first four variants of hCRBPII, the rest of the mutants belong to CRABPII) .................... 14 Table I-2. Other mutational studies on M1 construct ..................................................... 35 Table I-3. PCR protocol ................................................................................................. 44 Table I-4. PCR program. ............................................................................................... 45 Table I-5. Crystallographic data collection and refinement statistics for various dark and light states of all-trans bound M1-L121E ................................................................ 51 Table I-6. Crystallographic data collection and refinement statistics for 13-cis bound M1- L121E:A32Y, and all-trans bound M1-L121W, M1-L121Y, and M1-L121Q dark states . 52 Table I-7. Crystallographic data collection and refinement statistics for different light states of all-trans bound M1-L121Q .............................................................................. 53 Table I-8. Crystallographic data collection and refinement statistics for dark and light states of all-trans bound M1-L121E:P39Q ..................................................................... 54 Table II-1. Summary of apo and holo domain-swapped dimer crystal structures for several hCRBPII mutants. ......................................................................................................... 71 Table II-2. Variants constructed in search of an allosteric metal binding site. ............... 87 Table II-3. Insertion in the hinge loop region and other variants constructed in search of an allosteric metal binding site ....................................................................................... 88 Table II-4. PCR condition ............................................................................................ 114 Table II-5. PCR program ............................................................................................. 115 Table II-6. X-ray crystallography data and refinement statistics of the holo dimer Q108K:T51D, Q108K:K40L:T51F, Q108K:K40L:T51W structures. ............................. 126 Table II-7. X-ray crystallographic data and refinement statistics for holo dimer hCRBPII Q108K:K40L:T51F:Y60A Q108K:T51D:A28H, hCRBPII-WT, and Q108K:T51D:A28C mutants.. ...................................................................................................................... 127 ix and zinc-bound Q108K:T51D:A28C:L36C:F57H, Table II-8. X-ray crystallographic data and refinement statistics for apo dimer hCRBPII Q108K:T51D, Q108K:K40L:T51F, Q108K:K40L:T51W, and Q108K:T51D:A28H mutants. .................................................................................................................................... 128 Table II-9. X-ray crystallographic data and refinement statistics for apo dimer Q108K:T51D:A28C, reduced Q108K:T51D:A28C:L36C:F57H hCRBPII mutants ...................................................... 129 Table II-10. X-ray crystallography data and refinement statistics of the apo dimer 1H, 3H, 6H, and 3W mutants of hCRBPII........................................................................... 130 Table II-11. X-ray crystallographic data and refinement statistics for apo hCRBPII 3H’, WT-3H, WT-6H, and Q108K:T51D:A28C:L36C hCRBPII mutants .............................. 131 Table II-12. X-ray crystallographic data and refinement statistics for domain-swapped dimer of A1 mutant in apo, retinal bound, and merocyanine bound forms................... 132 Table II-13. X-ray crystallographic data and refinement statistics for Q108R:T51D mutant bound with fatty acids .................................................................................................. 133 Table III-1. Ratios of monomer, dimer, and trimer formations in different variants ...... 153 Table III-2. Thermal stability assay for different variants ............................................. 159 Table III-3. X-ray crystallographic data and refinement statistics for apo trimer Q108K:T51D:A28C:I32C, A1 and Zinc-bound A1-F57H hCRBPII mutants ................. 175 Table III-4. X-ray crystallographic data and refinement statistics for trimer hCRBPII mutants Q108K:T51D, A2, and Q108K:K40L:T51K..................................................... 176 Table III-5. X-ray crystallographic data and refinement statistics for apo dimer hCRBPII Q108K:T51D:I32C and A1-F57H mutants ................................................................... 177 Table III-6. X-ray crystallographic data and refinement statistics for apo side-by-side dimer of hCRABPII I31C mutant .................................................................................. 178 Table IV-1. Spectroscopic properties of ThioFluor-PSB analogs coupled with L1:L117E. (The following data was prepared by Dr. Santos.) ....................................................... 189 Table IV-2. Effect of L117D versus L117E in the development toward a single fluorescent protein ratiometric sensor ............................................................................................ 202 Table IV-3. X-ray crystallographic data and refinement statistics for hCRBPII L1:L117E bound with ThioFluor2 , ThioFluor3, and ThioFluor4 analogs ................................. 216 x Table IV-4. X-ray crystallographic data and refinement statistics for hCRBPII L1:L117E variants bound with ThioFluor6 , ThioFluor7, and ThioFluor8 analogs .................... 217 Table IV-5. X-ray crystallographic data and refinement statistics for hCRBPII variants bound with ThioFluor analogs. ................................................................................... 218 Table IV-6. X-ray crystallographic data and refinement statistics for hCRBPII variants L1:L117E, L1:L117D, and L1:L117D:Q4F bound with ThioFluor analogs.................. 219 Table IV-7. X-ray crystallographic data and refinement statistics for hCRBPII L2, L3, and Q108K:K40D:T53A:R58L:Q38F:Q4F variants bound with ThioFluor and FR1, and ThioFluor3 analogs..................................................................................................... 220 xi LIST OF FIGURES Figure I-1. a. The LOV domain bound with flavin chromophore. b. Phytochrome bound with a bilin chromophore. c. Rhodopsin that is covalently attached to the protein through Lsy216. The chromophores have been shown with stick representation (carbons in blue, nitrogen in dark blue, oxygen in red, and phosphorous in orange). ................................. 1 Figure I-2. Photoisomerization of the chromophore-bound protein in: a. microbial rhodopsin; b. animal rhodopsin ........................................................................................ 3 Figure I-3. Two different possible conformations for 11-cis retinal in solution: 11-cis-12- s-trans and 11-cis-12-s-cis. 11-cis-12-s-trans conformations generate a steric between the hydrogen of C10 and methyl group of C13 ................................................................ 4 Figure I-4. Bacteriorhodopsin binding pocket contains aromatic residues such as W182, W138, W189, W86, and Y185, that sandwich the chromophore through aromatic interaction. D85 and D212 act as a counter ion for the iminium, stabilizing the positive charge. The water molecules shown in red mediate the hydrogen bond between the nitrogen of the iminium and D85 and D212. Retinal is shown in cyan, and the protein backbone carbon, nitrogen, and sulfur atoms are shown in purple-blue, dark blue, and dark yellow, respectively (PDB code:1C3W) ................................................................... 6 Figure I-5. a. Photocycle of bacteriorhodopsin, where different intermediates form during the photocycle. The lifetime of each is shown underneath the retinal structure. b. The residues that are important in proton-pumping activity, protonation and deprotonation of iminium, are shown as transparent purple spheres. Retinal is shown as cyan spheres. R82, E194, and E204 are referred to the proton releasing complex, facilitating the proton transfer to the bulk. .......................................................................................................... 8 M1 Figure retinal. The photoinduced (R111K:R134F:T54V:R132Q:P39Y:R59Y) bound isomerization of the imine functionality leads to changes in its pKa and consequently its protonation state. The hydrophilic and hydrophobic environment of the imine nitrogen atom for the cis and trans geometry, respectively, is highlighted in the figure obtained from the crystal structure of each form (PDB IDs: 4YFP and 4YFQ) ............................. 12 Figure I-7. a. Electron density of M1 bound with 13-cis retinal showed electron density that is more similar to all-trans retinal. b. HPLC chromatogram of retinals extracted from 13-cis bound M1 after 24 h of incubation in the dark demonstrated the dark isomerization of 13-cis to all-trans retinal. c. UV-vis spectrum of M1-L121E bound with all-trans retinal. d. UV-vis spectrum of M1-L121E bound with 13-cis retinal. e. Photo-irradiation of 13-cis retinal with UV (B.P. filter, 300-400nm), green light (L.P. filter > 500 nm) ..................... 15 hexamutant hCRABPII to I-6. Photocycle of xii Figure I-8. a. Binding pocket of bacteriorhodopsin, where W86 is located in the vicinity of C13-C14. b. hCRABPII binding pocket. The L121 position is analogous to the W86 position in bacteriorhodopsin relative to retinal .............................................................. 15 Figure I-9. Overlay of M1 (yellow) and M1-L121W (cyan). Introducing Trp in the L121 position impinges on the bound retinal, which leads to the concomitant change in retinal trajectory ........................................................................................................................ 16 Figure I-10. UV-vis spectra of M1 and M1-L121E upon incubation with: a. all-trans and b.13-cis retinal. The SB (Schiff Base, Imine) and PSB (Protonated Schiff Base, Iminium) peaks are highlighted..................................................................................................... 16 Figure I-11. a. UV-vis absorption spectrum of all-trans-retinal bound M1-L121E in the dark state (black) and after green light irradiation (long pass filter > 500 nm) (green), blue- light irradiation using band pass filter (440 nm ± 20 nm) (blue), and UV irradiation (UV band pass filter (300-400 nm)) (cyan). b. Hydrogen bonding network between the imine hydrogen of R111K and L121E, P39Y, H2O-303, R132Q, and S12. c. Overlay of M1 (carbon atoms in cyan) with all-trans-retinal bound M1-L121E (carbon atoms in green). d. Electron density (contoured at 1σ) of all-trans-retinylidene in M1-L121E in the dark state (green carbons) vs. 13-cis-15-syn-retinal imine (magenta carbons) generated after five- minute laser irradiation at ~400 nm. e. Overlay of all-trans-retinylidine-bound M1-L121E in the dark (green) with the 13-cis-15-syn imine (magenta) generated after five-minute laser irradiation at ~399 nm, showing the movement of Lys111, Glu121, and the rotation of the β-ionone ring in 13-cis upon isomerization. f. Overlay of all-trans-retinal bound M1- L121E imine (green) and all-trans-retinal bound bacteriorhodopsin (yellow carbons, PDB code: 1C3W), in bacteriorhodopsin relative to retinal. All heteroatoms are colored by type, with N in blue and O in red ................................................................................................................... 18 Figure I-12. HPLC chromatogram of retinal isomers extracted from all-trans retinal bound M1-L121E (recorded at 363 nm). a. Dark, after irradiation with mercury lamp- green light, blue light, UV, and coinjection with the pure 13-cis retinal (red).The ratios were corrected using the extinction coefficient of each isomer (all-trans 48,000 M-1cm-1, 13-cis 38,770 M- 1cm-1, and 9,13-dicis retinal 32,380 M-1cm-1) ................................................................. 19 Figure I-13. HPLC chromatogram of M1-L121E bound with all-trans retinal before irradiation, after 30 sec laser irradiation, and after 5 min laser irradiation (recorded at 363 nm). The ratios were corrected using the extinction coefficient of each isomer. The ratios were corrected using the extinction coefficient of each isomer (all-trans 48,000 M-1cm-1, 13-cis 38,770 M-1cm-1, and 9,13-dicis retinal 32,380 M-1cm-1) ....................................... 20 Figure I-14. 1H-NMR spectrum of retinal extracted from the protein complex recorded in CDCl3 solvent ................................................................................................................ 21 Figure I-15. a. Overlay of the dark state M1-L121E (pink) and after 30 sec laser irradiation (green-13-cis) revealing the changes in the polyene and ionone ring conformation of interaction mimics Trp86 illustrating how the P39Y-L121E xiii retinal after laser irradiation. Electron density belongs to the crystal irradiated for 30 sec (countered at 1σ). b. Overlay of 30 sec irradiation structure (green) with five min irradiation with laser (cyan) of M1-L121E, demonstrating the similar ionone ring and polyene conformation and verifying the formation of 13-cis after 30 sec laser irradiation. Electron density belongs to the crystal irradiated for 30 sec (countered at 1σ). .......................... 23 Figure I-16. a. All-trans retinal-bound M1-L121E crystal after 30 sec irradiation with laser (399 nm) and subsequent incubation in the dark for 10 minutes (electron density contoured at 1σ). This crystal structure indicates the regeneration of all-trans retinal through the thermal isomerization of 13-cis. Although the chromophore adopted the all- trans retinal conformation, L121E does not reset completely. b. Structure from an all-trans retinal-bound M1-L121E crystal after 30 sec irradiation with laser (399 nm) and subsequent incubation in the dark for 25 minutes (electron density contoured at 1σ), showing the regeneration of all-trans retinal and L121E adopted the same conformation as the initial dark state conformation. c. Overlay of structures from all-trans bound M1- L121E crystal (dark state, in pink) and M1-L121E irradiated with the laser (399nm) for 30 sec and subsequently incubated in the dark for 25 minutes (purple) verifies the similarity of the retinal trajectory in both structures, showing the thermal isomerization of the 13-cis created after 30 sec laser irradiation to produce all-trans retinal (RMSD=0.1). The electron density (contoured at 1σ) was calculated from the structure after 25 minutes incubation in the dark ...................................................................................................................... 24 Figure I-17. a. Aromatic residues located in the binding pocket of 13-cis bound M1- L121E:A32Y; b. Bacteriorhodopsin binding pocket where the chromophore is sandwiched with aromatic residues ................................................................................................... 25 Figure I-18. a. 13-cis electron density in M1-L121E:A32Y variant countered at 1.5 σ. The atoms are colored by type, with the C atoms in yellow. b. UV-vis spectrum of M1- L121E:A32Y bound with the 13-cis retinal, depicting the stable PSB after 10 hours of incubation. c. Overlay of 13-cis bound M1:L121E:A32Y (yellow) with all-trans bound M1:L121E (purple) without irradiation of either. d. Overlay of 13-cis retinal bound M1- L121E-A32Y (yellow) with the structure from a crystal of all-trans bound M1-L121E that was irradiated with laser (5 min) and immediately frozen (cyan) .................................. 26 Figure I-19. Overlay of M1-L121E (purple-blue) with M1-L121E:P39Q (pink). The P39Q mutation does not increase the steric interaction in the vicinity of the C13-C14 double bond. Furthermore, in the absence of P39Y, the chromophore does not move to a trajectory seen for M1-L121E, highlighting the essential role of P39Y for defining the chromophore trajectory in M1-L121E and presumably directing the isomerization toward 13-cis isomer formation ................................................................................................. 27 Figure I-20. a. UV-vis spectrum of M1-P39Q:L121E in the dark, after UV (B.P.300-400 nm) irradiation, and after subsequent green-light irradiation (L.P 500 nm). b. Drastic color change in the crystal upon irradiation with the UV. c. Overlay of dark state (pink) and UV- irradiated state (yellow) of M1-P39Q:L121E crystal structures, depicting the all-trans xiv conformation for both states. d. Overlay of UV-irradiated crystal structure of M1- P39Q:L121E vs. the light state of M1-L121E ................................................................. 29 Figure I-21. a. UV-vis spectra of M1-L121Q before and after irradiation with green light, showing a high/low pKa system where green-light irradiation leads to decreased PSB and increased SB absorption. b. Crystal structure of M1-L121Q obtained in the dark. L121Q has a direct hydrogen bond with R111K. c. Crystal structure of M1-L121Q after 5 min irradiation with 399 nm laser. L121Q and P39Y adopted two conformations but no isomerization occurred. This shows the importance of sidechain conformation stability in directing the photoisomerization product. L121Q in one conformation is far away from iminium (4.8 Å). d. After 10 min laser irradiation of crystals with 532 nm laser (close to PSB maximum absorption), the L121Q swings away from iminium, lowering the pKa of the chromophore and leading to the SB formation. e. Overlay of the all-trans-bound M1- L121Q imine before (green carbons) and after (salmon carbons) laser irradiation at 528 nm, clearly showing the Gln121 movement. f. Drastic color change of crystals upon irradiation with 532 nm laser. ......................................................................................... 30 Figure I-22. a. Dark state of M1-L121E:A32Y bound with 13-cis retinal, showing well- ordered water molecules mediated between the retinal and protein (electron density is contoured to 1 s). b. After 399 nm laser irradiation, no isomerization occurs; however, the Lys111 side chain and retinal polyene move vertically ~ 0.7 Å, which leads to Glu121 movement and disordering of the water molecules (electron density is contoured to 1 s). c. Overlay of dark state (purple-blue) vs. light state demonstrated the subtle movement of the chromophore, Lys111 side chain, and Glu121, disordering the water molecules after irradiation. .............................................................................................................. 32 Figure I-23. a. Crystal structure of M1-L121N/all-trans retinal in the dark. b. Crystal structure of M1-L121N/all-trans retinal after 3 min laser irradiation with 399 nm laser. c. Crystal structure of M1-L121M /all-trans retinal in the dark (PDB code:6NNX). d. Crystal structure of M1-L121E:P39F/ all-trans retinal in the dark. e. Crystal structure of M1- L121E:P39E in the dark (PDB code:6NNY). f. Crystal structure of M1-L121E:I63D in the dark (PDB code: 6NOE). The presence of I63D can potentially generate a hydrogen bond network on the other side of the chromophore opposed to L121E. The electron density was contoured to 1s for the 2FO-FC map. ...................................................................... 34 Figure I-24. UV-vis photocycle spectra of different all-trans retinal-bound variants: a. M1- L121N, b. M1-L121D, c. M1-L121E:P39N, d. M1-L121E:P39F, and e. M1-L121E:P39E. UV (B.P 300-400 nm) and green light irradiation (Y filter L.P>500 nm) were performed with a mercury lamp as the source of irradiation. .......................................................... 37 Figure I-25. Overlay of M1-L121E:A32Y-CRABPII (yellow) and Q108K:K40L-hCRBPII (blue) showing the equivalent positions between the two proteins in the retinal binding pocket b. Crystal structure of H1 bound with 13-cis retinal. The electron density is contoured to 1s. c. Overlay of H1 (blue) and M1-L121E:A32Y bound with 13-cis retinal, highlighting the aromatic interactions holding the chromophore. ................................... 38 xv Figure I-26. Simplified Potential Energy Surface for the forward (top) and reverse (bottom) dynamics of M1 initiated with UV excitation (forward) or green-light (reverse) excitation. Wavelengths are indications of the true peak wavelength for the intermediates (provided by Adam Jenkins). ......................................................................................... 39 Figure I-27. The pump-probe method was used for the time-resolved x-ray experiment. The laser was used to initiate the reactions in the crystals, and the x-ray with microsecond-duration pulses was used to monitor the changes in the crystals. The x-ray and laser are positioned perpendicular to each other to increase the efficiency of the experiment. Photo credit: Prof. Geiger........................................................................... 40 Figure I-28. a. A 2FO-FC Electron density map of the dark state of M1 collected at room temperature using the Laue method, contoured at 1σ for. The time-dependent difference maps were subtracted from the dark state: b. 50 µs, c. 100 µs, d. 500 µs, and e.1 ms . 41 Figure II-1. Natural protein conformational switches. a. The overlay of apo (cyan) and holo (pink) maltose binding protein. b. The crystal structure of calcium-bound conformation of calmodulin. c. The GPCR transmembrane crystal structure. ............... 63 Figure II-2. Schematic picture demonstrating the FRET between GFPs can measure Ca2+. The GFPs are drawn as cylinders. Top: calmodulin (CaM) without Ca2+ in the elongated conformation, disordered unbound M13 (calcium binding peptide) Bottom: The Ca2+-bound calmodulin in the closed conformation and Ca2+/M13 complex in the structured conformation. ................................................................................................ 64 Figure II-3. A circularly-permutated PDZ domain is linked to a fibronectin type III (FN3)- based enhancer domain that detects the ligand-bound form of the PDZ domain. A significant conformational change ensues upon PDZ ligand binding with the concomitant formation of a sandwich complex that can be measured through the FRET between the two fluorescent proteins YPet and CyPet. ..................................................................... 65 Figure II-4. Crystal structure of circularly-permutated GFP (purple) inserted inside the maltose-binding protein (cyan). PDB code: 3OSQ. ....................................................... 66 Figure II-5. Inserting maltose-binding protein (MBP) into a circularly permutated version of β-lactamase (BLA) giving rise to creation of a maltose sensor which regulates the activity of β-lactamase (BLA) as a function of maltose binding. ..................................... 67 Figure II-6. a. The crystal structure of LOV domain in dark. PDB code: 3T50. b. LOV domain photocycle is detected by a ground state oxidize flavin generating a flavin- cysteine adduct upon blue-light irradiation. The adduct reset to the ground state through UV irradiation or dark adaptation. .................................................................................. 68 Figure II-7. a. The monomeric form of hCRBPII bound with retinol (shown with cyan). The two alfa-helices decorating the ligand binding site were highlighted with green. (PDB xvi code: 4QZT) b. The domain-swapped dimer of hCRBPII. Each chain is color coded (PDB code: 4ZH9). .................................................................................................................. 70 Figure II-8. Overlay of WT-hCRBPII (holo) DS dimer chains H and D shown by dark blue and WT-hCRBPII (apo) (PDB code: 4ZH9) shown by green. b. Overlay of WT-hCRBPII (holo) chains F and J shown by yellow and WT-hCRBPII (apo) shown by green. c. Overlay of WT-hCRBPII (holo) chains B and L shown by red and WT-hCRBPII (apo) shown by green. d. Overlay of WT-hCRBPII (holo) chains E and L shown by orange and WT- hCRBPII (apo) shown by green. e. Overlay of WT-hCRBPII (holo) chains A and K shown by pink and WT-hCRBPII (apo) shown by green. f. Overlay of WT-hCRBPII (holo) chains E and I is shown by brown and WT-hCRBPII (apo) shown by green. ............................ 72 Figure II-9. a. Overlay of one chain each of four apo dimers: WT (cyan), Q108K:T51D (purple-blue), Q108K:K40L:T51F (yellow), and Q108K:K40L:T51W (pink), showing a relatively rigid conformation for apo dimers. b. Overlay of holo (purple) and apo (yellow) Q108K:T51D structures showing a large change in the relative orientations of the two domains upon ligand binding, leading to a 16 Å movement (inset) in the α-helix. Retinal (green) is shown in both sphere and stick representations. c. The retinal (green) binding pocket of apo (yellow C atoms) and retinal-bound (blue C atoms) Q108K:T51D. Allosteric conformational change is driven by the orientation of the Tyr60 and Asn59 sidechains. A flipped-in Asn59 would sterically clash with the bound retinylidene, leading to the flipped- out conformation of Asn59 and flipped-in conformation of Tyr60. All other atoms colored by type throughout: nitrogen (blue), oxygen (red). d. Overlay of apo Q108K:K40L:T51F (purple) and holo Q108K:K40L:T51F:Y60A (yellow) showing no change in the relative orientation of the two domains upon ligand binding. ...................................................... 74 Figure II-10. Retinylidene-bound hCRBPII undergo dramatic conformational change upon ligand binding. a. An overlay of apo Q108K:K40L:T51F (KLF) (purple) and holo KLF (chains A and C, both shown in pink), showing the large motion of helix 1 upon ligand binding. b. An overlay of holo KLF (pink) and holo Q108K:T51D (cyan), showing the modest conformational difference in the two holo structures. The relative distance between Cα of Arg 30 in the overlaid holo/apo structures is 11.3 Å while the same distance in the overlaid holo structures is 1 Å. *Bound retinal molecules in holo structures are shown as transparent spherical models. The complete dimer in the apo structures was generated by crystallographic two-fold symmetry operation. ......................................... 75 Figure II-11. Overlay of WT apo monomer (gray, PDB code: 2RCQ) and WT apo hCRBPII dimer (purple, PDB code: 4ZH9). The “flipped in” Asn59 conformation is required for the relative orientation of the two halves of an open monomer to adopt the orientation required to form the symmetric DS dimer. ................................................................................... 76 Figure (purple) and Q108K:K40L:T51F:Y60A (holo) (yellow) shows no change in the relative orientation of two domains upon ligand binding. b. The retinal bound Q108K:K40L:T51F:Y60A DS dimer. The electron density is observed only for the first 5 carbons of retinal illustrating that Y60A mutation makes the chromophore disordered inside of the binding pocket. II-12. a. The overlay of Q108K:K40L:T51F (apo) xvii Therefore, Asn 59 keeps its flipped-in conformation which does not lead to the change in the relative orientation of the two domains.*Bound retinal molecules in the holo structure are shown as transparent spherical models. The complete dimer in the symmetric DS dimers was generated by crystallographic two-fold symmetry operation. ..................... 77 Figure II-13. The difference between phi/psi angles of holo and apo DS dimers and WT hCRBPII monomer phi/psi angles. For holo structures, one of the chains from each mutant has been considered. Phi angle differences are shown in blue and psi angle differences are shown in red. a. Tyr 60. b. Asn59. c. Arg58. d. Phe57. e. Thr60. As shown from a to d, a large difference in the phi/psi angles between holo and apo structures is observed. ....................................................................................................................... 78 Figure II-14. Engineering a cross-link in the interface of two domains by A28C mutation: a. Overlay of Q108K:T51D-Apo (cyan) and Q108K:T51D:A28C-Apo (blue) showing the new conformation enforced by disulfide bond formation. b. A close-up showing the disulfide bond and the flipped-out conformation for Asn59 and flipped-in conformation for Tyr60. Inset: The disulfide bond electron density contoured at 1σ. c) SDS PAGE gel electrophoresis under reducing (1) and non-reducing (2) conditions, confirming disulfide bond formation in solution. ............................................................................................. 80 Figure II-15. An overlay of apo (orange) and holo Q108K:T51D:A28C (yellow), both featuring the Cys28 disulfide bond. On the right side, the bound retinal sterically forces the α-helices to move, inducing the ligand-induced conformational change. ................. 81 Figure II-16. Fluorescent quenching assay for Q108K:T51D and Q108K:T51D:A28C shows dissociation constants of 220 nM and 368 nM respectively for retinal. ............... 82 Figure II-17. Structural overlays of a. Q108K:T51D (KD) apo (yellow) versus KDA28C apo (blue). b. KD-A28C-holo (green) versus KD-A28C-Apo (blue). c. KD-A28C-holo (green) versus KD-holo (orange). All of these overlays show a large conformational change originated from either a disulfide bond or ligand binding or both. d. A summary of conformations potentially addressable by ligand binding, redox environment, or both. Only the c-termini of the proteins are represented. The distance between apo versus holo retinal bound structures is 12 Å, apo versus disulfide bond-apo is 16 Å, apo versus disulfide bond holo is 7.5 Å, and the apo with disulfide bond versus holo with disulfide bond is 10 Å. The apo and holo structures are represented by Q108K:T51D (cyan and blue, respectively) and the apo and holo disulfide-bonded structures are represented by Q108K:T51D:A28C (orange and yellow, respectively)................................................... 83 Figure II-18. The CD spectra of different variants: Q108:T51D dimer, Q108K:K40L:T51F:Y60A dimer, Q108K:T51D:A28C dimer, Q108K:K40L:T51V monomer, and Q108K:K40L:Y60L:A28C upon adding retinal or/and reducing agent. ... 84 Figure II-19. a. Designing a new metal-binding site using Q108K:T51D:A28C crystal structure as a guide. The proximity of F57H and L36C is critical to providing the metal- binding site. b. The crystal structure of holo retinal-bound Q108K:T51D chains G and I, xviii showing the large separation between two F57s from each chain upon ligand binding. c. Atomic resolution crystal structure of Q108K:T51D:A28C:L36C:F57H bound with zinc showing tetrahedral geometry for zinc coordinated with two histidines coming from two adjacent chains of dimer, a cysteine, and a water molecule. The electron density of the 2FO-FC map is countered to 1σ. The conformations of F57H that did not contribute to the metal-binding site have been omitted for clarity. ............................................................ 85 Figure II-20. a. The binding curve for apo AG1 using a radioactive-labeled 65Zn showed Kd=1.62 ± 0.14 µM. b. Comparison of association constants for different conformations calculated in 40 µM total 65ZnCl2 : 1) Apo; 2) in the presence of reducing agent only (Red- Apo), 3) in the presence of retinal (Holo). 4) in the presence of all-trans retinal and reducing agent (Red-Holo); As controls, Q108K:T51D (control 1) and Q108K:T51D:A28C (control 2) were used and led to a negligible association constant with Zn. c. ITC experiment shows the dissociation constant of 1.60 ± 0.22 µM for apo. d. The ITC experiment shows the dissociation constant of 29.3±11µM in the presence of all-trans retinal and reducing agent. e. Q108K:T51D:A28C was used as a control. ...................................................................................................................................... 89 Figure II-21. a. The overlay of Q108K:T51D:A28C:L36C:F57H reduced form (green) and KD-Met (purple-blue) showing no significant change in the conformation upon reducing the disulfide bond. b. Overlay of KD-Met (yellow) and holo Q108K:T51D (purple-blue) crystal structures demonstrating a large change anticipated upon retinal binding, giving rise to reorganization of the metal binding site to a conformation which is not optimal for metal binding. F57 residues of each chain in Q108K:T51D structure have been mutated using PyMOL. c. Overlay of zinc-bound apo Q108K:T51D:A28C:L36C:F57H (purple) (Kd=1.60 ± 0.22 µM) with holo and reduced form structure (Kd=29 ± 8 µM) modeled using Q108K:T51D (green) crystal structure demonstrating a large change anticipated upon retinal binding, giving rise to reorganization of the metal binding site to a conformation which is not optimal for metal binding. .......................................................................... 90 Figure II-22. a. The overlay of holo linoleic acid bound Q108R:T51D (purple) and Q108K:T51D (apo) domain-swapped dimer crystal structures. b. The same mechanism of dispensation of N59 and Y60 demonstrates that the large conformational change occurs upon fatty acid binding. c. The binding pocket of linoleic acid and the interacting residues were highlighted. d. The linoleic acid electron density was shown. (The 2FO-FC map was countered to 1σ.) Q108R is far away from the carboxylic acid group of linoleic acid and it not involved in the binding event. ................................................................. 91 Figure II-23. a. Overlay of arachidonic acid bound Q108R:T51D crystal structure (purple) with apo Q108K:T51D (yellow) showing the large conformational change in the relative orientation of the two dimer units. b. Overlay of linoleic acid bound structure vs. arachidonic acid bound structure demonstrates a similar trajectory for these two fatty acids in the binding pocket of the Q108R:T51D DS dimer variant of hCRBPII. c. Overlay of Q108R:T51D crystal structure bound with linoleic acid (pink) with FABP5 bound with N-(2-hydroxyethyl) icosanamide (PDB code: 4AZR). d. Overlay of Q108R:T51D crystal xix structure bound with linoleic acid (pink) with FABP4 bound with linoleic acid (yellow), PDB code: 2Q9S. .......................................................................................................... 92 Figure II-24. The structure of merocyanine chromophore with aldehyde functionality designed in Prof. Borhan’s lab. ...................................................................................... 94 Figure II-25. a. Overlay of apo DS dimer of A1 (purple) with WT-hCRBPII DS dimer (orange) showing a symmetric dimer (RMSD: 0.3.). b. Apo DS dimer of A1 (purple) with holo retinal-bound DS dimer of A1 (orange) revealing the large conformational change in the relative orientation of the two dimer domains. c. Overlay of holo retinal-bound DS dimer of A1 (orange) with holo retinal-bound DS dimer of Q108K:T51D (cyan). ........... 95 Figure II-26. a. The merocyanine-bound A1 crystal structure showing the swapped-back DS dimer. Chains are shown in green and cyan. The merocyanine ligand is represented by a combination of sphere and stick representation (pink). b. The overlay of merocyanine-bound A1 (green) and apo (purple-blue). c.The electron density highlighted for the ligand and K40D contoured to 1σ. d. Left: The normal domain-swapped dimer where half of the protein undergo the domain-swapping. The swapped open monomers showed with different colors. Right: The “swap back” dimer where the α-helix and beta strand 1 from each chain is swapped back highlighted with a box and light orange and purple colors. e. K40D has to move two residues along the strand to interact directly with the nitrogen of iminuim, giving rise to a change in the registry of the amino acids that make up strand 2 forcing the melting of the α-helix. f. Different positions of K40D in apo (purple-blue) vs. merocyanine-bound A1 DS dimer (green) in the overlaid structures. g. The positions of K40D in merocyanine-bound A1 DS dimer (green) vs. retinal-bound A1 (orange), chain I. h. The positions of K40D in merocyanine-bound A1 (green) vs. retinal- bound A1 (pink), chain i. The overlay of merocyanine-bound A1 DS dimer (green) vs. merocyanine-bound Q108K:K40L:T51V (purple blue). ................................................. 95 Figure II-27. a. Overlay of A1 DS dimer bound with merocyanine (green) and A1-K40F monomer bound with merocyanine (pink) portraying a significant change in the trajectory of the ligand. b. Overlay of apo A1 vs. ThioFluor9 chromophore-bound A1 DS dimer crystal structure shows no conformational change. c. UV. Vis spectrum of A1 bound with the merocyanine ligand. The protonated Schiff base absorbs at 584 nm. d. UV. Vis spectrum of A1-K40F bound with the merocyanine ligand. The protonated Schiff base absorbs at 595 nm.. ....................................................................................................... 98 Figure II-28. a. Q108:T51D:A28C:L36C crystal structure. Each chain is shown in a different color and is connected by the disulfide bond on the top part of the α-helices. b. Overlay of Q108:T51D:A28C:L36C crystal structure (purple) with WT hCRBPII domain- swapped dimer (green) showing a similar conformation to the symmetric dimer for Q108:T51D:A28C:L36C. c. Overlay of Q108:T51D:A28C (orange) with WT hCRBPII dimer showing a drastic change in the relative orientation of the two dimer units, leading to a large change in the c-termini distance between the two structures. d. Arrangement of the residues connecting strands 3 and 4 in the monomer; left panel shows the crystal structure of the monomer where the aforementioned strands were highlighted. e. xx Arrangement of the residues connecting strands 3 and 4 in the dimer to make a continuous strand leading to a phase problem; left panel shows the crystal structure of the dimer where the aforementioned strands were highlighted... ................................ 101 Figure II-29. a. Overlay of 1H (orange) and Q108K:T51D:A28C:L36C (purple) crystal structures. The c-terminus has been moved ~ 8 Å after introducing His between F57 and R58. b. Top: The hinge loop regions and part of the two β-strands (3 and 4) were highlighted in Q108K:T51D:A28C:L36C where both Y60 and N59 are located inside the binding pocket and cause a “phase problem”. Bottom: After insertion of one His between F57 and R58, a continuous β-strand is formed where every residue is “in phase”. c. Overlay of 3H (red) and Q108K:T51D:A28C:L36C (purple) crystal structures, representing a large movement of the c-termini and the relative orientation of the two dimer units. d. The “zoom-in” demonstration of the hinge region after the addition of three His, solving the “phase issue”. ..................................................................................... 102 Figure II-30. Overlay of 6H (yellow) and Q108K:T51D:A28C:L36C (purple) crystal structures. The c-terminus has been moved more than 23 Å after introducing six His between F57 and R58. b. Adding the even number of residues depicts the “phase problem,” where the two neighboring His are outside the binding pocket and “out of phase.” Also, three His are in the right conformation for generating a metal-binding site. c. Overlay of two open monomers of 3H generating a domain-swapped dimer (chain A, pink, and chain B, cyan). d. Overlay of two open monomers of 6H generating a domain- swapped dimer (chain A, green, and chain B, purple). e. Overlay of 3W (cyan) and 3H (pink) crystal structures, showing a novel conformation was created where the nature of amino acid insertion is different. f. Close-up view of the 3W hinge region, where three Trp residues were inserted between F57 and R58 and every residue is “in phase.” g. Overlay of 3H’ vs. 3H crystal structures, demonstrating the position of insertion is also highly important in the creation of a new conformation in the protein. h. Close-up view of 3H’ hinge region, where three His residues were inserted between R58 and N59 and every residue is “in phase”. ................................................................................................... 104 Figure II-31. a. Crystal structure of WT-3H, chain B. The melted α-helix is shown in purple. b. Crystal structure of WT-3H, chain A. The melted α-helix is shown in purple. c. Overlay of two chains of WT-3H. d. Crystal structure of WT-6H. e. Overlay of WT-3H (cyan), and WT-hCRBPII (orange) (PDB code: 2RCQ) The inserted histidines were shown with blue. f. Overlay of WT-6H (dark blue), and WT-hCRBPII (orange). .......... 108 Figure II-32. a. EGFP that is fused to the c-terminus of A1 hCRBPII. The gel filtration result demonstrates the production of the dimer in the presence of GFP when the retention time is compared to the biorad gel filtration standard. b. Biorad gel filtration standard components. ................................................................................................ 111 Figure II-33. The schematic design of HIV-protease fused to hCRBPII. a. Apo form of hCRBPII, where presumably HIV protease is in active form. b. Ligand-bound form of hCRBPII, which presumably causes a perturbation in the orientations of two dimer units. c. HIV protease crystal structure. PDB code: 2pc0. .................................................... 113 xxi Figure III-1. Proposed mechanism of domain-swapping. Instead of turning and making a closed monomer, the hinge loop region elongates in the open monomer conformation. If two or more open monomers with proper C- and N- termini conformations find each other, they can generate dimers, trimers, or higher-order oligomers. ................................... 143 Figure III-2. Schematic demonstration of different domain-swapped forms. a. Closed monomer. The hinge loop region and primary interface are highlighted. b. Domain- swapped dimer. The extended conformation of the hinge loop and secondary interface created by two open monomers is represented. c. Cyclical oligomerization. d. Open- ended or “run-away” oligomerization. ......................................................................... 144 Figure III-3. Several examples of domain-swapped proteins. a. Diphteria toxin monomeric form (PDB code: 1MDT). b. Diphteria toxin domain-swapped dimer form (PDB code: 1DDT). c. Monomeric structure of RNase A (PDB code: 1RTB). d. RNaseA N-terminal domain-swapped dimer crystal structure (PDB code: 1A2W). e. RNaseA C-terminal domain-swapped dimer crystal structure (PDB code: 1F0V). f. RNaseA C-terminal cyclical domain-swapped trimer crystal structure (PDB code: 1JS0). g. GB1 monomeric structure (PDB code: 1Q10). h. GB1 monomeric structure (PDB code: 3V3X). i. amyloid-like cystatin domain-swapped dimer (PDB code:1TIJ). j. Human prion protein monomer (PDB code: 1QM0). k. Human prion protein domain-swapped dimer (PDB code: 1I4M). ..... 147 Figure III-4. The formation of domain-swapped dimer (previous study) and trimer (current study) based on open-monomer theory in hCRBPII. .................................................. 149 for: a. A1: Figure Q108K:K40L:T53A:Q38F:R58L, 23°C, 20 hours expression. b. A1, 19°C, 40 hours expression. c. A1: V62E, 19°C, 40 hours expression. d. Gel filtration chromatogram of high-salt-eluted protein samples for A1 (corresponding to the second peak in source Q) expressed at 19°C, 40 hours (black) vs. A1 expressed at 23°C for 20 hours (blue). ... 151 Figure III-6. a. Native PAGE gel for 1st and 2nd peaks shown in the source Q column. b. SDS PAGE gel for 1st and 2nd peaks shown in the source Q column. c. Dynamic light scattering for monomer, dimer, and trimer (Q108K:K40L:T51V was used as a monomer for comparison). ........................................................................................................... 152 Figure III-7. a. The crystal structure of domain-swapped trimer of A1 (chains A, B, and D). The close monomers highlighted with different colors. b. The overlay of three closed monomers show similar conformations for all of them. c. Domain-swapped trimer structure of A1 (the same chains as mentioned above) where the open monomers made up trimer are highlighted by different colors. d. Overlay of the three open monomers demonstrated the similar conformations for chains A, B, and D. ................................ 154 Figure III-8. Overlay of open monomer of apo domain-swapped trimer A1 (purple), chain D, with: a. open monomer of apo WT hCRBPII domain-swapped dimer (blue); b. open monomer of Q108K:K40L:Y60W, chain A, apo asymmetric domain-swapped dimer III-5. Source Q and gel filtration chromatograms xxii (yellow); c. open monomer of Q108K:K40L:Y60W, chain B, apo asymmetric domain- swapped dimer (yellow); and d. open monomer of Q108K:K40L:T51F holo domain- swapped dimer, chain A (green). ................................................................................. 155 Figure III-9. a. Subtraction of N59 torsional angles for different trimer structures vs. the apo symmetric dimer. b. Overlay of dimer (blue) vs. trimer (purple) demonstrates that by rotating around N59, the dimer structure can be converted to trimer and vice versa. .. 156 Figure III-10. Gel filtration chromatogram of Q108K:T51D dimer and trimer. The trimer was kept at room temperature and reloaded to Gel filtration column to examine the stability of trimer. The intensity for the trimer kept for 30 days at room temperature was normalized. .................................................................................................................. 157 Figure III-11. Hydrophobic core generated by hydrophobic residues at the center of the domain-swapped trimer. F27 from each chain is shown in green, I32 in yellow, L36 in orange, and F57 in cyan. The pseudo-C3 symmetry brings residues from the different chains close together. ................................................................................................. 158 Figure III-12. a. Overlay of apo (cyan) and holo (orange) ligand-bound trimer (A1 variant). b. Top: electron density of the ligand in one of the trimer chains. Bottom: overlay of ligands in three different chains. c. Overlay of apo (yellow) and holo ligand-bound trimer of A1 variant (purple). N59 is pointing outside the binding cavity in both apo and holo structures where it does not sterically impinge on the chromophore. d. Overlay of apo (orange) and retinal-bound domain-swapped dimer (purple). N59 in apo conformation sterically clashes with the retinal trajectory. The result of this steric interaction is the “flipped-out” conformation for N59 in holo. e. Overlay of apo (green) and holo (purple) retinol-bound structure of WT-hCRBPII (PDB codes: 2RCQ and 2RCT). f. Overlay of retinylidene Q108K:T51D dimer (purple) with the synthesized chromophore used for trimer crystallization with A1 variant (cyan)............................................................................ 160 Figure III-13. Ramachandran plots for a. Q108K:T51D:I32C and b. Q108K:T51D:A28C. Q108:T51D:I32C shows some of residues as outliers. c. Overlay of Q108K:T51D:A28C (purple-blue) vs. Q108K:T51D:I32C (yellow) crystal structures. .. 162 Figure III-14. The crystal structure of the domain-swapped trimer of KD:A28C:I32C showing the disulfide bond formation between different chains. ................................. 163 Figure III-15. The zinc-bound apo A1-F57H mutant: a. The 2Fo-Fc map shows the presence of zinc coordinated with three histidines (contoured to 1σ.). b. The domain- swapped dimer of A1-F57H shows two histidines are far away from each other to generate the metal-binding site (contoured to 1σ.). ..................................................... 164 Figure III-16. a. The anomalous difference map collected above zinc x-ray absorption edge contoured to 8σ. The saturation-binding curve for A1-F57H showing the Kd: 0.70 ±0.21 µM...................................................................................................................... 166 the xxiii thiophene-based Figure III-17. a. Left: The gel electrophoresis under the reducing and no-reducing conditions. Right: Source Q chromatogram of I31C showing two peaks. b. The crystal structure of I31C showing two close monomers connected by cross linked disulfide bridge. ......................................................................................................................... 167 Figure IV-1. The maturation of the GFP chromophore contains cyclization, oxidation, and dehydration steps to generate an emissive ligand. ..................................................... 187 Figure IV-2. ThioFluor analogs synthesized by Prof. Borhan’s lab to investigate structure-property correlations of fluorophores with aldehyde functionality inside the binding pocket of the protein. .................................................. 188 Figure IV-3. ThioFluor3 bound with the L1:L117E variant of hCRBPII. The aromatic residues, Y19W, R58W, and W106, and other proximal amino acids are highlighted in the binding pocket. ........................................................................................................... 190 Figure IV-4. Crystal structures of Q108K:K40L:T51V:T53S:R58W:Y19W:L117E (L1:L117E) bound with ThioFluor analogs. a. L1:L117E crystal structure bound with ThioFluor. b. ThioFluor2. c. ThioFluor3. d. ThioFluor4. e. ThioFluor6. f. ThioFluor. g. ThioFluor8. h. ThioFluor9. i. ThioFluor10. ............................................................... 191 Figure IV-5. Overlay of L1:L117E bound with ThioFluor6 (cyan) and L1:L117E bound with ThioFluor4 (pink). (RMSD: 0.1) ........................................................................... 192 Figure IV-6. a. L1:L117E crystal structure bound with ThioFluor with two presentations. The crystal structure depicted the almost-planarity of the N,N-dimethyl group with the chromophore’s backbone. b. L1:L117E crystal structure bound with ThioFluor3. The crystal structure depicted the almost-planarity of the azetidine group with the chromophore’s backbone. ........................................................................................... 193 Figure IV-7. The binding pocket of ThioFluor7 bound with L1:L117E demonstrated the interacting residues. Two C-H…F interactions trifluoromethyl group and V62 and V6, S… π interaction which can be seen in all of the ThioFluor analogs with L1:L117E. and finally he low barrier hydrogen bond with L117E and nitrogen of iminium. ............................ 194 Figure IV-8. a. Approximate coplanarity of the two thiophenes in ThioFluor8. b. Position of two thiophenes provides a suitable conformation for metal-binding. This ligand can be used in the future as a metal sensor, where metal-binding can tune the absorption and emission properties of ThioFluor8. The residues that are in proximity with the two rings are highlighted. Metal binding site that bridges chromophore and protein can be envisioned by mutation of these residues. .................................................................. 195 Figure IV-9. a. Color changes upon: iminium formation of ThioFluor8/L1:L117E (Protein concentration ~35 µM - 0.5 equivalent ThioFluor was added., Adding 100 µM KAuCl4 to the mixture of ThioFluor8/L1:L117E, incubating the previous mixture for two minutes, readding 100 µM KAuCl4. b. The UV-vis spectrum of ThioFluor8/L1:L117E pre and post xxiv 100 µM KAuCl4 addition. c. The UV-vis spectrum of ThioFluor8/L1:L117E pre and post 100 µM CuSO4 addition as a control. d. The UV-vis spectrum of ThioFluor/L1:L117E pre and post 100 µM KAuCl4 addition as a control. ........................................................... 196 Figure IV-10. Representation of the geometric arrangement of the donor and acceptor groups upon excitation in: a. twisted intramolecular charge transfer (TICT) and b. planarized intramolecular charge transfer (PLICT) c. Dihedral angle between the two phenyl rings in ThioFluor10 measured 31.5o. ............................................................ 197 Figure IV-11. Overlay of L1:L117E crystal structures bound with ThioFluor (pink) and ThioFluor9 (green), showing a small change in trajectory, which affects the position of Y60. ............................................................................................................................ 198 Figure IV-12. a. Binding pocket of L1:L117E bound with ThioFluor2, depicting how the aromatic residues sandwich the chromophore in the binding pocket. b. Overlay of L1:L117E bound with ThioFluor2 (cyan) and ThioFluor5 (pink) crystal structures, showing a new conformation of thiophene for ThioFluor2. ......................................... 199 Figure IV-13. Crystal structure of ThioFluor bound with L1:L117E, showing how L117E acts as a counter-ion for the nitrogen of iminium. b. UV-vis spectrum of pH titration for the L1:L117E/ThioFluor complex. c. Plot of abs. vs. pH reveals two apparent pKas, 6.6 and 11.2 for L117E and iminium, respectively. d. Color changes upon acidification of the L1:L117E/ThioFluor complex. (The data for UV-vis. and pKa was provided by Dr. Santos.) ....................................................................................................................... 201 Figure IV-14. The crystal structure of L1:L117E:A33W/ThioFluor depicts the cis-iminium cis-iminium in L1:L117E:A33W/ThioFluor further stabilized through a water mediated hydrogen bond between Q4 and iminium (the electron density is contoured to 1σ.). b. The Overlay of L1:L117E/ThioFluor (orange) vs. L1:L117E:A33W/ThioFluor (cyan) revealed the presence of A33W sterically push the chromophore and locate it deeper in the binding pocket. ......................................................................................................................... 203 Figure IV-15. a. Crystal structure of L1:L117D/ThioFluor shows a cis-iminium configuration. Unfortunately, in this case the acetate ion coming from the crystallization conditions occupied the water molecule site between Q4 and the nitrogen of imine. b. Overlay of L1:L117D/ThioFluor (purple) and L1:L117E/ThioFluor depicts a different conformation of L117D vs. L117E, which is very far from the nitrogen of iminium. c. Overlay of two cis-iminium structures coming from L1:L117D/ThioFluor (purple) and L1:L117E:A33W/ThioFluor highlights the chromophore deeper in the binding pocket. ................................................................. 204 Figure IV-16. a. The crystal structure of L1:L117D:Q4F/ThioFluor demonstrates the trans-imine configuration and the same conformation of L117D as observed for L1:L117E/ThioFluor. b. The overlay of L1:L117D:Q4F/ThioFluor (yellow) and L1:L117E/ThioFluor (purple). Interestingly the position of nitrogen of imine is almost the same for both structures. importance of A33W the in locating xxv IV-17. UV-vis and spectra emission In the pH of crystallization (~4.5) both aspartate and iminium would be anticipated to be protonated. ................................................................................................................. 206 Figure of Q108K:K40E:T53A:R58W:Q38F:Q4F:Y19W/ThioFluor SB (pH 7.1) and PSB (pH:4.5). .................................................................................................................................... 207 IV-18. Crystal structure of Q108K:K40E:T53A:R58W:Q38F:Q4F:Y19W Figure (L2/ThioFluor complex shows four molecules in the asymmetric units, where the electron density of the chromophore in three chains indicates ligand binding. a. ThioFluor bound to the protein in chain A depicts the trans imine. b. ThioFluor bound to the protein in chain C depicts an electron density more correlated with a cis iminine. c. ThioFluor bound to the protein in chain B depicts an electron density more correlated with a mixture of cis and trans iminine. d. K40E conformation that is not interacting with the nitrogen of imine. It is expected that this conformation would cause a suppressed pKa for imine, leading to an SB population. K40E is held in this conformation through a water-molecule interaction with the T53 and Y60 main chain. The electron density is contoured to 1σ for all the structures. ................................................................................................................... 209 Figure IV-19. Crystal structure of ThioFluor3 bound with a trimeric form of Q108K:K40D:T53A:R58L:Q4F:Q38F reveals a K40 conformation that is far away from the nitrogen of iminium. .............................................................................................. 210 Figure IV-20. Crystal structure of Q108K:K40H:T53A:R58L:Q4F:Q38F (L3) bound with FR1 shows four molecules in the asymmetric unit that are fully occupied by FR1. a. Chains A and B show the same conformation of K40H, in which it directly interacts with the nitrogen of iminium. b. Chains C and D show the same conformation of K40H, in which it does not interact with the nitrogen of iminium (4.8 Å distance). Instead, K40H gets locked into this conformation by hydrogen bond networks containing the residues highlighted in the Figure. c. Left: Overlay of chain A and chain C, indicating that the only difference between the two chains is a different conformation of K40H. The K40H conformation that is closer to the nitrogen of imine would be expected to stabilize the iminium, and the other conformation would be an active ESPT ground-state conformation. Right: Overlay of L3/FR1 (green) and L2/ThioFluor (purple) crystal structures further confirms the ESPT active conformer of K40H. ............................................................ 211 Figure IV-21. UV-vis and emission spectra of L3/FR1 complex. ................................. 212 xxvi KEY TO ABBREVIATIONS alanine, Ala cysteine, Cys aspartic acid, Asp glutamic acid, Glu phenylalanine, Phe glycine, Gly histidine, His isoleucine, Ile lysine, Lys leucine, Leu methionine, Met asparagine, Asn proline, Pro glutamine, Gln arginine, Arg serine, Ser threonine, Thr valine, Val tryptophan, Trp tyrosine, Tyr Angstrom xxvii A C D E F G H I K L M N P Q R S T V W Y Å M μΜ nM mM ε h m sec t1/2 φ molar micromolar nanomolar millimolar extinction coefficient hour minute second maturation half-life quantum yield CRABPII hCRBPII cellular retinoic acid binding protein II human cellular retinol binding protein II EGFP FABP iLBP HPLC SB PSB TF EF 7TM XFEL TLC enhanced GFP fatty acid binding protein intracellular lipid binding protein high pressure liquid chromatography Schiff base protonated Schiff base transcription factor elongation factor seven transmembrane helix x-ray free electron laser thin layer chromatography xxviii UV-vis FTIR PCR DS FRET MBP BLA LOV BLUF FLARE TEV CD ITC WT ultraviolet visible spectroscopy Fourier transform infrared spectroscopy polymerase chain reaction domain-swapped Förster resonance energy transfer maltose-binding protein β-lactamase light-oxygen-voltage-sensing domain blue-light using FAD fast light and activity-regulated expression tobacco etch virus circular dichroism isothermal titration calorimetry wild type hCRBPII-HIV HIV-protease connected to the c-terminus of hCRBPII DT RNase A BS-RNase CV-N PAGE ESPT KLK KLF diphteria toxin ribonucleases bovine seminal ribonucleases Cyanovirin-N polyacrylamide gel electrophoresis excited-state proton transfer Q108K:K40L:T51K Q108K:K40L:T51F xxix A1 KD M1 H1 AG1 L1 L2 L3 Q108K:R58L:T53A: K40D:Q38F:Q4F Q108K:T51D R111K:Y134F:T54V:R132Q:P39Y:R59Y Q108K:K40L:T51V:T53S:R58W:Y19W:A33W:L117D:Q38Y Q108K:T51D:A28C:L36C:F57H Q108K:K40L:T53S:Y19W:R58W Q108K:K40E:T53A:R58W:Q38F:Q4F:Y19W Q108K:K40H:T53A:R58L:Q4F:Q38F xxx CHAPTER I: MIMICKING MICROBIAL RHODOPSIN ISOMERIZATION IN A SINGLE CRYSTAL I-1. Rhodopsin and retinal isomerization The conversion of light into chemical energy is one of the most essential processes in nature. Photoreceptor proteins are at the heart of many biological functions mandating the conversion of light energy into chemical energy. These proteins encompass all kingdoms of life; eukaryotes, bacteria, and archaea, and they are responsible for sensing, intracellular and intercellular signaling, as well as energy production in lower organisms. 1-2 These proteins’ ability to respond with light arises from a chromophore enclosed inside a. b. c. Figure I-1. a. The LOV domain bound with flavin chromophore. b. Phytochrome bound with a bilin chromophore. c. Rhodopsin that is covalently attached to the protein through Lsy216. The chromophores have been shown with stick representation (carbons in blue, nitrogen in dark blue, oxygen in red, and phosphorous in orange). the binding cavity of proteins, such as bilin proteins bound to bilin, flavoproteins bound 1 with flavin chromophore, and retinylidene proteins bound with retinal (Figure I-1). Rhodopsin is one of the earliest and most important classes of photoreceptors, employing retinal as a covalently-bound chromophore.1-4 The rhodopsin family is a large and functionally diverse group of proteins that are found in all kingdoms of life. Their functional diversity comprises proton or ion pumps, 5 ion channels,6 signaling proteins,7 vision,8 and sensing.9 The apo-proteins, referred to as opsins, are seven transmembrane helix (“7TM”) integral membrane proteins, and form an imine (Schiff Base) between retinal, or a closely-related analog, and a buried lysine residue to generate the pigment.1,10 Vast literature that includes time-resolved spectroscopy, structural biology — most notably X- ray crystallography 11-14 — and biochemistry 15-16 describes the structures, spectroscopy and lifetimes of the various intermediates in the process and identifies the key protein residues essential for its function.1, 17-19 There are two pivotal classes of rhodopsin proteins: microbial or type I rhodopsin, and animal, or type II rhodopsin. Structurally, these proteins share a seven transmembrane helix (7-TM) fold, but share little sequence similarity with one another. While microbial rhodopsins are employed in lower organisms for energy production and intracellular and intercellular signaling, animals employ a G- protein-coupled receptor (GPCR) family of rhodopsins mostly for signaling processes. Color vision, non-vision transduction, and maintenance of the circadian clock are some important examples of where these proteins are utilized.1 Animal rhodopsin mostly uses 11-cis retinal as a chromophore (Figure I-2). Upon absorption of a light photon, the chromophore undergoes selective isomerization to generate all-trans retinal, a thermodynamically more stable isomeric form. Then, at the end of the photocycle, after the protein adopts the active meta II state, which is essential 2 for G-protein activation 20 and where the chromophore gets deprotonated, the all-trans retinal becomes hydrolyzed from the protein binding pocket. Through a catalytic cycle in which many proteins and enzymes cooperate together, the 11-cis retinal is regenerated and delivered to the opsin (apo protein) to restore the pigmented rhodopsin. 21-22 Although hydrolyzing the chromophore after isomerization is the hallmark of many rhodopsins’ functions, some bistable rhodopsins have recently been identified that do not require the hydrolysis of the chromophore, and through irradiation with light, the chromophore becomes reset to the original state.23 However, in microbial rhodopsin, the isomerization ensues from all-trans to 13-cis retinal (Figure I-2a), in which a thermodynamically more stable isomeric form undergoes selective isomerization to produce a thermodynamically less stable isomer. Also, in contrast to most animal rhodopsins, the chromophore does not get hydrolyzed, but instead, through thermal adaptation, the chromophore and conformation of the proteins revert back to their original states.24 A recent study showed that in the case of animal a. b. 13 11 14 12 N R H Light Dark Light N R H 13 14 N R H 11 12 N R Figure I-2. Photoisomerization of the chromophore-bound protein in: a. microbial rhodopsin; b. animal rhodopsin. rhodopsin, the only role of protein in the isomerization of retinal is to preselect the active conformer of retinal, which is 11-cis-12-s trans versus 11-cis 12-s-cis. 25 3 According to this study, the isomerization of 11-cis to all-trans retinal happens both 11 12 H O 11-cis-12-s-trans 11 12 H O 11-cis-12-s-cis Figure I-3. Two different possible conformations for 11-cis retinal in solution: 11-cis-12- s-trans and 11-cis-12-s-cis. 11-cis-12-s-trans conformations generate a steric between the hydrogen of C10 and methyl group of C13. solution and the protein binding pocket with 100% selectivity. However, inside the proteins this isomerization ensues with a high quantum yield, since the binding pocket is well- engineered to bind with the active conformer, 11-cis-12-s trans, where the methyl group of C13 and hydrogen of C10 sterically clash with each other, causing the fast photoisomerization of retinal (Figure I-3).25 In contrast, in microbial rhodopsin photoisomerization, the proteins need to perform three important tasks. First, they must overcome the energy barrier for isomerization. Second, they need to prevent the formation of other isomers, since there is more than one double bond that can potentially get isomerized. The irradiation study of all-trans retinal in solution leads to the distribution of different isomers highlighting the role of the protein in the selective isomerization of retinal.26 And third, the proteins must reset the chromophore at the end of isomerization to the original state through a thermal process, where the energy of the absorbing photon is gone. These tasks demonstrate the nature and complexity of the biophysical processes that need to be performed in retinal isomerization by proteins in the microbial rhodopsin family. 4 One of the long-lasting, remaining questions that has attracted many researchers in the past few decades is how microbial rhodopsin can selectively perform the isomerization of retinal around one double bond, while retinal photoisomerization in solution gives rise to the distribution of different isomers. 27-28 To answer this challenging question, research groups have taken different approaches, such as the chemical modification of retinal 27, 29 or mutational studies of the protein.17, 30 Although these approaches unravel different biophysical features and thermodynamic driving forces of rhodopsin isomerization, they have some inherent limitations: first, some of the key residues in the photocycle of rhodopsin are conserved residues, and mutations of them can lead to insoluble expressions of the protein; second, the chemical modification of retinal might lead to a completely different trajectory of the chromophore inside the binding cavity of the protein. Creating a new photoactive model system using a soluble protein, which is evolutionary naive, not only provides us with an easier system to study the isomerization pathway of retinal in comparison with a membrane protein, but it would also reveal a new insight into how evolution develops such photoactive proteins from scratch. In this chapter, the photocycle of microbial rhodopsin and the key residues in this event will be discussed first, and then we will describe our strategy to create a new model system that mimics all the important features of rhodopsin isomerization. I-2. Microbial rhodopsin features The microbial rhodopsin family encompasses more than 7,000 photoactive proteins in different organisms from different kingdoms of life.31 This family offers an explicit example of nature’s diverse evolution of a single protein scaffold, giving rise to 5 functionally-distinct proteins ranging from ion/proton pumps, sensors, channels, signaling proteins, etc. 1-2, 8 Although all the proteins in this family share a 7-TM helical scaffold and a conserved Lys that generates a covalent link to retinal through protonated Schiff base (PSB) formation, there are many variations in the residues, the orientation of loops, and M118 W182 T90 W138 W189 Y185 D212 K216 W86 406 D85 402 401 Figure I-4. Bacteriorhodopsin binding pocket contains aromatic residues such as W182, W138, W189, W86, and Y185, that sandwich the chromophore through aromatic interaction. D85 and D212 act as a counter ion for the iminium, stabilizing the positive charge. The water molecules shown in red mediate the hydrogen bond between the nitrogen of the iminium and D85 and D212. Retinal is shown in cyan, and the protein backbone carbon, nitrogen, and sulfur atoms are shown in purple-blue, dark blue, and dark yellow, respectively (PDB code:1C3W). the water molecules’ positions in the protein structures. The binding pocket of retinal perhaps is the most conserved part among all the different proteins in this family.1 Bacteriorhodopsin, one of the most well-studied systems in the microbial rhodopsin family, highlights the role of aromatic residues, as shown in Figure I-4, for holding the chromophore in the specific trajectory needed for isomerization and a net conformational 6 change in the protein that gives rise to its functional regulation. The presence of these aromatic residues is diffuse in the family of rhodopsin proteins.1 For instance, many spectroscopic studies, along with recent time-resolved X-ray Free Electron Laser (XFEL) experiments,32 revealed that a conserved Trp182 acts as a mechanical transformer, leading to a net conformational change in the protein upon isomerization of retinal by tilting the α-helices in response to retinal photoisomerization. Other studies speculated the importance of Trp83 in directing the photoisomerization towards the C13-C14 double bond by providing a tuned steric interaction in the vicinity of the chromophore.17,30 Furthermore, another conserved residue, Tyr185, forms a hydrogen bond with the nitrogen of iminium, which is important for stabilizing the PSB and holding the trajectory of the chromophore during the photocycle. Interestingly, this residue is conserved in animal rhodopsin as well.33 Since the chromophore is protonated inside the binding cavity, there are some acidic residues, such as Asp85 and Asp212, inside the binding cavity that act as counter ions to stabilize the positive charge on the nitrogen of the iminium (Figure I-4). Notably, these residues are conserved between all the proton pumps in this family. Additionally, mutational studies of these residues, specially Asp212, showed they are important not only for proton pumping activity, but also for preventing other isomerizations.5, 34 In addition to these residues, the ordered water molecules in the binding pocket, mostly residing between acidic residues and the chromophore, are crucial for adjusting the pKa of the chromophore and surrounding residues and dictating the directionality of the ion/proton channels from the cytoplasmic side to the extracellular side and vice versa. 35 A very recent time-resolved x-ray experiment on bacteriorhodopsin demonstrated that the 7 water molecules between the chromophore and acidic residues become disordered in less than 500 fs, prior to retinal isomerization, where they can provide space for the isomerization of the imine bond and C13-C14 double bond.36 I-2-1. Bacteriorhodopsin photocycle The photocycle of the microbial rhodopsin family has been extensively studied using many different spectroscopic and biophysical techniques.6, 24, 26, 37-38 In all the H+ Cytoplasmic side a. b. R + N H Ground'state' 10'ms' 5'ms' O' N' 3'ms' R + N H + H N R hν' K' L' 50'μs' 1'μs' + H + H N R N R M' N R D96 D85 R82 E204 E194 Extracellular side Figure I-5. a. Photocycle of bacteriorhodopsin, where different intermediates form during the photocycle. The lifetime of each is shown underneath the retinal structure. b. The residues that are important in proton-pumping activity, protonation and deprotonation of iminium, are shown as transparent purple spheres. Retinal is shown as cyan spheres. R82, E194, and E204 are referred to the proton releasing complex, facilitating the proton transfer to the bulk. proteins in the microbial rhodopsin family, the photocycle begins with selective isomerization of the chromophore, and after the formation of several intermediates and conformational changes of the protein, the structure of the chromophore resets to the original state. The bacteriorhodopsin photocycle is one of the most well-studied 8 photocycles in this protein family. This protein acts as a unidirectional proton pump, in which it translocates the proton from the cytoplasmic side to the extracellular side. This activity, regulated by light, is crucial in halobacteria cells, where it leads to large gradient potential between the cells and the environment, resulting in energy production within the cells.39 However, this process, like other photo-induced systems,3 passes through a series of short-lived intermediate events, as shown in Figure I-5a, where the lifetime of each event has been tuned through many years of evolution to create one of the most efficient photosystems in nature.40 The photocycle of bacteriorhodopsin begins with the photoisomerization of the chromophore to generate 13-cis-15-anti from all-trans retinal less than 500 fs. The K intermediate that is formed initially (13-cis 15-anti) is red-shifted (590 nm vs. 570 nm, which is an original absorption of the retinal/protein complex) where the chromophore conformation is highly twisted. This intermediate possesses extra free energy for the net photoinduced proton pump activity in the protein.41 After about a microsecond, the K intermediate relaxes to the blue-shifted L intermediate (550 nm), where the chromophore is planar and creates a hydrogen bond network with the neighboring residues. The K-to- L transitions are escorted by reorientations of the Schiff base pointing toward the intracellular side, producing a local conformational change in the protein.38 Remarkably, the L intermediate is the incipient platform for proton-pumping activity, where the proton of the iminium switches to the primary acceptor group (D85) resulting in the formation of the significantly blue-shifted M intermediate (412 nm), which is not absorbing at the wavelength in which the irradiation occurs.38 At this step, as a result of a change in the pKa of the chromophore and subsequently Asp85, the Arg82 that originally maintains a 9 salt bridge with Asp 85, swings its conformation away from the cytoplasmic site and toward the proton releasing complex (E194 and E204) in the extracellular site.1, 42 The reconfiguration of Arg82 during the photocycle has been detected using recent XFEL time-resolved crystallography as well.12 This local conformational change induces the proton release through the proton-releasing complex by altering the pKa of E204 and E194 that now are deprotonated. After this step, the N intermediate (520 nm) is generated where the largest conformational change in the protein backbone ensues. This conformational change (tilting the α-helices) situates the chromophore in the cytoplasmic side, where the chromophore becomes reprotonated again through D96, Figure I-5b.43 During the N-to-O transition (PSB λmax = 640 nm), the chromophore reverts back to the original state and the D96 positioned in the cytoplasmic site becomes reprotonated again.44 Finally, in the last step, the D85 delivers its proton to the proton-releasing complex and the protein conformation relapses back to the original state.1 The mechanism of the unidirectionality of the proton pump in bacteriorhodopsin, which makes the ion/proton channel different from ion/proton pumps, has been controversial. However, one of the most popular mechanisms proposed is that through the photocycle, the proton accessibility of the imine varies between the extracellular site and cytoplasmic site that prevents the proton backflow.45 Furthermore, the crystal structure of bacteriorhodopsin, along with time-resolved FTIR, depicts seven ordered water molecules in the extracellular half while only two molecules have been identified in the cytoplasmic side.16, 46 This disconnection in the water-mediated hydrogen bond networks between the cytoplasmic and extracellular sites makes the dark inactive state and demands the light-induced 10 conformational change to occur for a proton uptake from the cytoplasmic side and a fast proton release to the extracellular side.1, 16 It is important to mention that a wealth of cryo-kinetic data detected the structure of different intermediates formed in the photocycle by prolonging the lifetime of these intermediates through certain mutations, irradiation with a specific wavelength of light, or a change in pH. 38, 41, 43, 47 Additionally, recent X-ray free electron laser (XFEL) studies detected different stages of the photocycle with femtosecond time resolution. 12, 36 I-3. Engineering hCRABPII to mimic microbial rhodopsin isomerization 48 I-3-1. Choosing an initial protein template We have had a long-standing interest in understanding the protein-chromophore interactions giving rise to wavelength regulation in rhodopsins,49-59 using protein mimics to investigate the latter phenomena. We have now turned our attention to understanding the factors that govern the isomerization events by investigating the minimal requirements for bacteriorhodopsin in a model system to mimic the hallmarks of the natural system. Human Cellular Retinol Binding Protein II (hCRBPII) and Human Cellular Retinoic Acid Binding Protein II (hCRABPII), two small, soluble cytosolic proteins from the intracellular lipid binding protein family, were used as templates for protein reengineering.49-59 To initiate our investigation, we required a platform that would support photoinduced isomerization of the bound chromophore. As described in previous studies by Nosrati et al, a number of hCRABPII variants exhibited photoinduced changes in the protonation state of a retinylidene-bound protein; namely, an initially-observed pigmented protein- chromophore complex reverts to a colorless state in time.49 X-ray crystallography and UV- 11 vis spectroscopy showed that imine bond isomerization from 15-syn to 15-anti is W109 initial state L121 R111K I43 V41 1a-H, cis-iminium λmax = 555 nm pigmented heat or visible light high pKa H N Lys111 N Lys111 1a, cis-imine λmax = 360 nm UV light W109 final state R111K L121 V41 I43 N Lys111 1b, trans-imine λmax = 360 nm colorless low pKa N Lys111 H 1b-H, trans-iminium λmax = 555 nm Figure I-6. Photocycle of hCRABPII hexamutant M1 (R111K:R134F:T54V:R132Q:P39Y:R59Y) bound to retinal. The photoinduced isomerization of the imine functionality leads to changes in its pKa and consequently its protonation state. The hydrophilic and hydrophobic environment of the imine nitrogen atom for the cis and trans geometry, respectively, is highlighted in the figure obtained from the crystal structure of each form (PDB IDs: 4YFP and 4YFQ). responsible for this change. The initial kinetically-formed, 15-cis isomer orients the imine nitrogen in a polar environment that supports a high pKa regime, and thus yields a protonated Schiff base (PSB) that is a colored pigment. Time-dependent 15-cis to 15- trans imine isomerization yields a thermodynamic product that places the nitrogen atom in a low pKa regime, leading to the unprotonated Schiff base (SB) that is colorless (Figure I-6) (Figures I-6 to I-22 in this chapter along with the context were reproduced by the permission of Journal of American Chemical Society)48. Additionally, green and UV light irradiation interconverts between the PSB and SB, respectively, in solution and crystalline states.48-49 We previously demonstrated that the pKa difference between the two isomers 12 is the driving force for 15-syn (high pKa) to 15-anti (low pKa) isomerization, where the imine nitrogen experiences two different environments in each isomeric form (Figure I- 6).49 Having shown the ability to photochemically isomerize the imine bond, the next goal was to design a protein capable of isomerizing the C13 double bond, reminiscent of the action of microbial rhodopsin. In general terms, this would be achieved by mimicking the placement of critical residues in bacteriorhodopsin — one of the most well-studied systems in the microbial rhodopsin family — in the binding pocket of our surrogate protein. Having high-resolution structures on hand, along with the high structural tolerance of these proteins to mutation, provides us the opportunity to explore their capabilities as an isomerase. The very first goal was to obtain a high resolution structure of protein-bound 13-cis retinal in our system. However, attempts to obtain the crystal structure of 13-cis bound protein using different variants of hCRABPII and hCRBPII were not fruitful (As shown in Table I-1). Finally the crystal structure of M1 incubated with 13-cis showed an electron density in the binding cavity that was more similar to the all-trans bound form (Figure I-7a). The UV-vis spectra of M1 with all-trans retinal and 13-cis revealed that M1 gives little PSB with 13-cis at physiological pH, in contrast to all-trans (Figure I-7). However, the fine structure of the retinal SB in the UV-vis spectrum (at ~369 nm) proved that 13-cis forms a robust SB within the protein.48 Consistent with this result, the HPLC extraction of M1 after incubation with 13-cis in the dark (~24 hours) revealed that almost half of the 13-cis was converted to all-trans at room temperature, (Figure I-7b), indicating slow thermal isomerization of 13-cis to all- 13 trans in the M1 binding pocket. Additionally, light irradiation of the chromophore led to the Table I-1. The initial efforts at obtaining a 13-cis bound crystal structure (Other than the first four variants of hCRBPII, the rest of the mutants belong to CRABPII). Variants Consequence Q108K:K40L-(hCRBPII) Q108K:K40L-T5V-(hCRBPII) Q108K:K40L-T5V:R58F-(hCRBPII) Q108K:K40L-T5V:R58F:L117E-(hCRBPII) No crystal No crystal No crystal No crystal R111K:Y134F:T54V:R132Q:P39Q:R59Y:A32W:F3Q No electron density for retinal R111K:Y134F:T54V:R132Q:P39Q:R59Y:A32W:F3Q:L121W No electron density for retinal R111K:Y134F:T54V:R132Q:P39Q:R59Y:A32W:F3Q:L121H No electron density for retinal R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121M R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121Y R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121W R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121Q R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121E R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121D R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121N R111K:Y134F:T54V:R132Q:P39N:R59Y:L121E R111K:Y134F:T54V:R132Q:P39E:R59Y:L121E R111K:Y134F:T54V:R132Q:P39F:R59Y:L121E R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121E:A32W R111K:Y134F:T54V:R132Q:P39F:R59Y:L121E:I31W R111K:Y134F:T54V:R132Q:P39N:R59Y:M121E No crystal No crystal No crystal No crystal No crystal No crystal No crystal No crystal No crystal No crystal No crystal No crystal No crystal R111K:Y134F:T54V:R132Q:P39F:R59Y electron density was similar to all-trans 14 a. R111K b. 13-cis all-trans c. SB PSB d. SB e. SB PSB PSB Figure I-7. a. Electron density of M1 bound with 13-cis retinal showed electron density that is more similar to all-trans retinal. b. HPLC chromatogram of retinals extracted from 13-cis bound M1 after 24 h of incubation in the dark demonstrated the dark isomerization of 13-cis to all-trans retinal. c. UV-vis spectrum of M1-L121E bound with all-trans retinal. d. UV-vis spectrum of M1-L121E bound with 13-cis retinal. e. Photo-irradiation of 13-cis retinal with UV (B.P. filter, 300-400 nm), green light (L.P. filter > 500 nm). same imine bond isomerization (Figure I-7e). Since we were able to show isomerization of the imine bond in crystals of R111K:Y134F:T54V:R132Q:P39Y:R59Y (M1),49 we decided to employ this construct for further design. a. b. W109 K216 R111K W86 3.3)Å C13 C13 3.6*Å L121 Figure I-8. a. Binding pocket of bacteriorhodopsin, where W86 is located in the vicinity of C13-C14. b. hCRABPII binding pocket. The L121 position is analogous to the W86 position in bacteriorhodopsin relative to retinal. 15 To mimic the microbial rhodopsin isomerization, it was first necessary to prevent the imine isomerization pathway described above. This was essential to keeping the imine protonated in all isomeric forms, and thus maintaining its bathochromicity for absorption R111K L121W L121 Figure I-9. Overlay of M1 (yellow) and M1-L121W (cyan). Introducing Trp in the L121 position impinges on the bound retinal, which leads to the concomitant change in retinal trajectory. of visible light to initiate isomerization of the C13 double bond. We feared that the imine bond isomerization could potentially lead to a conformation resulting in a low pKa, and therefore, an unprotonated imine, which would siphon away a portion of the pigment without C13 isomerization. In essence, the unprotonated retinylidene would be an “off- a. b. SB PSB PSB Figure I-10. UV-vis spectra of M1 and M1-L121E upon incubation with: a. all-trans and b.13-cis retinal. The SB (Schiff Base, Imine) and PSB (Protonated Schiff Base, Iminium) peaks are highlighted. 16 cycle” well that removes the isomerizing species from the photocycle due to the lack of absorptivity at the wavelength of irradiation. Furthermore, to perform a specific C13 isomerization, we needed to tune the steric interaction in the vicinity of the C13-C14 double bond to make this double bond distinguishable for the system in comparison with other double bonds. Using the all-trans retinal-bound structures as a guide, bulky residues were introduced in the vicinity of the C13-C14 double bond to introduce steric interactions. Inspired by bacteriorhodopsin, our first screening effort was centered on mutating L121 to aromatic residues to mimic the steric interactions of the natural protein’s Trp 86.1, 60 Leu121 is ~ 3.6 Å away from C13, making it a suitable candidate to fulfill this goal (Figure I-8). Therefore, L121W, L121Y, and L121F were incorporated in the M1 construct. M1-L121F did not lead to soluble expression, but M1-L121W and M1-L121Y were solubly expressed and purified.48 The all-trans retinal-bound crystal structures of M1-L121W and L121Y were determined, but attempts at crystallizing the 13-cis bound variants were unsuccessful (Table I-1). The overlay of all-trans retinal-bound M1-L121W and M1 showed that the chromophore had rotated about its C14-C15 and C6-C7 torsion angles such that the polyene methyl groups point in the opposite direction from those of all-trans retinal-bound M1, with a concomitant translation in the direction approximately perpendicular to the polyene axis (Figure I-9). M1-L121Y revealed some changes in retinal trajectory along the polyene but poor electron density for the β-ionone ring, indicating some disorder of the retinal inside the binding pocket. The failure in crystallizing 13-cis-bound proteins and the all-trans retinal chromophore disorder in these variants leads to the supposition that aromatic residues at 17 position 121 may be too large to comfortably accommodate retinal. We envisioned the a. d. Dark Green light DARK-state 13-trans LIGHT-state R111K R111K 13-cis c. f. b. e. S12 2.8 A P39Y R132Q 2.1 A 2.7 A 2.8 A L121E 2.9 A R111K S12 P39Y W109 R132Q L121E L121 R111K R132Q L121E L121 P39Y R111K W109 P39Y L121E W86 R111K K216 Figure I-11. a. UV-vis absorption spectrum of all-trans-retinal bound M1-L121E in the dark state (black) and after green light irradiation (long pass filter > 500 nm) (green), blue-light irradiation using band pass filter (440 nm ± 20 nm) (blue), and UV irradiation (UV band pass filter (300-400 nm)) (cyan). b. Hydrogen bonding network between the imine hydrogen of R111K and L121E, P39Y, H2O-303, R132Q, and S12. c. Overlay of M1 (carbon atoms in cyan) with all-trans-retinal bound M1-L121E (carbon atoms in green). d. Electron density (contoured at 1σ) of all-trans-retinylidene in M1-L121E in the dark state (green carbons) vs. 13-cis-15-syn-retinal imine (magenta carbons) generated after five-minute laser irradiation at ~400 nm. e. Overlay of all-trans-retinylidine-bound M1-L121E in the dark (green) with the 13-cis-15- syn imine (magenta) generated after five-minute laser irradiation at ~399 nm, showing the movement of Lys111, Glu121, and the rotation of the β-ionone ring in 13-cis upon isomerization. f. Overlay of all-trans- retinal bound M1-L121E imine (green) and all-trans-retinal bound bacteriorhodopsin (yellow carbons, PDB code: 1C3W), illustrating how the P39Y-L121E interaction mimics Trp86 in bacteriorhodopsin relative to retinal. All heteroatoms are colored by type, with N in blue and O in red. placement of acidic residues could stabilize the protonated species to an extent that would prevent imine bond isomerization. The proximity of Leu121 to the SB nitrogen atom led to the introduction of the L121E mutation in M1 yielding the R111K:Y134F:T54V:R132Q:P39Y:R59Y:L121E (M1-L121E) variant, Figure I-10, which successfully abolished the imine isomerization pathway. As shown in Figure I-11a, irradiation with either visible or UV light no longer switches the absorption between visible 18 light-absorbing (high pKa) and UV light-absorbing (low pKa) forms, but instead leads to a decrease in the absorbance of both forms independent of the wavelength of light, presumably due to changes in the structure of the chromophore (vida infra). 9 13 CHO A - all-trans-retinal B - 13-cis-retinal CHO CHO C - 9,13-di-cis-retinal a, dark A b. A (95%) B (5%) B c. co-injection with 13-cis-retinal B blue light irradiation A A (37%) B (27%) C (36%) BC A C d. green light irradiation A e. UV light irradiation A A (61%) B (18%) C (21%) B C A (34%) B (26%) C (40%) C B Figure I-12. HPLC chromatogram of retinal isomers extracted from all-trans retinal bound M1-L121E (recorded at 363 nm). a. Dark, after irradiation with mercury lamp- green light, blue light, UV, and coinjection with the pure 13-cis retinal (red).The ratios were corrected using the extinction coefficient of each isomer (all-trans 48,000 M-1cm-1, 13-cis 38,770 M-1cm-1, and 9,13-dicis retinal 32,380 M-1cm-1). 19 The structure of all-trans retinal-bound M1-L121E (Figure I-11b) was successfully determined and revealed a trans-imine, with the newly-installed Glu121 engaged as a salt 9 13 CHO A - all-trans-retinal B - 13-cis-retinal CHO CHO C - 9,13-di-cis-retinal a. dark A b. 30 sec laser irradiation A c. 5 min laser irradiation A A (91%) B (9%) B A (49%) B (19%) C (32%) C B A (37%) B (20%) C (43%) C B Figure I-13. HPLC chromatogram of M1-L121E bound with all-trans retinal before irradiation, after 30 sec laser irradiation, and after 5 min laser irradiation (recorded at 363 nm). The ratios were corrected using the extinction coefficient of each isomer. The ratios were corrected using the extinction coefficient of each isomer (all-trans 48,000 M-1cm-1, 13-cis 38,770 M-1cm-1, and 9,13-dicis retinal 32,380 M-1cm-1). bridge with the iminium (Figure I-11b). The overlay of M1 and M1-L121E also reveals a serendipitous structural rearrangement (Figure I-11c). Tyr39, which hydrogen-bonds with Gln132 in M1, adopts a new conformation, presumably as a result of a stronger hydrogen bond with Glu121 (Figure I-11b). In this conformation, Tyr39 sterically impinges on the bound retinal, causing a translation of the polyene in the vertical axis, sandwiching the chromophore between Trp109, located on the opposite side of the chromophore, and Tyr39, while tilting the chromophore relative to the trajectory in M1 (Figure I-11c). Figure I-11c shows that these changes occur in the polyene part of the chromophore, leaving the β-ionone ring in approximately the same position seen in M1. The direct salt bridge between Glu121 and the nitrogen atom of the trans-iminium gives a substantially 20 blue-shifted spectrum (468 nm, compared to 568 nm of the cis-iminium of M1),49 by localizing the positive charge on the polyene (Figures I-7c and I-10a).59 Glu121 seems to Hexane impurity all-trans 13-cis 9,13-dicis Figure I-14. 1H-NMR spectrum of retinal extracted from the protein complex recorded in CDCl3 solvent. not only shut off the imine bond isomerization pathway by substantially increasing the pKa of the trans-iminium, but it also provides a steric interaction that orders the chromophore in the binding pocket by sandwiching the chromophore between Tyr39 and Trp109. Unfortunately, crystals of 13-cis-bound M1-L121E in the dark could not be obtained. Intrigued by the large drop in the absorption of both SB and PSB upon irradiation with light of any wavelength, we extracted the chromophore from the M1-L121E samples and analyzed them by HPLC (Figures I-12 and I-13). Two peaks are apparent in the HPLC traces for the light-irradiated M1-L121E samples: one that co-elutes with authentic 13-cis retinal, and another that was assigned as 9,13-di-cis-retinal, based on a 1H-NMR aldehyde chemical shift comparison with the reported isomer (Figure I-14).48, 61 21 We surmised that the drop in absorption of our retinal-bound M1-L121E upon irradiation is due to both the generation of a new species with a lower extinction coefficient, along with some potential photobleaching. To gain further insight into the process, the irradiation of all-trans retinal-bound M1-L121E crystals with a 399 nm laser (Figure 11d) was performed. The crystals were frozen in liquid nitrogen after various time intervals of irradiation (30 seconds and 5 minutes), and their structures were determined. After five minutes and 30 seconds of laser irradiation, the electron density of the retinal shows no evidence of all-trans retinal and instead is consistent with its quantitative photo- conversion to 13-cis (Figures I-11d and I-15). As shown in Figure I-11e, the overlay of 13- cis generated after laser irradiation with all-trans shows both substantial β-ionone ring rotation and polyene movement. Furthermore, the conformational change of Glu121, which follows the movement of the imine nitrogen, and Lys108, is apparent. Interestingly, in all these structures, the position of Tyr39 remains fixed before and after irradiation. Tyr39 may act as a “steric pillar” that dictates the path of isomerization in a specific direction (Figure I-11e). The role of Tyr39 is reminiscent of Trp86 in bacteriorhodopsin, which sterically packs against the retinal and assists in directing the photoisomerization of the C13-C14 double bond 62. As shown in the overlay of M1-L121E and bacteriorhodopsin (Figure I-11F), Glu121 and Tyr39 occupy a similar space relative to the retinal, as does Trp86 in bacteriorhodopsin. Once again, the fortuitous, yet critical, role of the L121E mutation is apparent, as it aids in orienting Tyr39 in a conformation that mimics Trp86 in bacteriorhodopsin, leading to specific 13-cis photoisomerization.48 22 I-3-2. Thermal stability of 13-cis retinal in the crystal With photoisomerization to 13-cis confirmed, its thermal stability was investigated next. To find the shortest exposure time that can lead to 13-cis formation in the crystal, the crystal was irradiated with the indicated 399 nm laser (please see section I-7-10 for the details) for 30 seconds, and then the crystals were immediately flash-frozen in liquid N2. The 13-cis generated after 30 seconds of laser irradiation depicted the same β-ionone ring rotation as observed for the five-minute laser-irradiated crystal (M1-L121E light state) (Figure I-15), while the overlay of this structure with the all-trans retinal-bound structure of M1-L121E (dark state) shows drastic differences in the retinal trajectory and β-ionone ring. a. b. Figure I-15. a. Overlay of the dark state M1-L121E (pink) and after 30 sec laser irradiation (green-13-cis) revealing the changes in the polyene and ionone ring conformation of retinal after laser irradiation. Electron density belongs to the crystal irradiated for 30 sec (countered at 1σ). b. Overlay of 30 sec irradiation structure (green) with five min irradiation with laser (cyan) of M1-L121E, demonstrating the similar ionone ring and polyene conformation and verifying the formation of 13-cis after 30 sec laser irradiation. Electron density belongs to the crystal irradiated for 30 sec (countered at 1σ). To this end, photochemically-generated 13-cis-bound M1-L121E crystals were left in the dark at room temperature for 10 and 25 minutes. We anticipated the thermal reversion of the 13-cis isomer to all-trans, since it appears to be the thermodynamically more stable isomer of retinal in M1-L121E. This would complete the photocycle that is 23 a. b. c. Figure I-16. a. All-trans retinal-bound M1-L121E crystal after 30 sec irradiation with laser (399 nm) and subsequent incubation in the dark for 10 minutes (electron density contoured at 1σ). This crystal structure indicates the regeneration of all-trans retinal through the thermal isomerization of 13-cis. Although the chromophore adopted the all-trans retinal conformation, L121E does not reset completely. b. Structure from an all-trans retinal-bound M1-L121E crystal after 30 sec irradiation with laser (399 nm) and subsequent incubation in the dark for 25 minutes (electron density contoured at 1σ), showing the regeneration of all-trans retinal and L121E adopted the same conformation as the initial dark state conformation. c. Overlay of structures from all-trans bound M1- L121E crystal (dark state, in pink) and M1-L121E irradiated with the laser (399nm) for 30 sec and subsequently incubated in the dark for 25 minutes (purple) verifies the similarity of the retinal trajectory in both structures, showing the thermal isomerization of the 13-cis created after 30 sec laser irradiation to produce all-trans retinal (RMSD=0.1). The electron density (contoured at 1σ) was calculated from the structure after 25 minutes incubation in the dark. essentially identical to that of microbial rhodopsin, though on a significantly slower time scale. The structures of all the thermally-equilibrated crystals (after 10 and 25 minutes of equilibration) revealed electron density consistent with all-trans-bound M1-L121E, confirming the completion of a microbial rhodopsin photocycle (Figure I-16). The main characteristics of the structure, most notably the rotation and trajectory of the β-ionone ring, are clearly observed (Figure 1-16c). This is consistent with the fact that the extraction of irradiated M1-L121E shows the majority of the chromophore to be all-trans (Figures I- 24 12 and I-13). Thus, a complete photocycle, analogous to that of microbial rhodopsin, has been generated in an orthogonal protein and confirmed at atomic resolution.48 I-3-3. Atomic resolution crystal structure of 13-cis retinal in the dark We were intrigued by the ionone ring rotation (Figure I-11e) accompanying the photoconversion to 13-cis and hypothesized that the stabilization of the rotated ionone ring conformation would also stabilize 13-cis in the binding pocket. To this end, Ala32, located at the mouth of the binding pocket, was mutated to aromatic amino acids to discourage rotation of the ring due to steric compression afforded by the larger amino acids. Gratifyingly, the structure of 13-cis-retinal M1-L121E:A32Y clearly shows 13-cis a. F15 b. P39Y A32Y R59Y R111K W138 W189 Y83 W182 Y185 K216 W86 Figure I-17. a. Aromatic residues located in the binding pocket of 13-cis bound M1-L121E:A32Y; b. Bacteriorhodopsin binding pocket where the chromophore is sandwiched with aromatic residues. exclusively bound in the binding cavity (Figures I-17a and I-18a). Analysis of the high- resolution structure indicates that the 13-cis isomer is stabilized by the packing of several aromatic residues, including Tyr39, Tyr32, Tyr59, and Trp109, inside the binding pocket, mostly through “aromatic sandwiches” (Figure I-17a). Interestingly, this is similar to bacteriorhodopsin, where the chromophore finds itself well-packed in a sphere of aromatic residues, Figure I-17b. The overlay of this structure with the 13-cis-retinal M1- 25 L121E post-irradiation (13-cis-isomer) shows both the β-ionone and polyene moieties to be quite similar, while the overlay of this structure with the all-trans retinal-bound structures shows obvious structural variations (Figure I-18). These results demonstrate that by increasing the pKa (using the L121E mutation) and altering the β-ionone ring packing (via the A32Y mutation), a 13-cis “unfriendly” binding pocket (M1) is converted into a 13-cis-specific protein, nestling a well-ordered 13-cis inside the binding pocket.48 a. b. PSB c. R111K R111K R111K d. R111K Figure I-18. a. 13-cis electron density in M1-L121E:A32Y variant countered at 1.5 σ. The atoms are colored by type, with the C atoms in yellow. b. UV-vis spectrum of M1-L121E:A32Y bound with the 13- cis retinal, depicting the stable PSB after 10 hours of incubation. c. Overlay of 13-cis bound M1:L121E:A32Y (yellow) with all-trans bound M1:L121E (purple) without irradiation of either. d. Overlay of 13-cis retinal bound M1-L121E-A32Y (yellow) with the structure from a crystal of all-trans bound M1- L121E that was irradiated with laser (5 min) and immediately frozen (cyan). 26 I-3-4. Mechanistic insight and prediction Our interpretation of the results described above naturally leads to predictions regarding the mechanism in which both Glu121 and Tyr39 are essential for providing a W109 R111K L121E P39Q/Y Figure I-19. Overlay of M1-L121E (purple-blue) with M1-L121E:P39Q (pink). The P39Q mutation does not increase the steric interaction in the vicinity of the C13-C14 double bond. Furthermore, in the absence of P39Y, the chromophore does not move to a trajectory seen for M1-L121E, highlighting the essential role of P39Y for defining the chromophore trajectory in M1-L121E and presumably directing the isomerization toward 13-cis isomer formation. 13-cis isomerizing system. To test this prediction, mutations were made to both Tyr39 and Glu121. As predicted, the P39Q mutation that led to M1-L121E:P39Q resulted in a protein complex incapable of isomerizing the C13 double bond. Instead, it resulted in yet another imine isomerizing system. The crystal structure of this mutant was obtained in the dark at 2.2 Å resolution. As shown in the overlay of M1:L121E P39Q (dark state) vs. M1-L121E (dark state), replacing Tyr39 with Gln causes no steric interaction in the vicinity of the chromophore around the C13-C14 double bond. Furthermore, the chromophore is not pushed to the retinal trajectory as seen for M1-L121E, a conformation that is 27 presumably necessary for retinal isomerization around a C13-C14 double bond (Figure I- 19). Additionally, the UV spectrum of M1-P39Q:L121E, in contrast with M1-L121E, showed the PSB absorption (457 nm) red-shifted to a new PSB absorbing at 507 nm upon irradiation with UV (B.P 300-400 nm) (Figure I-20a). Interestingly, irradiation with green light (L.P.>500 nm) retrieves the original PSB (457nm). The irradiation of the M1- P39Q:L121E crystal with UV showed a drastic change in color, presumably correlating with the shift observed in the UV-vis spectrum (Figure I-20-b). The crystal structure after UV irradiation with a handheld TLC UV lamp was determined at 2.6 Å resolution. The conformation of the polyene and β-ionone ring confirmed the presence of all-trans retinal after five minutes of UV irradiation (Figure I-20c). Furthermore, as shown in Figure I-20d, the overlay of the UV-irradiated structure with M1-L121E bound with 13-cis, generated after five minutes of laser irradiation, further confirmed that the product of M1:L121E P39Q irradiation is all-trans. (It is not correlated with the 13-cis retinal trajectory.) This demonstrates the importance of the steric interaction between Tyr39 and the chromophore in directing isomerization to the 13 bond. To verify the role of the L121E mutation as an iminium-stabilizing species, we examined the photochemical characteristics of the structurally conservative M1-L121Q mutant. As shown in Figure I23a, irradiation of M1-L121Q in solution leads to behavior similar to mutants that undergo imine isomerization with two apparent pKa’s for each isomer. UV light irradiation (<400 nm) results in a loss of intensity of the 374 nm SB absorption, concomitant with a large increase in the PSB absorption at 522 nm. Conversely, irradiation with visible light (>500 nm) reduces the PSB absorption at 522 nm, and increases the 374 nm SB absorption. At first, this result was not unexpected, 28 since the loss of the salt bridge afforded through the interaction of the iminium with Glu121 could reestablish the imine cis/trans isomerization observed before, leading to two states a) SB b) PSB (457nm) PSB (507nm) Dark UV irradiated c) W109 R111K L121E P39Q d) R111K R111K Figure I-20. a. UV-vis spectrum of M1-P39Q:L121E in the dark, after UV (B.P.300-400 nm) irradiation, and after subsequent green-light irradiation (L.P 500 nm). b. Drastic color change in the crystal upon irradiation with the UV. c. Overlay of dark state (pink) and UV-irradiated state (yellow) of M1- P39Q:L121E crystal structures, depicting the all-trans conformation for both states. d. Overlay of UV- irradiated crystal structure of M1-P39Q:L121E vs. the light state of M1-L121E. with distinct pKa’s via iminium cis/trans isomerization. Nonetheless, crystal structure analysis reveals a surprising and unanticipated mechanism for the observed pKa change: photo-irradiation leads to side-chain motion.48 29 a. c. e. b. Dark P39Y R111K 3 Å L121Q After Laser Irradiation (~400 nm) d. After Laser Irradiation (~532 nm) R111K 4.8 Å 2.7 Å L121Q 4.7 Å R111K P39Y L121Q f. Dark ~532 nm laser P39Y P39Y L121Q 3.0 Å R111K Figure I-21. a. UV-vis spectra of M1-L121Q before and after irradiation with green light, showing a high/low pKa system where green-light irradiation leads to decreased PSB and increased SB absorption. b. Crystal structure of M1-L121Q obtained in the dark. L121Q has a direct hydrogen bond with R111K. c. Crystal structure of M1-L121Q after 5 min irradiation with 399 nm laser. L121Q and P39Y adopted two conformations but no isomerization occurred. This shows the importance of sidechain conformation stability in directing the photoisomerization product. L121Q in one conformation is far away from iminium (4.8 Å). d. After 10 min laser irradiation of crystals with 532 nm laser (close to PSB maximum absorption), the L121Q swings away from iminium, lowering the pKa of the chromophore and leading to the SB formation. e. Overlay of the all-trans- bound M1-L121Q imine before (green carbons) and after (salmon carbons) laser irradiation at 528 nm, clearly showing the Gln121 movement. f. Drastic color change of crystals upon irradiation with 532 nm laser. As illustrated in Figure I-21, the dark-state structure of M1-L121Q is almost 30 identical to that of M1-L121E, with Gln121 making a direct hydrogen bond with the iminium nitrogen atom. However, irradiation with the 399 nm laser shows no evidence of chromophore isomerization or conformational change. Instead, Gln121 adopts two conformations, the original conformation seen in the dark, and a new position that is rotated away from the iminium, resulting in a naked imine nitrogen atom with no other interacting residues (Figure I-21c). Such an imine would be expected to have a substantially-suppressed pKa relative to the dark state. Subsequent irradiation with a 532 nm laser leads to a complete conversion to the low pKa form, where Gln121 makes no interaction with the imine (Figures I-21d and I-21e). In contrast with the expected isomerization of the chromophore’s double bond that would lead to a change in the environment of the imine, and thus its pKa, the excited state of the M1-L121Q retinal complex instead leads to a change in the conformation of an interacting amino acid sidechain, effectively altering the environment. This is reminiscent, albeit on a much longer timescale, of the sub-picosecond X-ray laser time-resolved study of bacteriorhodopsin, which showed that the critical water molecule moves away from the chromophore in response to the change in the polarity of the excited state, before the chromophore itself displays substantial motion.19 Such subtle changes in the structure of the surroundings upon irradiation would be difficult to identify in the absence of high- resolution structural interrogation of the irradiated systems. This may explain the complex hydrogen-bonding network holding the two key aspartates in place in bacteriorhodopsin.5, 63 Perhaps this prevents the interacting residues from moving during photo excitation, resulting in a low pKa form, which would sabotage the bacteriorhodopsin photocycle.48 31 I-3-5. Irradiation study of M1-L121E:A32Y As cited above, we were able to confine the 13-cis retinal in the binding pocket of our protein by placing A32Y in the vicinity of the β-ionone ring, in the context of M1-L121E, a. b. R111K L121E R111K 2.7 Å 3.7 Å L121E P39Y d. 2.4 Å R132Q PSB: 472 nm PSB: 499 nm P39Y R132Q c. R111K O40 O49 O128 L121E P39Y R132Q Figure I-22. a. Dark state of M1-L121E:A32Y bound with 13-cis retinal, showing well-ordered water molecules mediated between the retinal and protein (electron density is contoured to 1 s). b. After 399 nm laser irradiation, no isomerization occurs; however, the Lys111 side chain and retinal polyene move vertically ~ 0.7 Å, which leads to Glu121 movement and disordering of the water molecules (electron density is contoured to 1 s). c. Overlay of dark state (purple- blue) vs. light state demonstrated the subtle movement of the chromophore, Lys111 side chain, and Glu121, disordering the water molecules after irradiation. giving rise to a well-ordered chromophore inside the binding cavity and well-diffracting crystals. The next question is what would happen to the M1-L121E-A32Y/13-cis retinal complex upon irradiation. To test this, a 399 nm laser was used as a source of irradiation. 32 As shown in Figure I-22, upon irradiation with the laser, the chromophore still holds the 13-cis retinal configuration; however, the position of Lys111, the retinal polyene, P39Y, and L121E change subtly, which disorders the ordered water molecules around the iminium and increases the distance between L121E and the nitrogen of the imine (from 2.8 to 3.7 Å). The movement of L121E and disordering of the water molecules around the polyene are expected to suppress the pKa of the chromophore and shift the retinal absorption to an SB population. This observation is consistent with the UV irradiation of the M1-L121E:A32Y/retinal complex in solution, revealing that upon irradiation with UV, the SB population increases. Additionally, the UV-vis spectra depicts that upon irradiation with UV, a new, short-lived red-shifted PSB forms. (Due to the short lifetime of the red- shifted PSB, we were not able to completely detect the absorption wavelength by our method, but based on the color of the solution right after irradiation, it would be expected that the new PSB possesses a higher absorption wavelength than 499 nm.) As shown in Figure I-22c, increasing the distance between L121E and nitrogen of iminium (from 2.8 Å to 3.7 Å), along with dispelling the water molecules around the iminium, might lead to the new red-shifted PSB by allowing the delocalization of charge on the polyene. Overall, the UV-vis spectra and crystal structure together reveal another important photoswitching mechanism for the retinal/protein complex upon interaction with light. The irradiated chromophore can avoid the absorption of light through subtle movement in the binding pocket, which leads to the changes in the surrounding residues and water molecules, causing a low pKa form of the chromophore and a bathochromic shift from the irradiated wavelength. Presumably, in this case, the binding pocket restricts any vibrational flexibility of the chromophore for isomerization where A32Y is perpendicular to the ionone ring, and 33 in contrast to L121Q, the charged residue of L121E does not reorient without the impulsive motion of the Lys111 side chain. However, we cannot rule out another possible mechanism for this observation, in which the chromophore undergoes a fast photoisomerization and thermal reversion to the original state (representing the bacteriorhodopsin photocycle in reverse), which is not detectable within the time window of our method of study. I-3-6. Other mutational studies on the M1 construct a. d. b. R111K L121N e. R111K R111K L121N c. f. R111K L121M I63D L121E P39F L121E P39E R111K Q97 T107 E73 I52 R111K L121E R132Q S12 Figure I-23. a. Crystal structure of M1-L121N/all-trans retinal in the dark. b. Crystal structure of M1- L121N/all-trans retinal after 3 min laser irradiation with 399 nm laser. c. Crystal structure of M1-L121M /all-trans retinal in the dark (PDB code:6NNX). d. Crystal structure of M1-L121E:P39F/ all-trans retinal in the dark. e. Crystal structure of M1-L121E:P39E in the dark (PDB code:6NNY). f. Crystal structure of M1-L121E:I63D in the dark (PDB code: 6NOE). The presence of I63D can potentially generate a hydrogen bond network on the other side of the chromophore opposed to L121E. The electron density was contoured to 1 s for the 2FO-FC map. Other than the mutational studies cited above, other variants using the M1-L121E construct were made, as shown in Table I-2. 34 The M1-L121N/retinal photocycle UV-vis spectra and the crystal structure were obtained, as shown in Figures I-23a and I-24a. The photoirradiation of M1-L121N/retinal depicted a two-pKa system reminiscent of M1-L121Q . The crystal structure of the dark and laser- Table I-2. Other mutational studies on M1 construct. Variants M1-L121N M1-L121M M1-L121D M1-L121E:P39F M1-L121E:P39N M1-L121E:P39E M1-L121E:F65Y M1-L121E:I63D M1-L121E:F3Q Crystal structure Available Available (PDB code:6NNX) Was not crystallized Available Available Available (PDB code:6NNY) Was not solubly expressed Available (PDB code:6NOE) Was not solubly expressed M1-L121E:A33W:F3Q Was not solubly expressed M1-L121E:I31W M1-L121E:A32W M1-L121E:T56Y M1-L121E:I52E Was not solubly expressed Available Was not solubly expressed Was not solubly expressed M1-L121E:A23Y:I52E Was not solubly expressed irradiated forms were obtained (Figures I-23A and I-23b), demonstrating that the chromophore does not undergo isomerization upon 399 nm laser irradiation. However, the exact position of L121N in the dark and light state is not as clear as L121Q in elucidating another photoinduced side-chain motion mechanism. M1-L121D was also 35 created, which shows some photocycle behavior with a high photobleaching effect (Figure I-24b). The crystallization attempt for this variant was not successful. M1-121M/retinal was prepared and crystallized (Figure I-23c), and showed no tangible spectroscopic or structural variation upon light irradiation. The M1-L121E:P39F variant was made and crystallized in the presence of retinal. The crystal structure of this construct revealed a rotation of the β-ionone ring as a result of the P39F mutation, which possesses a conformation different from P39Y held by L121E. The chromophore possessed a low pKa without showing a photocycle (Figure I-23d). M1-L121E:P39E was successfully crystallized (Figure I-23e), and the spectroscopic data was measured. The UV-vis spectrum demonstrates the same feature as M1-L121E, where upon irradiation with different wavelengths, both the SB and PSB population decrease; however, the spectrum also shows some recovery of the SB population over time using UV-vis spectroscopy (Figure I-24e). The chromophore trajectory was very similar to M1-L121E, as shown in Figure I-23. The next mutant was M1-L121E:P39N, which depicted the photocycle of the protein/retinal complex upon irradiation with green and UV light, Figure I-24c. However, this mutant could not be crystallized. Some new variants were also made with the hope of stabilizing the 13-cis-15-anti configuration by imposing F3Q, I63D, F65Y,I52E, mutations to yield M1-L121E:F3Q, M1-L121E:A33W:F3Q, M1-L121E:I63D, M1- L121E:F65Y, M1-L121E:I52E, and M1-L121E:A32Y:I52E. Unfortunately, only M1- L121E:I63D was solubly expressed. The crystal structure of M1-L121E:I63D bound with all-trans was obtained (Figure I-23f); however, no interesting spectroscopic behavior or deviation from M1-L121E was observed. Finally, to elucidate the effect of aromatic residues in the vicinity of the β-ionone ring, M1-L121E:I31W, M1-L121E:A32W, and M1- 36 L121E:T56Y were created. Unfortunately, M1-L121E:I31W and M1-L121E:T56Y were not crystallized; however, M1-L121E:A32W was crystallized with retinal and showed no electron density for the chromophore, indicating a disordered ligand in the binding pocket. a. b. SB PSB c. SB PSB SB PSB d. SB e. SB PSB Figure I-24. UV-vis photocycle spectra of different all-trans retinal-bound variants: a. M1-L121N, b. M1- L121D, c. M1-L121E:P39N, d. M1-L121E:P39F, and e. M1-L121E:P39E. UV (B.P 300-400 nm) and green light irradiation (Y filter L.P>500 nm) were performed with a mercury lamp as the source of irradiation. 37 I-4. 13-cis bound hCRBPII crystal structure Having demonstrated the key elements of stabilizing 13-cis retinal in the protein binding pocket using the M1-L121E:A32Y variant of CRABPII, the next goal was to a. Q38 A33 P39Y A32Y R58 R59Y L121E L117 b. R58W Q108K R111K A33W W106 Q108K Q38Y L117D W106 W109 c. R59Y R58W A32Y A33W W109 W106 L121E Q38Y P39Y L117D Figure I-25. Overlay of M1-L121E:A32Y-CRABPII (yellow) and Q108K:K40L-hCRBPII (blue) showing the equivalent positions between the two proteins in the retinal binding pocket b. Crystal structure of H1 bound with 13-cis retinal. The electron density is contoured to 1s. c. Overlay of H1 (blue) and M1-L121E:A32Y bound with 13-cis retinal, highlighting the aromatic interactions holding the chromophore. introduce hCRBPII binding residues in pocket, these key the which possesses almost 40% sequence similarity with CRABPII. The Q108K:K40L:T51V:T53S:R58W:Y19W:A33W:L117D:Q38Y (H1) variant was employed as our template, which contains several aromatic residues (R58W, Y19W, A33W, and Q38Y) with an acidic residue, L117D, in the vicinity of nitrogen of iminium. Notably, the 33, 38, 58, and 117 positions in CRBPII correlate with the 32, 39, 59, and 121 positions in CRABPII (Figure I-25a). To our delight, the crystal structure of 13-cis bound H1 was obtained at 1.1 Å resolution, depicting a well-ordered ligand in the binding pocket (Figure 38 I-25b). Based on the overlay of M1-L121E:A32Y/13-cis and H1/13-cis, as anticipated, the H1 construct creates aromatic sandwiches surrounding 13-cis retinal, leading to a well- ordered ligand in the binding pocket (Figure I-25c). This is in contrast to the all-trans bound H1 crystal structure that led to a disordered chromophore. This once again demonstrates the critical role of the aromatic residues in stabilizing the 13-cis in the rhodopsin mimic. I-5. Time-resolved x-ray crystallography experiment with rhodopsin mimic As mentioned above, using CRABPII as an engineering template, we observed four distinct processes that occur upon irradiation of the retinal chromophore in the Figure I-26. Simplified Potential Energy Surface for the forward (top) and reverse (bottom) dynamics of M1 initiated with UV excitation (forward) or green-light (reverse) excitation. Wavelengths are indications of the true peak wavelength for the intermediates (provided by Adam Jenkins). binding cavity of the protein: imine bond isomerization (M1), 13-cis specific isomerization (M1-L121E), photoinduced side-chain motion without a retinal conformational change (M1-L121Q), and retinylidene movement without isomerization, leading to the side-chain 39 motion (M1-L121E:A32Y). Elucidating the detailed mechanism of these systems requires a time-resolved experiment. As a starting point, we decided to use the M1 system, where the imine bond undergoes photoisomerization for the time-resolved experiment. In collaboration with Prof. Larsen’s group at UC Davis (Mr. Adam Jenkins), we were able to identify several intermediates using cryokinetics and ultrafast spectroscopy experiments. For M1, Mr. Adam Jenkins calculated the simple potential-energy surfaces of the ground- state intermediates for both forward (SB) and reverse (PSB) reaction directions Figure I-27. The pump-probe method was used for the time-resolved x-ray experiment. The laser was used to initiate the reactions in the crystals, and the x-ray with microsecond-duration pulses was used to monitor the changes in the crystals. The x-ray and laser are positioned perpendicular to each other to increase the efficiency of the experiment. Photo credit: Prof. Geiger. (Figure I-26). The ultrafast data depicted one or more structural evolutions before a protonation (or deprotonation) reaction ensues. For the forward direction (SB excitation), the excited-state lifetime was calculated to be 30 ps. For the reverse direction (PSB 40 excitation), the excited-state kinetics were much faster (4.4 ps). This data revealed the possibility of observing intermediates during the photoisomerization of M1. The next step was to study the time-resolved x-ray experiment in M1 crystals, where we could identify the structure of intermediates at atomic resolution. Two important features of M1 crystals made them great choices for these experiments: first, the crystals are not mosaic; second, the reactions in these crystals are highly reversible when sequentially irradiated with green and UV light, and are accompanied by concomitant, drastic color changes offering a visible signature for each isomeric form. To perform the time-resolved experiment, the synchrotron resource for time-resolved X-ray science at BioCARS was chosen. Many M1 crystals were grown in a sitting-drops for the first experiment (setup shown in Figure I- 27). The “pump-probe” method was selected for the time-resolved x-ray crystallography a. Dark R111K b. 50 μs W109 c. 100 μs W109 R111K L121 R111K L121 d. 500 μs W109 R111K L121 e. 1 ms W109 using the Laue method, contoured at 1σ for. The time-dependent difference maps were subtracted from Figure I-28. a. A 2FO-FC Electron density map of the dark state of M1 collected at room temperature the dark state: b. 50 µs, c. 100 µs, d. 500 µs, and e. 1 ms. R111K L121 of the M1-retinal complex. The UV laser was used to initiate the reaction in the crystals, and the X-ray with microsecond-duration pulses was used to monitor the reaction in the 41 crystals. The X-ray and laser were positioned perpendicular to each other to increase the efficiency of the data collection. In contrast with a normal synchrotron experiment, where monochromatic X-rays are used and the crystal is rotated on a goniometer at 100 K, in this experiment, the crystal was located in a precisely stationary position, and a “pink” polychromatic X-ray beam is used. The result is Laue diffraction, which gives significantly more data on a single image. The data must also be collected at room temperature as cryo temperatures will completely freeze out any molecular motion. The Laue diffraction experiment on the M1 crystals was performed, and the data was kindly processed using the “linear analytical” integration method by Dr. Srajer from BioCARS. The initial data refinement showed promising preliminary results, with an excellent electron density for retinal (Figure I-28a). The next step was to investigate the changes occurring in the crystals during the photoisomerization using the pump-probe method. The 370 nm laser with a ns pulse width was chosen as the irradiation source (toward PSB formation-cis iminium). An LED (570 nm) illumination source was used to reset the crystals to the SB state. In this experiment, performed by Prof. Geiger and Dr. Srajer, several data points were collected (Dark, 50 µs, 100 µs, 0.5 ms, and 1ms). As shown in Figure I-28, the time- dependent difference map was obtained by subtraction: ΔF(hkl,t) = |F(hkl, t)| - |F(hkl, dark)|. Additionally, the PDB model from the dark state was represented at each time point. Essentially, the presence of a positive and negative map in the vicinity of each atom demonstrates the movement of those atoms at each time point relative to the dark-state PDB model. As shown in Figure I-28, the change in the polyene is not significant, which leads to two hypotheses: first, the time scale we are looking at is not appropriate, and we should study the shorter time points where larger changes might 42 be visible; or second, we simply need to increase the exposure time of laser irradiation, since the quantum yield of photoisomerization is not quite high enough in the crystal. These are the studies that will be performed in the future. However, these preliminary time-resolved experimental results, both in solution and crystal, demonstrate the potential of our engineered model system for studying retinal photoisomerization in a simpler system than membrane proteins. The advantage of this model system vs. membrane proteins is that our proteins can be easily expressed, crystallized, and studied using simple or time-resolved techniques. I-6. Conclusion and future direction In conclusion, an artificially-designed protein that recapitulates the photocycle of microbial rhodopsins, specifically and quantitatively photoisomerizing all-trans-retinal to 13-cis-retinal in a well-diffracting crystal, has been created. Because of its slow rate of isomerization, our engineered rhodopsin mimic can be exploited using relatively simple time-resolved spectroscopic and crystallographic methods. Our mimic also introduced an ideal system for time-resolved crystallography. Using this system, a mechanism for protein photoswitching that involves no net conformational change of the chromophore was identified at atomic resolution. This observation, which would be difficult to observe using other techniques, opens the door to a novel strategy for the design and study of photoactive and photoswitchable protein systems. Our ultimate goal is to employ our system for time-resolved x-ray and FT-IR experiments to elucidate the intermediates of the isomerization and compare them with natural microbial rhodopsin. Another exciting avenue would be to explore the development of an isomerase, where the chromophore is hydrolyzed and released from 43 the protein after isomerization. Exploring other isomerization events, such as mimicking eukaryotic rhodopsin isomerization where the 11-cis retinal isomerizes to all-trans retinal, using our model system, is another potential avenue of future investigation. Finally, if we can alter the conformation of the protein during photoisomerization, we can generate a new optogenetic tool in this system that can be exploited for a variety of protein- engineering applications. I-7. Experimental 48 I-7-1. Materials and methods All-trans retinal, 13-cis retinal, and other chemicals used in this study were purchased from Sigma Aldrich unless otherwise specified. A BioLogic DuoFlow (BioRad) chromatography instrument was used for the protein purifications. Source Q and Fast Q Table I-3. PCR protocol. PCR program 94 oC 94 oC temperature 3-5 oC lower than primer melting temperature
 72o C
 72 oC 10 o C 3 min 20 sec 50 sec 4 min 30 sec 10 min 5 min 1x 20x 1x 1x anion exchange resins were purchased from GE Health Care. An Ultrasonic Homogenizer from Biologics, Inc. was used for sonication of the samples. 44 Table I-4. PCR program. 1x 20x PCR program 94 oC 94 oC temperature 3-5 oC lower than primer melting temperature
 3 min 20 sec 50 sec 72o C
 4 min 30 sec (3 min 30 Sec for 1x 72 oC CRBPII) 10 min I-7-2. List of primers: The hCRABPII-pET17b plasmid previously described was employed for mutagenesis.64 L121W Forward: 5’- GAACTGATCTGGACCATGACG-3’ Reverse: 5’-CGTCATGGTCCAGATCAGTTC-3’ L121E Forward: 5’- GAACTGATCGAAACCATGACG-3’ Reverse: 5’-CGTCATGGTTTCGATCAGTTC-3’ L121Y Forward: 5’- GAACTGATCTACACCATGACG-3’ Reverse: 5’- CGTCATGGTGTAGATCAGTTC-3’ 45 L121Q Forward: 5’- GAACTGATCCAGACCATGACG-3’ Reverse: 5’- CGT CAT GGT CTG GAT CAG TTC-3’ P39Y Forward: 5'- GCAGCGTCCAAGTATGCAGTGG -3' Reverse: 5'- CCACTGCTTGCTTGGACGCTGC - 3' P39Q Forward: 5’- GCAGCGTCCAAGCAAGCAGTGG-3’ Reverse: 5’- CCACTGCAAACTTGGACGCTGC-3’ A32Y Forward: 5’-GAGGAAGATTTATGTGGCTGC-3’ Reverse: 5’-GCAGCCACATAAATCTTCCTC-3 T54V Forward:5’-CTACATCAAAGTCTCCACCACCGTGCG-3’ Reverse: 5’- CGCACGGTGGTGGAGACTTTGATGTAG -3’ R111K Forward: 5’-CCCAAGACCTCGTGGACCAAAGAACTGACCAACGATGGG-3’ Reverse: 5’-CCCATCGTTGGTCAGTTCTTTGGTCCACGAGGTCTTGGG-3’ R132Q:Y134F Forward: 5’-GTTGTGTGCACCCAGGTCTTCGTCCG-3’ Reverse: 5’-CGGACGAAGACCTGGGTGCACACAAC-3’ R59Y Forward: 5’-CCTCCACCACCGTGTACACCACAGAG-3’ 46 Reverse: 5’- CTCTGTGGTGTACACGGTGGTGGAGG -3’ I-7-3. Transformation of PCR product 5 μL of PCR product was added to DH5α competent cells (50 μL) thawed on ice. The sample was kept 30 minutes in ice. After heat-shock at 42 ̊C for 45 seconds, 450 μL of LB was added to the sample. The sample and agar plate containing ampicillin (100 μg/mL) were incubated at 37 ̊C for 30 minutes. Then 100 μL of the solution was spread on the LB agar plate and incubated at 37 ̊C overnight. The PCR product was purified using the Promega DNA purification Kit. I-7-4. Expression of hCRABPII variants The gene was transformed into BL21 (DE3) pLysE E. coli competent cells. The same protocol mentioned for PCR was exploited, with the only difference being the addition of 100 μg/mL ampicillin and 28 μg/mL chloramphenicol as antibiotics. To inoculate 1 L of LB with ampicillin (100 μg/mL) and chloramphenicol (28 μg/mL), a single colony was used. After the cell culture was grown at 37 ̊C while shaking for 8 h, (0D~0.8- 0.9) the overexpression was started by adding 1 mL of 1M IPTG solution into 1 L cell culture (overall concentration 1.0 mM of IPTG). The solution was shaken at 19 ̊C for 36 hours.49, 53 I-7-5. Protein isolation and purification By centrifuging for 20 min at 5,000 rpm (4 ̊C), the cells were collected. The supernatant was discarded and the cells were resuspended in 60 mL of Tris buffer (10 mM Tris.HCl, pH=8.0). Then ultrasonication was used to lyse the cells (Power 60%, 1 min x 3), and the sonicated mixture centrifuged at 16 ̊C (10,000 rpm, 20 min). 47 Purification by ion exchange chromatography using Q SepharoseTM, Fast Flow resin was next performed at 4 ̊C. The protein was bound to the column by gravity flow, the column washed with 50 mL of 10 mM Tris. HCl buffer (pH=8.0). The bound protein was then eluted with 10 mM Tris. HCl, 150 mM NaCl, pH = 8.0, the pure fractions were desalted using EMD Millipore centriprep centrifugal units (cutoffs:10 kDa) with 10 mM Tris. HCl pH=8.0 buffer, three times, and loaded on a second anion exchange column (15Q, GE Health Sciences, BioLogic DuoFlow system). The Source Q program, as mentioned previously,64 was used for the purification. I-7-6. UV-vis measurements UV-vis spectroscopy was performed using a Cary 300 and Cary 100 Bio Win UV UV-Vis spectrophotometer (Varian Instruments). The samples were prepared in a phosphate buffer purchased from Sigma Aldrich. The concentration of retinal was kept at a 0.5 equivalent of the protein concentration in all experiments. The PSB has absorptions corresponding to λmax > 450 nm, while deprotonated imine peaks (SB) appear at λmax ~370 nm. The all-trans retinal extinction coefficient in ethanol is 48,000 M-1cm-1 at 368 nm, and the 13-cis retinal extinction coefficient in ethanol is 38,770 M-1cm-1 at 363 nm, as reported. 65 The protein extinction coefficient for different variants was calculated using the Gill and Hippel method.66 M1 ε280nm = 22,400 M-1cm-1 M1-L121E ε280nm= 22,283 M-1cm-1 M1-L121Q ε280nm= 20,293 M-1cm-1 M1-L121E:P39Q ε280nm= 16,776 M-1cm-1 48 M1-L121E:A32Y ε280nm= 24,500 M-1cm-1 I-7-7. pKa determination of mutants The pKa was determined by titration (absorbance vs. pH) and was fitted by the protocol previously described.53 The following equation was used for fitting the data and calculating pKa by employing KaleidaGraph: ΔA = ΔA0 /(1+10[pH−pKa]) I-7-8. Irradiation of protein/retinal complexes in solution An Oriel Illuminator (Model 66142, Oriel Instruments) attached to a power supply (Model 668820, Oriel Instruments, 500 W Mercury (Xenon) lamp) was employed for the light irradiation of samples. A combination of two filters was utilized for all light irradiations. A glass filter (6 mm thickness) was utilized to filter UV light below 320 nm. The second filter for UV irradiations was a U-360 (UV) 2” square band-pass filter (center wavelength (CWL) = 360 nm, full width at half-maximum height (FWHM) = 45 nm, purchased from Edmund Optics). For the visible light green-light irradiations, a Y-50 2” square long-pass filter (cut-off position = 500 ± 6 nm, purchased from Edmund Optics) was used. The third filter was a 440 ± 20 nm UV-VIS Bandpass Filter purchased from Edmund Optics for blue light irradiation.50 I-7-9. Protein crystallization The proteins were concentrated to ~20-25 mg/ml using EMD Millipore centriprep concentrators (cutoffs: 10 kDa), and ~4 equiv. of ligands (all-trans retinal or 13-cis retinal) were added in the dark. The vapor diffusion method was used for crystallization using 24 well plates in the dark (purchased from Hampton Research) with 1 ml solution in the reservoir. M1-L121E, M1-L121Q bound with all-trans and M1-L121E:A32Y bound with 49 13-cis retinal were crystallized under the following conditions: 20% PEG 3350 and 0.1 M DL-malic acid (titrated with NaOH to pH=6.0 and 6.5) at 4o C. For M1-L121W, M1-L121Y the crystals grew at 20% PEG 3350, 0.1 M sodium malonate (pH=6.0-6.5) 4o C. For M1- L121E:P39Q, the crystals grew at 20% PEG 3350, 0.1 M bis-tris propane (pH=6.0-6.5), 0.2 M sodium fluoride. All crystals appeared after 1-3 days and were then frozen using liquid nitrogen in a cryogenic buffer containing their mother liquor along with 20% glycerol in the dark. Diffraction data were collected at the Advanced Photon Source (APS) (Argonne National Laboratory IL) LS-CAT, (sector 21-ID-D,F, and G) using Eiger 9M, and MAR300 detectors, and 1.00Å wavelength radiation at 100K. The initial diffraction data was indexed, processed and scaled using the HKL2000 software package.67 The structures were solved by molecular replacement using PHASER_ENREF_568 in PHENIX 69 and hCRABPII (PDB entry 4YFP) as a search model. The initial electron density map was produced using Phaser-MR in PHENIX. Model rebuilding, placement of water molecules, etc. were performed using COOT.70 The structures were refined using the PHENIX program package. The chromophore was built using the COOT and PHENIX programs to generate restraints. 50 Table I-5. Crystallographic data collection and refinement statistics for various dark and light states of all-trans bound M1-L121E. M1-L121E bound with all-trans retinal in dark M1-L121E bound with all-trans retinal, 5 minutes irradiation with 399nm laser M1-L121E bound with all- trans retinal, 30 seconds irradiation with 399 nm laser P 3121 58.986 58.986 100.314 90 90 120 1 P 3121 58.567 58.567 98.974 90 90 120 1 P 3121 58.343 58.343 99.865 90 90 120 1 M1-L121E bound with all-trans retinal, 30 seconds irradiation with 399 nm laser, 10 minutes incubation in dark P 3121 58.586 58.586 100.178 M1-L121E bound with all-trans retinal, 30 seconds irradiation with 399 nm laser, 25 minutes incubation in dark P 3121 58.501 58.501 99.982 90 90 120 1 90 90 120 1 Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge Resolution (Å) (Last Shell) 405325 503530 777250 478475 463610 16473 (1618) 12474 (1214) 16010 (1552) 12273 (1205) 99.92 (100.00) 34.4 (2.15) 0.05 (0.87) 35.79 - 1.90 (1.93 - 1.90) 99.92 (99.92) 26 (1.87) 0.078 (0.95) 45.14 - 2.07 (2.14 - 2.07) 99.48 (98.66) 99.83 (100.00) 35.5 (3.3) 0.047 (0.36) 28.0 -1.90 (1.97 - 1.90) 36.75 (3.80) 0.067 (0.68) 29.29-2.09 (2.16 - 2.09) 13681(1345) 99.54 (99.78) 44.63 (3.3) 0.065 (0.62) 29.25 - 2.01 (2.01-2.08) Rwork/ Rfree 19.50/22.74 19.31 /25.21 0.009 0.87 40.13 101 Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID 0.008 1.06 40.03 45 18.93/21.70 18.30/22.02 19.87/23.40 0.008 0.88 32.83 42 0.007 0.87 39.40 54 0.008 0.86 52.38 38 6MOP 6MQZ 6MQY 6MQW 6MQX a Values in the parenthesis refer to the last resolution shell. 51 Table I-6. Crystallographic data collection and refinement statistics for 13-cis bound M1-L121E:A32Y, and all-trans bound M1-L121W, M1-L121Y, and M1-L121Q dark states. M1-L121E- A32Y bound with 13-cis retinal in dark M1-L121W bound with all-trans retinal in dark M1-L121Y bound with all- trans retinal in dark M1-L121Q bound with all-trans retinal in dark Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness Average I/s (%) Rmerge Resolution (Å) (Last Shell) P 3121 58.761 58.761 99.174 90 90 120 1 P 3121 58.117 58.117 99.629 90 90 120 1 P 3121 58.469 58.469 100.118 90 90 120 1 556609 27796 (2743) 99.77 (100) 36 (2.18) 0.065 (0.60) 35.51 - 1.58 (1.64 - 1.58) 1093404 10327 (1019) 98.99 (98.25) 22.8 (1.6) 0.077 (0.58) 27.90 - 2.20 (2.28 - 2.20) 395603 19270 (1890) 99.94 (99.89) 39.9 (3.6) 0.044 (0.51) 35.60 - 1.79 (1.85 - 1.79) P 3121 59.219 59.219 99.837 90 90 120 1 552392 20945 (2022) 99.84 (98.92) 45 (2.4) 0.055 (0.87) 35.77 - 1.75 (1.81 -1.75) Rwork/ Rfree (%) 20.29/22.18 22.62/26.57 22.31/24.72 21.47/23.34 Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID 0.006 0.86 32.72 111 0.009 1.04 56.32 29 0.007 0.95 30.86 99 6MPK 6MOQ 6MOR a Values in the parenthesis refer to the last resolution shell. 0.009 0.92 32.99 67 6MOV 52 Table I-7. Crystallographic data collection and refinement statistics for different light states of all-trans bound M1-L121Q. M1-L121Q bound with all-trans retinal, 5 minutes irradiation with laser (399 nm) M1-L121Q- bound with all-trans retinal, 10 minutes irradiation with laser (532 nm) Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge Resolution (Å) (Last Shell) Rwork/ Rfree (%) Bond Length (Å) Bond Angle (°) Average B factor Number of water Molecules PDB ID P 3121 59.075 59.075 99.814 90 90 120 1 378799 17237 (1682) 99.94 (100.00) 40.6 (2.55) 0.044 (0.734) 35.72 - 1.87 (1.937 - 1.87) P 3121 59.238 59.238 99.393 90 90 120 1 545064 11843 (1167) 99.82 (98.90) 19.3 (1.8) 0.06 (0.51) 35.69 - 2.12 (2.20 - 2.12) 21.52/25.96 18.85/24.31 0.009 0.95 35.95 50 6MQI 0.009 0.94 41.35 43 6MQJ a Values in the parenthesis refer to the last resolution shell. 53 Table I-8. Crystallographic data collection and refinement statistics for dark and light states of all-trans bound M1-L121E:P39Q. M1:L121E :P39Q bound with all-trans in dark M1:L121E:P39Q all-trans 5 minutes UV irradiation bound with Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge P 3121 58.546 58.546 99.525 90 90 120 1 258241 10769 (1059) 99.94 (99.81) 14.5 (2.2) 0.089 (0.64) P 3121 58.476 58.476 99.552 90 90 120 1 534615 6064 (570) 99.59 (97.43) 11.25 (2) 0.15 (0.89) Resolution (Å) 45.18 - 2.18 (2.26 - 2.18) 33.18 - 2.65 (2.74 - 2.65) (Last Shell) Rwork/ Rfree Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID 20.35/25.01 19.12/27.34 0.009 0.94 31.76 25 6MOX 0.010 1.01 32.46 12 6MR0 a Values in the parenthesis refer to the last resolution shell. 54 REFERENCES 55 REFERENCES 1. Ernst, O. P.; Lodowski, D. T.; Elstner, M.; Hegemann, P.; Brown, L. S.; Kandori, H., Microbial and Animal Rhodopsins: Structures, Functions, and Molecular Mechanisms. Chemical Reviews 2014, 114 (1), 126-163. Kurihara, M.; Sudo, Y., Microbial rhodopsins: wide distribution, rich diversity and 2. great potential. Biophysics and Physicobiology 2015, 12, 121-9. 3. Schotte, F.; Cho, H. S.; Kaila, V. R. I.; Kamikubo, H.; Dashdorj, N.; Henry, E. R.; Graber, T. J.; Henning, R.; Wulff, M.; Hummer, G.; Kataoka, M.; Anfinrud, P. A., Watching a signaling protein function in real time via 100-ps time-resolved Laue crystallography. Proceedings of the National Academy of Sciences of the United States of America, 2012, 109 (47), 19256-19261. 4. Sudo, Y.; Ihara, K.; Kobayashi, S.; Suzuki, D.; Irieda, H.; Kikukawa, T.; Kandori, H.; Homma, M., A Microbial Rhodopsin with a Unique Retinal Composition Shows Both Sensory Rhodopsin II and Bacteriorhodopsin-like Properties. Journal of Biological Chemistry 2011, 286 (8), 5967-5976. 5. Butt, H. J.; Fendler, K.; Bamberg, E.; Tittor, J.; Oesterhelt, D., Aspartic acids 96 and 85 play a central role in the function of bacteriorhodopsin as a proton pump. The EMBO Journal 1989, 8 (6), 1657-1663. Kandori, H., Ion-pumping microbial rhodopsins. Frontiers in Molecular Biosciences 6. 2015, 2, 52. 7. Gordeliy, V. I.; Labahn, J.; Moukhametzianov, R.; Efremov, R.; Granzin, J.; Schlesinger, R.; Buldt, G.; Savopol, T.; Scheidig, A. J.; Klare, J. P.; Engelhard, M., Molecular basis of transmembrane signalling by sensory rhodopsin II-transducer complex. Nature 2002, 419 (6906), 484-487. Govorunova, E. G.; Koppel, L. A., The road to optogenetics: Microbial rhodopsins. 8. Biochemistry-Moscow, 2016, 81 (9), 928-940. 9. Royant, A.; Nollert, P.; Edman, K.; Neutze, R.; Landau, E. M.; Pebay-Peyroula, E.; Navarro, J., X-ray structure of sensory rhodopsin II at 2.1-Å resolution. Proceedings of the National Academy of Sciences of the United States of America 2001, 98 (18), 10131- 10136. 10. Kandori, H., Protein-Controlled Isomerization in Rhodopsins. In Chemical Science of π-Electron Systems, Akasaka, T.; Osuka, A.; Fukuzumi, S.; Kandori, H.; Aso, Y., Eds. Springer Japan: Tokyo, 2015; pp 695-713. 56 11. Hirai, T.; Subramaniam, S., Protein Conformational Changes the Bacteriorhodopsin Photocycle: Comparison of Findings from Electron and X-Ray Crystallographic Analyses. PLOS ONE 2009, 4 (6), e5769. in 12. Nango, E.; Royant, A.; Kubo, M.; Nakane, T.; Wickstrand, C.; Kimura, T.; Tanaka, T.; Tono, K.; Song, C.; Tanaka, R.; Arima, T.; Yamashita, A.; Kobayashi, J.; Hosaka, T.; Mizohata, E.; Nogly, P.; Sugahara, M.; Nam, D.; Nomura, T.; Shimamura, T.; Im, D.; Fujiwara, T.; Yamanaka, Y.; Jeon, B.; Nishizawa, T.; Oda, K.; Fukuda, M.; Andersson, R.; Båth, P.; Dods, R.; Davidsson, J.; Matsuoka, S.; Kawatake, S.; Murata, M.; Nureki, O.; Owada, S.; Kameshima, T.; Hatsui, T.; Joti, Y.; Schertler, G.; Yabashi, M.; Bondar, A.-N.; Standfuss, J.; Neutze, R.; Iwata, S., A three-dimensional movie of structural changes in bacteriorhodopsin. Science 2016, 354 (6319), 1552-1557. 13. Kato, H. E.; Zhang, F.; Yizhar, O.; Ramakrishnan, C.; Nishizawa, T.; Hirata, K.; Ito, J.; Aita, Y.; Tsukazaki, T.; Hayashi, S.; Hegemann, P.; Maturana, A. D.; Ishitani, R.; Deisseroth, K.; Nureki, O., Crystal structure of the channelrhodopsin light-gated cation channel. Nature 2012, 482 (7385), 369-74. Vogeley, L.; Sineshchekov, O. A.; Trivedi, V. D.; Sasaki, J.; Spudich, J. L.; Luecke, 14. H., Anabaena Sensory Rhodopsin: A Photochromic Color Sensor at 2.0 Å. Science 2004, 306 (5700), 1390-1393. 15. Maresca, J. A.; Keffer, J. L.; Miller, K. J., Biochemical Analysis of Microbial Rhodopsins. Current Protocols in Microbiology 2016, 41, 1F 4 1-1F 4 18. Shibata, M.; Kandori, H., FTIR Studies of Internal Water Molecules in the Schiff 16. Base Region of Bacteriorhodopsin. Biochemistry 2005, 44 (20), 7406-7413. 17. Hatanaka, M.; Kashima, R.; Kandori, H.; Friedman, N.; Sheves, M.; Needleman, R.; Lanyi, J. K.; Maeda, A., Trp86 → Phe Replacement in Bacteriorhodopsin Affects a Water Molecule near Asp85 and Light Adaptation. Biochemistry 1997, 36 (18), 5493- 5498. 18. Otto, H.; Marti, T.; Holz, M.; Mogi, T.; Stern, L. J.; Engel, F.; Khorana, H. G.; Heyn, M. P., Substitution of amino acids Asp-85, Asp-212, and Arg-82 in bacteriorhodopsin affects the proton release phase of the pump and the pKa of the Schiff base. Proceedings of the National Academy of Sciences 1990, 87 (3), 1018-1022. 19. Nogly, P.; Weinert, T.; James, D.; Carbajo, S.; Ozerov, D.; Furrer, A.; Gashi, D.; Borin, V.; Skopintsev, P.; Jaeger, K.; Nass, K.; Båth, P.; Bosman, R.; Koglin, J.; Seaberg, M.; Lane, T.; Kekilli, D.; Brünle, S.; Tanaka, T.; Wu, W.; Milne, C.; White, T.; Barty, A.; Weierstall, U.; Panneels, V.; Nango, E.; Iwata, S.; Hunter, M.; Schapiro, I.; Schertler, G.; Neutze, R.; Standfuss, J., Retinal isomerization in bacteriorhodopsin captured by a femtosecond x-ray laser. Science 2018, 361, eaat0094. 57 Palczewski, K., G., Protein-coupled receptor rhodopsin. Annual review of 20. biochemistry 2006, 75, 743-67. Zhou, X. E.; Melcher, K.; Xu, H. E., Structure and activation of rhodopsin. Acta 21. Pharmacologica Sinica 2012, 33 (3), 291-9. 22. Kiser, P. D.; Golczak, M.; Lodowski, D. T.; Chance, M. R.; Palczewski, K., Crystal structure of native RPE65, the retinoid isomerase of the visual cycle. Proceedings of the National Academy of Sciences of the Unitd States of America 2009, 106 (41), 17325- 17330. Tsukamoto, H.; Terakita, A., Diversity and functional properties of bistable 23. pigments. Photochemical & Photobiological Sciences 2010, 9 (11), 1435-1443. 24. Govorunova, E. G.; Sineshchekov, O. A.; Li, H.; Spudich, J. L., Microbial Rhodopsins: Diversity, Mechanisms, and Optogenetic Applications. Annual Review of Biochemistry 2017, 86 (1), 845-872. 25. Bassolino, G.; Sovdat, T.; Soares Duarte, A.; Lim, J. M.; Schnedermann, C.; Liebel, M.; Odell, B.; Claridge, T. D. W.; Fletcher, S. P.; Kukura, P., Barrierless Photoisomerization of 11-cis Retinal Protonated Schiff Base in Solution. Journal of the American Chemical Society 2015, 137 (39), 12434-12437. Liu, R. S. H.; Asato, A. E., Photochemistry and synthesis of stereoisomers of 26. vitamin-A. Tetrahedron 1984, 40 (11), 1931-1969. Schapiro, I., The origin of bond selectivity and excited-state reactivity in retinal 27. analogues. The Journal of Physical Chemistry A 2016, 120 (19), 3353-3365. 28. Ben-Nun, M.; Molnar, F.; Schulten, K.; Martínez, T. J., The role of intersection topography in bond selectivity of cis-trans photoisomerization. Proceedings of the National Academy of Sciences of the United States of America, 2002, 99 (4), 1769-1773. 29. Jang, G. F.; Kuksa, V.; Filipek, S.; Bartl, F.; Ritter, E.; Gelb, M. H.; Hofmann, K. P.; Palczewski, K., Mechanism of rhodopsin activation as examined with ring-constrained retinal analogs and the crystal structure of the ground state protein. The journal of Biological Chemistry 2001, 276 (28), 26148-53. 30. Mogi, T.; Marti, T.; Khorana, H. G., Structure-function studies on bacteriorhodopsin. tryptophan residues affect protein-retinal interactions in bacteriorhodopsin. The Journal of Biological Chemistry 1989, 264 (24), 14197-201. IX. Substitutions of Sharma, A. K.; Spudich, J. L.; Doolittle, W. F., Microbial rhodopsins: functional 31. versatility and genetic mobility. Trends in Microbiology 2006, 14 (11), 463-469. 58 32. Nango, E.; Royant, A.; Kubo, M.; Nakane, T.; Wickstrand, C.; Kimura, T.; Tanaka, T.; Tono, K.; Song, C.; Tanaka, R.; Arima, T.; Yamashita, A.; Kobayashi, J.; Hosaka, T.; Mizohata, E.; Nogly, P.; Sugahara, M.; Nam, D.; Nomura, T.; Shimamura, T.; Im, D.; Fujiwara, T.; Yamanaka, Y.; Jeon, B.; Nishizawa, T.; Oda, K.; Fukuda, M.; Andersson, R.; Bath, P.; Dods, R.; Davidsson, J.; Matsuoka, S.; Kawatake, S.; Murata, M.; Nureki, O.; Owada, S.; Kameshima, T.; Hatsui, T.; Joti, Y.; Schertler, G.; Yabashi, M.; Bondar, A. N.; Standfuss, J.; Neutze, R.; Iwata, S., A three-dimensional movie of structural changes in bacteriorhodopsin. Science 2016, 354 (6319), 1552-1557. 33. Duñach, M.; Marti, T.; Khorana, H. G.; Rothschild, K. J., Uv-visible spectroscopy of bacteriorhodopsin mutants: substitution of Arg-82, Asp-85, Tyr-185, and Asp-212 results in abnormal light-dark adaptation. Proceedings of the National Academy of Sciences of the United States of America 1990, 87 (24), 9873-9877. 34. Otto, H.; Marti, T.; Holz, M.; Mogi, T.; Stern, L. J.; Engel, F.; Khorana, H. G.; Heyn, M. P., Substitution of amino acids Asp-85, Asp-212, and Arg-82 in bacteriorhodopsin affects the proton release phase of the pump and the pKa of the Schiff base. Proceedings of the National Academy of Sciences of the United States of America 1990, 87 (3), 1018- 1022. Kandori, H., Role of internal water molecules in bacteriorhodopsin. Biochimica et 35. Biophysica Acta 2000, 1460 (1), 177-91. 36. Nogly, P.; Weinert, T.; James, D.; Carbajo, S.; Ozerov, D.; Furrer, A.; Gashi, D.; Borin, V.; Skopintsev, P.; Jaeger, K.; Nass, K.; Båth, P.; Bosman, R.; Koglin, J.; Seaberg, M.; Lane, T.; Kekilli, D.; Brünle, S.; Tanaka, T.; Wu, W.; Milne, C.; White, T.; Barty, A.; Weierstall, U.; Panneels, V.; Nango, E.; Iwata, S.; Hunter, M.; Schapiro, I.; Schertler, G.; Neutze, R.; Standfuss, J., Retinal isomerization in bacteriorhodopsin captured by a femtosecond x-ray laser. Science 2018, eaat0094. 37. Govorunova, E. G.; Sineshchekov, O. A.; Janz, R.; Liu, X.; Spudich, J. L., Natural light-gated anion channels: A family of microbial rhodopsins for advanced optogenetics. Science 2015, 349 (6248), 647-650. 38. Kouyama, T.; Nishikawa, T.; Tokuhisa, T.; Okumura, H., Crystal structure of the L intermediate of bacteriorhodopsin: evidence for vertical translocation of a water molecule during the proton pumping cycle. Journal of Molecular Biology 2004, 335 (2), 531-546. 39. W Stoeckenius, a.; Bogomolni, R. A., Bacteriorhodopsin and related pigments of halobacteria. Annual Review of Biochemistry 1982, 51 (1), 587-616. 40. Wagner, N. L.; Greco, J. A.; Ranaghan, M. J.; Birge, R. R., Directed evolution of bacteriorhodopsin for applications in bioelectronics. Journal of the Royal Society, Interface 2013, 10 (84), 20130197. 41. Schobert, B.; Cupp-Vickery, J.; Hornak, V.; Smith, S. O.; Lanyi, J. K., crystallographic structure of the K intermediate of bacteriorhodopsin: Conservation of free 59 energy after photoisomerization of the retinal. Journal of Molecular Biology 2002, 321 (4), 715-726. 42. Sass, H. J.; Buldt, G.; Gessenich, R.; Hehn, D.; Neff, D.; Schlesinger, R.; Berendzen, J.; Ormos, P., Structural alterations for proton translocation in the M state of wild-type bacteriorhodopsin. Nature 2000, 406 (6796), 649-53. Fodor, S. P. A.; Ames, J. B.; Gebhard, R.; Van den Berg, E. M. M.; Stoeckenius, 43. W.; Lugtenburg, J.; Mathies, R. A., Chromophore structure in bacteriorhodopsin's N intermediate: implications for the proton-pumping mechanism. Biochemistry 1988, 27 (18), 7097-7101. 44. Wang, T.; Sessions, A. O.; Lunde, C. S.; Rouhani, S.; Glaeser, R. M.; Duan, Y.; Facciotti, M. T., Deprotonation of D96 in bacteriorhodopsin opens the proton uptake pathway. Structure 2013, 21 (2), 290-7. 45. Subramaniam, S.; Henderson, R., Crystallographic analysis of protein conformational changes in the bacteriorhodopsin photocycle. Biochimica et biophysica acta (BBA) - Bioenergetics 2000, 1460 (1), 157-165. 46. Luecke, H.; Schobert, B.; Richter, H.-T.; Cartailler, J.-P.; Lanyi, J. K., Structure of bacteriorhodopsin at 1.55 Å resolution, Edited by D. C. Rees. Journal of molecular biology 1999, 291 (4), 899-911. 47. Wickstrand, C.; Dods, R.; Royant, A.; Neutze, R., Bacteriorhodopsin: Would the real structural intermediates please stand up? Biochimica et Biophysica Acta 2015, 1850 (3), 536-53. 48. Ghanbarpour, A.; Nairat, M.; Nosrati, M.; Santos, E. M.; Vasileiou, C.; Dantus, M.; Borhan, B.; Geiger, J. H., Mimicking Microbial Rhodopsin Isomerization in a Single Crystal. Journal of the American Chemical Society 2019, 141 (4), 1735-1741. 49. Nosrati, M.; Berbasova, T.; Vasileiou, C.; Borhan, B.; Geiger, J. H., A Photoisomerizing Rhodopsin Mimic Observed at Atomic Resolution. Journal of the American Chemical Society 2016, 138 (28), 8802-8808. Berbasova, T.; Santos, E. M.; Nosrati, M.; Vasileiou, C.; Geiger, J. H.; Borhan, B., 50. Light-Activated Reversible Imine Isomerization: Towards a Photochromic Protein Switch. Chembiochem 2016, 17 (5), 407-414. 51. Yapici, I.; Lee, K. S. S.; Berbasova, T.; Nosrati, M.; Jia, X.; Vasileiou, C.; Wang, W.; Santos, E. M.; Geiger, J. H.; Borhan, B., “Turn-On” Protein Fluorescence: In Situ Formation of Cyanine Dyes. Journal of the American Chemical Society 2015, 137 (3), 1073-1080. 60 52. Wang, W.; Geiger, J. H.; Borhan, B., The photochemical determinants of color vision. Bioessays 2014, 36 (1), 65-74. 53. Berbasova, T.; Nosrati, M.; Vasileiou, C.; Wang, W.; Lee, K. S. S.; Yapici, I.; Geiger, J. H.; Borhan, B., Rational Design of a Colorimetric pH Sensor from a Soluble Retinoic Acid Chaperone. Journal of the American Chemical Society 2013, 135 (43), 16111-16119. 54. Lee, K. S. S.; Berbasova, T.; Vasileiou, C.; Jia, X.; Wang, W.; Choi, Y.; Nossoni, F.; Geiger, J. H.; Borhan, B., Probing Wavelength Regulation with an Engineered Rhodopsin Mimic and a C15-Retinal Analogue. Chempluschem 2012, 77 (4), 273-276. 55. Vasileiou, C.; Lee, K. S. S.; Crist, R. M.; Vaezeslami, S.; Goins, S. M.; Geiger, J. H.; Borhan, B., Dissection of the critical binding determinants of cellular retinoic acid binding protein II by mutagenesis and fluorescence binding assay. Proteins: Structure, Function, and Bioinformatics 2009, 76 (2), 281-290. 56. Vaezeslami, S.; Mathes, E.; Vasileiou, C.; Borhan, B.; Geiger, J. H., The Structure of Apo-wild-type Cellular Retinoic Acid Binding Protein II at 1.4 Å and its Relationship to Ligand Binding and Nuclear Translocation. Journal of Molecular Biology 2006, 363 (3), 687-701. 57. Crist, R. M.; Vasileiou, C.; Rabago-Smith, M.; Geiger, J. H.; Borhan, B., Engineering a Rhodopsin Protein Mimic. Journal of the American Chemical Society, 2006, 128 (14), 4522-4523. 58. Assar, Z.; Nossoni, Z.; Wang, W.; Santos, Elizabeth M.; Kramer, K.; McCornack, C.; Vasileiou, C.; Borhan, B.; Geiger, James H., Domain-Swapped Dimers of Intracellular Lipid-Binding Proteins: Evidence for Ordered Folding Intermediates. Structure, 2016, 24, 1590-1598. 59. Wang, W.; Nossoni, Z.; Berbasova, T.; Watson, C. T.; Yapici, I., Tuning the electronic absorption of protein-embedded all-trans-retinal. Science (American Association for the Advancement of Science) 338 (6112), 1340-1343. 60. Luecke, H.; Schobert, B.; Cartailler, J.-P.; Richter, H.-T.; Rosengarth, A.; Needleman, R.; Lanyi, J. K., Coupling photoisomerization of retinal to directional transport in bacteriorhodopsin11Edited by D. C. Rees. Journal of Molecular Biology 2000, 300 (5), 1237-1255. 61. Bolze, C. S.; Helbling, R. E.; Owen, R. L.; Pearson, A. R.; Pompidor, G.; Dworkowski, F.; Fuchs, M. R.; Furrer, J.; Golczak, M.; Palczewski, K.; Cascella, M.; Stocker, A., Human Cellular Retinaldehyde-Binding Protein Has Secondary Thermal 9- cis-Retinal Isomerase Activity. Journal of the American Chemical Society 2014, 136 (1), 137-146. 61 62. Rothschild, K. J.; Gray, D.; Mogi, T.; Marti, T.; Braiman, M. S.; Stern, L. J.; Khorana, H. G., Vibrational spectroscopy of bacteriorhodopsin mutants: Chromophore isomerization perturbs tryptophan-86. Biochemistry 1989, 28 (17), 7052-7059. 63. Needleman, R.; Chang, M.; Ni, B.; Váró, G.; Fornés, J.; White, S. H.; Lanyi, J. K., Properties of Asp212----Asn bacteriorhodopsin suggest that Asp212 and Asp85 both participate in a counterion and proton acceptor complex near the Schiff base. Journal of Biological Chemistry 1991, 266 (18), 11478-11484. 64. Nosrati, M.; Berbasova, T.; Vasileiou, C.; Borhan, B.; Geiger, J. H., A Photoisomerizing Rhodopsin Mimic Observed at Atomic Resolution. Journal of the American Chemical Society 2016, 138 (28), 8802-8. 65. Maeda, T.; Van Hooser, J. P.; Driessen, C. A.; Filipek, S.; Janssen, J. J.; Palczewski, K., Evaluation of the role of the retinal G protein-coupled receptor (RGR) in the vertebrate retina in vivo. Journal of Neurochemistry 2003, 85 (4), 944-56. 66. Gill, S. C.; von Hippel, P. H., Calculation of protein extinction coefficients from amino acid sequence data. Analytical Biochemistry 1989, 182 (2), 319-26. 67. Otwinowski, Z.; Minor, W., [20] Processing of X-ray diffraction data collected in oscillation mode. Methods in Enzymology 1997, 276, 307-326. 68. Murshudov, G. N.; Vagin, A. A.; Dodson, E. J., Refinement of Macromolecular Structures by the Maximum-Likelihood Method. Acta Crystallographica Section D 1997, 53 (3), 240-255. 69. Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis, I. W.; Echols, N.; Headd, J. J.; Hung, L.-W.; Kapral, G. J.; Grosse-Kunstleve, R. W.; McCoy, A. J.; Moriarty, N. W.; Oeffner, R.; Read, R. J.; Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H., PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallographica Section D 2010, 66 (2), 213-221. Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K., Features and development of 70. Coot. Acta Crystallographica Section D-Biological Crystallography 2010, 66, 486-501. 62 CHAPTER II: CREATING A NEW CLASS OF PROTEIN CONFORMATIONAL SWITCHES II-1. Protein conformational switches Understanding the concept of a protein conformational switch is woven tightly with appreciating the meaning of allostery, “the second secret of life”.1 Classically, allostery is defined as transferring the effect of binding on one site to a remote site from where the binding ensues, accompanied by a concomitant change in the function.1-2 a. Protein conformational switches can be broadly defined as a family of allosteric b. c. Figure II-1. Natural protein conformational switches. a. The overlay of apo (cyan) and holo (pink) maltose binding protein. b. The crystal structure of calcium-bound conformation of calmodulin. c. The GPCR transmembrane crystal structure. proteins that undergo structural rearrangement upon detecting input signals, such as small molecules, proteins, light, hormones, changes in pH, temperature, the reduction potential of the environment, etc.3 Protein conformational switches are ubiquitous in biology (Figure II-1) and play essential roles in almost all crucial biological processes including translation, transcription, signaling, cellular transport, enzymatic activity and 63 others.3-8 This means the presence of these proteins is vital for organisms to respond properly to the transitional environment in which they live.1 The ability to control the conformational readouts of a system makes these proteins highly interesting for various protein-engineering applications such as sensing and functional regulation.7, 9-19 For example, these proteins have been used in the field of synthetic circuits where the arrangement of highly tunable and well-characterized modules can be managed to execute different complicated programs such as creating cellular memory, conducting cellular behavior, and converting analog to digital, wherein the continuous chemical gradient is converted into a digital output. 20-21 370 or 440 nm 440 or 480 nm CaM M13 BFP or CFP -Ca2+ +4 Ca2+ GFP or YFP 370 or 440 nm FRET CaM BFP or CFP GFP or YFP 510 or 535 nm +4 Ca2+ Figure II-2. Schematic picture demonstrating the FRET between GFPs can measure Ca2+. The GFPs are drawn as cylinders. Top: calmodulin (CaM) without Ca2+ in the elongated conformation, disordered unbound M13 (calcium binding peptide) Bottom: The Ca2+-bound calmodulin in the closed conformation and Ca2+/M13 complex in the structured conformation. 64 Furthermore, these proteins have been employed as a sensor to identify and respond to certain signals. For instance, in a research study by Miyawaki et al 22 GFP and YFP were linked to the c- and n- termini of calmodulin. Upon sensing the calcium, calmodulin undergoes the conformational change leading to Förster resonance energy transfer (FRET), a distance-dependent fluorescence (Figure II-2). Therefore, the calcium concentration inside the cell can be a measurable readout throughout the change in FRET. The developments of these types of protein conformational switches leads to significant progress in developing novel calcium sensors, featuring the low-concentration detection of calcium release during the potential firing in neurons.9, 23 Another application of this strategy is to create novel affinity clamps by connecting two proteins through a flexible linker, which generates a new complex with a novel Fn3 High FRET YPet CyPet + PDZ peptide PDZ Fn3 e d i t p e p Z D P PDZ YPet Low FRET CyPet Figure II-3. A circularly-permutated PDZ domain is linked to a fibronectin type III (FN3)-based enhancer domain that detects the ligand-bound form of the PDZ domain. A significant conformational change ensues upon PDZ ligand binding with the concomitant formation of a sandwich complex that can be measured through the FRET between the two fluorescent proteins YPet and CyPet. conformation upon sensing a ligand of interest. In this case, the signal can be monitored through intermolecular FRET activated through a large conformational change occurring from an open state to a closed state. Connecting a circularly-permutated PDZ to 65 fibronectin through a flexible linker that detects the ligand-bound form of the PDZ domain is an example of this approach (Figure II-3).24 Developing a maltose sensor by inserting the circularly-permutated GFP inside the domain of the maltose binding protein is another example of this strategy being used. In this case, upon sensing the maltose, a large protein conformational change eventuates, pulling on the circularly-permutated GFP and changing the fluorescence activity (Figure II-4).3, 25 In addition to a large conformational change occurring from well-structured protein domains, there are other examples where intrinsically disordered protein fragments have been employed. The sensing of a specific ligand causes the partial unfolding/folding Figure II-4. Crystal structure of circularly-permutated GFP (purple) inserted inside the maltose-binding protein (cyan). PDB code: 3OSQ. states in the intrinsically disordered protein accompanied by a large change in the conformation, which is detectable using fluorescent proteins.26-27 Synthetic protein conformational switches can be designed through alternative frame folding. 28-29 The ligand-binding site of the protein is replicated and linked to the opposite side of the protein to create a continuous polypeptide where the ligand binding causes reequilibration between native and circularly-permutated conformations. Eliciting a thrombin sensor that, 66 upon activation by light, turns the fluorescent output from green to yellow, is a successful example of this design approach.30 Protein conformational switches have been utilized as functional regulators to control and manage different cellular events. In most cases, the protein of interest is inserted inside the domain of the protein conformational switch, and the change in the conformation of the protein conformational switch leads to modifying the protein of interest’s activity. This approach has been exercised extensively in eliciting enzyme-based protein switches with the ability to enlarge the signal transduction inside the cell, where the ability and compatibility of fluorescent-based sensors to identify and respond to certain cellular events are restrained.3, 15, 31 The creation of maltose-based β-lactamase with a large number of response times,30 Figure II-5, protease-based proximity sensors of protein-protein interactions,32-33 and allosteric protease receptors 34 are some examples where the domain-insertion strategy has been practiced. a. b. + M Maltose BLA MBP M MBP BLA β-lactamase Figure II-5. Inserting maltose-binding protein (MBP) into a circularly permutated version of β- lactamase (BLA) giving rise to creation of a maltose sensor which regulates the activity of β-lactamase (BLA) as a function of maltose binding. 67 The idea of controlling function by light also led to creating new light-sensitive protein switches enabling spatio and temporal resolution in live cell imaging and fast response times in optogenetic applications.18, 35-36 For the most part, this strategy has been performed using LOV domains 37 and BLUF domains where the blue light drives the flavin-based chromophore to form a covalent bond with an active cysteine inside the protein (Figure II-6) causing the unfolding of the c-terminal helix.38 These proteins have been applied to create new classes of optogenetic molecular tools to turn on transcription factors and regulate neural cell circuitry. Wang et al demonstrated the hallmark of these studies, where a light-and-calcium-gated transcription factor (TF) system called FLARE (Fast Light and Activity-Regulated Expression) has been developed by reengineering the LOV domain, TEV protease, and calmodulin binding peptide. In this system, the TF, which is located in the plasma membrane, cannot perform the gene transcription of a reporter gene situated in the nucleus.36 However, upon blue light treatment and in the presence a. b. HO OH OH OH N O O NH (LOV) N N SH CH2 Cys 450 nm Light Dark, UV HO OH OH OH N O S CH2 N N H O NH Cys (LOV) 390 nm Figure II-6. a. The crystal structure of LOV domain in dark. PDB code: 3T50. b. LOV domain photocycle is detected by a ground state oxidize flavin generating a flavin-cysteine adduct upon blue- light irradiation. The adduct reset to the ground state through UV irradiation or dark adaptation. of high calcium concentration, the TF is cleaved from the membrane and gets translocated to the nucleolus where it leads to the transcription of a reporter gene. This 68 strategy provides a powerful tool for optogenetics, where rhodopsin proteins 39 and azobenzene-based photoswitches are the only other alternatives.40 As mentioned above, natural protein conformational switches have been amply operated to create new molecular tools in synthetic biology; 3, 9, 11, 15, 29-31 however, there are few examples where a new protein conformational switch has been developed. Producing an allosteric site in an antibody using a computational- approach 41 and the de novo design of an allosteric metalloprotein through the mechanical strain of disulfide bonds42 are some of the current efforts towards creating new allosteric behavior. So far, most attempts to create allosteric regulation centered on biophysical models comparing the two states — the presence and absence of the input signals — both concentrating on understanding the thermodynamic impetus of this phenomena. There is a paucity of information for grasping the molecular basis of allostery. Essentially, the connectivity in the protein scaffold leading to allosteric responses on a molecular level is poorly understood.43-44 The goal of this chapter is to create allostery and elucidate the mechanism of that through atomic resolution crystal structures and rational mutagenesis using the domain-swapped (DS) dimer of human cellular retinol binding protein II (hCRBPII) as an engineering template. This study accentuates the design elements of contriving novel allosteric proteins based on high-resolution structural interrogation and evaluation in a protein scaffold that has not been evolved to work as such. II-2. Using human cellular retinol binding protein II as an engineering scaffold Our strategy to create a novel protein conformational switch is focused on using small cytosolic proteins belonging to intracellular Lipid Binding Protein (iLBP).45,46 This protein family, as mentioned in the previous section, is responsible for trafficking 69 hydrophobic ligands inside the cells. hCRBPII is responsible for carrying retinal and retinol within the cell.47 These proteins structurally contain ten β-sheets and two α- helices embellishing the ligand binding site (Figure II-7). Very recently, Assar et al 48 identified several key mutations in this protein that give rise to a substantial amount of domain- swapped dimer. Furthermore, they elucidated the mechanism giving rise to DS dimer formation. A thorough definition and examples of domain-swapping are covered in a. b. Figure II-7. a. The monomeric form of hCRBPII bound with retinol (shown with cyan). The two alfa-helices decorating the ligand binding site were highlighted with green. (PDB code: 4QZT) b. The domain-swapped dimer of hCRBPII. Each chain is color coded (PDB code: 4ZH9). Chapter III. Generally, domain swapping can be defined as a mechanism where two identical protein segments exchange part of their structures to generate dimer or higher- order oligomers. In our previous work,48 we elaborated that a single mutation in hCRBPII leads to the breakage of the hydrogen bond between Y60 and E72, which then leads to destabilizing the closed monomer formation, increasing the lifetime of open-monomer intermediates, thereby increasing the population of the DS dimer of hCRBPII. Dr .Zahra Assar further confirmed that the WT hCRBPII can lead to negligible dimer formation, and fortuitously, she obtained the crystal structure of this protein as well.48 Other than hCRBPII, Fatty Acid Binding Protein 5, another protein in this family that is a carrier for long-chain fatty acids and other endocannabinoids, has been recognized to generate a 70 domain-swapped dimer.49 In this particular case, the crystal structures revealed different Table II-1. Summary of apo and holo domain-swapped dimer crystal structures for several hCRBPII mutants. Symm./ Asymm. Space group Mutants No. of Molecules Q108K:T51D Q108K:K40L: T51F Q108K:K40L: T51W Q108K:T51D:A28H Q108K:T51D* Q108K:K40L: T51W Q108K:K40L: T51F Q108K:K40L:T51F:Y60A WT-hCRBPII Q108K:T51D:A28C Q108K:T51D:A28C Q108K:T51D:A28H Q108K:K40L:T51H Q108K:K40L:T51N Holo/ Apo Apo Apo Apo Apo Holo Holo Holo Holo Holo Apo Holo Holo Holo Holo P21212 P21212 P21212 P21212 P212121 P21 P21 P21212 P212121 P212121 P212121 P212121 P212121 P212121 1 1 1 1 12 12 12 1 12 2 4 12 12 12 Symmetric Symmetric Symmetric Symmetric Asymmetric Asymmetric Asymmetric Symmetric Asymmetric Asymmetric Asymmetric Asymmetric Asymmetric Asymmetric *The crystal structure of the holo Q108K:T51D domain-swapped dimer was obtained by Dr. Nosrati. protein conformations upon binding with two different ligands.49 It has been shown that domain-swapping can sometimes alter the original functions of the protein by increasing the concentration of active/binding sites, generating binding sites in the interface of swapped units, and creating the possibility of allosteric regulation.50 Having a large number of hCRBPII variants on hand, we decided to explore the possibility of creating allostery in hCRBPII mutants. 71 II-2-1. Crystallization of apo and holo retinal-bound variants of hCRBPII In our previous work 48 two positions, Tyr60 residing in the hinge-loop region, and Glu72 which directly interacts with Tyr60, were found to promote DS dimerization when mutated. Two structural forms, symmetric and asymmetric DS dimers, of hCRBPII, were a. c. e. b. d. f. Figure II-8. Overlay of WT-hCRBPII (holo) DS dimer chains H and D shown by dark blue and WT-hCRBPII (apo) (PDB code: 4ZH9) shown by green. b. Overlay of WT-hCRBPII (holo) chains F and J shown by yellow and WT-hCRBPII (apo) shown by green. c. Overlay of WT-hCRBPII (holo) chains B and L shown by red and WT-hCRBPII (apo) shown by green. d. Overlay of WT-hCRBPII (holo) chains E and L shown by orange and WT-hCRBPII (apo) shown by green. e. Overlay of WT-hCRBPII (holo) chains A and K shown by pink and WT-hCRBPII (apo) shown by green. f. Overlay of WT-hCRBPII (holo) chains E and I is shown by brown and WT-hCRBPII (apo) shown by green. 72 observed, depending on the amino acid at position 60. While the wild-type, Y60L and E72A variants produced almost identical symmetric dimers, both the Q108K:K40L:Y60W and Y60W variants produced almost identical asymmetric dimers. Disappointingly however, we were unable to obtain any ligand-bound structures of these mutants. Dr. Nosrati and Dr. Santos identified additional mutational “hot spots” outside the hinge region that predominantly fold as domain-swapped dimers. We have now crystallized and determined the structures of five retinylidene PSB-bound variants, all of which produce mostly DS dimers, along with four of their apo counterparts. Furthermore, the crystal structure of the retinal-bound WT hCRBPII holo dimer has been determined (Figure II-8). The apo structures demonstrated a symmetric dimer as previously described 48 (Figure II-9a), though some differed in space group and crystal packing (Table 1). Together, the data indicates that the symmetric fold is the most common DS dimer structure. These results further indicate that the domain-swapped dimer has a dominant and robust structure in the absence of any perturbation, regardless of crystal packing. 73 a. b. 16 Å d. Y60-Holo N59-Apo Y60A 2.5 Å N59-Holo Y60-Apo c. E72 Figure II-9. a. Overlay of one chain each of four apo dimers: WT (cyan), Q108K:T51D (purple- blue), Q108K:K40L:T51F (yellow), and Q108K:K40L:T51W (pink), showing a relatively rigid conformation for apo dimers. b. Overlay of holo (purple) and apo (yellow) Q108K:T51D structures showing a large change in the relative orientations of the two domains upon ligand binding, leading to a 16 Å movement (inset) in the α-helix. Retinal (green) is shown in both sphere and stick representations. c. The retinal (green) binding pocket of apo (yellow C atoms) and retinal-bound (blue C atoms) Q108K:T51D. Allosteric conformational change is driven by the orientation of the Tyr60 and Asn59 sidechains. A flipped-in Asn59 would sterically clash with the bound retinylidene, leading to the flipped-out conformation of Asn59 and flipped-in conformation of Tyr60. All other atoms colored by type throughout: nitrogen (blue), oxygen (red). d. Overlay of apo Q108K:K40L:T51F (purple) and holo Q108K:K40L:T51F:Y60A (yellow) showing no change in the relative orientation of the two domains upon ligand binding. In contrast, the structures of the retinylidene-bound dimers reveal that none lead to the symmetric structure cited above. In fact, the retinal-bound holo displayed a different relative orientation of the two domains in comparison with the apo symmetric dimers 74 (Figure 11-9b). Remarkably, most of the holo DS dimer crystal structures depicted four to twelve molecules in the asymmetric unit in contrast to the apo symmetric dimer that mostly crystallized with one molecule in the asymmetric unit, Table II-1. However, the similarity between the relative orientations of the two dimer domains is much closer in the holo structures than the apo symmetric dimers (Figure II-10). The distance between the main-chain nitrogen atoms of Arg30, located in the ligand portal region, is much larger in the overlaid holo/apo structures of Q108K:K40L:T51F than in a. b. Arg30-apo 11 Å Arg30-holo Arg30’-holo Arg30-holo ~1 Å close-up view close-up view Figure II-10. Retinylidene-bound hCRBPII undergo dramatic conformational change upon ligand binding. a. An overlay of apo Q108K:K40L:T51F (KLF) (purple) and holo KLF (chains A and C, both shown in pink), showing the large motion of helix 1 upon ligand binding. b. An overlay of holo KLF (pink) and holo Q108K:T51D (cyan), showing the modest conformational difference in the two holo structures. The relative distance between Cα of Arg 30 in the overlaid holo/apo structures is 11.3 Å while the same distance in the overlaid holo structures is 1 Å. *Bound retinal molecules in holo structures are shown as transparent spherical models. The complete dimer in the apo structures was generated by crystallographic two-fold symmetry operation. 75 the holo/holo Q108K:K40L:T51F and Q108K:T51D structures (~11 Å in apo/holo vs. ~1.0 Å in holo/holo). II-2-2. Elucidating the mechanism of the protein switch The major cause of this large relative twist in the two domains upon ligand binding involves the dispensation of Asn59 and Tyr60 (Figure II-9). In virtually all of the apo- N59-dimer N59-monomer Figure II-11. Overlay of WT apo monomer (gray, PDB code: 2RCQ) and WT apo hCRBPII dimer (purple, PDB code: 4ZH9). The “flipped in” Asn59 conformation is required for the relative orientation of the two halves of an open monomer to adopt the orientation required to form the symmetric DS dimer. symmetric dimers, the Asn59 side-chain is positioned inside the binding cavity while Tyr60 follows the anticipated β-strand register and is directed outside of the binding cavity. However, in the hCRBPII monomer (with more than 40 structures determined so far), Asn59 and Tyr60 are directed outside and inside of the binding pocket, respectively (Figure II-11). Presumably, the flipped-in Asn59 conformation facilitates the relative orientation of the two halves of an open monomer required to form the symmetric domain- swapped dimer during folding (Figure II-11).48 Conversely, the Asn59 side-chain is flipped out of the binding cavity in the seven holo dimer structures, similar to the monomer. The cause of this is clearly seen in Figure II-9, where an overlay of the Q108K:T51D holo and apo 76 a. b. Q108K N59 II-12. overlay (apo) (purple) a. The of Q108K:K40L:T51F Figure and Q108K:K40L:T51F:Y60A (holo) (yellow) shows no change in the relative orientation of two domains upon ligand binding. b. The retinal bound Q108K:K40L:T51F:Y60A DS dimer. The electron density is observed only for the first 5 carbons of retinal illustrating that Y60A mutation makes the chromophore disordered inside of the binding pocket. Therefore, Asn59 keeps its flipped-in conformation which does not lead to the change in the relative orientation of the two domains. *Bound retinal molecules in the holo structure are shown as transparent spherical models. The complete dimer in the symmetric DS dimers was generated by crystallographic two-fold symmetry operation. dimers shows that the flipped-in Asn59 sterically impinges on the bound retinylidene. As a result, retinal binding sterically forces the Asn59 side-chain out of the binding cavity, leading to the large change in the relative orientation of the two domains. The flipped-out Asn59 in holo domain-swapped dimers also correlates with the Tyr60 conformation, in which the flipped-out Asn59 causes a flipped-in conformation for Tyr60, resulting in a low- barrier hydrogen bond between the side-chains of Tyr60 and Glu72, which influences the trajectory of the bound retinylidene (Figure II-9). Therefore, the interactions between the ligand and both Tyr60 and Asn59 act as the “allosteric spring” propelling the large conformational change of the two domains upon ligand binding. This same structural change is observed for the WT-hCRBPII domain-swapped dimer bound non-covalently with retinal (Figure II-8). The hypothesis that the allosteric conformational change between apo and holo dimers is driven solely by the orientation of the Tyr60 and Asn59 side-chains naturally gives rise to the prediction that the “allosteric spring” could be broken by the mutation of either of these residues. 77 b. d. a. c. e. Figure II-13. The difference between phi/psi angles of holo and apo DS dimers and WT hCRBPII monomer phi/psi angles. For holo structures, one of the chains from each mutant has been considered. Phi angle differences are shown in blue and psi angle differences are shown in red. a. Tyr 60. b. Asn59. c. Arg58. d. Phe57. e. Thr60. As shown from a to d, a large difference in the phi/psi angles between holo and apo structures is observed. The latter hypothesis was tested by mutating Tyr60 to Ala, resulting in the 78 Q108K:K40L:T51F:Y60A hCRBPII tetramutant. As predicted the retinylidene-bound holo dimer structure is very similar to the apo-symmetric dimers (Figure II-9d). In the other holo dimers, the flipped-in conformation of Tyr60 stabilizes the β-ionone ring and places it in a defined trajectory that sterically clashes with Asn59, resulting in its flipped-out conformation. However, the removal of this interaction (Y60A mutation) leads to ligand disorder, demonstrated by the loss of electron density for all but the four carbons closest to the imine bond (Figure II-12). This allows Asn59 to maintain its flipped-in conformation, resulting in a similar orientation for the two domains as compared to apo dimers, even with the ligand bound (Figure II-12). The observation above supports the central role played by Tyr60 and Asn59 in the mechanism for allosteric conformational change seen in hCRBPII domain-swapped dimers and further illustrates the symmetric apo dimer conformation as the default structure for the dimer in the absence of any perturbation. The rotation of the phi and psi angles of residues 58 to 60 is primarily responsible for both flipping Asn59 from inside to outside of the binding pocket and significantly changing the relative orientation of the two dimer domains (Figures II-9 and II-13). This illustrates the almost mechanical nature of the structural changes seen in the dimer. 79 II-3. Creating a disulfide bond as a trigger for allosteric behavior With the structural and mechanical details regulating the conformational change well understood, we then pursued strategies for enlarging the library of conformational readouts available to the system. As shown in Figure II-14, the relative orientations of two a. c. 37#KD# 20#KD# 15#KD# 1" 2" b. A28C’ A28C N59-KD Y60-KD-A28C Y60-KD N59-KD-A28C A28C’ A28C Figure II-14. Engineering a cross-link in the interface of two domains by A28C mutation: a. Overlay of Q108K:T51D-Apo (cyan) and Q108K:T51D:A28C-Apo (blue) showing the new conformation enforced by disulfide bond formation. b. A close-up showing the disulfide bond and the flipped-out conformation for Asn59 and flipped-in conformation for Tyr60. Inset: The disulfide bond electron density contoured at 1σ. c) SDS PAGE gel electrophoresis under reducing (1) and non-reducing (2) conditions, confirming disulfide bond formation in solution. α- helices in the holo versus apo domain-swapped dimers are proximal and undergo a large twist such that the introduction of a disulfide bond between them will necessarily alter the conformational readout of ligand binding, with the Ala28 position particularly well- positioned for the linkage. Our prediction is experimentally verified, as the Q108K:T51D:A28C variant yields a DS dimer with a disulfide bond between the Cys28 residues of the two chains. This was shown by both reducing and non-reducing SDS 80 PAGE gel electrophoresis and crystal structure analysis (Figure II-14). The inter-subunit linkage causes the apo structure of Q108K:T51D:A28C to adopt a relative orientation of domains different from apo Q108K:T51D (Figure II-14a), resulting in a flipped-out conformation for Asn59 and a flipped-in conformation for Tyr60 in the Q108K:T51D:A28C variant. In contrast, the retinal-bound structure of this mutant revealed yet another unique and distinguishable conformation. This demonstrates the potential use of the DS dimer as a conformational molecular switch, where the switching can be tuned to give distinct conformational outcomes. The overlay of the apo and holo Q108K:T51D:A28C structures clearly shows the driving force behind this new conformational change to be the steric interactions between residues located in the α-helices and retinal that drive a conformational change in the opposite direction relative to the holo domain-swapped 4.4 Å Figure II-15. An overlay of apo (orange) and holo Q108K:T51D:A28C (yellow), both featuring the Cys28 disulfide bond. On the right side, the bound retinal sterically forces the α-helices to move, inducing the ligand-induced conformational change. dimers. Additionally, the retinal dissociation constant for Q108K:T51D and Q108K:T51D:A28C was calculated using a fluorescent quenching assay showing that the 81 disulfide bond has a minor effect on the retinal dissociation constant (Figure II-16). Collectively, four unique conformations are identified: apo and holo DS dimers and apo and holo disulfide-bonded DS dimers, with each revealing a distinct conformation (Figure II-17). Furthermore, we employed CD spectroscopy as a convenient tool to explore the conformational readouts of our system in solution (Figure II-18). In this assay, retinal binding causes an inc rease in the CD at 218 nm in the dimer (Q108K:T51D), whereas II-16. Fluorescent quenching assay Figure for Q108K:T51D shows dissociation constants of 220 nM and 368 nM respectively for retinal. and Q108K:T51D:A28C the monomer (Q108K:K40L:T51V) and Y60 mutated dimer (Q108K:K40L:T51F:Y60A) showed minor changes upon the addition of retinal. Furthermore, a decrease in CD absorption at 218 nm was observed for the disulfide-bonded dimer (Q108K:T51D:A28C) upon adding the reducing agent (1mM DTT). The additon of retinal after reducing the disufide bond causes an increase in CD absorption at 218 nm correlated with Q108K:T51D retinal addition. However, the unreduced forms of Q108K:T51D:A28C (dimer) and Q108K:K40L:Y60L:A28C (dimer), used as controls, led to minor changes in CD absorption at 218 nm with the addition of retinal and reducing agents, respectively. 82 These four conformations can be addressed via ligand binding and/or redox potential, resulting in a versatile protein switch. Conveniently, each conformation gives rise to a a. His#81)KD)Apo# 12#Å# His#81)KDA28C)Apo# b. His$81'KD'A28C'holo$ 12$Å$ His$81'KDA28C'Apo$ c. d. His#81)KD)A28C)holo# 17#Å# His#81)KD)holo)Chains#B#and#J# Apo Disulfide bond-holo Holo Disulfide bond-apo Figure II-17. Structural overlays of a. Q108K:T51D (KD) apo (yellow) versus KDA28C apo (blue). b. KD-A28C-holo (green) versus KDA28C-Apo (blue). c. KD-A28C-holo (green) versus KD-holo (orange). All of these overlays show a large conformational change originated from either a disulfide bond or ligand binding or both. d. A summary of conformations potentially addressable by ligand binding, redox environment, or both. Only the c-termini of the proteins are represented. The distance between apo versus holo retinal bound structures is 12 Å, apo versus disulfide bond-apo is 16 Å, apo versus disulfide bond holo is 7.5 Å, and the apo with disulfide bond versus holo with disulfide bond is 10 Å. The apo and holo structures are represented by Q108K:T51D (cyan and blue, respectively) and the apo and holo disulfide-bonded structures are represented by Q108K:T51D:A28C (orange and yellow, respectively). 83 change in the c-terminus (by as much as 10 Å), allowing the protein switch to be attached at the c-terminus instead of being inserted. This is in contrast to most naturally-occurring Figure II-18. The CD spectra of different variants: Q108:T51D dimer, Q108K:K40L:T51F:Y60A dimer, Q108K:T51D:A28C dimer, Q108K:K40L:T51V monomer, and Q108K:K40L:Y60L:A28C upon adding retinal or/and reducing agent. 84 conformational protein switches, where there is relatively little motion of the termini. II-4. Generating an allosterically-regulated zinc binding site A protein switch is only useful if the conformational change can be coupled with a change in function. We chose metal binding as a convenient and potentially applicable function to test in our system. To this end, a metal binding site was designed in the interface of the two domains because residues that are proximal in the apo conformation can undergo significant separation upon ligand binding, with the predicted result being a substantial decrease in metal binding affinity. b. d. a. Retinal3 binding A28C A28C’ Retinal3 binding c. H2O Zn N22 F57H N12 F57H’ L36C S1 F57’ L36’ L36 F57 A28C A28C’ Zn F57H F57H’ L36C F57’ L36’ ~12 Å F57 L36 A28C A28C’ L36C 14(Å F57H L36C’ F57H’ Figure II-19. a. Designing a new metal-binding site using Q108K:T51D:A28C crystal structure as a guide. The proximity of F57H and L36C is critical to providing the metal-binding site. b. The crystal structure of holo retinal-bound Q108K:T51D chains G and I, showing the large separation between two F57s from each chain upon ligand binding. c. Atomic resolution crystal structure of Q108K:T51D:A28C:L36C:F57H bound with zinc showing tetrahedral geometry for zinc coordinated with two histidines coming from two adjacent chains of dimer, a cysteine, and a water molecule. The electron density of the 2FO-FC map is countered to 1σ. The conformations of F57H that did not contribute to the metal-binding site have been omitted for clarity. 85 Though a number of potential binding sites were investigated, Listed in Table II-2 and Table II-3, Leu36 and Phe57 were chosen due to their favorable relative orientation in the apo Q108K:T51D:A28C structure (Figures II-19a and II-19b), and their separation upon retinal binding (II-19b), resulting in the Q108K:T51D:A28C;L36C;F57H (AG1) variant. The metal ion affinity of apo AG1 was evaluated using two assays, a radioactive 65Zn- binding assay (with a help of Dr. Pinger, Prof. Spence’s lab) and isothermal titration calorimetry (ITC) (collaboration with Prof. Jin) (Figure II-20). In both cases, a dissociation constant of 1.6 µM was obtained. In contrast, the metal ion affinity was substantially reduced through the addition of the ligand (all-trans retinal) (Figure II-20b). The addition of the reductant only moderately reduces the affinity either in the presence or absence of all-trans retinal (Figure II-20b). Two variants that do not contain the metal binding site (Q108K:T51D and Q108K:T51D:A28C) show more than one hundred-fold weaker metal ion affinity, demonstrating the specificity of metal binding to the engineered site. This demonstrates the allosteric control of metal binding in the DS dimer, as structurally predicted. To verify the allosteric mechanism of Zn-bound AG1 was determined, revealing a tetrahedrally coordinated zinc with one His57 from each protomer, one Cys36 and one water molecule define the coordination sphere of the bound zinc (Figure II-19c). The coordination distances are consistent with those found in other zinc-bound proteins (bond 86 Table II-2. Variants constructed in search of an allosteric metal binding site. Results Variants Q108K:T51D:T56C Q108K:T51D:F57H Q108K:K40D:T53A:R58L:Q38F:Q4FV62E:F57H Q108K:T51D:F57C Q108K:T51D:F57E Q108K:T51D:F57C:S55C Q108K:T51D:A28C Q108K:T51D:I32C Q108K:T51D:A28H Q108K:T51D:L36C Q108K:T51D:I32H Q108K:K40L:T51H Q108K:K40H:I42H Q108K:K40H:I42H:T51H Q108K:T51D:A28C:L36C Q108K:T51D:F27Y:A28H Q108K:T51D: F27H:A28H Q108K:T51D:A28H:D26H Q108K:T51D:I32C:L36C Q108K:T51D:I32C:F57H Q108K:T51D:L36H:F57H Q108K:T51D:A28C:F57H Q108K:T51D:A28C:L36H Q108K:T51D:I32C:L36H:F57H Q108K:T51D:I32H:L36H:F57H Q108K:T51D:I32G:L36C:F57H Q108K:T51D:A28C:I32A:L36C Q108K:T51D:A28C:I32H:L36H:F57H Q108K:T51D:S55C:F57H:L36H Q108K:T51D:S55H:F57H trimer. However, in vitro refolding in vitro refolding formation of the the stable during Insoluble expression The protein was expressed solubly but does not generate a metal-binding site in the dimer form. Insoluble expression Insoluble expression Insoluble expression Insoluble expression (The experiment also led to protein aggregation) The protein was expressed solubly but does not generate a metal-binding site in the dimer form shown by the crystal structure. The protein was expressed solubly but does not generate a metal-binding site in the dimer form shown by the crystal structure. The protein was expressed solubly but does not generate a metal-binding site in the dimer form shown by the crystal structure. Insoluble expression Insoluble expression (The experiment also showed monomer, dimer, and refolded protein was not concentrating. The protein was expressed solubly but does not generate a metal-binding site in the dimer form shown by the crystal structure. Insoluble expression Insoluble expression (The experiment also led to protein aggregation) The protein was expressed solubly but does not generate a metal-binding site in the dimer form shown by the crystal structure. The protein was expressed solubly. The crystallization of dimer were not successful. The protein was expressed solubly.It led to more oligomer formation whose crystallization was unsuccessful. Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression Insoluble expression in vitro refolding distances for Zn-S: 2.2 Å, Zn-N1: 2.2 Å, Zn-N2: 2.1 Å, and Zn-O: 2.1 Å.). It should be noted 87 Variants Q108K:K40L:T51F:Y60H:E72C Q108K:T51D:K31:H:H:I32 Q108K:T51D:K31:C:C:I32 Q108K:T51D:K31:H:H:H:I32 Q108K:T51D:F57:H:H:H:H:H:H:R58 Q108K:T51D:A28C:F57:H:R58 Q108K:T51D:A28C:F57:H:H:H:R58 Q108K:T51D:A28C:F57:H:H:H:H:R58 Q108K:T51D:A28C:F57:H:H:H:H:H:R58 Q108K:T51D:A28C:F57:H:H:H:H:H:H:R58 Q108K:T51D:A28C:F57:L:H:H:H:L:L:R58 Q108K:T51D:A28C:F57:C:C:C:C:C:C:R58 Q108K:T51D:A28C :R58:H:H:H:N59 Q108K:T51D:A28C:L36C:F57H Q108K:T51D:A28C:L36C:F57H:N59D Q108K:T51D:A28C:L36C:F57C Table II-3. Insertion in the hinge loop region and other variants constructed in search of an allosteric metal binding site. Result The protein was expressed solubly but does not generate a metal-binding site in the dimer form as shown by the crystal structure. Insoluble expression Insoluble expression Insoluble expression The protein was expressed solubly and was purified using nickel column but disappointingly it was not crystallized. The protein was expressed solubly but does not generate a metal-binding site in the dimer form as shown by the crystal structure. The protein was expressed solubly but does not generate a metal-binding site in the dimer form as shown by the crystal structure. The protein was expressed solubly. The protein was expressed solubly and it was crystallized in the apo form. Radioactive labelling zinc assay revealed a Kd: 20 nM to zinc. The protein was expressed Disappointingly it was not crystallized. Insoluble expression The protein was expressed solubly but does not generate a metal-binding site in the dimer form as shown by the crystal structure. The protein was expressed solubly and it was crystallized in the zinc-bound form. Radioactive labelling zinc assay and ITC revealed a Kd: 1.6µM to zinc. The protein was expressed solubly but does not generate a metal-binding site in the dimer form as shown by the crystal structure. The protein was expressed solubly but led to a formation. The large trimer crystallization of dimer and trimer were not successful. solubly but amount of that the zinc geometry revealed here resembles that seen for many natural zinc binding proteins such as carbonic anhydrase. 51 88 Further, the structure of the reduced disulfide form of AG1 was also determined, showing the disulfide bond to be successfully reduced, but with little conformational change a. b. c. d. d. e. Figure II-20. a. The binding curve for apo AG1 using a radioactive-labeled 65Zn showed Kd=1.62 ± 0.14 µM. b. Comparison of association constants for different conformations calculated in 40 µM total 65ZnCl2 : 1) Apo; 2) in the presence of reducing agent only (Red-Apo), 3) in the presence of retinal (Holo). 4) in the presence of all-trans retinal and reducing agent (Red-Holo); As controls, Q108K:T51D (control 1) and Q108K:T51D:A28C (control 2) were used and led to a negligible association constant with Zn. c. ITC experiment shows the dissociation constant of 1.60 ± 0.22 µM for apo. d. The ITC experiment shows the dissociation constant of 29.3±11µM in the presence of all-trans retinal and reducing agent. e. Q108K:T51D:A28C was used as a control. relative to metal-bound AG1, consistent with the modest change in metal affinity upon reduction (Figure II-19d). As shown in Figure II-21b, the overlay of the zinc-bound AG1 and retinal-bound Q108K:T51D shows the extensive conformational change anticipated upon retinal binding, resulting in a rearrangement of the metal binding site to a conformation incompetent for metal binding (Figures II-21b and II-21c), consistent with the allosteric behavior detected in the metal binding assays (Kd (Holo + reduced form): 29 ± 8 µM vs Kd (apo):1.60 ± 0.22 µM). (Figure II-20). 89 b. a. c. F57H’ L36C 2.2 Å F57H’ L36C 2.2 Å 2.1 Å Zn F57H F57H F57H L36C 6 Å 1 F57H’ L36C’ Figure II-21. a. The overlay of Q108K:T51D:A28C:L36C:F57H reduced form (green) and KD-Met (purple-blue) showing no significant change in the conformation upon reducing the disulfide bond. b. Overlay of KD-Met (yellow) and holo Q108K:T51D (purple-blue) crystal structures demonstrating a large change anticipated upon retinal binding, giving rise to reorganization of the metal binding site to a conformation which is not optimal for metal binding. F57 residues of each chain in Q108K:T51D structure have been mutated using PyMOL. c. Overlay of zinc-bound apo Q108K:T51D:A28C:L36C:F57H (purple) (Kd=1.60 ± 0.22 µM) with holo and reduced form structure (Kd=29 ± 8 µM) modeled using Q108K:T51D (green) crystal structure demonstrating a large change anticipated upon retinal binding, giving rise to reorganization of the metal binding site to a conformation which is not optimal for metal binding. 90 II-5. Using fatty acids as input signals So far, we created a novel protein conformational switch that can be activated by retinal or the reduction potential of the environment. To further expand the number of input signals in our system, we decided to use fatty acids, which are highly crucial for a. c. b. d. E72 apo/holo 2.5 Å Y60-Holo N59-Apo N59-Holo Y60-Apo Q108R 7.2 Å Q97 Y60 E72 Figure II-22. a. The overlay of holo linoleic acid bound Q108R:T51D (purple) and Q108K:T51D (apo) domain-swapped dimer crystal structures. b. The same mechanism of dispensation of N59 and Y60 demonstrates that the large conformational change occurs upon fatty acid binding. c. The binding pocket of linoleic acid and the interacting residues were highlighted. d. The linoleic acid electron density was shown. (The 2FO-FC map was countered to 1σ.) Q108R is far away from the carboxylic acid group of linoleic acid and it not involved in the binding event. regulating many biological processes, such as inflammation, signal transduction, energy production, etc. 52. To create the binding site for fatty acids — inspired by fatty acid binding protein families and cellular retinoic acid binding protein II — Q108R:T51D was created 91 with the hope that Q108R could interact with fatty acids and stabilize them in the binding pocket (Figure I-22c). The crystal structures of Q108R:T51D bound with linoleic acid and arachidonic acid were obtained at 2.4 Å and 1.97 Å resolution. The overlay of these structures with an apo domain-swapped dimer revealed the large conformational change in the relative orientations of two dimer domains as observed for holo retinal-bound a. c. b. d. Figure II-23. a. Overlay of arachidonic acid bound Q108R:T51D crystal structure (purple) with apo Q108K:T51D (yellow) showing the large conformational change in the relative orientation of the two dimer units. b. Overlay of linoleic acid bound structure vs. arachidonic acid bound structure demonstrates a similar trajectory for these two fatty acids in the binding pocket of the Q108R:T51D DS dimer variant of hCRBPII. c. Overlay of Q108R:T51D crystal structure bound with linoleic acid (pink) with FABP5 bound with N-(2-hydroxyethyl) icosanamide (PDB code: 4AZR). d. Overlay of Q108R:T51D crystal structure bound with linoleic acid (pink) with FABP4 bound with linoleic acid (yellow), PDB code: 2Q9S. 92 domain-swapped dimers (Figure II-22a). The close-up overlay of apo and holo in the hinge region revealed the conformational switch follows the same mechanism as the retinal-bound holo dimer, where the dispensation of Asn59 and Tyr60 causes a large change in the relative orientations of the two dimer domains (Figure II-22b). Furthermore, the crystal structure revealed a novel binding site for both fatty acids. As shown in Figures II-20C and II-20d, the structure of Q108R:T51D bound with linoleic acid shows that E72 and Y60 are interacting with the carboxylate group of fatty acids, and Q108R does not participate in fatty acid binding. This is in contrast with the typical fatty acid protein binding pocket, which shows that the caboxylate group of fatty acids usually interacts with a conserved Arg and Tyr.53 The crystal structure of arachidonic acid bound with Q108R:T51D shows a similar conformational change and trajectory of fatty acid binding as cited above (Figure II-23). This observation begs the question of whether or not the WT monomeric form of hCRBPII, traditionally thought to merely bind with retinal and retinol,54 also has an afffinity toward fatty acids. Since these proteins are largely expressed within the intestine,54 they might have a secondary function to the one originally determined. In this case, a fluorescence-quenching competition assay was performed for the WT-hCRBPII-monomer, revealing that linoleic acid is able to replace retinal in the binding pocket in a concentration-dependent manner. Additionally, the crystal structure of WT-hCRBPII bound with linoleic acid was obtained at 1.6 Å resolution, further depicting the reproducibility of the binding pocket toward linoleic acid binding in WT-hCRBPII (The structure and data has not been shown here). 93 II-6. The observation of a ligand-induced “swap back” at atomic resolution To expand the conformational diversity of our domain-swapped dimer, we used merocyanine, a synthetic fluorophore with aldehyde functionality (Figure II-24). Previously, Yapici et al. demonstrated new “turn-on” fluorescent protein tags using O N II-24. The Figure merocyanine aldehyde Prof. Borhan’s lab. chromophore functionality designed structure of with in merocyanine, exciting at 594 nm and emitting at 619 nm upon generation of a protonated Schiff base with the monomeric hCRABPII.55 Furthermore, in other unpublished results a. b. C. Figure II-25. a. Overlay of apo DS dimer of A1 (purple) with WT-hCRBPII DS dimer (orange) showing a symmetric dimer (RMSD: 0.3.). b. Apo DS dimer of A1 (purple) with holo retinal-bound DS dimer of A1 (orange) revealing the large conformational change in the relative orientation of the two dimer domains. c. Overlay of holo retinal-bound DS dimer of A1 (orange) with holo retinal- bound DS dimer of Q108K:T51D (cyan). Dr. Santos described the photobleaching feature of this chromophore. She showed that during excitation, the fluorophore is photobleached leading to the loss of fluorescence; then, by re-adding the fluorophore, the protein rebinds with new molecules, and fluorescence is retrieved. If the bleaching resulted in loss of bound ligand, this feature could give us a convenient optogenetic with conformational switching driven by 94 b. a. d. c. e. Q108K K40D Normal symmteric dimer “swap back” dimer f. g. h. Q108K 2.7 Å K40D-apo 4.5 Å 6.7 Å K40D - merocyanine-bound 4.7 Å K40 5.4 Å K40 K40D-merocyanine K40-retinal 4.6 Å 3.4 Å K40D-apo Q108K K40D-merocyanine i. Q108K K40D K40L Figure II-26. a. The merocyanine-bound A1 crystal structure showing the swapped-back DS dimer. Chains are shown in green and cyan. The merocyanine ligand is represented by a combination of sphere and stick representation (pink). b. The overlay of merocyanine-bound A1 (green) and apo (purple-blue). c.The electron density highlighted for the ligand and K40D contoured to 1σ. d. Left: The normal domain-swapped dimer where half of the protein undergo the domain-swapping. The swapped open monomers showed with different colors. Right: The “swap back” dimer where the α-helix and beta strand 1 from each chain is swapped back highlighted with a box and light orange and purple colors. e. K40D has to move two residues along the strand to interact directly with the nitrogen of iminuim, giving rise to a change in the registry of the amino acids that make up strand 2 forcing the melting of the α-helix. f. Different positions of K40D in apo (purple-blue) vs. merocyanine-bound A1 DS dimer (green) in the overlaid structures. g. The positions of K40D in merocyanine-bound A1 DS dimer (green) vs. retinal-bound A1 (orange), chain I. h. The positions of K40D in merocyanine-bound A1 (green) vs. retinal-bound A1 (pink), chain i. The overlay of merocyanine-bound A1 DS dimer (green) vs. merocyanine-bound Q108K:K40L:T51V (purple blue). chromophore bleaching, resulting in a return to the apo conformation. Therefore, we decided to screen the crystallization of the merocyanine ligand with many variants of hCRBPII that form domain-swapped dimers. The comprehensive screening led us to the following variants: Q108K:T51D and Q108K:R58L:T53A: K40D:Q38F:Q4F (A1). Unfortunately merocyanine-bound Q108K:T51D yielded poorly diffracting crystals (beyond 4 Å resolution). However, the crystal structures of apo, merocyanine bound and 95 retinal bound A1 were determined successfully. The apo structure of the protein, as shown in Figure II-25a, depicted the expected apo symmetric dimer. Furthermore, the retinal-bound crystal structure showed 12 molecules in asymmetric unit and retinal binding led to a large twist in the relative orientations of the two dimer domains, similar to that observed for other retinal-bound dimers (Figures II-25b and II-25c). In contrast, the crystal structure of A1/merocyanine (obtained at 1.81 Å) gave a surprisingly different structure. A new swapping occurred in the protein. In the normal domain-swapped dimer of hCRBPII, half of the protein (55-133) underwent domain swapping. Here, n-terminal residues (1-25) of the protein were swapped back, leading to a large conformational change in the protein (Figures II-26a-II-26d). Having the atomic- resolution crystal structures of this variant in our hands, we were interested in understanding the mechanism of this conformational change. As shown in Figures II-26e and II-26f, the overlay of the merocyanine-bound structure with retinal-bound holo dimers reveals that in the normal domain-swapped dimer, the K40 residue does not interact directly with the nitrogen of iminium. However, in the merocyanine-bound structure this residue moves two positions along the β-strand to create a direct hydrogen bond with the nitrogen of the iminium (2.7 Å), altering the registration of 26-46 which include β-strand 2 and α-helix 2. This shift in registry apparently causes the melting of α-helix (26-36) in the structure. Presumably, loss of helix/helix interactions due to melting the second helix would also facilitate the helix 1 and strand 1 swap back. The reason behind this abnormal behavior of the ligand can be seen in Figures II-26g and II-26h. The merocyanine ligand, which is a shorter ligand than all-trans retinal, adopts almost the same configuration of the imine bond for all the structures obtained so. In all of these conformations, 96 merocyanine lacks the ability to interact directly with K40D, Figure II-26I. However, the position of the imine bond can be altered in retinylidiene structures, as seen in some chains of the all-trans retinal-bound A1 crystal structure, where retinal generates a direct interaction with K40D, Figure II-26h. Another difference is that merocyanine is a highly- conjugated ligand in comparison with all-trans retinal, leading to a naturally high pKa of the chromophore. Additionally, in the case of the A1 variant, the binding pocket is more hydrophobic as a result of hydrophilic residue mutations, such as Q4F, Q38F, and T53A. Therefore, there is a high pKa iminium in the merocyanine-bound structure — possessing a positive charge that is distant from the negatively-charged K40D — surrounded by many hydrophobic residues in the binding pocket. As mentioned earlier, merocyanine does not have the ability to alter the position of the imine bond to create a direct interaction with the counter ion. As a result, K40D has to move two residues along the strand, giving rise to a change in the registry of the amino acids that make up strand 2 forcing the melting of the α-helix. To investigate the effect of K40D, three new variants: A1-K40, A1-K40L, and A1-K40F, were constructed, all of which were dominantly expressed as monomers. This indicates that the presence of K40D is important for dimer formation. Only the monomer of A1-K40F bound to merocyanine led to well-diffracting crystals. As shown in Figure II-27a, the crystal structure demonstrates that in the absence of K40D there is no change in registry of the β-strand. This is consistent with the idea that the change in registry is driven by the formation of the salt bridge between Asp40 and the Iminium. Instead, the chromophore is rotated almost 180° allowing the iminium nitrogen to form a cation-π interaction with Trp106. This cation-π interaction between iminium and Trp106 (or its equivalent) has been seen in several other PSB-forming ligand-bound 97 structures such as Q108K:K40L:T51V:R58W:Y19W/retinal (PDBcode: 5FFH), Q108K:K40L:T51V:R58W:Y19W/retinal (PDB code: 5F7G), and Q108K:K40L:T51V:T53C:R58W:T29L:Y19W:A33W/retinal (PDB code: 4EDE). Finally, the ability to create such a large conformational change through merocyanine can a. c. 2.7 Å K40D W106 3.4 Å K40F 584 nm b. d. 595 nm Figure II-27. a. Overlay of A1 DS dimer bound with merocyanine (green) and A1-K40F monomer bound with merocyanine (pink) portraying a significant change in the trajectory of the ligand. b. Overlay of apo A1 vs. ThioFluor9 chromophore-bound A1 DS dimer crystal structure shows no conformational change. c. UV. Vis spectrum of A1 bound with the merocyanine ligand. The protonated Schiff base absorbs at 584 nm. d. UV. Vis spectrum of A1-K40F bound with the merocyanine ligand. The protonated Schiff base absorbs at 595 nm. be used to generate a new class of molecular switches in optogenetics, a critical application that should be pursued in the future. As shown above, retinal and merocyanine led to complete different conformations on A1. This characteristic would allow us to control the conformational state of one protein scaffold by different ligands (retinal and merocyanine). We also used the shorter ligands as a proof of principle — ThioFluors, thiophene-based fluorophores synthesized by Dr. 98 Santos (see Chapter IV for details) — where the ligand binding does not disturb the apo- symmetric dimer and therefore does not alter the protein conformation (Figure II-27b). This demonstrates the power of our protein conformational switch to “surf” various conformational outcomes upon the introduction of different input signals. The latter application can be especially pursued for certain in vivo studies when the protein conformational change is unnecessary at the time of the experiment. In this case, the goal is to merely monitor the protein of interest’s localization, which is linked to the protein conformational switch using fluorescence signals. II-7. Hinge loop region: nature’s economic way to create conformationally-distinct proteins? Nature creates functionally and conformationally-distinct proteins from a common scaffold through convergent or divergent evolution.56 Domain-swapping, where two identical segments of a protein exchange to create dimers or higher-order oligomers, was originally speculated to be the natural mechanism for surfing different oligomerization states in proteins, and therefore regulating the final function. Based on previous studies, variables leading to a large amount of domain-swapped oligomer formation include a change in pH, protein concentration, ligands, temperature, and certain mutations in the hinge loop region.57,58 Specifically, many research groups have demonstrated the hinge loop role as an effective tool for controlling the oligomerization states in proteins.59 For example, studies of domain swapping in staphylococcal nuclease, suc1, ckshs1, CD2, and single-chain Fv showed that reducing the hinge-loop length causes destabilization of monomeric forms 99 and promotion of domain swapped dimer formation. 60-64 Other studies revealed that the insertion of specific amino acids in the hinge region could favor domain swapping.65 However, to our knowledge, there are no examples showing, in a systematic manner, the effect of the insertion in the hinge loop region on the overall conformational states of the domain-swapped oligomers and their monomeric forms with atomic- resolution crystal structures. This study illustrates the essential role the hinge loop region plays to explore new conformational states without imposing a drastic change in the sequence of proteins. To investigate the role of the hinge region, we needed a suitable platform, meaning a protein scaffold that has a high tolerance toward mutations and can be easily expressed and crystallized. II-7-1. Choosing a suitable construct As mentioned earlier, the role of the engineered disulfide bond was highlighted as a tool to create new conformations in a protein. To investigate the role of amino acid insertion and to avoid changing the monomer/dimer ratio, we decided to use a disulfide- bonded construct as our initial template. The presence of the disulfide bond would further stabilize the scaffold and make it more stable towards mutation, but also, since the position of the disulfide bond is selected on top of the α-helices, it would not prevent conformational flexibility to produce and investigate different conformations in the dimer. The Q108K:T51D:A28C:L36C variant was chosen as a starting point. Q108K is a mutation that is required for Schiff base formation, T51D favors dimer formation, A28C generates the disulfide bond, and finally, L36C enforces the overall protein conformation as a normal symmetric dimer, as defined in our previous sections. This construct also gives rise to well-diffracting crystals, allowing us to observe the changes at atomic 100 resolution. II-7-2. “Phase problem" in the domain-swapped dimer It’s important to highlight that in all of our domain-swapped dimer structures, domain swapping is the result of the loop connecting β-strands 3 and 4 of the monomer, leading to the formation of a single β-strand stretching the length of the dimer (Figure II- 28). However, this results in a “phase problem” because, while the odd sidechains of β- a. disulfide bond A28CA28C’ b. c. ~13Å d. Monomer D63 V62 T51 K50 D61 Y60 T53 K52 N59 R58 S55 T54 F57 T56 e. Dimer T51 K50 “Phase problem” T53 K52 S55 T54 F57 T56 N59 R58 Y60 D61 V62 D63 = Carbonyl C = α C = Nitrogen = β Sheet, Side Chain In= β Sheet, Side Chain Out= Loop Figure II-28. a. Q108:T51D:A28C:L36C crystal structure. Each chain is shown in a different color and is connected by the disulfide bond on the top part of the α-helices. b. Overlay of Q108:T51D:A28C:L36C crystal structure (purple) with WT hCRBPII domain-swapped dimer (green) showing a similar conformation to the symmetric dimer for Q108:T51D:A28C:L36C. c. Overlay of Q108:T51D:A28C (orange) with WT hCRBPII dimer showing a drastic change in the relative orientation of the two dimer units, leading to a large change in the c-termini distance between the two structures. d. Arrangement of the residues connecting strands 3 and 4 in the monomer; left panel shows the crystal structure of the monomer where the aforementioned strands were highlighted. e. Arrangement of the residues connecting strands 3 and 4 in the dimer to make a continuous strand leading to a phase problem; left panel shows the crystal structure of the dimer where the aforementioned strands were highlighted. 101 strand 3 face inside the binding pocket, the even sidechains of β-strand 4 face outward. This means that the single β-strand formed in the domain-swapped dimer must, at some point, re-phase the strand, which would then put the even sidechains of the c-terminal half of the strand “in phase” with the odd-numbered sidechains of the first half. II-7-3. Amino acid insertion: odd vs. even number of insertions According to the “phase theory,” we would expect the insertion to show a new trend Phase problem N59 F57 T56 S55 T54 T53 a. c. b. d. ~ 8 Å 21 Å Y60 D61 R58 Y60 D61 Y60 F57 R58 N59 H1 S55 T56 T53 T54 R58 H2 F57 S55 T53 D61 N59 H3 H1 T56 T54 Figure II-29. a. Overlay of 1H (orange) and Q108K:T51D:A28C:L36C (purple) crystal structures. The c-terminus has been moved ~ 8 Å after introducing His between F57 and R58. b. Top: The hinge loop regions and part of the two β-strands (3 and 4) were highlighted in Q108K:T51D:A28C:L36C where both Y60 and N59 are located inside the binding pocket and cause a “phase problem”. Bottom: After insertion of one His between F57 and R58, a continuous β-strand is formed where every residue is “in phase”. c. Overlay of 3H (red) and Q108K:T51D:A28C:L36C (purple) crystal structures, representing a large movement of the c- termini and the relative orientation of the two dimer units. d. The “zoom-in” demonstration of the hinge region after the addition of three His, solving the “phase issue”. in the overall structure of the protein, where the number of residues connecting the two strands is different from the original domain-swapped dimer. 102 The result of the insertion could lead to two predictions: 1. Odd insertion would give rise to a continuous β-strand; even insertion would lead to loop formation between the two strands; 2. The number, nature, and location of the insertions would give rise to conformationally-distinct protein scaffolds, since they can influence the torsional angles of the key residues in the hinge region in different ways. In the initial effort, histidines were added between F57 and R58, the residues located in the middle of the hinge loop. Histidine residues were chosen for three reasons: first, they are soluble amino acids; second, their pKas are pretty close to the PI of the protein (~5) so the insertion does not lead to a large change in the charge, and subsequently, the purification strategy that uses the ion exchange column; and third, during the amino acid insertion experiment, a novel metal-binding site can possibly be generated, which portrays how the new conformation can impose new functions in the protein. In our initial effort, two constructs were created: Q108K:T51D:A28C:L36C:F57:H:R58 (1H) and Q108K:T51D:A28C:L36C:F57:H:H:H:R58 (3H). These constructs were expressed and purified, and their crystal structures were determined. As shown in Figure II-29, the crystal structures of these three constructs were overlaid with Q108K:T51D:A28C:L36C, which is structurally very similar to the apo- symmetric dimer. Adding a histidine between F57 and R58 creates an ideal β-strand connecting the two separated strands, fixing the phase problem mentioned above and satisfying the first prediction. Upon adding this residue, as shown in Figures II-29a and II-29b, the relative orientations of the two dimer domains were altered, leading to 8 Å distance between the c-termini of the two structures (before and after insertion). 103 As shown in Figures II-29c and II-29d, the crystal structure of the 3H insertion depicted a. c. e. e. g. b. Y60 Preorganized for a metal-binding site R58 H4 H5 H1 F57 S55 T53 D61 N59 H6 H3 H2 T56 “Phase problem” T54 23.6 Å d. f. 13.5 Å Y60 R58 W2 F57 S55 T53 D61 N59 W3 W1 T56 T54 h. Y60 H3 H1 F57 S55 T53 D61 N59 H2 R58 T56 T54 10 Å 7.3 Å 9 Å Figure II-30. Overlay of 6H (yellow) and Q108K:T51D:A28C:L36C (purple) crystal structures. The c-terminus has been moved more than 23 Å after introducing six His between F57 and R58. b. Adding the even number of residues depicts the “phase problem,” where the two neighboring His are outside the binding pocket and “out of phase.” Also, three His are in the right conformation for generating a metal-binding site. c. Overlay of two open monomers of 3H generating a domain- swapped dimer (chain A, pink, and chain B, cyan). d. Overlay of two open monomers of 6H generating a domain-swapped dimer (chain A, green, and chain B, purple). e. Overlay of 3W (cyan) and 3H (pink) crystal structures, showing a novel conformation was created where the nature of amino acid insertion is different. f. Close-up view of the 3W hinge region, where three Trp residues were inserted between F57 and R58 and every residue is “in phase.” g. Overlay of 3H’ vs. 3H crystal structures, demonstrating the position of insertion is also highly important in the creation of a new conformation in the protein. h. Close-up view of 3H’ hinge region, where three His residues were inserted between R58 and N59 and every residue is “in phase”. 104 the same features of a continuous strand, formed as a result of an odd number of residues added in the β-sheet. Furthermore, the overlay of the 3H-insertion vs. no-insertion structures demonstrated a larger conformational change in the relative orientation of the two dimer units. It almost looks as if the structure goes from a “cis” conformation, in the case of no insertion, to a “trans” form, in the case of 3H insertion, giving rise to a drastic change in the c-termini of the protein (21 Å distance). Overall, the odd number of insertions satisfied the phase theory, as mentioned earlier, and it leads to a substantial change in the relative orientation of the two dimer units. The next step was to test the effect of an even number of insertions. We tested this by introducing four histidines and six histidines between F57 and R58, leading to the formation of Q108K:T51D:A28C:L36C:F57:H:H:H:H:R58 (4H) and Q108K:T51D:A28C:L36C:F57:H:H:H:H:H:H:R58 (6H). The four histidines did not produce high-quality crystals; however, the six histidines diffracted to 2.54 Å. As shown in Figure II-30b, the structure delineated the formation of loops in the hinge loop region where three histidines were preorganized to bind zinc. Radioactive 65Zn labeling (with the help of Dr. Pinger) further confirmed a 20.8 ± 3.8 nM dissociation constant with zinc. This shows that adding a certain number of histidines in the binding pocket can not only affect the conformation, but also leads to the creation of a new function in the protein. Additionally, the overlay of this structure with no H insertion demonstrates a new conformational change in the relative orientation of the two dimer units, leading to a ~24 Å change in the c-terminal distance (Figure II-30a). The overlay of the two chains of 6H generating the domain-swapped dimer revealed that structurally, the two open-monomer 105 units are very different in comparison with the two open monomers forming the 3H domain-swapped dimer (Figures II-30c and II-30d). This demonstrates that for a 6H domain-swapped dimer, a large conformational change should occur in the c- and n- termini of the open monomers to form a dimer molecule, simply because more structural flexibility occurs as a result of loop formation in the hinge region vs. the strand in the 3-H structure. Perhaps the presence of a disulfide bond would assist with the formation of a domain-swapped dimer, where the two open monomers are drastically different in their conformations. The in-depth screening to obtain a crystal — in the context of the Q108K:T51D:F57:H:H:H:H:H:H:R58 construct that does not provide the disulfide bond — was not successful. II-7-4. The effect of the nature of amino acid insertion The next step is to elucidate how the nature of amino acid insertion affects the overall conformation of the protein. This time, tryptophan was chosen as a bulkier amino acid that would be expected to achieve different conformational states of the protein. To test this hypothesis, three and six tryptophan residues were added to the Q108K:T51D:A28C:L36C construct to create Q108K:T51D:A28C:L36C:F57:WWW:R58 (3W) and Q108K:T51D:A28C:L36C:F57:W:W:W:W:WW:R58 (6W) constructs. The 3W constructs were expressed solubly and successfully purified; however, 6W did not lead to soluble expression. 3W was also successfully crystallized, and its structure was determined at 2.38 Å resolution. As shown in Figures II-30e and II-30f, reminiscent of other odd insertions, this structure also showed a continuous β-strand formation as a result of solving the phase problem. However, it leads to a conformationally-distinct orientation of the two dimer units relative to one another, in comparison with 3H and 6H. 106 This further demonstrates the high capacity of the loop region to “surf” different conformational states in the protein, not only by the length of insertion, but also by the nature of the amino acid insertion in the hinge region. II-7-5. Investigating the effect of the insertion position on the conformational state The next question is how a change in the position of insertion can affect the conformations of these domain-swapped dimer proteins. To tackle this question, we decided to change the position of insertion to between R58 and N59 instead of between F57 and R58. The results of the insertion yields Q108K:T51D:A28C:L36C:R58:H:H:H:N59 (3H’) and Q108K:T51D:A28C:L36C:R58:H:H:H:H:H:H:N59 (6H’). The first construct was not solubly expressed; however, the second construct was successfully expressed and purified, and its crystal structure was determined at 2.19 Å resolution. As shown in Figures II-30g and II-30h, the overlay of the construct with 3-H insertion between F57 and R58 demonstrated a novel conformation, which showed that in addition to the length and nature of the amino acid, the position of insertion also effects the overall conformation of the proteins. II-7-6. The effect of insertion on the monomer fold Our final question is how these insertions would affect the monomeric form of the protein. We added three histidines and six histidines to WT-hCRBPII, leading to F57:H:H:H:R58 (WT-3H) and F57:H:H:H:H:H:H:R58 (WT-6H) constructs. The constructs were purified and yielded substantial quantities of monomer (The dimer was not detected.). The crystal 107 a. c. e. f. b. d. Close-up -Front view H1 H2 H3 2.1 Å L36-WT L36-WT-3H Close-up L36-WT-6H H1 L36-WT 5.3 Å H6 H5 H4 8.8 Å H3 H2 Figure II-31. a. Crystal structure of WT-3H, chain B. The melted α-helix is shown in purple. b. Crystal structure of WT-3H, chain A. The melted α-helix is shown in purple. c. Overlay of two chains of WT- 3H. d. Crystal structure of WT-6H. e. Overlay of WT-3H (cyan), and WT-hCRBPII (orange) (PDB code: 2RCQ) The inserted histidines were shown with blue. f. Overlay of WT-6H (dark blue), and WT-hCRBPII (orange). structures of both of these proteins were determined at atomic resolution (1.3 and 1.1 Å, 108 respectively). As shown in Figures II-31a and II-31b, the crystal structure of WT-3H shows alpha helix 2 to be melted. To our knowledge, this is the first report of such a structure in the iLBP family, in which an α-helix, which is responsible for covering the ligand-binding site, melted, and that this drastic change was observed at atomic resolution. The 6H inserted structure represented a similar α-helix, as expected for hCRBPII, but with a different orientation in the loop region (Figure II-31d). As shown in the overlay of the wild type (PDB code: 2RCQ) vs. WT-3H and wild type vs. WT-6H, the orientation of the loop is highly different, and perhaps in WT-3H, this would require unfolding the α-helix where it sterically clashes with L36 in the α-helix conformation (Figure II-31e). However, the longer length of insertion gives rise to higher flexibility in the hinge loop region, which leads to no effect on the overall structure of the protein, Figure I31f. The take-home lesson from this study is that the hinge-loop region in proteins, especially those prone to domain swapping, can be used as a tool to explore different conformational states. This leads to a new, evolving scaffold with novel conformations and functional features produced only through economic changes in the protein sequence. The effects of the number, location, and nature of amino acid insertion, as observed in atomic-resolution crystal structures, are quite drastic. This study highlights the high potency of the hinge-loop region to engineer novel proteins with new functions using a protein redesign strategy for protein engineering communities. II-8. Using the hCRBPII domain-swapped dimer as a functional regulator As mentioned in the introduction, section II-1, many natural protein conformational switches have been employed as functional regulators. Most of the efforts to create such 109 systems, as cited earlier, are centered on domain insertion, where the circularly permutated versions of the target proteins are inserted inside the domains of the protein conformational switches, where the largest conformational changes occur. However, this strategy suffers from an important pitfall, in which many proteins might not fold properly when another protein is inserted inside their domains. As a result, this strategy may increase the risk of protein aggregation and complicate the protein design strategy.66 Creating new protein conformational switches with a large conformational change in the c- and n-termini would address this problem. As mentioned in sections II-3 to II-6, we created a large conformational change in the c-termini of the protein upon the addition of retinal, fatty acid, merocyanine, and a change in the reduction potential of environment (Figures II-17d, II-22, II-26). The next step toward creating a functional regulator is to first investigate whether or not the domain-swapped dimer is still formed after fusing another protein to the c-termini of our protein conformational switch. To perform this investigation, a A1-GFP construct was created in pET17b. (The fusion construct of hCRBPII was made first by Dr. Santos, and a few mutations were imposed to create the A1-GFP construct). The protein was solubly expressed and, fortuitously, as mentioned in the experimental section, it was purified using ion exchange chromatography and size exclusion chromatography. The UV-vis spectrum of the fused protein and the green color of the purified protein reveal that the GFP expressed solubly. To elucidate whether hCRBPII expressed as a dimer or not, the biorad standard was carried out on the column and the retention times of standard proteins were compared to the A1-GFP retention time. As shown in Figure II-32, the result demonstrates dimer as well as monomer expression of A1-GFP (86 kDa) in the presence of n-terminal fused eGFP. This observation further 110 highlights that the domain-swapped dimer is a robust scaffold and does not dissociate in the presence of a fusion protein in the c-terminus. Therefore, as a result of a large conformational change observed in the c-terminus, the dimer can be used as a convenient platform for functional switch-related applications. II-9. Does the domain-swapped dimer of hCRBPII possess a novel functionality in humans? N- a. hCRBPII (A1) Dimer C- N- C- eGFP b. Monomer Biorad gel filteraction standard components A&B C D E F Figure II-32. a. EGFP that is fused to the c-terminus of A1 hCRBPII. The gel filtration result demonstrates the production of the dimer in the presence of GFP when the retention time is compared to the biorad gel filtration standard. b. Biorad gel filtration standard components. Recently allosteric regulation for some fatty acid binding proteins (FABPs) and cellular retinoic acid binding proteins (CRABPs), namely FABP453 and CRABPII,67 was proposed. It was suggested in both FABP4 and CRABPII that ligand binding gives rise to a conformational change in the protein, resulting in a translocation of FABP4 and CRABPII to the nucleus. This then facilitates the delivery of the fatty acid and retinoic acid to their respective nuclear receptors.68-69 Nonetheless, the crystal structures of the apo and holo iLBP monomers show small conformational variations upon ligand binding.46, 70- 111 74 Based on the high-resolution crystal structures cited above, a large reproducible conformational change happens in the relative orientation of two dimer domains in the hCRBPII DS dimer upon ligand binding, demonstrating high sensitivity of this fold for ligand binding. Based on these findings, we wondered, based on the inherent allostery of the DS dimer, if the DS dimer fold could have some physiological relevance. Subsequently we identified a single mutation, T51A that substantially increases the dimer/monomer ratio of hCRBPII. Intriguingly, the T51A variant is found in the human population, though no phenotype has been attributed to it thus far. We successfully determined the crystal structure of the dimer at 1.50 Å resolution and found it to form the same symmetric dimer every other apo-DS dimer adopts. This observation, along with the discovery of fatty acid binding for hCRBPII, as mentioned in section II-5, further suggest that the DS dimer of hCRBPII might have a distinct biological function in humans. The validation of such a mechanism will require further functional exploration of the iLBP family members in vivo. II-10. Conclusion and future direction Using the high-resolution structural evaluation and interrogation approach, we created a novel protein conformational switch that can be activated by ligand binding. We elucidated the mechanism of this protein conformational switch and then, by employing an engineered disulfide cross-link, we expanded the number of conformational readouts of the system. Furthermore, to define a function for the protein conformational switch, we generated a novel metal binding site in the interface of two dimer domains that can be regulated by the conformational readouts of the system. Using biochemical and 112 biophysical techniques, we showed the structure and binding affinity of the metal in each conformational state of the protein. Additionally, we demonstrated the ability to drive the conformational change using fatty acids and synthetic chromophores. In the latter case, the effect of binding was investigated using a merocyanine fluorophore, where the second swapping observed at atomic resolution gave rise to the largest conformational change observed so far. The special nature of this chromophore as a photobleachable ligand introduces the possibility of creating novel optogenetic conformational switches. Further improvement of such a concept can be achieved by using retinal isomerization, where the two isomers can generate different conformational outcomes in the protein, or by using other light-sensitive ligands, such as azobenzene-based chromophores. a. b. c. hCRBPII HIV-protease Figure II-33. The schematic design of HIV-protease fused to hCRBPII. a. Apo form of hCRBPII, where presumably HIV protease is in active form. b. Ligand-bound form of HCRBPII, which presumably causes a perturbation in the orientations of two dimer units. c. HIV protease crystal structure. PDB code: 2pc0. We can take advantage of our highly-flexible protein conformational switch to make a new classes of functional regulators. For instance, a new construct has been made with the Q108K:T51D variant of hCRBPII in the n-terminus and HIV-protease in the c- terminus, connected by three Gly residues. HIV-protease is an important enzyme for the maturation of the HIV virus and functions as a homo dimer (Figure II-33). Through side- 113 by-side protein/protein interaction, the active site of the enzyme is created when the two aspartates are in close proximity to one another. The connection of HIV-protease to the c-terminus of our protein (hCRBPII-HIV) may allow us to develop an allosterically regulated protease. Ligand binding would alter the conformations of our protein and subsequently the two monomeric forms of HIV-protease and perturb the active site of the enzyme. The hCRBPII-HIV construct would aid us in understanding if we can alter the function of an enzyme through our protein conformational switch. Additionally, this phenomenon can be investigated in the future by attaching our protein conformational switch to transcription factors, especially those creating homo dimers to elicit novel molecular tools in biology with higher flexibility of signal implementation. II-11. Experimental Procedures II-11-1. PCR protocol Table II-4. PCR condition. Template (DNA plasmid) Primer forward Primer reverse dNTP 10X pfu buffer pfu Turbo (DNA polymerase) 70 ng (× μl) 20 pmol (y μl) 20 pmol (z μl) 1 μl 5 μl 1 μl DI water 43-(x+y+z) μl 114 Table II-5. PCR program. PCR program 1x 20x 94 oC 94 oC temperature 3-5 oC lower than primer melting temperature
 3 min 20 sec 50 sec 72o C
 4 min 30 sec (3 min 30 Sec for 1x 72 oC CRBPII) 10 min II-11-2. List of PCR primers: Site-directed mutagenesis was performed using K-CRBPII or KL-CRBPII-PET17b, which were generated in our previous work.48, 75 T51D Forward: 5’-GGTGATAACTTCAAGGATAAAACCACTAGCAC-3’ Reverse: 5’-GTGCTAGTGGTTTTATCCTTGAAGTTATCACC-3’ T51F Forward: 5’-GGTGATAACTTCAAGTTCAAAACCACTAGCAC-3’ Reverse: 5’-GTGCTAGTGGTTTTGAACTTGAAGTTATCACC-3’ T51W Forward: 5’- GGTGATAACTTCAAGTGGAAAACCACTAGCAC -3’ Reverse: 5’- GTGCTAGTGGTTTTCCACTTGAAGTTATCACC -3’ A28C 115 Forward: 5’- GATATTGATTTTTGCACCCGCAAG-3’ Reverse: 5’- CTTGCGGGTGCAAAAATCAATATC-3’ A28H Forward: 5’- GATATTGATTTTCATACCCGCAAG-3’ Reverse: 5’- CTTGCGGGTATGAAAATCAATATC-3’ Y60A Forward: 5’-CACATTCCGCAACGCTGATGTGGATTTCAC-3’ Reverse: 5’-GTGAAATCCACATCAGCGTTGCGGAATGTG-3’ F57H Forward: 5’-ACCACTAGCACACATCGCAAC-3’ Reverse: 5’-GTTGCGATGTGTGCTAGTGGT -3’ L36C Forward: 5’-GCAGTACGTTGTACTCAGACG -3’ Reverse: 5’-CGTCTGAGTACAACGTACTGC -3’ II-11-3. Hinge loop insertion primers and PCR protocol Site-directed mutagenesis was performed using Q108K:T51D:A28C:L36C-CRBPII- PET17b. The optimized primers for Q5 site directed mutagenesis were designed using the NEBase Changer. F57:H:R58 Forward: 5’- CATCGCAACTATGATGTGGATTTC-3’ Reverse: 5’- GAATGTGCTAGTGGTTTTTG-3’ F57:H:H:H:R58 Forward: 5’- TCATCGCAACTATGATGTGGATTTC-3’ 116 Reverse: 5’- TGATGGAATGTGCTAGTGGTTTTTG-3’ F57:H:H:H:H:H:H:R58 Forward: 5’- CACCACCACCGCAACTATGATGTGGATTTC-3’ Reverse: 5’- ATGATGATGGAATGTGCTAGTGGTTTTTG -3’ F57:W:W:W:R58 Forward: 5’- GTGGCGCAACTATGATGTGGATTTC -3’ Reverse: 5’- CACCAGAATGTGCTAGTGGTTTTTG-3’ F57:W:W:W:W:W:W:R58 Forward: 5’- TGGTGGTGGCGCAACTATGATGTGGATTTC-3’ Reverse: 5’- CCACCACCAGAATGTGCTAGTGGTTTTTG-3’ R58:H:H:H:N59 Forward: 5’- TCATAACTATGATGTGGATTTCACTGTTG-3’ Reverse: 5’- TGATGGCGGAATGTGCTAGTGGT-3’ R58:H:H:H:H:H:H:N59 Forward: 5’-CATCATCATAACTATGAT-3’ Reverse: 5’-ATGATGATGGCGGAATGTGCTAGTGGT-3’ II-11-4. Transformation of PCR product DH5α competent cells (50 μL) were thawed on ice, and then 8 μL of PCR product was added to the sample. After 30 minutes incubation on ice, the samples were heat- shocked at 42 ̊C for 45 seconds and 450 μL of LB was added. Then the mixture solution was incubated at 37 ̊C for 30 minutes. The cells were spread on an LB agar plate containing ampicillin (100 μg/mL) and incubated at 37 ̊C overnight. II-11-5. DNA Purification 117 A single colony was utilized to inoculate 5 mL of LB solution containing ampicillin (100 μg/mL). The cell culture was grown at 37 ̊C overnight. Then it was centrifuged at 13,000 rpm for one minute. DNA isolation was performed by employing Promega Wizard® Plus SV Miniprep (A1330). The sample was then sequenced by the Research Technology Support Facility at Michigan State University. A T7 primer was used for sequencing. II-11-6. Expression of hCRBPII mutants For the expression, the gene was transformed into BL21 (DE3) pLysS (InvitrogenTM) E. coli competent cells. A protocol similar to the one mentioned for the transformation of PCR products was employed. The only difference was the use of ampicillin 100 μg/mL as antibiotic. To inoculate 1 L of LB with ampicillin (100 μg/mL), a single colony was used. After the cell culture was grown at 37 ̊ C while shaking for 8 hours, the overexpression was induced by adding 1 mL of 1M IPTG solution into 1 L cell culture (overall concentration 1.0 mM of IPTG). The solution was shaken at 16 ̊C for 36 hours. Q108K:T51D:A28C and Q108R:T51D were expressed overnight at 23 ̊ C and 16 ̊ C respectively. II-11-7. Protein isolation and purification The cells were harvested by centrifuging for 20 min at 5,000 rpm at 4 ̊C. After removal of the supernatant, the cells were resuspended in 60 mL of Tris buffer (10 mM Tris. HCl, pH=8.0). Then the resuspended cells lysed by ultrasonication (Power 80%, 1 min x 3) and centrifuged at 4 ̊C (10,000 rpm, 20 min). The ion exchange chromatography Q SepharoseTM Fast Flow resin, was used for purification of proteins at 4 ̊C. Further purification was performed with Fast Protein Liquid 118 Chromatography. In the first step of purification with Q Sepharose Fast Flow resin (GE Health Sciences), the bound protein was washed with 10 mM Tris.HCl, pH = 8.0 and then eluted with 10 mM Tris. HCl, 150 mM NaCl, pH = 8.0. The pure fractions desalted using concentrators against 10 mM Tris pH=8.0 buffer, and loaded on a second anion exchange column (15Q, GE Health Sciences, BioLogic DuoFlow system). The fractions were separately concentrated at 4°C to 8 mg/mL and further purified by size exclusion chromatography (Superdex 75 16/760 column) in a buffer containing 10 mM Tris. HCl, 150 mM NaCl, pH = 8.0. 119 II-11-8. Crystallization, data collection, and refinement Each mutant protein was concentrated to between 6 to 8 mg/mL in a buffer containing 10 mM Tris.HCl, 150 mM NaCl, pH = 8.0. Crystals were grown using the hanging drop vapor diffusion method using 1 μL of protein solution and 1 μL of crystallization solution in the drop and 1 mL of crystallization solution in the reservoir. Q108K:K40L:T51F and Q108K:K40L:T51W apo crystals were obtained in the presence of 2 eq. retinol. However, the overall conformations of these mutants show the normal apo symmetrical dimer, and no evident electron density for retinol has been observed. For the crystallization of holo DS dimer, proteins were incubated with 4 equivalents of retinal. All of the crystallization trials with retinal were wrapped in aluminum foil to protect the light sensitive chromophore. The crystallization was done under following conditions: solution of 25-30% PEG 4000, 0.1 M sodium acetate pH=4.0-4.8, 0.1 M ammonium acetate. The crystals were soaked in a cryoprotectant solution (30% PEG 4000, 0.1 M sodium acetate pH = 4.5, 0.1 M ammonium acetate with 20% glycerol), flash frozen in loops in liquid nitrogen and stored in a liquid nitrogen Dewar prior to data collection. For crystallization of zinc bound Q108K:T51D:A28C:L36C:F57H, 7.8 mg/ml (252 μM) of the protein was incubated with 540 μM zinc acetate dihydrate (purchased from Fluka). After about 20 minutes of incubation, the crystallization was performed. The protein crystals grew after 2 days at 0.2 M ammonium acetate, 0.1 M Bis-Tris:HCl, pH 5.5, and 25% PEG 3350 and 0.1 M Bis-Tris:HCl, pH 5.5, 25% PEG 3350. To obtain the reduced form crystals of Q108K:T51D:A28C:L36C:F57H, the same concentration of the protein (252μM) was incubated with 540 μM of iron (II) sulfate hepta 120 hydrate for 5 minutes (purchased from J.T. Baker Chemical Company). The protein crystals grew after 2-4 days at 0.1 M Bis-Tris:HCl, pH 5.5, 25% PEG3350. Fatty acid bound crystal structures For the crystallization of Q108R:T51D and WT, 8-10 mg/ml of protein was incubated with 4 equivalents of fatty acid (either linoleic and arachidonic acid). The best crystals grew using a crystallization solution of 25-30% PEG 4000, 0.1 M sodium acetate pH=4.0-4.8, 0.1 M ammonium acetate, 28% PEG 400. A1 crystal structures The crystal structures of apo and holo retinal bound A1, as well as Q108K:T51D (~8mg/ml) was crystallized with 4 equivalent of merocyanine under 25-30% PEG 4000, and 0.1 M sodium acetate pH=4.0-4.8, 0.1 M ammonium acetate. The merocyanine bound crystal of A1 was obtained by incubation of 4 equivalent of ligand with 10 mg/ml protein under two conditions using robot screening: 0.2 M calcium acetate, 0.1 M MES:NaOH, pH=6, 20% PEG 8000 and 0.2 M calcium chloride, 0.1 M HEPES, NaOH pH=7.5. A1-K40F crystals bound with merocyanine grew at 4 M sodium formate. Hinge loop insertion mutants crystallization conditions Q108K:T51D:A28C:L36C, 1H, 6H, 3W WT-3H, and WT-6H were crystallized under 25-30% PEG 4000, 0.1 M sodium acetate pH=4.0-4.8, 0.1 M ammonium acetate. The 3H crystals grew under following conditions: 200 µl of protein (~11.7 mg/ml) was incubated with 500 µM Cobalt (II) acetate tetrahydrate and crystallized with 0.2 M ammonium acetate, 0.1 M Bis-Tris: HCl, pH 6.5, and 25% PEG 3350. The 3H’ and 6H also crystallized under 0.2 M sodium Fluoride, 20% PEG 3350. 121 For Diffraction data were collected at the Advanced Photon Source (APS) (Argonne National Laboratory IL) LS-CAT, (sector 21-ID-D, F, G) using a MAR300 and MAR350 detectors, 1.00Å wavelength radiation at 100K. Also, for the zinc bound protein anomalous data was collected above the zinc edge. The initial diffraction data were indexed, processed and scaled using the HKL2000 software package.5 The structures were solved by molecular replacement using PHASER in PHENIX and hCRBPII (PDB entry 2RCQ) as a search model.6 The initial electron density map was produced Phaser- MR in PHENIX.7, 8 Model rebuilding, placement of water molecules etc. were done using COOT.9 The structures were refined using the PHENIX program packages. The chromophore was built using the JLIGAND program to generate restraints.10,11 II-11-9. UV-vis measurements UV-vis spectroscopy was performed using a Cary 300 and Cary 150 Bio WinUV UV-Vis spectrophotometer (Varian Instruments). The samples were prepared in phosphate buffer. The experiment was performed using 0.5 equivalents of retinoid. The protonated Schiff Base (PSB) has Absorptions corresponding to λmax> 450 nm, while deprotonated imine peaks (SB) shows up at λmax ~368 nm. The unbound retinal absorbs at λmax~380 nm. All-trans-retinal (purchased from Sigma, ε = 48,000 M-1 cm-1, λmax =380 nm.), Q108K:T51D ε280nm =57,050 M-1cm-1, Q108K:K40L:T51F ε280nm= 64,540 M-1cm-1; Q108K:K40L:T51F=75,400 M-1cm-1; Q108K:K40D ε280nm=58,690 M-1cm-1; Q108K:T51D:A28C:L36C:F57H; ε280nm = 51,485 M-1cm-1; Q108K:T51D:A28C ε280nm = 56,720 M-1cm-1, Q108K:K40L:Y60L:A28C ε280nm =49,397 M-1cm-1. 122 II-11-10. Fluorescence quenching assay The assay was done as previously reported.76 0.01% gelatin containing PBS (4 mM NaH2PO4, 16 mM Na2HPO4, 150 mM NaCl, pH=7.3) was used as a buffer. 3 mL of 500 nM protein in PBS buffer was added to the cell. The sample was excited at 283 nm with a slit width about 1.5 nm. The fluorescence is measured at the peak maximum, about 345 nm. The chromophore was added at different equivalents from a ~1.5 mM stock (in DMSO) sample maintained in the dark. The results are plotted as concentration of chromophore versus relative fluorescence intensity. II-11-11. CD spectroscopy CD spectra were recorded on a JASCO J-810 spectropolarimeter. The protein concentrations were adjusted in PBS buffer (purchased from Sigma Aldrich). The ethanolic stock solution of retinal and PBS stock solution of DTT were prepared. The CD changes (mdeg) were monitored upon adding retinal and DTT by using CD spectrometer. Protein concentrations were as follows: Q108K:T51D (dimer): 4.76 µM, Q108K:K40L:T51V (monomer): 5.30 µM, Q108K:T51D:A28C (dimer): 4.05 µM, Q108K:K40L:T51F:Y60A (dimer): 5.6 µM, and Q108K:K40L:Y60L:A28C (dimer): 2.3 µM. II-11-12. Radioactive 65ZnCl2 labeling assay For ultrafiltration zinc-binding experiments, radioactive 65ZnCl2 (radionuclide purity: 99%) was purchased from PerkinElmer and was diluted with buffer and non- radioactive ZnCl2 (10 mM Tris. HCl/150 mM NaCl, pH 7.0) to make a radioactive 259 µM ZnCl2 stock solution. 123 Samples were prepared in 1.5 mL centrifugation tubes to be 250 µM total and contain 50.0 µM protein and varying concentrations of 65ZnCl2 in buffer. To create a saturation- binding curve, the 65Zn2+ concentrations were varied from 10-100 µM. For statistical comparisons of the relative binding affinities between different modifications of proteins for 65Zn2+, samples were made with 40 µM 65Zn2+ and 50 µM protein, with three to five replicates of each. Then, 220 µL of each sample was pipetted into an ultrafiltration centrifugation device with a 10 kDa MWCO size-exclusion membrane. The samples were centrifuged at 8000 rpm for 3 minutes to allow ~20 µL of ultrafiltrate to collect in the bottom of the tube, the concentration of which is considered the free-Zn2+ in the sample. The volume separated from the sample is low enough so as to not disturb the equilibrium of the binding system. Then, a 15 µL aliquot of ultrafiltrate sample was transferred into a separate tube for analysis. Calibration standards containing 15 µL of known Zn2+ concentrations were also prepared in separate tubes. Then, all standards and samples were analyzed by counting the emitted gamma radiation at (1115 keV) for 5 minutes per sample on a PerkinElmer 2480 Automatic Gamma Counter. The concentration of Zn2+ in the ultrafiltrate was quantified by comparing the counts-per-minute (CPM) for each sample to an external standards calibration curve. The concentration of Zn2+ in the ultrafiltrate was considered to be the free-Zn2+ concentration in the sample, while the bound-Zn2+ concentration was calculated as the free-Zn2+ subtracted from the total- Zn2+. For the saturation-binding curve, free-Zn2+ vs bound-Zn2+ was plotted and analyzed via non-linear regression software (SigmaPlot 13.0) to calculate the equilibrium binding constant (Kd) and stoichiometry (Bmax). 124 For comparing the different binding affinities between proteins, the global affinity constant, or nKa of the binding system, was calculated from the free concentration of ligand. Reduced AG1 (Red-apo) was prepared by adding 1mM BME in the buffer (10mM Tris.HCl, 150 mM NaCl, pH=7.0). Holo AG1 (Holo) was obtained by adding 2 equivalents of ethanolic solution of all-trans retinal to AG1. Holo/reduced AG1 (Red-holo) then was obtained by adding 2 equivalents of ethanolic solution of retinal to the reduced AG1. II-11-13. ITC experiment Isothermal titration calorimetry (ITC) experiments were carried out using a PEAQ- ITC calorimeter (Malvern). To determine the Kd of apo AG1 to Zn+2, 1mM zinc acetate was titrated into the cell containing ~50 µM AG1 proteins. Titration experiments were performed at 25 °C. Both the protein and ligand solutions were centrifuged for 2 min at 10,000 g before each titration. In a typical experiment, the reference power was set at 10 μcal/sec, an initial 0.2 μl injection (disregarded in analyses) was followed by 19, 2-μl injections of titrant (Δt = 2.5 min). Experiments and controls with each combination were repeated at least once. The thermograms were blank-subtracted and analyzed using the MicroCal PEAQ ITC analysis software and fitted to single-site binding models. 125 Table II-6. X-ray crystallography data and refinement statistics of the holo dimer Q108K:T51D, Q108K:K40L:T51F, Q108K:K40L:T51W structures. Q108K:T51D Q108K:K40L:T51F Q108K:K40L:T51W P 21 36.33 63.14 118.22 90 96 90 4 1638156 29781 96.25 (89.85) a 13.6 (2.2)a 4.5 (40.0)a 39.17-2.11 (2.183-2.11)a 19.70/27.13 0.008 1.19 38.3 228 6MKV Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflections Unique Completeness(% Reflection Average I/s ) Rmerge(%) Resolution (Å) (Last Shell) P212121 66.19 73.79 353.40 90 90 90 12 4184309 99384 91.86 (82.88)a 22 5 (80)a 45.46 - 2.061 (2.13 - 2.06)a P 21 36.48 63.28 117.78 90 96 90 4 1580011 2164 98.1(96.6)a 32.2(2.5)a 6.8 (40.0) a 14.87 - 2.17 (1.85 - 1.80) a Rwork/ Rfree (%) 22.23/27.35 20.49/ 29.13 RMSD from ideal value Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID 0.009 1.13 40.00 650 6E5S 0.008 1.12 51.40 174 6MLB a Values in the parenthesis refer to the last resolution shell. 126 Table II-7. X-ray crystallographic data and refinement statistics for holo dimer hCRBPII Q108K:K40L:T51F:Y60A Q108K:T51D:A28H, hCRBPII-WT, and Q108K:T51D:A28C mutants. Q108K:K40L:T51 Q108K:T51D: hCRBPII WT Q108K:T51D: F:Y60A A28H P 2121 2 36.57 60.55 63.76 90 90 90 1 P212121 65.87 73.99 349.99 90 90 90 12 846912 11665 1539077 49017 99.72 (99.38)a 90.86 ( 90.57)a P212121 65.82 74.4 352.55 90 90 90 12 6589507 24088 83.47 (91.81) 13 (2.20)a 11.2 (93.5) Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness Average I/s Rmerge(%) (%) Resolution (Å) A28C P212121 67.74 72.06 112.25 90 90 90 4 1021462 15658 99.96 (100.00) 11.5 (2.0)a 14.6 (88.5) 9.36 - 2.70 (2.80 - 2.70)a 21.99-29.23 0.004 0.89 40.08 39 16.1(2.2)a 6.0 (74) 31.88 -1.90 12 (2.07) 9.8 (69) 14.99 - 2.60 14.99 - 3.30 (2.69 - 2.60)a 21.0/30.04 (3.42 - 3.30)a 25.7/30.52 0.009 1.11 38.5 171 0.002 0.51 74.6 89 (Last Shell) (1.97 - 1.90)a Rwork/ Rfree (%) 21.04/25.59 RMSD from ideal value Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID 0.007 0.82 44.16 43 6E6L 6MCU 6MCV 6E7M a Values in the parenthesis refer to the last resolution shell. 127 Table II-8. X-ray crystallographic data and refinement statistics for apo dimer hCRBPII Q108K:T51D, Q108K:K40L:T51F, Q108K:K40L:T51W, and Q108K:T51D:A28H. Q108K:T51D Q108K:K40L :T51F Q108K:K40L: Q108K:T51D :A28H P212121 60.218 63.757 36.739 90 90 90 1 P 21 212 63.80 36.72 60.51 90 90 90 1 T51W P 21 212 36.89 60.86 64.16 90 90 90 1 P 21 212 36.7 60.27 63.40 90 90 90 1 Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge (%) Resolution (Å) (Last Shell) Rwork/ Rfree (%) RMSD from ideal value Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID 540742 16231 1023617 19468 1145094 6023 195446 101134 99.71 (97.45)a 99.90 99.90 (100.00)a 99.91(99.30)a 32 (1.0) 8.5 (94) 36.74 - 1.70 (1.76 1.70)a 19.58/23.30 0.012 1.31 36.70 60 6E5E (100.00)a 23.8 (2.8)a 5.7(43) 28.12 - 1.96 (2.036 - 1.96) 20.30/26.88 )a 0.007 1.05 38.20 86 6E50 11(2.9)a 9.1 (53) 32.08 - 2.4 (2.49- 2.40)a 24.94/17.8 25 (2.3)a 8.9 (73) 36.73 - 1.99 (2.06-1.99)a 21.74/27.64 0.008 1.12 48.01 49 6E51 0.008 0.92 34.23 30 6E5Q a Values in the parenthesis refer to the last resolution shell. 128 Table II-9. X-ray crystallographic data and refinement statistics for apo dimer Q108K:T51D:A28C, zinc-bound Q108K:T51D:A28C:L36C:F57H, and reduced Q108K:T51D:A28C:L36C:F57H hCRBPII mutants. Q108K:T51D:A28C Q108K:T51D:A28C: L36C:F57H Zinc bound Q108K:T51D: A28C:L36C:F57H Reduced form P212121 60.42 61.76 72.66 90 90 90 2 4246024 32802 (3193) 96.19 (95.68) 22 (2.30) 15.4 (180) 37.12 - 1.64 (1.70 - 1.64)a 20.30/24.29 0.007 0.93 25.50 211 TBD P 2 2121 36.97 60.55 63.78 90 90 90 1 366691 5891 (501) 99.5(99.7) 22.6 (2.30) 6.6 (47) 31.98 - 2.40 (2.48 - 2.40) a 18.29/25.92 0.007 1.03 57.39 12 TBD Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge (%) Resolution (Å) (Last Shell) P212121 30.86 68.18 130.04 90 90 90 2 685586 8997 99.94 (100.00)a 10.63 (2.30) 18 (81) 36.58 - 2.6 (2.69 - 2.6)a Rwork/ Rfree (%) 24.18/30.54 RMSD from ideal value Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID 0.003 0.75 26.70 6E5R a Values in the parenthesis refer to the last resolution shell. 129 Table II-10. X-ray crystallography data and refinement statistics of the apo dimer 1H, 3H, 6H, and 3W mutants of hCRBPII. 3W 1H 3H 6H P212121 35.80 76.58 110.52 90 96 90 2 I4 140.65 140.65 63.95 90 90 90 4 580019 10638 (947) 98.48 (91.59) a 725079 25561(2499) 99.83 (98.85)a 13.76 (2.01) 9.0 (45.8) 36.18 - 2.52 (2.612 - 2.52) 30.2 (2.81) 8.8 (70.1) 44.48-2.37 (2.45 - 2.37) 18.13/25.07 20.08/28.01 0.008 0.93 37.35 44 TBD 0.009 0.94 50.07 35 TBD P212121 39.71 56.58 136.197 90 90 90 2 Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflections Unique Reflection 22274 (2048) 98.15 (91.58)a Completeness (%) Average I/s Rmerge (%) Resolution(Å) (Last Shell) 33.3 (2.52) 10.2 (93.8) 38.12-1.96 (2.03 -1.96) 2496404 P212121 36.93 69.98 106.34 90 90 90 2 1156335 21400 (2032) 98.51(95.71)a 21.7 (2.4) 7.3 (68.6) 33.24- 1.92 (1.991 - 1.92) 22.26/27.15 20.57/24.92 Rwork/ Rfree (%) RMSD from ideal value Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID a Values in the parenthesis refer to the last resolution shell. 0.007 0.81 37.80 67 0.009 1.01 33.01 52 TBD TBD 130 Table II-11. X-ray crystallographic data and refinement statistics for apo hCRBPII 3H’, WT-3H, WT-6H, and Q108K:T51D:A28C:L36C hCRBPII mutants. 3H’ WT-3H WT-6H Q108K:T51D: A28C:L36C P21 36.50 60.12 63.07 90.00 93.70 90.00 2 851776 17684 (1739) 92.63 (91.57) 18.03 (2.8) 7.0 (42.0) 31.47-1.98 (2.052 - 1.98) 24.86/29.77 0.008 1.05 31.30 30 TBD Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge (%) Resolution (Å) (Last Shell) Rwork/ Rfree (%) Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID P 2121 21 36.95 69.90 107.25 90.00 90.00 90.00 2 3054247 14150 (1272) 94.80 (87.36) 11.5 (2.8) 7.5 (26.4) 34.94 - 2.19 (2.27 - 2.19) 19.84/27.17 0.008 0.84 23.86 53 P21 48.52 47.81 53.22 90.00 92.61 90.00 2 P1 29.61 36.93 68.51 73.44 83.92 66.06 2 728017 2700610 55387 (5299) 101926 (9598) 96.44 (92.43) 93.17 (87.33) 3.3 (2.1) 6.2 (62.1) 29.09 - 1.319 (1.37 - 1.32) 113 (3.3) 6.0 (50.7) 27.06- 1.08 (1.12 - 1.08) 21.68/19.42 0.005 0.77 17.17 339 21.35/20.70 0.005 0.80 15.10 295 TBD TBD TBD a Values in the parenthesis refer to the last resolution shell. 131 Table II-12. X-ray crystallographic data and refinement statistics for domain-swapped dimer of A1 mutant in apo, retinal bound, and merocyanine bound forms. A1/retinal A1/merocyanine A1-Apo Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge (%) Resolution (Å) (Last Shell) Rwork/ Rfree (%) Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID P2121 2 60.78 64.03 36.80 90 90 90 1 765009 4903 (477) 99.94 (99.79) 24.7 (2.44) 10 (81) P2121 21 65.51 73.80 350.12 90 90 90 12 2700610 55387 (5299) 96.44 (92.43) 13.8 (1.6) 15 (180) P2121 21 60.15 60.63 76.95 90 90 90 2 1971397 25564 (2465) 97.53 (95.35) 25.2 (2.3) 11.3 (144.2) 36.8 - 2.57 48.99-3.031 47.39-1.81 (2.66 - 2.57) (3.14-3.03) (1.88 - 1.81) 21.78/27.92 23.55/32.56 18.84/22.98 0.008 1.09 51.82 10 TBD 0.006 0.62 59.01 4 TBD 0.008 0.93 25.03 233 TBD a Values in the parenthesis refer to the last resolution shell. 132 II-13. X-ray crystallographic data and refinement statistics for Table Q108R:T51D mutant bound with fatty acids. Q108R:T51D Q108R:T51D/ Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge (%) Resolution (Å) (Last Shell) Rwork/ Rfree (%) Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID /Linoleic acid Arachidonic acid P2121 21 64.425 72.802 353.76 90 90 90 12 3432087 57772 (5222) 86.83 (79.48) 24.7 (13.8) 7.0 (51.2) 43.49 - 2.40 P2121 21 63.90 72.16 353.50 90 90 90 12 5903084 55387 (5299) 96.44(92.43) 33.63 (2.15) 78 (183) 47.84 -1.96 (2.48 - 2.40) 19.42/26.91 0.008 (2.034 - 1.96) 22.21/26.34 0.008 0.98 36.25 134 TBD 0.93 39.72 22 TBD a Values in the parenthesis refer to the last resolution shell. 133 REFERENCES 134 REFERENCES Ha, J. H.; Loh, S. N., Protein conformational switches: from nature to design. Andrianantoandro, E.; Basu, S.; Karig, D. K.; Weiss, R., Synthetic biology: new Raman, S., Systems approaches to understanding and designing allosteric Fenton, A. W., Allostery: an illustrated definition for the 'second secret of life'. 1. Trends in Biochemical Sciences 2008, 33 (9), 420-5. 2. proteins. Biochemistry 2018, 57 (4), 376-382. 3. Chemistry (Weinheim an der Bergstrasse, Germany) 2012, 18 (26), 7984-99. 4. engineering rules for an emerging discipline. Molecular Systems Biology 2006, 2 (1). 5. Silvaroli, J. A.; Arne, J. M.; Chelstowska, S.; Kiser, P. D.; Banerjee, S.; Golczak, M., Ligand binding induces conformational changes in human cellular retinol-binding protein 1 (CRBP1) revealed by atomic resolution crystal structures. The Journal of Biological Chemistry 2016, 291 (16), 8528-40. 6. Meister, G. E.; Joshi, N. S., An engineered calmodulin-based allosteric switch for Peptide biosensing. Chembiochem : a European journal of chemical biology 2013, 14 (12), 1460-7. 7. Ke, W.; Laurent, A. H.; Armstrong, M. D.; Chen, Y.; Smith, W. E.; Liang, J.; Wright, C. M.; Ostermeier, M.; van den Akker, F., Structure of an Engineered β-Lactamase Maltose Binding Protein Fusion Protein: Insights into Heterotropic Allosteric Regulation. PLOS ONE 2012, 7 (6), e39168. 8. Choi, J. H.; Zayats, M.; Searson, P. C.; Ostermeier, M., Electrochemical activation of engineered protein switches. Biotechnology and Bioengineering 2016, 113 (2), 453- 456. 9. DeGrave, A. J.; Ha, J.-H.; Loh, S. N.; Chong, L. T., Large enhancement of response times of a protein conformational switch by computational design. Nature Communications 2018, 9 (1), 1013. 10. Kim, J. R.; Ostermeier, M., Modulation of effector affinity by hinge region mutations also modulates switching activity in an engineered allosteric TEM1 β-lactamase switch. Archives of Biochemistry and Biophysics 2006, 446 (1), 44-51. 11. Banala, S.; Aper, S. J. A.; Schalk, W.; Merkx, M., Switchable Reporter Enzymes Based on Mutually Exclusive Domain Interactions Allow Antibody Detection Directly in Solution. ACS Chemical Biology 2013, 8 (10), 2127-2132. 135 12. Teasley Hamorsky, K.; Ensor, C. M.; Wei, Y.; Daunert, S., A Bioluminescent Molecular Switch For Glucose. Angewandte Chemie International Edition 2008, 47 (20), 3718-3721. 13. Dueber, J. E.; Yeh, B. J.; Chak, K.; Lim, W. A., Reprogramming Control of an Allosteric Signaling Switch Through Modular Recombination. Science 2003, 301 (5641), 1904-1908. 14. Guntas, G.; Mansell, T. J.; Kim, J. R.; Ostermeier, M., Directed evolution of protein switches and their application to the creation of ligand-binding proteins. Proceedings of the National Academy of Sciences of the United States of America 2005, 102 (32), 11224- 11229. 15. Ostermeier, M., Engineering allosteric protein switches by domain insertion. Protein Engineering, Design and Selection 2005, 18 (8), 359-364. 16. Baird, G. S.; Zacharias, D. A.; Tsien, R. Y., Circular permutation and receptor insertion within green fluorescent proteins. Proceedings of the National Academy of Sciences of the United States of America 1999, 96 (20), 11241-6. 17. Grünewald, F. S., Periplasmic binding proteins in biosensing applications. In Advances in Chemical Bioanalysis, Matysik, F.-M., Ed. Springer International Publishing: Cham, 2014; pp 205-235. 18. Akerboom, J.; Carreras Calderon, N.; Tian, L.; Wabnig, S.; Prigge, M.; Tolo, J.; Gordus, A.; Orger, M. B.; Severi, K. E.; Macklin, J. J.; Patel, R.; Pulver, S. R.; Wardill, T. J.; Fischer, E.; Schuler, C.; Chen, T. W.; Sarkisyan, K. S.; Marvin, J. S.; Bargmann, C. I.; Kim, D. S.; Kugler, S.; Lagnado, L.; Hegemann, P.; Gottschalk, A.; Schreiter, E. R.; Looger, L. L., Genetically encoded calcium indicators for multi-color neural activity imaging and combination with optogenetics. Frontiers in Molecular Neuroscience 2013, 6, 2. 19. Opinion in Structural Biology 2006, 16 (4), 525-30. 20. Mukherji, S.; van Oudenaarden, A., Synthetic biology: understanding biological design from synthetic circuits. Nature Reviews Genetics 2009, 10, 859. 21. Lormeau, C.; Rybiński, M.; Stelling, J., Multi-objective design of synthetic biological circuits**This publication is supported by the Swiss National Science Foundation as part of the NCCR Molecular Systems Engineering. IFAC-PapersOnLine 2017, 50 (1), 9871- 9876. 22. Miyawaki, A.; Llopis, J.; Heim, R.; McCaffery, J. M.; Adams, J. A.; Ikura, M.; Tsien, R. Y., Fluorescent indicators for Ca2+ based on green fluorescent proteins and calmodulin. Nature 1997, 388 (6645), 882-887. Ambroggio, X. I.; Kuhlman, B., Design of protein conformational switches. Current 136 Koide, S.; Huang, J., Generation of high-performance binding proteins for peptide 23. Sulmann, S.; Dell'Orco, D.; Marino, V.; Behnen, P.; Koch, K.-W., Conformational changes in calcium-sensor proteins under molecular crowding conditions. Chemistry – A European Journal 2014, 20 (22), 6756-6762. 24. motifs by affinity clamping. Methods Enzymology 2013, 523, 285-302. 25. Marvin, J. S.; Schreiter, E. R.; Echevarría, I. M.; Looger, L. L., A genetically encoded, high-signal-to-noise maltose sensor. Proteins: Structure, Function, and Bioinformatics 2011, 79 (11), 3025-3036. 26. Motlagh, H. N.; Wrabl, J. O.; Li, J.; Hilser, V. J., The ensemble nature of allostery. Nature 2014, 508 (7496), 331-9. 27. Garcia-Pino, A.; Balasubramanian, S.; Wyns, L.; Gazit, E.; De Greve, H.; Magnuson, R. D.; Charlier, D.; van Nuland, N. A.; Loris, R., Allostery and intrinsic disorder mediate transcription regulation by conditional cooperativity. Cell 2010, 142 (1), 101-11. 28. Mitrea, D. M.; Parsons, L. S.; Loh, S. N., Engineering an artificial zymogen by alternate frame protein folding. Proceedings of the National Academy of Sciences the United States of America 2010, 107 (7), 2824-2829. 29. Ha, J.-H.; Loh, S. N., Construction of Allosteric Protein Switches by Alternate Frame Folding and Intermolecular Fragment Exchange. In Synthetic Protein Switches: Methods and Protocols, Stein, V., Ed. Springer New York: New York, NY, 2017; pp 27- 41. 30. Do, K.; Boxer, S. G., GFP variants with alternative β-strands and their application as light-driven protease sensors: a tale of two tails. Journal of the American Chemical Society 2013, 135 (28), 10226-9. 31. Ha, J.-H.; Karchin, J. M.; Walker-Kopp, N.; Castañeda, C. A.; Loh, S. N., Engineered domain swapping as an on/off Switch for protein function. Chemistry & Biology 2015, 22 (10), 1384-1393. 32. Stein, V.; Alexandrov, K., Protease-based synthetic sensing and signal amplification. Proceedings of the National Academy of Sciences of the United States of America 2014, 111 (45), 15934-9. 33. Greenwood, C.; Ruff, D.; Kirvell, S.; Johnson, G.; Dhillon, H. S.; Bustin, S. A., Proximity assays for sensitive quantification of proteins. Biomolecular Detection and Quantification 2015, 4, 10-6. 34. Canto, I.; Soh, U. J.; Trejo, J., Allosteric modulation of protease-activated receptor signaling. Mini Reviews in Medicinal Chemistry 2012, 12 (9), 804-11. 35. Deisseroth, K., Optogenetics. Nature Methods 2011, 8 (1), 26-29. 137 Beharry, A. A.; Woolley, G. A., Azobenzene photoswitches for biomolecules. Khersonsky, O.; Fleishman, S. J., Incorporating an allosteric regulatory site in an 36. Wang, W.; Wildes, C. P.; Pattarabanjird, T.; Sanchez, M. I.; Glober, G. F.; Matthews, G. A.; Tye, K. M.; Ting, A. Y., A light- and calcium-gated transcription factor for imaging and manipulating activated neurons. Nature Biotechnology 2017, 35 (9), 864- 871. 37. Pudasaini, A.; El-Arab, K. K.; Zoltowski, B. D., LOV-based optogenetic devices: light-driven modules to impart photoregulated control of cellular signaling. Frontiers in Molecular Biosciences 2015, 2, 18. 38. Anderson, S.; Dragnea, V.; Masuda, S.; Ybe, J.; Moffat, K.; Bauer, C., Structure of a novel photoreceptor, the BLUF domain of AppA from Rhodobacter sphaeroides. Biochemistry 2005, 44 (22), 7998-8005. 39. Glock, C.; Nagpal, J.; Gottschalk, A., Microbial Rhodopsin Optogenetic Tools: Application for Analyses of Synaptic Transmission and of Neuronal Network Activity in Behavior. Methods in Molecular Biology (Clifton, N.J.) 2015, 1327, 87-103. 40. Chemical Society Reviews 2011, 40 (8), 4422-37. 41. antibody through backbone design. Protein Science 2017, 26 (4), 807-813. 42. Churchfield, L. A.; Medina-Morales, A.; Brodin, J. D.; Perez, A.; Tezcan, F. A., De novo design of an allosteric metalloprotein assembly with strained disulfide bonds. Journal of the American Chemical Society 2016, 138 (40), 13163-13166. 43. Mitternacht, S.; Berezovsky, I. N., Binding leverage as a molecular basis for allosteric regulation. PLOS Computational Biology 2011, 7 (9), e1002148. 44. Burger, W. A. C.; Sexton, P. M.; Christopoulos, A.; Thal, D. M., Toward an understanding of the structural basis of allostery in muscarinic acetylcholine receptors. The Journal of General Physiology 2018, 150 (10), 1360-1372. 45. Schaap, F. G.; van der Vusse, G. J.; Glatz, J. F., Evolution of the family of intracellular lipid binding proteins in vertebrates. Molecular and Cellular Biochemistry 2002, 239 (1-2), 69-77. 46. Vaezeslami, S.; Mathes, E.; Vasileiou, C.; Borhan, B.; Geiger, J. H., The Structure of Apo-wild-type cellular retinoic acid binding protein II at 1.4 Å and its relationship to ligand binding and nuclear translocation. Journal of Molecular Biology 2006, 363 (3), 687- 701. 47. Nossoni, Z.; Assar, Z.; Yapici, I.; Nosrati, M.; Wang, W.; Berbasova, T.; Vasileiou, C.; Borhan, B.; Geiger, J., Structures of holo wild-type human cellular retinol-binding 138 Bennett, M. J.; Eisenberg, D., The evolving role of 3D domain swapping in proteins. protein II (hCRBPII) bound to retinol and retinal. Acta Crystallographica. Section D, Biological Crystallography 2014, 70 (Pt 12), 3226-32. 48. Assar, Z.; Nossoni, Z.; Wang, W.; Santos, E. M.; Kramer, K.; McCornack, C.; Vasileiou, C.; Borhan, B.; Geiger, J. H., Domain-swapped dimers of intracellular lipid- binding proteins: evidence for ordered folding intermediates. Structure 2016, 24 (9), 1590- 8. 49. Sanson, B.; Wang, T.; Sun, J.; Wang, L.; Kaczocha, M.; Ojima, I.; Deutsch, D.; Li, H., Crystallographic study of FABP5 as an intracellular endocannabinoid transporter. Acta Crystallographica Section D 2014, 70 (2), 290-298. 50. Structure 2004, 12 (8), 1339-1341. Liang, J. Y.; Lipscomb, W. N., Binding of substrate CO2 to the active site of human 51. carbonic anhydrase II: a molecular dynamics study. Proceedings of the National Academy of Sciences of the United States of America 1990, 87 (10), 3675-9. 52. Das, U. N., Biological significance of essential fatty acids. The Journal of the Association of Physicians of India 2006, 54, 309-19. 53. Gillilan, R. E.; Ayers, S. D.; Noy, N., Structural basis for activation of fatty acid- binding protein 4. Journal of Molecular Biology 2007, 372 (5), 1246-60. 54. Napoli, J. L., Cellular retinoid binding-proteins, CRBP, CRABP, FABP5: Effects on retinoid metabolism, function and related diseases. Pharmacology & Therapeutics 2017, 173, 19-33. 55. Yapici, I.; Lee, K. S.; Berbasova, T.; Nosrati, M.; Jia, X.; Vasileiou, C.; Wang, W.; Santos, E. M.; Geiger, J. H.; Borhan, B., "Turn-on" protein fluorescence: in situ formation of cyanine dyes. Journal of the American Chemical Society 2015, 137 (3), 1073-80. 56. Gilson, A. I.; Marshall-Christensen, A.; Choi, J.-M.; Shakhnovich, E. I., The Role of Evolutionary Selection in the Dynamics of Protein Structure Evolution. Biophysical Journal 2017, 112 (7), 1350-1365. 57. Science 2002, 11 (6), 1285-1299. 58. Mechansim for Oligomer Assembly. Protein Science 1995, 4 (12), 2455-2468. 59. Nandwani, N.; Surana, P.; Udgaonkar, J. B.; Das, R.; Gosavi, S., Amino-acid composition after loop deletion drives domain swapping. Protein Science 2017, 26 (10), 1994-2002. Liu, Y.; Eisenberg, D., 3D domain swapping: As domains continue to swap. Protein Bennett, M. J.; Schlunegger, M. P.; Eisenberg, D., 3D DOMAIN SWAPPING - A 139 60. Green, S. M.; Gittis, A. G.; Meeker, A. K.; Lattman, E. E., One-step evolution of a dimer from a monomeric protein. Nature Structural Biology 1995, 2 (9), 746-751. 61. Rousseau, F.; Schymkowitz, J. W. H.; Wilkinson, H. R.; Itzhaki, L. S., Three- dimensional domain swapping in p13suc1 occurs in the unfolded state and is controlled by conserved proline residues. Proceedings of the National Academy of Sciences of the United States of America 2001, 98 (10), 5596-5601. 62. Seeliger, M. A.; Schymkowitz, J. W. H.; Rousseau, F.; Wilkinson, H. R.; Itzhaki, L. S., Folding and association of the human cell cycle regulatory proteins ckshs1 and ckshs2. Biochemistry 2002, 41 (4), 1202-1210. 63. Parker, M. J.; Dempsey, C. E.; Hosszu, L. L. P.; Waltho, J. P.; Clarke, A. R., Topology, sequence evolution and folding dynamics of an immunoglobulin domain. Nature Structural Biology 1998, 5 (3), 194-198. 64. Arndt, K. M.; Muller, K. M.; Pluckthun, A., Factors influencing the dimer to monomer transition of an antibody single-chain Fv fragment. Biochemistry 1998, 37 (37), 12918- 12926. 65. Rousseau, F.; Schymkowitz, J. W. H.; Itzhaki, L. S., The unfolding story of three- dimensional domain swapping. Structure 2003, 11 (3), 243-251. 66. engineering by domain insertion. Methods in Enzymology 2013, 523, 369-388. 67. retinoic acid binding protein-II. Molecular Cell 2005, 18 (3), 343-53. 68. Gillilan, R. E.; Ayers, S. D.; Noy, N., Structural Basis for Activation of Fatty Acid- binding Protein 4. Journal of Molecular Biology 2007, 372 (5), 1246-1260. 69. Retinoic Acid Binding Protein-II. Molecular Cell 2005, 18 (3), 343-353. 70. Nossoni, Z.; Assar, Z.; Yapici, I.; Nosrati, M.; Wang, W.; Berbasova, T.; Vasileiou, C.; Borhan, B.; Geiger, J., Structures of holo wild-type human cellular retinol-binding protein II (hCRBPII) bound to retinol and retinal. Acta Crystallographica Section D 2014, 70 (12), 3226-3232. 71. Folli, C.; Calderone, V.; Ramazzina, I.; Zanotti, G.; Berni, R., Ligand Binding and Structural Analysis of a Human Putative Cellular Retinol-binding Protein. Journal of Biological Chemistry 2002, 277 (44), 41970-41977. 72. dynamics of for conformational changes. Protein Engineering 1995, 8 (11), 1129-1135. Aalten, D. M. F. v.; Findlay, J. B. C.; Amadei, A.; Berendsen, H. J. C., Essential ligand-induced Sessler, R. J.; Noy, N., A Ligand-Activated Nuclear Localization Signal in Cellular the cellular retinol-binding protein evidence Kanwar, M.; Wright, R. C.; Date, A.; Tullman, J.; Ostermeier, M., Protein switch Sessler, R. J.; Noy, N., A ligand-activated nuclear localization signal in cellular 140 73. Silvaroli, J. A.; Arne, J. M.; Chelstowska, S.; Kiser, P. D.; Banerjee, S.; Golczak, M., Ligand binding induces conformational changes in human cellular retinol-binding protein 1 (CRBP1) revealed by atomic resolution crystal structures. Journal of Biological Chemistry 2016, 291 (16), 8528-8540. 74. Malpeli, G.; Stoppinf, M.; Zapponi, M. C.; Folli, C.; Berni, R., Interactions with retinol and retinoids of bovine cellular retinol-binding protein. European Journal of Biochemistry 1995, 229 (2), 486-493. 75. Wang, W.; Nossoni, Z.; Berbasova, T.; Watson, C. T.; Yapici, I.; Lee, K. S.; Vasileiou, C.; Geiger, J. H.; Borhan, B., Tuning the electronic absorption of protein- embedded all-trans-retinal. Science 2012, 338 (6112), 1340-3. Vasileiou, C.; Lee, K. S. S.; Crist, R. M.; Vaezeslami, S.; Goins, S. M.; Geiger, J. 76. H.; Borhan, B., Dissection of the critical binding determinants of cellular retinoic acid binding protein II by mutagenesis and fluorescence binding assay. Proteins: Structure, Function, and Bioinformatics 2009, 76 (2), 281-290. 141 CHAPTER III: DISCOVERY OF A DOMAIN-SWAPPED TRIMER AS A NOVEL FOLD FOR THE ILBP FAMILY III-1. Oligomerization in biology: an efficient regulator for many biological processes Oligomerization is perhaps one of the most efficient natural tools to create functionally distinct proteins from a common fold through evolutionary mechanisms.1 Oligomerization manifests its critical roles in many biological processes, such as signaling,2-3 enzymatic activities,4 and cargo transportation,5-6 where the self-associations of the proteins trigger different biological pathways. There are many services the oligomerization of proteins offers in cellular events, such as enhancing protein stability, inventing novel active sites or offering specificity to existing active centers, tuning the accessibility of proteins, etc.1, 7-10 Although oligomerization provides incredible advantages and diversity in protein functions, in some cases, it can also lead to protein aggregation and unwanted pathogenic frameworks.11-12 Due to the dual aspects of oligomerization in cellular events, thorough study of the proteins subjected to these processes is crucial. However, through a simple comparison of the number of proteins that have been structurally determined so far vs. the number of known oligomeric forms that have been reported, we can conclude that there is a dearth of information in this area. The lack of biophysical data and high- resolution crystal structures of the oligomeric forms of proteins is a missing link in the investigation of the function-structure relationship in protein oligomerization.10 1 These types of studies are essential due to the nature of oligomerization that can be exploited 142 in the field of protein engineering to broaden the ability of reprogramming functions and conformations in proteins. In this chapter, we will first mention previous efforts that have been performed to understand the oligomerization states of proteins. Then we will use our structural-based protein design approach to discover and elucidate the mechanism of oligomerization in our model system using hCRBPII as a template. III-2. Domain swapping Domain-swapping, where two protein domains trade identical parts of their structures to create dimers or higher-order oligomers, is one of the mechanisms leading Figure III-1. Proposed mechanism of domain-swapping. Instead of turning and making a closed monomer, the hinge loop region elongates in the open monomer conformation. If two or more open monomers with proper C- and N- termini conformations find each other, they can generate dimers, trimers, or higher-order oligomers. to self-associations and oligomer formation in proteins (Figure III-1). 13-15 In most cases, instead of the small, polypeptide part of a protein called the “hinge loop” turning to create a closed monomer, it straightens and adopts the elongated conformation in the domain- 143 swapped from. The majority of the proteins that undergo domain-swapping merely exchange small parts of their secondary structures with identical molecules.14, 16 However, there are several examples demonstrating that half of the protein is subjected to swapping.17-18 Domain swapping can ensue in two geometrically distinct modes: first, through the mutual or cyclical fashion to produce dimers, trimers, and higher-order oligomers; and second, in an open-ended or “run-away” fashion, where each protein molecule exchanges a. closed hinge loop b. extended hinge loop primary interface monomers secondary interface domain-swapped dimer c. d. Cyclic oligomer open ended oligomer Figure III-2. Schematic demonstration of different domain-swapped forms. a. Closed monomer. The hinge loop region and primary interface are highlighted. b. Domain-swapped dimer. The extended conformation of the hinge loop and secondary interface created by two open monomers is represented. c. Cyclical oligomerization. d. Open-ended or “run-away” oligomerization. a domain with a complementary part of the neighboring molecule (Figure III-2). The latter 144 mechanism of swapping is common in fibril formation although it has not been proven to be the origin of that.14 There are many variables playing roles to favor domain swapping, such as certain mutations in the hinge loop region, lyophilization, temperature, pH of the environment, protein concentration, ligand binding, etc.1, 13, 15, 19-22 Understanding these variables is crucial for controlling the domain-swapping mechanism. Below we will discuss some of the well-known domain-swapped dimers and the factors giving rise to domain-swapping. III-3. Reviewing several important examples of domain-swapped proteins Diphteria toxin (DT), the protein that causes diphtheria, 23 is the first example of bonafide domain-swapped dimers that were structurally characterized (Figure III-3a).24 DT performs three tasks in the course of cell intoxication using three folding segments: catalytic domain (C domain), transmembrane domain (T domain), and receptor domain (R domain). DT is endocytosed from coated pins when the R domain joins the cell-surface receptor. The protein then experiences a conformational change in the low-pH environment of the endosome and is transferred to the endosomal membrane. Finally, the C domain is transferred to the cytosol, which leads to the termination of protein synthesis through the deactivation of the eukaryotic elongation factor 2 (EF-2).22, 24-26 DT was first found to undergo domain swapping at low pH, mimicking the physiological endosomic environment (Figure III-3b). Additionally, based on the crystal structure obtained from the complex formation of DT and its receptor, it was shown that the DT domain-swapped dimer exists, in the presence of receptor binding, at neutral pH as well. 25,22 This study provides one of the best examples of how the domain-swapping mechanism leads to altered the protein function in biological systems. 145 Another excellent example of 3D domain-swapping is ribonucleases (RNase A), a bovine pancreatic enzyme whose exclusive function is RNA hydrolysis.27 This protein family is dominantly monomeric (Figure III-3c), with only a few RNases (BS-RNase) found in the dimeric form in nature.28 The dimeric conformation bestows proteins with incredible biological features, such as allosteric control and an increased concentration of active sites, among others.29 The BS-RNase dimer formation prompted researchers to characterize the domain-swapped dimer of the RNase A protein. Eventually, the lyophilization of RNase A (in the presence of acetic acid) led to the formation of the RNase domain-swapped dimer. One of the interesting features of RNase-A is its ability to generate two different domain-swapped dimers. One domain-swapped dimer was formed as a result of the N-terminal α- helices swapping (Figure III-3d),20 while another domain- swapped structure was formed where the c-terminal β-strands were exchanged (Figure III-3e).30-31 Later on, Park and Raines et al. discovered N-terminal domain swapping in physiological conditions.32 Aside from the domain-swapped dimer form of RNase A, Liu et al. could identify two domain-swapped trimers that are formed through “linear” or “open- ended swapping” and cyclic domain-swapping (Figure III-3f).31 Further studies demonstrated the formation of a domain-swapped tetramer for BS RNase through the incubation of the protein with acetic acid, followed by lyophilizing and re-dissolving the powder in “benign buffer”.33-34 The next example of domain-swapping is the immunoglobulin-binding B1 domain of streptococcal protein G (Figures III-3g and III-3h).35 The structural analysis of the domain-swapped dimer structure with the wildtype monomer illuminates that Phe34 is a 146 key amino acid triggering monomer-dimer conversions.35-36 The mutation of this residue to Ala causes a WT monomeric structure. Additionally, some variants of GB1 lead to the domain-swapped dimer resulting in fibril formation.37 Notably, Cyanovirin-N, typified as CV-N, is another protein producing domain- swapped dimers, in which half of the protein swaps. This protein has become an excellent a. d. g. j. c. f. I. b . e. h. k. Figure III-3. Several examples of domain-swapped proteins. a. Diphteria toxin monomeric form (PDB code: 1MDT). b. Diphteria toxin domain-swapped dimer form (PDB code: 1DDT). c. Monomeric structure of RNase A (PDB code: 1RTB). d. RNaseA N-terminal domain-swapped dimer crystal structure (PDB code: 1A2W). e. RNaseA C-terminal domain-swapped dimer crystal structure (PDB code: 1F0V). f. RNaseA C-terminal cyclical domain-swapped trimer crystal structure (PDB code: 1JS0). g. GB1 monomeric structure (PDB code: 1Q10). h. GB1 monomeric structure (PDB code: 3V3X). i. amyloid-like cystatin domain-swapped dimer (PDB code:1TIJ). j. Human prion protein monomer (PDB code: 1QM0). k. Human prion protein domain-swapped dimer (PDB code: 1I4M). 147 candidate for deactivating enveloped viruses, such as Influenza, Ebola, Hepatitis C, HIV, etc.38-40 The amount of domain-swapped dimer formation varies based on the experimental conditions. In neutral conditions, the monomer and domain-swapped dimer forms are in equilibrium with one another. Interestingly, the domain-swapped dimer was crystallized.40 It has been shown that the domain-swapped-dimer is a kinetically-trapped form of the protein in high concentrations, and in physiological conditions, it reverts back to a more stable monomeric form.41 The concurrence of the dimer and monomer in solution at normal conditions demonstrates the low barrier of interconversion between the two different forms of the protein. This equilibrium can be altered by certain mutations in the hinge loop region. For instance, the P51G mutation significantly stabilized the monomeric form.22, 42 However, since the WT dimer is not stable in the physiological conditions, the biological function of the domain-swapped dimer are still elusive. 22 As mentioned earlier, there are examples of domain-swapped proteins where the domain- swapping can trigger unwanted protein aggregations and pathogenic structures 43. The hallmarks of these disease-related proteins are cystatin44 and prion,45 whose domain- swapped dimer forms have been confirmed using crystal structures (Figures III-3i, III-3j, and III-3k). In contrast to the initial models of fibril amyloid formation that suggested the proteins are transformed entirely to β-sheet conformations from their native structures,46 these proteins indicated that many structural features of the native forms of the proteins are preserved during fibril amyloid formation.47 These proteins further show that fibril formation can be generated under ambient conditions in vivo and does not need harsh conditions with a high concentration of 148 denaturant, as is usually employed for in vitro fibril formation. Additionally, it was proposed that only a segment of the protein (mostly the terminal or middle sections of the domains) undergoes the β-stack formation, and the rest of the polypeptide chain usually remains intact in its native form. As shown above, many variables contribute to domain-swapped dimer formation in proteins, and domain-swapping can alter the functions of proteins and in some cases, lead to disease-related structures, such as Parkinson’s disease, diabetes, Huntington’s disease, Alzheimer’s disease, etc. 36, 48-50 Our next step is to understand the domain-swapping mechanism in the iLBP family. III-4. Domain swapping in the iLBP family: a novel tool for protein engineering As mentioned in the previous Chapter, Assar and Nossoni et al 17 reported that several variants of hCRBPII lead to large populations of domain-swapped dimers in which Figure III-4. The formation of domain-swapped dimer (previous study) and trimer (current study) based on open-monomer theory in hCRBPII. 149 half of the protein undergoes domain swapping. Furthermore, it has been reported that FABP5,51 another member of the iLBP family, produces a similar domain-swapped dimer, as observed for hCRBPII. In our previous study,17 we demonstrated that the domain-swapped dimer is a kinetically trapped form of the protein that occurs as a result of increasing the lifetime of open-monomer intermediates through certain mutations in the hinge region. Our previous work17 demonstrates the importance of the Y60 and E72 hydrogen bond in stabilizing the monomeric fold. The removal of the hydrogen bond between Y60 and E72 through mutations of either of these residues led to an increase in the dimer population as a result of enhancing the lifetime of open-monomer intermediates (Figure III-4). This is in contrast to the folding studies of this protein family conducted extensively by Gierasch’s lab52 (mainly focused on human cellular retinoic acid binding protein I (hCRABPI)), in which they showed that early β-barrel closure, driven by hydrophobic interaction, is pivotal in the folding pathway of this family, as it impedes protein aggregation and off-pathway oligomer formations such as dimers.17 As also shown in Chapter II, we brought the domain-swapped dimer to another level by illuminating the allosteric behavior of the hCRBPII domain-swapped dimer upon ligand binding, where a large conformational change was observed at atomic resolution. Through protein engineering, a new class of protein conformational switches was created that can be activated by retinal or reduction potential of the environment. Additionally, a novel metal-binding site was designed that can be tuned by the conformational readouts of the system. Furthermore, other studies have shown the power of domain-swapping that can be exploited in different protein engineering applications such as triggering the 150 domain swapping by ligand binding,53 generating a protein functional regulator using domain-swapped form of ribose binding protein,54 controlling protein splicing through domain-swapping,55 designing an optogenetic tool to regulate protein/protein interactions using domain-swapped form of photoactive yellow protein variant.56 These examples demonstrate how domain-swapping can be employed in protein engineering to produce new functions in the protein, an ability which is not attainable in the monomeric form. III-5. Discovery of the domain-swapped trimer The first observation of an hCRBPII domain-swapped trimer was made during the purification of the Q108K:Q4F:Q38F:K40L:T53A:R58L (A1) variant, which eluted at two a. Source Q b. Source Q 1st peak 1st peak 2nd peak 2nd peak c. Source Q 2nd peak d. Gel filtration Trimer 1st peak Trimer Dimer filtration chromatograms III-5. Source Q and gel Figure for: a. A1: Q108K:K40L:T53A:Q38F:R58L, 23°C, 20 hours expression. b. A1, 19°C, 40 hours expression. c. A1: V62E, 19°C, 40 hours expression. d. Gel filtration chromatogram of high-salt-eluted protein samples for A1 (corresponding to the second peak in source Q) expressed at 19°C, 40 hours (black) vs. A1 expressed at 23°C for 20 hours (blue). 151 distinct salt concentrations, with the first characteristic of the domain-swapped dimer (Figure III-5a). Further analysis using gel filtration, NativePAGE gel, Dynamic Light Scattering (DLS), and X-ray diffraction revealed that the species eluting at higher salt was an hCRBPII domain-swapped trimer (Figures III-5, III-6, and III-7a). Structural analysis of the a. Native PAGE gel monomer dimer 1st peak on source Q 2nd peak on source Q b. SDS PAGE gel c. DLS monomer dimer trimer 25 kDa 15 kDa 2nd peak on source Q 1st peak on source Q Figure III-6. a. Native PAGE gel for 1st and 2nd peaks shown in the source Q column. b. SDS PAGE gel for 1st and 2nd peaks shown in the source Q column. c. Dynamic light scattering for monomer, dimer, and trimer (Q108K:K40L:T51V was used as a monomer for comparison) . apo trimer of A1 showed that two trimer molecules (six chains) existed in the asymmetric unit. Each trimer contained three structurally similar domains, related to each other by pseudo C3 symmetry (RMSD~0.4) (Figure III-7). Additionally, the overlay of the open monomers of the trimer containing chains A, B, and D showed that these three open monomers are similar to each other, aside from a small change in the relative orientation of the two dimer domains (average RMSD~0.6) (Figure III-7). However, more structural 152 flexibility is observed between the open monomers of chain C, E, and F. In other words, Table III-1. Ratios of monomer, dimer, and trimer formations in different variants. Mutant Trimer Dimer Monomer Expression condition A1 A1 A1-V62E KD KLK KD-A28C KD-I32C KD-A28C:I32C A28C I32C A1-I32C KDA28W** KDI32M:L36C KDI32S 5% 30%* 58% 22% 17% - 95% 34% - 20% 91% 100% 64% 28% 95% 67% 42% 75% 70% 99% 5% 66% - 40% 9% 0 34% 15% <1% 2% - 3% 13% - - - 100% 40% - 0 - 57% 23°C - 20 hours 19°C - 40 hours 19°C - 40 hours 19°C - 40 hours 19°C - 40 hours 23°C - 20 hours 23°C - 20 hours 23°C - 20 hours 23°C - 20 hours 23°C - 20 hours 19°C - 40 hours 23°C - 20 hours 23°C - 20 hours 19°C - 40 hours ** The structure of the mutant was not obtained. It could be side by side protein oligomerization. while the ABD trimer has pseuo-three fold symmetry, the CEF trimer is significantly more asymmetric. Once the domain-swapped trimer was discovered, we subsequently identified and determined the structures of five other domain-swapped trimer-forming variants, including A1, Q108K:K40L:T53A:Q38F:R58L:V62E (A1-V62E), Q108K:T51D (KD), Q108K:K40L:T51K (KLK), Q108K:K40L:T53A:Q38F:R58L:F57H (trimer and dimer) (A1- 153 F57H), and Q108K:T51D:A28C:I32C (KD-A28C:I32C). In all cases, similar trimeric structures are evident, though there are significant differences between them. For example, the KD domain-swapped trimer contains 12 molecules (four distinct trimers; structure not shown here) in the asymmetric unit and shows more structural flexibility and heterogeneity than that seen in A1. This is in contrast to virtually all apo dimer structures (ten (the in all), which form very similar symmetric structures average RMSD between open monomers is ~0.3 Å). The conclusion to be drawn from this is that the domain-swapped trimer is more flexible a. c. b. d. Figure III-7. a. The crystal structure of domain-swapped trimer of A1 (chains A, B, and D). The close monomers highlighted with different colors. b. The overlay of three closed monomers show similar conformations for all of them. c. Domain-swapped trimer structure of A1 (the same chains as mentioned above) where the open monomers made up trimer are highlighted by different colors. d. Overlay of the three open monomers demonstrated the similar conformations for chains A, B, and D. than the domain-swapped dimer and is capable of adopting a number of distinct conformations. Additionally, as shown in Figure III-8, the overlay of the trimer’s open 154 monomer with the open monomers of the apo symmetric dimer, apo asymmetric dimer, and holo dimer reveals that the relative orientation of the c- and n-termini of the open monomers in the trimer form is very different from the apo and holo dimers, demonstrating the large displacement in the orientations of the c- and n-termini (relative to the symmetric, domain-swapped dimer) required to generate a domain-swapped trimer (Figure III-8). Furthermore, the torsional angle comparison between the dimer and trimer reveals that the torsional angles of N59, located on the β-strand, that runs between the domains in a. c. H81 b. 20.6 Å 48.5o T56 H81 H81 21 Å 70 o T56 H81 H81 11.6 Å d. H81 27o T56 32.3 Å H81 83.3o T56 H81 Figure III-8. Overlay of open monomer of apo domain-swapped trimer A1 (purple), chain D, with: a. open monomer of apo WT hCRBPII domain-swapped dimer (blue); b. open monomer of Q108K:K40L:Y60W, chain A, apo asymmetric domain-swapped dimer (yellow); c. open monomer of Q108K:K40L:Y60W, chain B, apo asymmetric domain-swapped dimer (yellow); and d. open monomer of Q108K:K40L:T51F holo domain-swapped dimer, chain A (green). the trimer are significantly different from the apo dimer, which may be one of the factors 155 contributing to this substantial conformational variation (Figure III-9). Based on our previous work, we proposed a folding pathway for hCRBPII that involved a common “open monomer” intermediate, where the n- and c-termini of the protein partially fold, followed a. 350 300 250 200 150 100 50 0 -50 A1-A A1-B A1-D A1-F57H-A A1-F57H-B A1-F57H-C KD-A28C:I32C-A KD-A28C:I32C-B KD-A28C:I32C-C b. N59-dimer Delta phi delta psi N59- trimer Figure III-9. a. Subtraction of N59 torsional angles for different trimer structures vs. the apo symmetric dimer. b. Overlay of dimer (blue) vs. trimer (purple) demonstrates that by rotating around N59, the dimer structure can be converted to trimer and vice versa. by closure to either the monomer, dimer, or now trimer, depending on the concentration and propensity of the polypeptide to adopt an orientation consistent with the oligomeric form (Figure III-4). Therefore, the mutations that provide a higher concentration of open monomers are expected to lead to a higher amount of dimer or trimer formation. To ponder this hypothesis, we used Q108K:T51D (KD), which was previously reported to yield a substantial amount of domain-swapped dimers. The protein’s monomer yield was very low. The expression of this protein at 19 degrees for ~40 hours depicted a large amount of domain-swapped dimers, along with a trimer peak (Table III-1). The KD and A1 variants 156 Figure III-10. Gel filtration chromatogram of Q108K:T51D dimer and trimer. The trimer was kept at room temperature and reloaded to Gel filtration column to examine the stability of trimer. The intensity for the trimer kept for 30 days at room temperature was normalized. only have the Q108K mutation in common. Additionally, the expression of a substantial number of mutants of hCRBPII containing Q108K showed that this mutation does not play a significant role in creating domain-swapped species. However, A1, A1-V62E, KD, and Q108K:K40L:T51K (KLK) all contain a number of mutations, such as K40D and T53A in A1, T51D in KD, and T51K in KLK, that lead to a significant amount of domain-swapped dimers. Adding a V62E mutation to the A1 construct resulted in more than 50% trimer, which is another example of the aforementioned effect. So far, all the constructs leading to trimer formation also make a substantial amount of domain-swapped dimers. Therefore, the formation of an open monomer in higher concentration might cause trimer formation as a byproduct, a process which is under the control of open-monomer concentration. Also remarkably the trimer form was stable after about a month kept at room temperature confirmed with gel filtration (Figure III-10). 157 III-6. Stability of dimer vs. trimer Next, the thermal stability and structural analysis of the dimer vs. trimer was conducted. (We are very thankful to Cayman Chemical and their employee, Dr. Assar, who aided us in performing this assay.) In terms of stability, as shown in Table III-2, the melting point of the KD dimer was calculated to be about 39° C and the trimer around a. b. Pseudo C3 c. F27 F27 I32 I32 I32 F27 L36 F57 F57 I32 I32 L36 I32 F57 L36 Figure III-11. Hydrophobic core generated by hydrophobic residues at the center of the domain-swapped trimer. F27 from each chain is shown in green, I32 in yellow, L36 in orange, and F57 in cyan. The pseudo-C3 symmetry brings residues from the different chains close together. 59°C. For A1, the melting point was calculated to be about 58° C for the trimer, whereas the dimer led to aggregation during the melting point experiment. Therefore, the trimer appears to be thermodynamically more stable than the dimer in these variants. A careful look at the structure of the domain-swapped trimer reveals the presence of an additional hydrophobic core at the center of the trimer structure. This additional interacting surface may provide more stability compared to the dimer. All three Phe57 residues in the trimer, 158 together with Leu36, Ileu32, and Phe27, provide the surface for this additional hydrophobic core (Figure III-11). Table III-2. Thermal stability assay for different variants. Monomer M.P. Trimer M.P. Variants Dimer M.P. A1 KD KD-A28C KD-I32C KD-A28C:I32C (°C) 58 59 No trimer* 46 Noisy A1-F57H Aggregated A1-F57H (Zn) 45, 61 (°C) (°C) Aggregated ~ No monomer* 39 51.5 ~ No monomer* ~No monomer* Noisy No monomer* 48.5 45.5 43.5 No monomer* 46 41 *The monomer or trimer populations were too low to be confidently isolated for performing the assay. III-7. Effect of ligand binding on trimer structure In chapter II, we showed that in the domain-swapped dimer, a large conformational change in the relative orientations of the two domains occurs upon retinal binding. Unfortunately, in contrast to many other efforts, the retinal-bound structure of the domain- swapped trimer was not obtained. Instead, we used a thiophene-based chromophore with aldehyde and azetidine functionalities (ThioFluor3) (check chapter IV for further information), designed and synthesized in Dr. Borhan’s lab as a similar ligand to retinal for crystallization purposes. Using the chromophore shown in Figures III-12a and III-12b, with A1 variant, we obtained a 2.54 Å resolution structure showing ligand binding in of the three binding sites, but with relatively different trajectories of ThioFluor3 (Figure III-12b). 159 Interestingly, as shown in Figure III-12a, ligand binding does not alter the conformation of Q108K b. d. Q108K N59-holo E72 N59-apo Q108K Y60-holo f. N59-apo N59-holo Y60-apo a. c. Q108K Y60-holo E72 Y60-holo E72 e. Q108 E72 Y60-holo Y60-apo N59-holo N59-apo Figure III-12. a. Overlay of apo (cyan) and holo (orange) ligand-bound trimer (A1 variant). b. Top: electron density of the ligand in one of the trimer chains. Bottom: overlay of ligands in three different chains. c. Overlay of apo (yellow) and holo ligand-bound trimer of A1 variant (purple). N59 is pointing outside the binding cavity in both apo and holo structures where it does not sterically impinge on the chromophore. d. Overlay of apo (orange) and retinal-bound domain-swapped dimer (purple). N59 in apo conformation sterically clashes with the retinal trajectory. The result of this steric interaction is the “flipped-out” conformation for N59 in holo. e. Overlay of apo (green) and holo (purple) retinol-bound structure of WT-hCRBPII (PDB codes: 2RCQ and 2RCT). f. Overlay of retinylidene Q108K:T51D dimer (purple) with the synthesized chromophore used for trimer crystallization with A1 variant (cyan). 160 the domain-swapped trimer considerably. As mentioned in chapter II, in the apo domain- swapped dimer (Figure III-12d), Asn59 is “flipped in” and Tyr60 is “flipped out” of the binding pocket. Asn59, in its flipped-in conformation, would sterically clash with the bound ligand, resulting in the flipped-out conformation for Asn59 and flipped-in conformation for Tyr60 within the holo dimer. The dispensation of these two residues causes a large conformational change within the holo domain-swapped dimer. However, as shown in Figures III-12c and III-12e, Asn59 in both the apo domain-swapped trimer and monomer is pointing outside of the binding cavity and Tyr60 is pointing inside. Asn59, in the flipped- out conformation, does not sterically impinge on the bound chromophore. Therefore, ligand binding does not affect the conformation of Asn59, and subsequently, Tyr60. Instead, there is a minor conformational change in the domain-swapped trimer that occurs upon ligand binding. This seems to indicate that the domain-swapped trimer might not be an allosterically regulated form of hCRBPII. 161 III-8. A cross-subunit disulfide bond drives trimerization Though we have identified mutational “hot spots” that lead to domain-swapped trimers during expression, control over the folding pathway is lacking as invariably a. b. c. A28C A28C’ I32C’ I32C Figure III-13. Ramachandran plots for a. Q108K:T51D:I32C and b. Q108K:T51D:A28C. Q108:T51D:I32C shows some of the residues as outliers. c. Overlay of Q108K:T51D:A28C (purple-blue) vs. Q108K:T51D:I32C (yellow) crystal structures. significant amounts of domain-swapped dimers are also formed. We therefore turned to a strategy previously exploited in controlling both domain-swapped dimer formation and allosteric behavior— the appropriate introduction of disulfide crosslinks that will bias the folding pathway toward domain-swapped trimer formation. In Chapter II, we showed that the introduction of cross-subunit disulfide linkages between Cys28 residues results in new allosteric behavior in the domain-swapped dimer and in essentially exclusive formation of domain-swapped dimers (Table III-1). Inspection of the domain-swapped trimer structure led us to conclude that a disulfide linkage in the middle of the α-helices may increase domain-swapped trimer formation, resulting in the KD-I32C and A1-I32C variants. The I32C position in the middle of the α-helices would lead to decreasing the flexibility of open monomers made up the dimer and presumably disfavoring this form. Gratifyingly, both variants, KD-I32C and A1-I32C, produced domain-swapped trimers almost exclusively. Similarly, the addition of the I32C mutation in the dimer-favored 162 Q108K:T51D:A28C construct again increased trimer formation (Table III-1). Even in the context of the wild-type background, which strongly favors monomers and produces no trimers, the addition of the I32C single mutation resulted in a mixture of monomer (40%), dimer (40%), and trimer (20%) (Table III-1). In this case, we could also obtain the monomer crystal structure. A comparison of the KD-A28C and KD-I32C domain-swapped dimer structures shows that the overall structures are very similar, with both adopting the apo symmetric dimer domain orientation. However, the Cys32 crosslink enforces an unfavorable A28C’’ A28C’ A28C I32C I32C’ I32C’’ Figure III-14. The crystal structure of the domain-swapped trimer of KD:A28C:I32C showing the disulfide bond formation between different chains. A28C close up I32C close up Ramachandran outlier for residue Ser55 (Figure III-13a), indicating decreased stability of the dimer due to the Cys32 crosslink. In contrast, the KD-A28C:I32C domain-swapped trimer structure shows inter-subunit Cys32 and Cys28 crosslinks comfortably accommodated in the center of the structure (Figure III-14). This demonstrates how the domain-swapped folding pathway can be manipulated by strategic placement of disulfide crosslinks, allowing control over the folding product. 163 III-9. Introduction of a metal binding site in the domain-swapped trimer As mentioned earlier, a characteristic feature of the domain-swapped trimer is the pseudo-C3 symmetry axis running through the buried surface between the three domains of the trimer. The three Phe57 residues are located in the center of the pseudo-C3 symmetry axis (Figure III-12). Previous studies demonstrated that domain-swapped oligomeric structures could result in binding sites for ions. For example, the domain- a. b. F57H F57H’’ F57H’ Zn F57H’ 14 Å F57H Figure III-15. The zinc-bound apo A1-F57H mutant: a. The 2Fo-Fc map shows the presence of zinc coordinated with three histidines (contoured to 1σ.). b. The domain-swapped dimer of A1-F57H shows two histidines are far away from each other to generate the metal-binding site (contoured to 1σ.). swapped trimer of RNase harbors a sulfate ion binding site,57 and the domain-swapped trimer of CksHs2 has a divalent metal ion bound in the trimer interface.58 164 In the cases mentioned above, the structures possess metal-binding residues in the interface; however, the hCRBPII domain-swapped trimers and dimers have hydrophobic interfaces devoid of metal-binding moieties. We therefore attempted to create a metal- binding site by introduction of the F57H mutation to the domain-swapped trimer-forming A1 variant to give the A1-F57H variant. Unfortunately, A1-F57H significantly reduced the quantity of trimers (<5%). This is probably because the trimer interface is less hydrophobic, and therefore less able to promote trimer formation. Therefore, we decided to add metal ion to the expression media in an effort to create a template for, and thereby promote, the trimer folding pathway. To our satisfaction, the addition of zinc acetate (100 μM) resulted in substantial quantities of trimers (15%), though monomers and dimers were also produced. The crystal structure of the Zn2+-bound A1-F57H domain-swapped trimer revealed a single trimer in the asymmetric unit, perhaps indicating a less-flexible trimer relative to the other, structurally-heterogeneous trimer structures previously obtained, Figure III-15a. In this structure, as shown in Figure III-15a, zinc was coordinated with the three His57 residues coming from different chains to generate a trigonal pyramidal structure. Although zinc usually makes a tetrahedral structure, trigonal pyramidal (TP) zinc-centered structures are not rare in biology. For example, Fructose- 1,6-bisphosphatase,59 an essential regulatory enzyme in gluconeogenesis, and NADP- Dependent Alcohol dehydrogenase60 both have zinc-centered TP structures. Furthermore, the domain-swapped dimer crystal structure of A1-F57H was obtained and showed that the two F57H residues are too far away from each other (~14 Å) to generate a metal-binding site, Figure III-15b. However, given the conditions of crystallization for the dimer, containing sodium malonate, the aforementioned structure for the dimer might be 165 due to the lack of available zinc for the dimer (malonate can act as a multidentate ligand and generate a zinc complex). The anomalous difference map obtained from the zinc- bound trimer crystal collected using X-ray near the zinc absorption edge confirmed the a. b. Figure III-16. a. The anomalous difference map collected above zinc x-ray absorption edge contoured to 8σ. The saturation-binding curve for A1-F57H showing the Kd: 0.70 ±0.21 µM. presence of zinc bound in the center of the trimer with a 8σ anomalous difference peak, Figure III-16a. We were also successful in determining the zinc binding affinity (through collaboration with Dr. Pinger from Dr. Spence’s lab) using a radioactive-zinc labeling assay. As shown in Figure III-16b, the Kd was found to be 0.70 ±0.21 µM for A1-F57H. Also, the melting points for this variant were calculated in the presence and the absence of zinc. As shown in Table III-2, the data further depicts the role of zinc in stabilizing the structure of the domain-swapped trimer protein where the absence of zinc causes protein aggregation during the thermal stability assay. These results suggest a novel protein fold in the iLBP family that could be used for a variety of protein engineering applications where the metal-binding site is required for triggering a special function in the protein. 166 III-10. Using “designed disulfide bond” to favor the domain-swapped oligomeric forms of other iLBP family members a. SDS PAGE gel Source Q 2nd peak 1st peak 37 kDa 15 kDa a b 1st peak a b 2nd peak a: in the presence of BME b: in the absence of BME b. Figure III-17. a. Left: The gel electrophoresis under the reducing and no- reducing conditions. Right:. Source Q chromatogram of I31C showing two peaks. b. The crystal structure of I31C showing two close monomers connected by cross linked disulfide bridge. The ability to favor different oligomeric forms through a designed disulfide bond piqued our attention and led us to pursue the same strategy and apply the mutation providing the disulfide bond in the equivalent position in CRABPII, another protein in iLBP, possessing about 40% sequence similarity with hCRBPII. The I31C mutation was made (equivalent to I32C in hCRBPII) and solubly expressed. The SDS PAGE gel demonstrated the presence of disulfide bond by comparing the results under reducing and non-reducing gel SDS PAGE gel electrophoresis, Figure III-17a. To our delight, the 167 crystal structure of I31C variant was obtained at 2.1 Å resolution, Figure III-17c. The structure depicted two close monomers connected to each other through the disulfide bond (I31C from each monomer) forming a side-by-side dimer. This is in contrast to hCRBPII that the same mutation led to domain-swapped dimer and trimer formation. This observation is consistent by Gierasch’s lab folding experiment 52 (as mentioned in III-4) and it further shows different proteins in the same family can have completely various folding pathways. III-11. Conclusion and future direction In conclusion, the first domain-swapped trimer in the iLBP protein family has been reported. The x-ray crystallographic data, along with melting-point measurements, showed the trimer structure has more thermal stability in comparison with the domain- swapped dimer. Furthermore, for the first time, an engineered metal-binding site and a disulfide bond have been used to control the oligomerization state in domain-swapped oligomers. The domain-swapped trimer provides a new structural template for protein design applications, including applications involving metal binding and ligand binding. In the future, further mutational studies can be done to detect other potential oligomerization states in the hCRBPII protein. This system could also be used as a potential candidate for studying the folding pathway in the iLBP family, where the high- resolution crystal structures of various folds, along with biochemical assays, would illuminate how domain-swapped dimers and trimers can evolve from the monomeric form. Finally, the methods used for regulating oligomerization states in hCRBPII can be performed in other protein families, such as CRBPI, FABP4, and FABP5, to unravel different conformational states in those proteins. 168 III-12. Experimental procedures III-12-1. DNA and protein purification The same procedure mentioned in the experimental section of chapter II was used here. III-12-2. PCR condition for hCRBPII The same procedure explained in the experimental section of chapter II was used here. III-12-3. List of primers Site-directed mutagenesis was performed using K-CRBPII or KL-CRBPII-PET17b, which were generated in our previous work.1 K40D Forward: 5’- CTCACTCAGACGGATGTTATTGATCAAGATGG-3’ Reverse: 5’- CCATCTTGATCAATAACATCCGTCTGAGTGAG -3’ T51D Forward: 5’-GGTGATAACTTCAAGGATAAAACCACTAGCAC-3’ Reverse: 5’-GTGCTAGTGGTTTTATCCTTGAAGTTATCACC-3’ K40L Forward: 5’-CTCACTCAGACGCTGGTTATTGATCAAGATGG-3 Reverse: 5’-CCATCTTGATCAATAACCAGCGTCTGAGTGAG-3’ Q108K Forward: 5’ -CCGCGGCTGGAAGAAGTGGATTGAGGGGG-3’ Reverse: 5’ -CCCCCTCAATCCACTTCTTCCAGCCGCGG-3’ Q4F Forward: 5'-GACGAGGGACTTCAATGGAACC-3' Reverse: 5'-GGTTCCATTGAAGTCCCTCGTC-3' 169 R58L Forward: 5'-CTAGCACATTCCTGAACTATGATGTG-3' Reverse: 5'-CACATCATAGTTCAGGAATGTGCTAG-3 R58L-57H Forward: 5'-AGCACACACCTGAACTATGATGTG-3' Reverse: 5'-CACATCATAGTTCAGGTGTGTGCT-3’ I32C Forward: 5'-AACCGCAAGTGTGCAGTACGT-3’ Reverse: 5’-ACGTACTGCACACTTGCGGGT-3’ A28C Forward: 5’- GATATTGATTTTTGCACCCGCAAG-3’ Reverse: 5’- CTTGCGGGTGCAAAAATCAATATC-3’ V62E Forward: 5’- AACTATGATGAGGATTTCACTGTTGGAGTA-3’ Reverse: 5’- TACTCCAACAGTGAAATCCTCATCATAGTT-3’ III-12-4. Transformation of PCR product DH5α competent cells (50 μL) were thawed on ice and then 5 μL of PCR product was added to the sample. The sample was kept 30 minutes in ice; then the sample was heat-shocked at 42 ̊C for 45 seconds and 450 μL of LB was added. The sample was then incubated at 37 ̊C for 30 minutes. The cells were spread on a LB agar plate containing ampicillin (100 μg/mL) and incubated at 37 ̊C overnight. III-12-5. DNA purification The same procedure mentioned in the experimental section of chapter II was used here. 170 III-12-6. Expression of hCRBPII mutants For the expression, the gene was transformed into BL21 (DE3) E. coli competent cells. A similar protocol mentioned for the transformation of PCR products was employed. To inoculate 1 L of LB with ampicillin (100 μg/mL), a single colony was used. After the cell culture was grown at 37 ̊C while shaking for~ 8 h, the overexpression was induced by adding 1 mL of 1M IPTG solution into the 1 L cell culture (overall concentration 1.0 mM of IPTG). The solution was shaken at mentioned temperature and cited time. III-12-7. Protein isolation and purification The cells were harvested by centrifuging for 20 min at 5,000 rpm at 4 ̊C. After removal of the supernatant, the cells were resuspended in 60 mL of Tris buffer (10 mM Tris.HCl, pH=8.0). Then the resuspended cells lysed by ultrasonication (Power 60%, 1 min x 3) and centrifuged at 16 ̊C (10,000 rpm, 20 min). The ion exchange chromatography using Q SepharoseTM, Fast Flow resin was used for the purification of proteins at 4 ̊C. Further purification was performed with Fast Protein Liquid Chromatography. In the first step of purification with Q Sepharose Fast Flow resin (GE Health Sciences), the bound protein was first washed with 10 mM Tris- HCl, pH = 8.0 and then eluted with 10 mM Tris. HCl, 150 mM NaCl, pH = 8.0, and the pure fractions were desalted by dialysis against a 10 mM Tris. HCl pH = 8.0 buffer and loaded on a second anion exchange column (15Q, GE Health Sciences, BioLogic DuoFlow system). The low and high salt fractions were separately concentrated at 4°C to 5-8 mg/mL and further purified by size exclusion chromatography (Superdex 75 16/75 column, GE Health Sciences, packed in-house) in a buffer containing 10 mM Tris, 150 mM NaCl, pH = 8.0. 171 III-12-8. Crystallization, data collection and refinement Each mutant protein was concentrated to between 6 to 8 mg/mL in a buffer containing 10 mM Tris, 150 mM NaCl, pH = 8.0. Crystals were grown via the hanging drop vapor diffusion method using 1 μL of protein solution and 1 μL of crystallization solution in the drop and 1 mL of crystallization solution in the reservoir. For the crystallization of the holo DS trimer A1/ThioFluor3, proteins were incubated with four equivalents of ThioFluor3. The best crystals for A1, KLK, KD grew using a crystallization solution of 25-30% PEG4000, 0.1 M sodium acetate pH 4.5-4.8, 0.1 M ammonium acetate. A1-F57H crystals grew under the following conditions: 0.2 M ammonium iodide, 20% PEG 3350. I31C (CRABPII) crystals grew under following conditions: 0.1 M sodium Malonate, 20% PEG 3350. The crystals were soaked in a cryoprotectant solution (the original conditions + 20% glycerol), flash frozen in loops in liquid nitrogen, and stored in a liquid nitrogen Dewar prior to data collection. Diffraction data was collected at the Advanced Photon Source (APS) (Argonne National Laboratory IL) LS-CAT, (sector 21- ID-D,F,G) using MAR300 and MAR350 detectors, 1.00 Å wavelength radiation at 100 K. The initial diffraction data was indexed, processed and scaled using the HKL2000 software package. The structures were solved by molecular replacement using PHASER in PHENIX and hCRBPII (PDB entry 2RCQ) as a search model. The initial electron density map was produced by Phaser-MR in PHENIX. Model rebuilding, placement of water molecules, etc. were done using COOT. The structures were refined using the PHENIX program packages. The chromophore was built using COOT and PHENIX to generate restraints. 172 III-12-9. Radioactive 65Zn labelling assay 65ZnCl2 was purchased from (radionuclide purity: 99%) PerkinElmer. The ZnCl2 was diluted in 18 µM distilled H2O to make a radioactive 100.0 µM 65ZnCl2 stock solution. Samples were prepared in 1.5 mL centrifugation tubes to contain a total volume of 210 µL and contain 15.0 µM protein and varying concentrations of ZnCl2 in buffer. To create a saturation binding curve, the Zn2+ concentrations were varied from 1-15 µM. Then, 200 µL of each sample were pipetted into an ultrafiltration centrifugation device with a 10 kDa MWCO size-exclusion membrane (Amicon Ultra 0.5 mL centrifugal filters). The samples were centrifuged at 8,000g for 3 minutes to allow 6-15 µL of ultrafiltrate to collect in the bottom of the tube. The concentration of Zn2+ in the ultrafiltrate is considered the free- Zn2+ in the sample, as described by the literature. The volume separated from the sample is small enough so as to not disturb the equilibrium of the binding system (<8% of the volume of the bulk sample). Then, a 9.0 µL aliquot of ultrafiltrate sample was transferred into a separate tube for analysis. Calibration standards containing 9.0 µL of known radioactive Zn2+ concentrations were also prepared in separate tubes. Then, all standards and samples were analyzed by counting the emitted gamma radiation at (1115 keV) for 5 minutes per sample on a PerkinElmer 2480 Automatic Gamma Counter. The concentration of Zn2+ in the ultrafiltrate was quantified by comparing the counts-per- minute (CPM) for each sample to an external standards calibration curve. The concentration of Zn2+ in the ultrafiltrate was considered to be the free-Zn2+ concentration in the sample, while the concentration of protein-bound Zn2+ was calculated as the free-Zn2+ subtracted from the total-Zn2+. The concentrations of free Zn2+ vs protein- bound Zn2+ for each sample were plotted and analyzed via non-linear regression software 173 (SigmaPlot 13.0) to create a saturation binding curve. This curve allowed the calculation of the equilibrium binding constant (Kd) and stoichiometry (Bmax). 174 Table III-3. X-ray crystallographic data and refinement statistics for apo trimer Q108K:T51D:A28C:I32C, A1 and Zinc-bound A1-F57H hCRBPII mutants. Q108K:T51D:A2 Q108K: K40L: T53A: Zinc-bound 8C: I32C Q38F: R58L (A1) A1-F57H Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge Resolution (Å) (Last Shell) Rwork/ Rfree (%) RMS Bond Length (Å) Bond Angle (°) Average B factor Number of water molecules PDB ID C2221 87.93 160.49 86.33 90 90 90 3 apo P 212121 67.05 72.68 164.80 90 90 90 6 1917398 35625 99.11 (94.26) 34.8 5 2352250 18585 99.49 (94.94) 23 17 29.39 - 2.10 14.97 - 2.90 (2.18- 2.10) 21.04 /24.85 0.008 (2.99 - 2.89) 32.16/25.46 0.001 0.92 48.04 110 TBD 1.49 58.70 16 TBD P 212121 60.51 74.45 94.34 90 90 90 90 3 2067465 14425 (1402) 99.87 (99.79) 27(2.3) 0.1 (1.1) 46.96 - 2.55 (2.64 - 2.55) 32.36/25.80 0.011 1.62 54.56 14 TBD a Values in the parenthesis refer to the last resolution shell. 175 Table III-4. X-ray crystallographic data and refinement statistics for trimer hCRBPII mutants Q108K:T51D, A2, and Q108K:K40L:T51K. Q108K: T51D Q108K: K40L: T53A: Q108K: K40L:T51K P212121 69.83 78.47 178.89 90 90 90 6 20546 (1916) 99.39 (94.38) 13.1 (2.3) 0.2 (1.1) 47.48 - 2.99 (3.09-2.99) 33.35/25.49 0.022 2.38 48.15 12 TBD Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge Resolution (Å) (Last Shell) Rwork/ Rfree (%) P21 84.56 111.27 101.83 90 112.89 90 12 2373724 42592 (4247) 98.99 (99.79) 22.5 (2.5) 0.098 (0.687) 15 - 2.80 (2.90 - 2.80) 33.80/27.28 Q38F: R58L:V62E (A2) P22121 37.12 101.03 109.47 90 90 90 3 555975 10793 (1015) 99.24 (93.81) 17.1(2.1) 0.119 (0.739) 45.87 - 2.79 (2.89 -2.79) 29.81/24.42 RMS Bond Length 0.014 (Å) RMS Bond Angle 1.96 (°) Average B factor Number of water molecules PDB ID 70.91 77 TBD 0.011 1.72 56.58 38 TBD a Values in the parenthesis refer to the last resolution shell. 176 Table III-5. X-ray crystallographic data and refinement statistics for apo dimer hCRBPII Q108K:T51D:I32C and A1-F57H mutants. Q108K: T51D:I32C (A1-F57H) apo Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge Resolution (Å) (Last Shell) Rwork/ Rfree (%) RMS Bond Length (Å) RMS Bond Angle (°) Average B factor Number of water molecules PDB ID P63 133.28 133.28 51.16 90 90 120 2 2082961 8637 (843) 98.82 (98.71) 13.8 (2.1) 0.17(1.40) P212121 60.09 63.84 72.83 90 90 90 2 902279 31490 (3092) 99.62 (99.84) 43.8 (2.2) 0.06 (0.90) 33.20- 3.20 (3.32- 3.20) 23.48/26.17 0.01 46.35 - 1.70 (1.76 - 1.70) 30.6-27.07 0.006 1.17 73.69 14 TBD 0.89 30.96 74 TBD a Values in the parenthesis refer to the last resolution shell. 177 Table III-6. X-ray crystallographic data and refinement statistics for apo side- by-side dimer of hCRABPII I31C mutant. I31C C21 118.18 61.57 71.42 90 122 90 3 1184723 22923 (2158) 98.90 (93.05) 29.06 (2.15) 6.2 (62) 33.17-2.17 (2.25 - 2.17) 21.75/30.98 0.008 0.96 54.84 14 TBD Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per Asymmetric Unit Total reflection Unique Reflection Completeness (%) Average I/s Rmerge Resolution (Å) (Last Shell) Rwork/ Rfree (%) RMS Bond Length (Å) RMS Bond Angle (°) Average B factor Number of water molecules PDB ID a Values in the parenthesis refer to the last resolution shell. 178 REFERENCES 179 REFERENCES Marianayagam, N. J.; Sunde, M.; Matthews, J. M., The power of two: protein 1. dimerization in biology. Trends in Biochemical Sciences 2004, 29 (11), 618-25. Smulski, C. R.; Decossas, M.; Chekkat, N.; Beyrath, J.; Willen, L.; Guichard, G.; 2. Lorenzetti, R.; Rizzi, M.; Eibel, H.; Schneider, P.; Fournel, S., Hetero-oligomerization between the TNF receptor superfamily members CD40, Fas and TRAILR2 modulate CD40 signalling. Cell Death & Disease 2017, 8 (2), e2601. 3. Needham, S. R.; Roberts, S. K.; Arkhipov, A.; Mysore, V. P.; Tynan, C. J.; Zanetti- Domingues, L. C.; Kim, E. T.; Losasso, V.; Korovesis, D.; Hirsch, M.; Rolfe, D. J.; Clarke, D. T.; Winn, M. D.; Lajevardipour, A.; Clayton, A. H.; Pike, L. J.; Perani, M.; Parker, P. J.; Shan, Y.; Shaw, D. E.; Martin-Fernandez, M. L., EGFR oligomerization organizes kinase- active dimers into competent signalling platforms. Nat Commun 2016, 7, 13307. 4. Fatmi, M. Q.; Chang, C. E., The role of oligomerization and cooperative regulation in protein function: the case of tryptophan synthase. PLoS Comput Biol 2010, 6 (11), e1000994. 5. Sato, K.; Nakano, A., Oligomerization of a cargo receptor directs protein sorting into COPII-coated transport vesicles. Molecular Biology of the Cell 2003, 14 (7), 3055-63. Klingenberg, M., Membrane protein oligomeric structure and transport function. 6. Nature 1981, 290 (5806), 449-54. 7. Hashimoto, K.; Panchenko, A. R., Mechanisms of protein oligomerization, the critical role of insertions and deletions in maintaining different oligomeric states. Proceedings of the National Academy of Sciences 2010, 107 (47), 20352-20357. Hurtley, S. M.; Helenius, A., Protein Oligomerization in the Endoplasmic Reticulum. 8. Annual Review of Cell Biology 1989, 5 (1), 277-307. 9. Fallas, J. A.; Ueda, G.; Sheffler, W.; Nguyen, V.; McNamara, D. E.; Sankaran, B.; Pereira, J. H.; Parmeggiani, F.; Brunette, T. J.; Cascio, D.; Yeates, T. R.; Zwart, P.; Baker, D., Computational design of self-assembling cyclic protein homo-oligomers. Nature chemistry 2017, 9 (4), 353-360. 10. Gabizon, R.; Friedler, A., Allosteric modulation of protein oligomerization: an emerging approach to drug design. Frontiers in Chemistry 2014, 2, 9. 11. Choi, M. L.; Gandhi, S., Crucial role of protein oligomerization in the pathogenesis of Alzheimer's and Parkinson's diseases. The FEBS journal 2018, 285 (19), 3631-3644. 180 12. Wilkaniec, A.; Czapski, G. A.; Adamczyk, A., Cdk5 at crossroads of protein oligomerization in neurodegenerative diseases: facts and hypotheses. Journal of Neurochemistry 2016, 136 (2), 222-33. Bennett, M. J.; Schlunegger, M. P.; Eisenberg, D., 3D domain swapping: a 13. mechanism for oligomer assembly. Protein Science 1995, 4 (12), 2455-68. 14. Rousseau, F.; Schymkowitz, J. W.; Itzhaki, L. S., The unfolding story of three- dimensional domain swapping. Structure 2003, 11 (3), 243-51. 15. Rousseau, F.; Schymkowitz, J.; Itzhaki, L. S., Implications of 3D domain swapping for protein folding, misfolding, and function. In Protein Dimerization and Oligomerization in Biology, Mattews, J. M., Ed. 2012; Vol. 747, pp 137-152. Koharudin, L. M.; Liu, L.; Gronenborn, A. M., Different 3D domain-swapped 16. oligomeric cyanovirin-N structures suggest trapped folding intermediates. Proceedings of the National Academy of Sciences of the United States of America 2013, 110 (19), 7702- 7. 17. Assar, Z.; Nossoni, Z.; Wang, W.; Santos, E. M.; Kramer, K.; McCornack, C.; Vasileiou, C.; Borhan, B.; Geiger, J. H., Domain-swapped dimers of intracellular lipid- binding proteins: evidence for ordered folding intermediates. Structure 2016, 24 (9), 1590- 8. 18. Barrientos, L. G.; Louis, J. M.; Botos, I.; Mori, T.; Han, Z.; O'Keefe, B. R.; Boyd, M. R.; Wlodawer, A.; Gronenborn, A. M., The domain-swapped dimer of cyanovirin-N is in a metastable folded state: reconciliation of X-ray and NMR structures. Structure 2002, 10 (5), 673-86. 19. Rousseau, F.; Schymkowitz, J. W.; Wilkinson, H. R.; Itzhaki, L. S., Intermediates control domain swapping during folding of p13suc1. The Journal of Biological Chemistry 2004, 279 (9), 8368-77. 20. Liu, Y.; Hart, P. J.; Schlunegger, M. P.; Eisenberg, D., The crystal structure of a 3D domain-swapped dimer of RNase A at a 2.1 Å resolution. Proceedings of the National Academy of Sciences of the United States of America 1998, 95 (7), 3437-42. 21. O'Neill, J. W.; Kim, D. E.; Johnsen, K.; Baker, D.; Zhang, K. Y., Single-site mutations induce 3D domain swapping in the B1 domain of protein L from peptostreptococcus magnus. Structure 2001, 9 (11), 1017-27. 22. Rousseau, F.; Schymkowitz, J.; Itzhaki, L. S., Implications of 3D domain swapping for protein folding, misfolding and function. Advances in Experimental Medicine and Biology 2012, 747, 137-52. 181 23. Zhang, J.; Wei, H.; Guo, X.; Hu, M.; Gao, F.; Li, L.; Zhang, S., Functional verification of the diphtheria toxin A gene in a recombinant system. Journal of Animal Science and Biotechnology 2012, 3 (1), 29. Bennett, M. J.; Eisenberg, D., Refined structure of monomeric diphtheria toxin at 24. 2.3 A resolution. Protein Science 1994, 3 (9), 1464-75. Zhukova Ia, F.; Dzhedzhula, H. H.; Radavs'kyi Iu, L., [The structural-functional 25. organization of the diphtheria toxin]. Mikrobiolohichnyi Zhurnal (Kiev, Ukraine : 1993) 1997, 59 (3), 91-101. Bennett, M. J.; Choe, S.; Eisenberg, D., Refined structure of dimeric diphtheria 26. toxin at 2.0 A resolution. Protein Science 1994, 3 (9), 1444-63. Liu, Y.; Hart, P. J.; Schlunegger, M. P.; Eisenberg, D., The crystal structure of a 27. 3D domain-swapped dimer of RNase A at a 2.1-Å resolution. Proceedings of the National Academy of Sciences of the United States of America 1998, 95 (7), 3437-3442. 28. Kim, J. S.; Soucek, J.; Matousek, J.; Raines, R. T., Catalytic activity of bovine seminal ribonuclease is essential for its immunosuppressive and other biological activities. The Biochemical Journal 1995, 308 ( Pt 2) (Pt 2), 547-50. Vitagliano, L.; Adinolfi, S.; Sica, F.; Merlino, A.; Zagari, A.; Mazzarella, L., A in bovine seminal 29. potential allosteric subsite generated by domain swapping ribonuclease. Journal of Molecular Biology 1999, 293 (3), 569-77. 30. Miller, K. H.; Marqusee, S., Propensity for C-terminal domain swapping correlates with increased regional flexibility in the C-terminus of RNase A. Protein Science 2011, 20 (10), 1735-44. 31. Miller, K. H.; Karr, J. R.; Marqusee, S., A hinge region cis-proline in ribonuclease A acts as a conformational gatekeeper for C-terminal domain swapping. Journal of Molecular Biology 2010, 400 (3), 567-78. Park, C.; Raines, R. T., Dimer formation by a "monomeric" protein. Protein Science 32. 2000, 9 (10), 2026-33. 33. Gotte, G.; Mahmoud Helmy, A.; Ercole, C.; Spadaccini, R.; Laurents, D. V.; Donadelli, M.; Picone, D., Double Domain Swapping in Bovine Seminal RNase: Formation of Distinct N- and C-swapped Tetramers and Multimers with Increasing Biological Activities. PLOS ONE 2012, 7 (10), e46804. Adinolfi, S.; Piccoli, R.; Sica, F.; Mazzarella, L., BS-RNase tetramers: an example 34. of domain-swapped oligomers. FEBS Lett 1996, 398 (2-3), 326-32. 182 35. Byeon, I. J.; Louis, J. M.; Gronenborn, A. M., A protein contortionist: core mutations of GB1 that induce dimerization and domain swapping. Journal of Molecular Biology 2003, 333 (1), 141-52. 36. Gronenborn, A. M., Protein acrobatics in pairs--dimerization via domain swapping. Current Opinion in Structural Biology 2009, 19 (1), 39-49. Louis, J. M.; Byeon, I. J.; Baxa, U.; Gronenborn, A. M., The GB1 amyloid fibril: 37. recruitment of the peripheral beta-strands of the domain swapped dimer into the polymeric interface. Journal of Molecular Biology 2005, 348 (3), 687-98. 38. Dey, B.; Lerner, D. L.; Lusso, P.; Boyd, M. R.; Elder, J. H.; Berger, E. A., Multiple antiviral activities of cyanovirin-N: blocking of human immunodeficiency virus type 1 gp120 interaction with CD4 and coreceptor and inhibition of diverse enveloped viruses. Journal of Virology 2000, 74 (10), 4562-9. 39. Matei, E.; Basu, R.; Furey, W.; Shi, J.; Calnan, C.; Aiken, C.; Gronenborn, A. M., Structure and Glycan Binding of a New Cyanovirin-N Homolog. The Journal of Biological Chemistry 2016, 291 (36), 18967-76. 40. Barrientos, L. G.; Gronenborn, A. M., The highly specific carbohydrate-binding protein cyanovirin-N: structure, anti-HIV/Ebola activity and possibilities for therapy. Mini Reviews in Medicinal Chemistry 2005, 5 (1), 21-31. Botos, I.; Wlodawer, A., Cyanovirin-N: a sugar-binding antiviral protein with a new 41. twist. Cellular and molecular life sciences : CMLS 2003, 60 (2), 277-87. Barrientos, L. G.; Matei, E.; Lasala, F.; Delgado, R.; Gronenborn, A. M., Dissecting 42. carbohydrate-Cyanovirin-N binding by structure-guided mutagenesis: functional implications for viral entry inhibition. Protein Engineering, Design & Selection : PEDS 2006, 19 (12), 525-35. Fontaine, S. N.; Brown, D. R., Mechanisms of prion protein aggregation. Protein 43. and Peptide Letters 2009, 16 (1), 14-26.44. Perlenfein, T. J.; Mehlhoff, J. D.; Murphy, R. M., Insights into the mechanism of cystatin C oligomer and amyloid formation and its interaction with beta-amyloid. The Journal of Biological Chemistry 2017, 292 (27), 11485- 11498. 45. Yang, S.; Levine, H.; Onuchic, J. N.; Cox, D. L., Structure of infectious prions: stabilization by domain swapping. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 2005, 19 (13), 1778-82. 46. Rambaran, R. N.; Serpell, L. C., Amyloid fibrils: abnormal protein assembly. Prion 2008, 2 (3), 112-7. 183 47. Shameer, K.; Sowdhamini, R., Functional repertoire, molecular pathways and diseases associated with 3D domain swapping in the human proteome. Journal of Clinical Bioinformatics 2012, 2 (1), 8. Jaskolski, M., 3D domain swapping, protein oligomerization, and amyloid 48. formation. Acta Biochimica Polonica 2001, 48 (4), 807-27. 49. Carey, J.; Lindman, S.; Bauer, M.; Linse, S., Protein reconstitution and three- dimensional domain swapping: benefits and constraints of covalency. Protein Science 2007, 16 (11), 2317-33. 50. Schlunegger, M. P.; Bennett, M. J.; Eisenberg, D., Oligomer formation by 3D domain swapping: a model for protein assembly and misassembly. Advances in Protein Chemistry 1997, 50, 61-122. 51. Sanson, B.; Wang, T.; Sun, J.; Wang, L.; Kaczocha, M.; Ojima, I.; Deutsch, D.; Li, H., Crystallographic study of FABP5 as an intracellular endocannabinoid transporter. Acta Crystallographica Section D 2014, 70 (2), 290-298. 52. Gershenson, A.; Gierasch, L. M.; Pastore, A.; Radford, S. E., Energy landscapes of functional proteins are inherently risky. Nat Chem Biol 2014, 10 (11), 884-891. 53. Karchin, J. M.; Ha, J. H.; Namitz, K. E.; Cosgrove, M. S.; Loh, S. N., Small Molecule-Induced Domain Swapping as a Mechanism for Controlling Protein Function and Assembly. Scientific Reports 2017, 7, 44388. 54. Ha, J. H.; Karchin, J. M.; Walker-Kopp, N.; Castaneda, C. A.; Loh, S. N., Engineered Domain Swapping as an On/Off Switch for Protein Function. Chem Biol 2015, 22 (10), 1384-93. 55. Aranko, A. S.; Oeemig, J. S.; Kajander, T.; Iwai, H., Intermolecular domain swapping induces intein-mediated protein alternative splicing. Nat Chem Biol 2013, 9 (10), 616-22. 56. Reis, J. M.; Burns, D. C.; Woolley, G. A., Optical control of protein-protein interactions via blue light-induced domain swapping. Biochemistry 2014, 53 (30), 5008- 16. Zegers, 57. I.; Deswarte, J.; Wyns, L., Trimeric domain-swapped barnase. Proceedings of the National Academy of Sciences of the United States of America 1999, 96 (3), 818-822. Tang, Y.; Reed, S. I., The Cdk-associated protein Cks1 functions both in G1 and 58. G2 in Saccharomyces cerevisiae. Genes & Development 1993, 7 (5), 822-832. 184 59. Zhang, Y.; Liang, J. Y.; Huang, S.; Ke, H.; Lipscomb, W. N., Crystallographic studies of the catalytic mechanism of the neutral form of fructose-1,6-bisphosphatase. Biochemistry 1993, 32 (7), 1844-1857. 60. Ramaswamy, S.; Eklund, H.; Plapp, B. V., Structures of Horse Liver Alcohol Dehydrogenase Complexed with NAD+ and Substituted Benzyl Alcohols. Biochemistry 1994, 33 (17), 5230-5237. 185 CHAPTER IV: ELUCIDATING THE MECHANISM OF HCRBPII-BASED FLUORSCENT PROTEIN TAGS USING HIGH-RESOLUTION CRYSTAL STRUCTURES IV-1. Fluorescent protein tags Fluorescence microscopy facilitates the in vitro studies of numerous crucial biological processes, including protein trafficking,1-3 protein-protein interactions,4-5 enzymatic activities,6-8 signaling,9 and others.10 The major tools in investigating the aforementioned processes are synthetic dyes that are post-translationally incorporated into the proteins11-13 and different fluorescent proteins.14 Although synthetic dyes can provide valuable labeling tools to target various organelles or proteins, this method suffers from the multiple washing steps needed to enhance the signal contrast. However, this problem can be solved using fluorescent protein tags, which do not require the exogenous chromophore for activation. Notably, fluorescent proteins can be genetically encoded to assist the labeling of a specific protein. The hallmark of this class of proteins is Green Fluorescent Protein (GFP),15 whose ligand is formed autocatalytically through the cyclization of three proximal residues (Ser65-Tyr66-Gly67) located inside the protein (Figure IV-1).16-17 Since the discovery of GFP (almost 60 years ago), this protein has been engineered to cover different colors ranging the full visible spectrum suitable for seven-color imaging of different cellular organelles.18 However, one of the pitfalls of GFP is that it demands molecular oxygen to maturate its intrinsic fluorophore, which makes it futile for studies that should be performed in a low-oxygen or anaerobic environment.19 186 Tyr66 O Gly67 HO HN Res65 R O O N H O N H Cyclization HO EGFP λex = 484 nm λem = 507 nm O O O N O N H N HO ±H+ HO O O N OH HN O HN R ) 2 O ( n o i t i a d x O -H2O O O N O N H N R Figure IV-1. The maturation of the GFP chromophore contains cyclization, oxidation, and dehydration steps to generate an emissive ligand. To address this problem and to increase the system’s flexibility, Prof. Borhan’s lab developed many fluorogenic chromophores 20-22 with aldehyde functionality that can ligate to the protein through protonated Schiff base (PSB) formation (in the same fashion as retinal) with an active Lys that was engineered inside the binding cavity (Q108K in hCRBPII and R111K in CRABPII).20, 23-25 Having demonstrated that the absorption maximum of all-trans-retinal bound hCRBPII can be rationally tuned over 220 nm through electrostatic perturbations,25 the focus is now on developing new fluorogenic chromophores whose emission can span the entire visible and near-IR spectrum. However, developing such a system requires understanding the microenvironment of the protein and managing the protein/ligand interactions. This can be facilitated using high- resolution crystal structures where the interaction can be visualized at atomic level. In this chapter, we will describe the power of using high-resolution protein crystallography and synthetic organic chemistry to create, interrogate, and examine novel functionalities in a new class of fluorescent protein tags. 187 IV-2. Elucidating the electronic effects on the photophysics of hCRBPII-based fluorescent proteins To elucidate the structure and electronic relationship of the chromophore in the rhodopsin mimic, Dr. Santos and Dr. Gholami from Prof. Borhan’s lab designed and synthesized a series of thiophene-based chromophore derivatives with aldehyde functionality, as shown in Figure IV-1. (Please see their Ph.D. theses for further information.) H O S N S H O F3C ThioFluor-5 ε360nm (EtOH) = 23,412 M-1cm-1 ThioFluor-9 ε416nm (EtOH) = 15,688 M-1cm-1 H O N S O H ThioFluor-10 ε369nm (EtOH) = 22,071 M-1cm-1 H O S N ThioFluor ε430nm (EtOH) = 27,021 M-1cm-1 H O N S ThioFluor-2 ε445nm (EtOH) = 29,763 M-1cm-1 H ε364nm (EtOH) = 17,708 M-1cm-1 ThioFluor-6 H O S N ThioFluor-3 ε418nm (EtOH) = 26,592 M-1cm-1 H O S O ThioFluor-4 ε384nm (EtOH) = 21,509 M-1cm-1 H O S CF3 N ThioFluor-7 ε340nm (EtOH/Et3N) = 20,682 M-1cm-1 H O S S N ThioFluor-8 ε472nm (EtOH) = 14,821 M-1cm-1 Figure IV-2. ThioFluor analogs synthesized by Prof. Borhan’s lab to investigate structure-property correlations of thiophene-based fluorophores with aldehyde functionality inside the binding pocket of the protein. The electron-donating groups were created on by substituting the para-N,N- dimethyl aniline found in ThioFluor with azetidine (ThioFluor-3) and methoxy (ThioFluor-4); the electron-withdrawing containing chromophore was synthesized with 188 trifluoromethyl in the para position (ThioFluor-5); and hydrogen was used as a reference (ThioFluor-6). The electron acceptor property of the iminium was altered by generating ThioFluor-7, in which a trifluoromethyl group was introduced in the β-position relative to the aldehyde. Additionally, to demonstrate the effect of conjugation across the π system, ThioFluor-8, ThioFluor-9, and ThioFluor-10 were synthesized. ThioFluor-2, as the shortest chromophore, was created to make a cyanine-reminiscent dye structure with high conjugation. To launch this study, we needed a protein platform to investigate the Table IV-1. Spectroscopic properties of ThioFluor-PSB analogs coupled with L1:L117E. (The following data was prepared by Dr. Santos.) Ligand λabs λem Φ ε (M-1cm-1) pKa ThioFluor-5 ThioFluor-6 ThioFluor-4 ThioFluor-10 ThioFluor-2 ThioFluor-3 ThioFluor ThioFluor-9 ThioFluor-8 ThioFluor-7 423 424 466 489 515 548 558 562 641 714 - 502 543 675 536 679 673 700 715 780 - - - 0.06 0.13 0.27 0.31 0.02 0.09 - 23,530 13,820 29,878 7,050 50,578 21,223 15,091 15,406 22,948 22,517 10.9 10.6 10.9 11.3 11.3 10.9 11.2 11.1 11.2 4.3 effects of various derivatizations on ThioFluor in terms of protein/chromophore behavior. We employed Q108K:K40L:T51V:T53S:R58W:Y19W:L117E (L1:L117E) as a template that can generate a high pKa form of chromophore-iminium through the acidic residue 189 (L117E) interaction with the iminium nitrogen. Furthermore, the presence of aromatic residues such as R58W and Y19W rigidifies the chromophore inside the binding cavity resulting in a well-ordered chromophore in the protein binding cavity (Figure IV-3). Using the aforementioned construct, Dr. Santos measured the Spectroscopic properties of the ThioFluor-PSB analogs, as enumerated in Table IV-1. Also, as shown in Figure Y60 T51V W106 Q108K L117E L119 Y19W R58W Figure IV-3. ThioFluor3 bound with the L1:L117E variant of hCRBPII. The aromatic residues, Y19W, R58W, and W106, and other proximal amino acids are highlighted in the binding pocket. IV-3, the crystal structures of the ThioFluor-PSB analogs were obtained at atomic resolution. Fortuitously, as expected, a well-ordered chromophore, characterized by the high quality of the electron density nestled in the binding pocket of the protein (Figure IV- 4).The absorption wavelength of each PSB-analogue portrayed the relative correlation between the maximum wavelength and degree of conjugation affected by the electron- withdrawing and electron-donating substituents, (Table IV-1). For instance, the most blue- shifted absorption wavelengths for chromophore/protein complexes were seen for 190 Q108K 2.8 Å L117E Q108K Q108K 3.0 Å L117E 2.5 Å L117E 2.8 Å L117E b. d. f. a. Q108K c. Q108K 2.9 Å L117E 2.8 Å L117E e. Q108K 3.0 Å L117E g. Q108K h. Q108K I. Q108K 2.7 Å L117E 2.5 Å L117E Figure IV-4. Crystal structures of Q108K:K40L:T51V:T53S:R58W:Y19W:L117E (L1:L117E) bound with ThioFluor analogs. a. L1:L117E crystal structure bound with ThioFluor. b. ThioFluor2. c. ThioFluor3. d. ThioFluor4. e. ThioFluor6. f. ThioFluor7. g. ThioFluor8. h. ThioFluor9. i. ThioFluor10. ThioFluor-5 (λabs = 423 nm) and ThioFluor-6 (λabs = 424 nm), containing the 191 trifluoromethyl group and hydrogen atom, respectively. Also, as shown in Table IV-I, the iminiums of these two complexes are non-emissive, but they both have a high pKa. Unfortunately, attempts at crystallizing ThioFluor5 were not successful. However, the crystal structure of ThioFluor-6 was obtained at atomic resolution. As shown in Figure IV-6e, the crystal structure depicts trans-iminium with L117E directly interacting with the nitrogen of iminium, explaining the high pKa of the chromophore. Table IV-1 shows that ThioFluor-4 contains an electron-donating methoxy group Q108K L117E Figure IV-5. Overlay of L1:L117E bound with ThioFluor6 (cyan) and L1:L117E bound with ThioFluor4 (pink). (RMSD: 0.1) that reveals a more red-shifted absorption wavelength (λabs = 466 nm) compared with hydrogen or trifluoromethyl. The high-resolution crystal structure of this mutant shows the same features as observed for ThioFluor6, such as trans-imine and L117E conformations, as mentioned above. The overlay of this structure with ThioFluor6 illustrates the similar trajectory of both chromophores (Figure IV-5). Therefore, the origin of the aforementioned difference in absorption wavelength should be exclusively due to the electron-donating character of the methoxy group. Nonetheless, the pigmented chromophore is still non-emissive (Table IV-1), highlighting the lack of ability of the methoxy group to create a push-pull system like the N, N-dimethyl group (ThioFluor). 192 The azetidine-based chromophore, ThioFluor-3, and the N, N-dimethyl chromophore, ThioFluor, depict larger red-shifted absorption wavelengths compared with ThioFluor6, Side veiw Back veiw a. b. dihedral: 24o dihedral: 5o dihedral: 24o dihedral: 5o dihedral: 22o dihedral: 13o dihedral: 22o dihedral: 13o Figure IV-6. a. L1:L117E crystal structure bound with ThioFluor with two presentations. The crystal structure depicted the almost-planarity of the N,N-dimethyl group with the chromophore’s backbone. b. L1:L117E crystal structure bound with ThioFluor3. The crystal structure depicted the almost-planarity of the azetidine group with the chromophore’s backbone. which only possesses a hydrogen atom in the para position. Looking at the crystal structures of both chromophores, ThioFluor and ThioFluor- 3 demonstrates that the origin of this change in absorption is solely due to the change in the electronic structure of the chromophore as a result of fixing two electron-donating groups in the para position that essentially extend the conjugation. Notably, the crystal structure portrays the near-planarity of both para-substituted analogs with the rest of 193 chromophore’s backbone, confirming the cited hypothesis. The planarity further explains the higher quantum yields for these two chromophores, as shown in Table IV-1. The placement of the trifluoromethyl group in the β-position, ThioFluor7 (λabs = 714 nm), to the aldehyde gave a more red-shifted absorption wavelength in comparison with ThioFluor (λabs = 558 nm). The high-resolution crystal structure obtained Y60 3.5 Å V62 3.2 Å V51 3.6 Å Q108K 2.5 Å L117E Figure IV-7. The binding pocket of ThioFluor7 bound with L1:L117E demonstrated the interacting residues. Two C-H…F interactions trifluoromethyl group and V62 and V6, S… π interaction which can be seen in all of the ThioFluor analogs with L1:L117E. and finally he low barrier hydrogen bond with L117E and nitrogen of iminium. for ThioFluor7 shows a cis double bond in the structure, with an overall trajectory similar to ThioFluor. Interestingly, the chromophore is not as emissive as ThioFluor, which might either be due to the cis conformation of the double bond or the presence of the trifluoromethyl group substitution. However, these hypotheses should be confirmed or ruled out through future experiments by synthesizing the trans isomer or photochemically producing the trans form from the cis isomer. As shown in Figure IV-7, the chromophore 194 also made S…π, C-H…F, and salt bridge interactions with Y60, V62, and L117E, respectively. As mentioned above, the bathochromic shifts were obtained either by installing electron-donating groups in the para position (ThioFluor and ThioFluor3) or introducing the electron-withdrawing group in the β-position relative to the aldehyde functional group (ThioFluor7). The next step was to investigate how changes in aromaticity can alter the photophysics of the chromophore. ThioFluor-8 was used to fulfil this purpose. As shown in the structure of the chromophore (Figure IV-4g), the phenyl ring was substituted with a thiophene, which maintains less resonance energy in comparison with phenyl. Therefore, the less-aromatic character of thiophene is expected to facilitate breaking the aromaticity and enhancing resonance through the π system. This would lead to lowering the energy gap between the HOMO and LUMO and a subsequent red-shifted absorption wavelength. a. 16.4 o b. Q108K Y60 3.7 Å 3.7 Å T53S 4.1 Å 4.5 Å Figure IV-8. a. Approximate coplanarity of the two thiophenes in ThioFluor8. b. Position of two thiophenes provides a suitable conformation for metal-binding. This ligand can be used in the future as a metal sensor, where metal-binding can tune the absorption and emission properties of ThioFluor8. The residues that are in proximity with the two rings are highlighted. Metal binding site that bridges chromophore and protein can be envisioned by mutation of these residues. As depicted in Table IV-1, ThioFluor-8 bound with the protein showed λabs = 641 nm, 195 which is much more red-shifted than ThioFluor. The high-resolution crystal structures of this chromophore were obtained at 1.6 Å resolution, demonstrating the near coplanarity between the two thiophene rings (Figure IV-8a). Therefore, the red-shifted absorption wavelength can be simply attributed to the electronic effects cited above. Another interesting feature of the chromophore, as shown in Figure IV-8b, is that the thiophene groups point in the same direction within the crystal structure. This trajectory is optimized a. ThioFluor8+protein ThioFluor8+protein+KAuCl4 Waiting 2 min Readding KAuCl4 b. c. d. 650 nm 650 nm 558 nm 545 nm Figure IV-9. a. Color changes upon: iminium formation of ThioFluor8/L1:L117E (Protein concentration ~35 µM - 0.5 equivalent ThioFluor was added., Adding 100 µM KAuCl4 to the mixture of ThioFluor8/L1:L117E, incubating the previous mixture for two minutes, readding 100 µM KAuCl4. b. The UV-vis spectrum of ThioFluor8/L1:L117E pre and post 100 µM KAuCl4 addition. c. The UV-vis spectrum of ThioFluor8/L1:L117E pre and post 100 µM CuSO4 addition as a control. d. The UV-vis spectrum of ThioFluor/L1:L117E pre and post 100 µM KAuCl4 addition as a control. to bind metals in a bidentate fashion. The metal-binding can further tune the fluorescence efficiency of the ligand by altering the aromaticity of the thiophenes. As a proof of principle, KAuCl4 was added to ThioFluor8/L1:L117E complex led to formation of short lived 545 nm species as shown in Figures IV-9a and IV-9b. The spectra of the controls including ThioFluor8/L1:L117E, as well as ThioFluor/L1:L117E delineated little changes upon 196 CuSO4 and KAuCl4 additions respectively. Although the complexation of protein with Au3+ was not stable, it shows a promising preliminary result of metal/ligand/protein interactions in an unoptimized binding pocket causing a hypsochromic shift in the absorption wavelength of ThioFluor8/L1:L117E complex. This approach can be further pursued by imposing novel mutations in the binding pocket that can create third or fourth coordination sites and increase the selectivity of the binding site for a specific metal, accomplished by both the ligand and the protein binding site. Specifically, to optimize this interaction, mutations were made on the Y60 and T53 (Y60C and T53C) positions in an ongoing project. Next, the replacement of thiophene with phenyl was investigated to yield ThioFluor-10 featuring two adjacent phenyl groups. In this case, a blue shift was observed in the absorption wavelength (λabs = 489 nm) comparable with ThioFluor (λabs = 558 nm), yet with a significant Stokes shift of 186 nm between the excitation and emission wavelengths (Table IV-1). a. twisted ICT (TICT) b. planarized ICT (PLICT) c. D’ D’ D’ D hu D rotation D A A A D’ D’ D’ D D D hu rotation A A A Figure IV-10. Representation of the geometric arrangement of the donor and acceptor groups upon excitation in: a. twisted intramolecular charge transfer (TICT) and b. planarized intramolecular charge transfer (PLICT) c. Dihedral angle between the two phenyl rings in ThioFluor10 measured 31.5o. 197 The crystal structure (at 1.8 Å resolution, Figure IV-4i) showed the trans-imine and other features similar to those observed for previous structures. The large change in absorption may be due to a decrease in delocalization of positive charge on the molecule as a result of enhancing the aromaticity by placing two phenyl rings instead of a phenyl and thiophene ring in ThioFluor. Looking at the crystal structure, the torsional angles between the two phenyl groups was 31.5o, resulting in the donor and acceptor groups being between planar and perpendicular orientations. The excitation of the chromophore similar to ThioFluor could lead to an intramolecular charge transfer (ICT) from donor to acceptor,26-28 with an ensuing rotation around the single bond linking the donor and acceptor groups. This reorientation in the excited state can proceed to achieve either a complete perpendicular conformation, which is characteristic of a twisted intramolecular charge transfer (TICT), or give rise to a planar orientation (PLICT), which could explain the large Stokes shifts observed (Figure IV-11). Although ruling out either of these mechanisms is not possible Y60 Q108K Figure IV-11. Overlay of L1:L117E crystal structures bound with ThioFluor (pink) and ThioFluor9 (green), showing a small change in trajectory, which affects the position of Y60. 198 with the ground state crystal structure, the TICT mechanism is more correlated with a low quantum yield (0.06), as listed in Table IV-1 for ThioFluor-10. Switching the order of the thiophene and phenyl groups (ThioFluor-9) did not lead to a significant change in the absorption wavelength of the chromophore. The crystal structure did not show a large difference from ThioFluor, as shown in Figure IV-11. Lastly, the crystal structure of the shortest chromophore (Figure IV-4b) surprisingly revealed a similar trajectory of the chromophore in the binding cavity, signifying the roles of aromatic residues in rigidifying the even-shorter chromophore (Figure IV-12). Additionally, for the first time, ThioFluor2 depicts a new conformation of thiophene. As shown in Figure IV-12b, the reason behind this new conformation might be due to the steric buttressing effect of Y60 (which can impinge on the sulfur atom of thiophene), along a. R58W b. Y60 W106 Y60 Q108K Y19W Q108K Figure IV-12. a. Binding pocket of L1:L117E bound with ThioFluor2, depicting how the aromatic residues sandwich the chromophore in the binding pocket. b. Overlay of L1:L117E bound with ThioFluor2 (cyan) and ThioFluor5 (pink) crystal structures, showing a new conformation of thiophene for ThioFluor2. with a higher flexibility of ThioFluor2 in the binding pocket. Since all of the crystal structures (other than ThioFluor-7, which shows a cis double bond) have a similar trajectory and imine configuration, we can confidently 199 conclude that the electronic effects are a significant factor contributing to the results observed in the absorption maxima, Stokes shifts, and quantum yields of the L1:L117E/ ThioFluor analog complexes. IV-3. Using high-resolution protein crystallography to optimize a ratiometric pH sensor Prof. Borhan’s lab was successful in creating ratiometric pH sensors, a prominent application of engineered fluorescent protein tags. The first attempt at creating such a system was achieved in hCRABPII, where two variants of the protein with more than 100 nm difference in absorption wavelength, as well as distinct pKas, were utilized for the absorption-based pH sensing. 24 Borhan’s lab further created a novel fluorescence-based ratiometric pH sensor in hCRBPII, using a julolidine fluorophore. 22 This design relies on a titratable amino acid side chain, L117E, that upon acidification, becomes protonated and alters the absorption of fluorophore, leading to a subsequent change in emission intensity at the same excitation. The mechanism of L117E controlling the photophysics of the chromophore is based on localizing and delocalizing the positive charge on the polyene, when L117E is deprotonated and protonated, respectively. 22 Although this approach reveals a concentration-independent pH measurement, it suffers from the fact that julolidine is not cell-permeable. 200 To address this problem, Dr. Santos employed the same system, this time using ThioFluor as a chromophore. As shown in Figure IV-13, the acidification of the L1:L117E/ThioFluor complex as a template led to a bathochromic change in wavelength from 558 nm to 614 nm. The plot of pH vs. absorption of L1:L117E/ThioFluor showed two pKas, 6.6. and 11.2, possibly related to L117E and iminium, respectively (Table IV- 2). Although the presence of L117E increased the rate of PSB formation and showed a ratiometric pH change with a fluorescent readout, it had the disadvantage of lowering the extinction coefficient by almost half. a. c. b. d. Q108K 2.9 Å L117E : 6.6 : 11.2 Figure IV-13. Crystal structure of ThioFluor bound with L1:L117E, showing how L117E acts as a counter- ion for the nitrogen of iminium. b. UV-vis spectrum of pH titration for the L1:L117E/ThioFluor complex. c. Plot of abs. vs. pH reveals two apparent pKas, 6.6 and 11.2 for L117E and iminium, respectively. d. Color changes upon acidification of the L1:L117E/ThioFluor complex. (The data for UV-vis. and pKa was provided by Dr. Santos.) 558 nm 614 nm 201 To increase the extinction coefficient of the L1:L117E / ThioFluor complex, A33W was introduced to L1:L117E to create L1:L117E:A33W. The presence of A33W led to a sufficient increase in the extinction coefficient (40,000 M-1cm-1) in a protonated form of L117E, while in the deprotonated state of L117E, the extinction was almost the same as L1:L117E (15,000 M-1cm-1). Furthermore, A33W insertion did not alter the quantum yield, resulting in a two-fold increase in brightness. The crystal structure of L1:L117E:A33W bound with ThioFluor was successfully determined at 1.3 Å resolution. Interestingly, the structure revealed a cis–iminium in contrast with the L1:L117E/ThioFluor structure. As shown in Figure IV-14, the overlay of L1:L117E/ThioFluor and L1:L117E:A33W/ ThioFluor shows that the bulky A33W forces the molecule more deeply into the binding pocket, resulting in the formation of the cis- iminium (Figure IV-14a). The deeper penetration into the binding cavity than Table IV-2. Effect of L117D versus L117E in the development toward a single fluorescent protein ratiometric sensor. hCRBPII mutant L1:L117E L1:L117E:A33W L1:L117D L1:L117D:Q4F λabsb λabsc ΔAbs pKad pKae Φf 558 539 525 529 614 628 618 623 56 89 93 94 11.2 11.3 10.9 11.0 6.6 8.0 6.7 5.7 0.31 0.28 0.30 0.31 t1/2 (min)g 82 241 14 8 E.R. h 7 - 21 16 L1 = Q108K:K40L:T51V:T53S:R58W:Y19W. b. λmax of absorbance at basic pH. c.λmax of absorbance at acidic pH. d. pKa corresponding to the iminium. e. pKa corresponding to L117D/E. f. Quantum yield measured at ~ pH 7.2. g. Kinetic measurements were performed at 23°C with 20 μM protein and 0.5 equiv ligand. PSB formation was monitored by UV-Vis at λmax for each complex over time at ~ pH 7.2. Data was then fit to second order kinetics. h. ER = the highest emission ratio collected as a result of excitation at 514 nm and 633 nm. The data for this table were provided by Dr. Santos. L1:L117E/ThioFluor, may enhance the electrostatic interaction of the chromophore and protein (Figure IV-14b). The pKa titration illustrated two pKas, 11.3 and 8, corresponding 202 to iminium and glutamate, respectively (Table IV-2). In this case, the high pKa of glutamate could be due to the cis-iminium formation, where the nitrogen of iminium is not in direct interaction with the cis-iminium. Also, in comparison with L1:L117E, a larger change between the two iminium absorption maxima was obtained upon acidification (89 nm vs. 56 nm). This might be due to the presence of A33W in the mouth of the binding cavity, which leads to the further delocalization of the charge on the chromophore by dispelling some of the ordered water molecules out of the binding cavity, as previously reported for a. b. Q4 W106 L117E L117E Q4 Q108K 2.9 Å 2.8 Å Q108K A33W A33W A Figure IV-14. The crystal structure of L1:L117E:A33W/ThioFluor depicts the cis-iminium cis-iminium in L1:L117E:A33W/ThioFluor further stabilized through a water mediated hydrogen bond between Q4 and iminium (the electron density is contoured to 1σ.). b. The Overlay of L1:L117E/ThioFluor(orange) vs. L1:L117E:A33W/ThioFluor (cyan) revealed the presence of A33W sterically pushes the chromophore and locates it deeper in the binding pocket. 203 retinylidene variants.25 Although this construct had the advantage of higher brightness and a larger separation between the two iminium maxima, the rate of PSB formation was slower than that for L1:L117E (Table IV-2). To further optimize the system, Dr. Santos investigated the effect of L117D vs. a. Acetate ion W106 b. Q108K Q4 Q108K L117D Q128 c. Q108K 5.1 Å L117E L117D 2.6 Å 2.2 Å 2.9 Å Q128 4.3 Å L117E 5.1 Å L117D Q128 Figure IV-15. a. Crystal structure of L1:L117D/ThioFluor shows a cis-iminium configuration. Unfortunately, in this case the acetate ion coming from the crystallization conditions occupied the water molecule site between Q4 and the nitrogen of imine. b. Overlay of L1:L117D/ThioFluor (purple) and L1:L117E/ThioFluor depicts a different conformation of L117D vs. L117E, which is very far from the nitrogen of iminium. c. Overlay of two cis-iminium structures coming from L1:L117D/ThioFluor (purple) and L1:L117E:A33W/ThioFluor highlights the importance of A33W in locating the chromophore deeper in the binding pocket. A33W L117E. L1:L117D (L1:L117D) was created and the pKa titration was performed, showing similar effects on the absorption wavelength of the imine upon acidification. The two pKas 204 were also measured at 11.0 and 7.0 for iminium and aspartate respectively. The crystal structure of this construct was achieved at 1.0 Å resolution, demonstrating the cis-iminium formation (Figure IV-15a). The overlay of this structure with L117E depicts the large change in the side chain of Lys108, which is a further indication of the different isomeric forms of imine. Additionally, L117D, in contrast to L117E, did not interact with the nitrogen of iminium; instead making a hydrogen bond with Gln128 (Figure IV-15b). This could either be due to the cis vs. trans configuration of iminium in L117D vs. L117E, leading to an increased distance between the L117D position and the nitrogen of iminium, or it could be due to the crystallization pH (4.0-4.5) leading to the protonation of L117D and a change in its conformation. To find an answer to this question, we needed to obtain the crystal structure of L117D containing a construct with the trans conformation of the imine bond. As shown in Figure IV-13, the cis-iminium in L1:L117D is stabilized through a water- mediated hydrogen bond with Q4. The hydrogen bond interaction between Q4 and the nitrogen of the cis-iminium was removed though the Q4F mutation to yield a L1:L117D:Q4F construct that presumably would lead to the formation of the trans form of iminium by destabilizing the cis-iminium isomer. The crystal structure of the L1:L117D:Q4F/ThioFluor complex was determined at 1.6 Å resolution, Figure IV-16. The presence of the Q4F mutation caused a one-unit decrease in the pKa of L117D (which makes this construct useful for physiological pH-sensing applications (working pH 5.7 to 7.7) (Table IV-2). As shown in Figure IV-15, the structure depicts a trans-iminium configuration with the exact same conformation of L117D as observed in L1:L117D, where L117D interacted directly with Q128. This observation suggests that this conformation of L117D might 205 correspond to the protonated form of L117D, which is produced as a result of a low pH of crystallization or L117D conformation remains in the same way in different pHs and remotely affects the chromophore photophysics. Unfortunately, attempts to grow the crystals in basic pH or soaking the crystals in higher pH cryogenic buffers were not a. Q4F Q108K 4.5 Å 2.6 Å L117D Q128 b. Q4F Q108K L117D Q128 Figure IV-16. a. The crystal structure of L1:L117D:Q4F/ThioFluor demonstrates the trans-imine configuration and the same conformation of L117D as observed for L1:L117E/ThioFluor. b. The overlay of L1:L117D:Q4F/ThioFluor (yellow) and L1:L117E/ThioFluor (purple). Interestingly the position of nitrogen of imine is almost the same for both structures. In the pH of crystallization (~4.5) both aspartate and iminium would be anticipated to be protonated. 206 successful. In the latter case, the structure did not show any tangible changes in different pHs of soaking. However, this hypothesis should be further confirmed by neutron diffraction, where the visualization of hydrogen atoms’ positions on L117D and the iminium nitrogen in different pHs can be accomplished. IV-4. Elucidating the mechanism of excited-state proton transfer using high- resolution crystal structures Excited-state proton transfer (ESPT) is a mechanism that largely ensues through intermolecular or intramolecular hydrogen bonding. Consider a ground state of A1– H···A2 (where A1 and A2 are electronegative atoms, such as N or O), in which two π- conjugated moieties or molecules are joined together through hydrogen bonding. Upon excitation at SB absorption emssion of PSB excitation at PSB absorption emssion of PSB pH 7.1 λabs = 392 nm λem = 621 nm 0.2 0.1 ) . u . a ( s b A 0 400 500 Wavelength (nm) 600 700 8 106 7 106 6 106 5 106 4 106 3 106 2 106 1 106 0 800 l F u o r e s c e n c e I n t e n s i t y ( c p s ) ) . u . a ( s b A pH 4.5 λabs = 612 nm λem = 687 nm 0.2 0.1 0 400 500 Wavelength (nm) 600 700 1.5 106 1 106 5 105 l F u o r e s c e n c e I n t e n s i t y ( c p s ) 0 800 N S N Lys108 pKa N S H N Lys108 IV-17. Figure of Q108K:K40E:T53A:R58W:Q38F:Q4F:Y19W/ThioFluor SB (pH 7.1) and PSB (pH:4.5). The data were provided by Dr. Santos. emission spectra UV-vis and excitation of the system with light, and as a result of ESPT, a new state is achieved where the hydrogen atom transfers from one moiety to another to generate an A1···H–A2 system. In this case, single or dual fluorescence can be observed for the ESPT system, 207 hinging upon the energy barrier. 29 This mechanism has been observed in many biological processes, such as photosystem II,30 green fluorescent proteins,31 bacteriorhodopsin, 32 and others.29 The first observation of ESPT in the hCRBPII binding pocket was made by Prof. Borhan’s lab. Dr. Santos showed that in certain mutants, when the protein/chromophore excited at the Schiff base (SB) wavelength, two bands appear. One of the bands corresponds to the SB emission and the other is more similar to the PSB emission. Further studies demonstrated that the placement of acidic residues in certain positions within the vicinity of the chromophore causes this dual emission, which leads to a large Stokes shift between the excitation and emission wavelength or ESPT. Additional exploration of this topic revealed some criteria for an ideal ESPT system: the chromophore should generate the imine, not iminium, with an active Lys 108, and the pKa of the chromophore should be kept low at all times to maintain the SB. To elucidate the mechanism of ESPT, after screening many variants and conditions for crystallization, we successfully obtained high resolution structures of two hCRBPII variants, bound to two thiophene-based chromophores that showed a high yield of ESPT (Figure IV-17), one in the monomeric fold and the other in the trimeric fold of hCRBPII. As shown in Figure IV-18, the crystal structure of ThioFluor bound to Q108K:K40E:T53A:R58W:Q38F:Q4F:Y19W (L2) was determined at 1.6 Å resolution. The crystal structure depicted four molecules in the asymmetric unit. Interestingly, the electron density for chain A was correlated with the trans-imine, chain B was refined as a mixture of the cis and trans-imine with about 0.5 occupancy for each, and chain C 208 corresponded with the cis-imine. The overlay of these three structures reveals similar a. Q108K 4.8 Å K40E c. 5.0 Å Q108K 4.8 Å K40E e. b. W106 Q108K 5.0 Å K40E d. Q108K 4.8 Å K40E T53 3.0 Å 2.8 Å Y60 3.0 Å Figure IV-18. Crystal structure of Q108K:K40E:T53A:R58W:Q38F:Q4F:Y19W (L2/ThioFluor complex shows four molecules in the asymmetric units, where the electron density of the chromophore in three chains indicates ligand binding. a. ThioFluor bound to the protein in chain A depicts the trans imine. b. ThioFluor bound to the protein in chain C depicts an electron density more correlated with a cis iminine. c. ThioFluor bound to the protein in chain B depicts an electron density more correlated with a mixture of cis and trans iminine. d. K40E conformation that is not interacting with the nitrogen of imine. It is expected that this conformation would cause a suppressed pKa for imine, leading to an SB population. K40E is held in this conformation through a water-molecule interaction with the T53 and Y60 main chain. The electron density is contoured to 1σ for all the structures. 209 trajectories of the chromophore for each chain (Figure IV-18). Remarkably, in all of these Q108K 5.0 Å K40D IV-19. Crystal structure of ThioFluor3 bound with a Figure form of Q108K:K40D:T53A:R58L:Q4F:Q38F reveals a K40 conformation that is far away from the nitrogen of iminium. trimeric chains, K40E does not interact directly with the nitrogen of iminium, which can explain the low pKa of the imine. As mentioned earlier, the ground-state imine should be kept at a low pKa for an efficient ESPT, a requirement that is clearly met in the atomic-resolution crystal structure of the L2/ThioFluor complex. How is K40E, an acidic residue, kept away from the nitrogen of imine? The crystal structure signifies that an ordered water molecule, mediated between Thr51, the oxygen atom of the Tyr60 main chain, and the nitrogen atom of the Ala53 main chain, holds K40E in that specific conformation. The crystal structure of Q108K:K40D:T53A:R58L:Q4F:Q38F trimer-bound ThioFluor3 elucidates a similar scenario (see chapter III for further information) where K40D is located in a remote position from the nitrogen of imine. 210 Additionally, the crystal structure of another chromophore-bound complex showing the ESPT was obtained (the chromophore was FR1 and was synthesized by Dr. Wei Sheng). Consistent with ThioFluor bound proteins demonstrarting ESPT, FR1 bound a. c. b. K40H 2.9 Å Q108K Q108K 4.8 Å K40H R111K Q108K Q128 2.7 Å 2.8 Å 3.9 Å 3.1 Å N13 3.1 Å R35 d. ESPT active conformation K40H PSB stabilizing conformation Q108K Q108K 4.8 Å 5.0 Å K40E K40H Figure IV-20. Crystal structure of Q108K:K40H:T53A:R58L:Q4F:Q38F (L3) bound with FR1 shows four molecules in the asymmetric unit that are fully occupied by FR1. a. Chains A and B show the same conformation of K40H, in which it directly interacts with the nitrogen of iminium. b. Chains C and D show the same conformation of K40H, in which it does not interact with the nitrogen of iminium (4.8 Å distance). Instead, K40H gets locked into this conformation by hydrogen bond networks containing the residues highlighted in the Figure. c. Left: Overlay of chain A and chain C, indicating that the only difference between the two chains is a different conformation of K40H. The K40H conformation that is closer to the nitrogen of imine would be expected to stabilize the iminium, and the other conformation would be an active ESPT ground-state conformation. Right: Overlay of L3/FR1 (green) and L2/ThioFluor (purple) crystal structures further confirms the ESPT active conformer of K40H. Q108K:K40H:T53A:R58L:Q4F:Q38F (L3) variant of hCRBPII showed an ESPT but with 211 a low quantum yield in comparison with L2/ThioFluor (Figure IV-21). Additionally, in contrast to the L2/ThioFluor complex that dominantly forms SB in the ground state, L3/FR1 shows some population of PSB as well as SB. The crystal structure of the L3/FR1 complex at 1.6 Å resolution depicts the presence of four molecules in the asymmetric unit with the same trajectory for the chromophore (Figure IV-19) in all four. However, as shown in Figure IV-19, the conformation of K40H varies in the different chains. In two chains (A and B), K40H maintains the interact ion with the nitrogen of iminium, and in the other two chains (C and D), K40H rotates away from the nitrogen of the iminium, generating a water- mediated hydrogen bond network with Q128, N13, and R35. It is expected that the K40H with the conformation proximal to the nitrogen of the iminium would stabilize the positive charge on the nitrogen of iminium and lead to PSB formation (Figure IV-20c). This observation is also is in agreement with the UV-vis spectrum of the L3/FR1 complex that depicts the mixture of SB and PSB in the ground state, Figure IV-21. As mentioned earlier, Figure IV-21. UV-vis and emission spectra of L3/FR1 complex. The data were provided by Dr. Sheng. 212 the K40H conformation which is proximal to the iminium is not favored for ESPT, because the chromophore is expected to have a high pKa and already be protonated in the ground state. On the other hand, in chains C and D, where the K40H is in a distal position relative to the nitrogen of iminium, it would be expected that the nitrogen of imine maintains a suppressed pKa. The latter conformation of K40H is anticipated to be an ideal conformation that gives ESPT. The overlay of L2/ThioFluor and L3/ThioFluor further proves the aforementioned points, in that K40E is always distant from the nitrogen of imine when L1/ThioFluor has a higher quantum yield of ESPT, Figure IV-20d. Based on the high-resolution crystal structures offered above and biophysical data, we propose the key residues (such as K40E, K40H, and K40D) should be kept distant from the nitrogen of iminium in the ground state where they maintain the low pKa imine. However, in response to the excitation of chromophore and altering its pKa, a transient proton translocation to the imine ensues, eventuating in a dual emission and what is called ESPT. IV-5. Conclusion and future direction In this chapter, several crystal structures for fluorophore bound hCRBPII variants were obtained. Through using atomic resolution crystal structures the microenvironment of the chromophore inside the protein binding pocket was visualized and the mechanism of hCRBPII fluorescent protein tags in pH ratiometric sensor and ESPT were discussed. The ability to use high-resolution protein crystallography to elucidate the mechanism of the engineered hCRBPII makes this protein a unique and well-studied engineered protein that can be further employed for a variety of protein-engineering applications. 213 The high-resolution crystal structures of ThioFluor analogs, as well as their low mosaicity, provide great opportunities to employ them for time resolved and neutron diffraction experiments. The ability to capture the intermediates of the ESPT process with time-resolved x-ray crystallography would provide valuable knowledge for the future design and improvement of fluorescent protein tags. Furthermore, neutron diffraction experiments would give us a clear picture of hydrogen atoms distribution in the protein binding pocket in different pHs. IV-6. Experimental procedures IV-6-1.ThioFluor analog synthesis Please see Dr. Elizabeth M. Santos and Dr. Hadi Gholami’s PhD theses. IV-6-2. FR1 synthesis Please see Dr. Wei Sheng’s PhD thesis. IV-6-3. Protein expression and purification The same protocol mentioned in Chapter II was used here with this modification that the expression was performed at OD~0.9-1.0 at 23°C for ~20 hours. IV-6-4. Crystallization conditions Each mutant protein was concentrated between 6 and 8 mg/mL in a buffer containing 10 mM Tris, 150 mM NaCl, pH = 8, and was incubated with four equivalent chromophores. Crystals were grown via the hanging drop vapor diffusion method using 1 μL of protein solution and 1 μL of crystallization solution in the drop and 1 mL of crystallization solution in the reservoir. The best crystals grew using a crystallization solution of 25-30% PEG 4000, 0.1 M sodium acetate, pH=4.0-4.8, 0.1 M ammonium acetate. The crystals were soaked in a 214 cryoprotectant solution (30% PEG 4000, 0.1 M sodium acetate pH = 4.5, 0.1 M ammonium acetate with 20% glycerol), flash frozen in loops in liquid nitrogen, and stored in a liquid nitrogen Dewar prior to data collection. 215 Table IV-3. X-ray crystallographic data and refinement statistics for hCRBPII L1:L117E bound with ThioFluor2 , ThioFluor3, and ThioFluor4 analogs. ThioFluor2-L1:L117E ThioFluor3-L1:L117E ThioFluor4- Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per asymmetric unit Total reflection Unique reflection Completeness (%) Average I/s Rmerge Resolution (Å) (last shell) Rwork/ Rfree (%) RMS bond length (Å) RMS bond angle (°) Average B factor Number of water molecules PDB ID P 21 29.26 66.72 63.95 90.00 90.78 90.00 2 1799743 37540 (3548) 96.34 (91.23) 37.3 (2.1) 0.069 (0.950) 26.74-1.51 (1.56-1.51) 19.69/22.88 0.008 0.93 24.17 265 TBD C21 29.41 66.65 64.15 90.00 90.81 90.00 2 685606 15679 (1524) 96.92 (93.30) 28.97(4.3) 0.047(0.250) 32.07-1.607 (1.65-1.61) 19.43/22.03 0.006 0.89 24.53 39 TBD L1:L117E C21 29.54 66.45 63.86 90.00 90.38 90.00 2 1468011 30407 (2898) 96.91 (93.15) 17.2 (2.2) 0.080 (0.534) 24.81- 1.28 (1.33- 1.28) 16.75/18.32 0.008 1.06 13.82 1.06 TBD a Values in parentheses refer to the last resolution shell. (L1:L117E: Q108K:K40L:T53S:R58W:Y19W:L117E) 216 Table IV-4. X-ray crystallographic data and refinement statistics for hCRBPII L1:L117E variants bound with ThioFluor6 , ThioFluor7, and ThioFluor8 analogs. ThioFluor6- ThioFluor7- ThioFluor8- Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per asymmetric unit Total reflection Unique reflection Completeness (%) Average I/s Rmerge Resolution (Å) (last shell) Rwork/ Rfree (%) Bond length (Å) Bond angle (°) Average B factor Number of water molecules PDB ID L1:L117E L1:L117E L1:L117E P21 29.30 66.84 64.07 90.00 90.75 90.00 2 1589460 30514 (2883) 96.47 (92.67) 37.8 (2.3) 0.058 (0.679) 33.42-1.62 C21 29.35 67.10 64.02 90.00 90.37 90.00 1 633336 25044 (2501) 96.04 (96.19) 31.4 (3.16) 0.042 (0.365) 29.72 - 1.37 C21 29.30 66.47 64.28 90.00 90.37 90.00 1 372414 14611 (1373) 97.20 (92.45) 28.6 (2.5) 0.069 (0.990) 29.52 -1.64 (1.68-1.62) (1.42 -1.37) 22.05/24.64 18.87/20.77 (1.70- 1.64) 17.26/21.89 0.008 0.96 25.41 121 TBD 0.007 0.85 18.20 134 TBD 0.008 0.91 21.11 66 TBD a Values in parentheses refer to the last resolution shell. (L1:L117E: Q108K:K40L:T53S:R58W:Y19W:L117E) 217 Table IV-5. X-ray crystallographic data and refinement statistics for hCRBPII variants bound with ThioFluor analogs. ThioFluor9- ThioFluor10- ThioFluor-L1:L117E Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) L1:L117E L1:L117E C21 29.396 66.751 64.067 90.00 90.30 90.00 2 287594 P1 29.40 66.74 64.22 89.73 90.28 90.10 4 984824 C21 29.618 66.614 64.119 90.00 90.95 90.00 1 661219 19317 (1927) 93.38 (92.91) 41.29 (4.8) 0.041(0.482) 33.38 - 1.48 17.52/19.35 (1.532-1.479) 0.008 44915 (4209) 96.48 (89.43) 24.07 (2.6) 0.065 (0.614) 33.37 - 1.79 21.21/24.78 (1.85 - 1.786) 0.008 0.94 17.29 45 TBD 0.83 26.03 330 TBD 18119 (1752) 98.07 (95.10) 2.16 (34.4) 0.05 (0.478) 33.31 - 1.54 (1.59 - 1.54) 17.72/21.21 0.006 0.85 23.78 84 TBD Molecules per asymmetric unit Total reflection Unique reflection Completeness (%) Average I/s Rmerge Resolution (Å) (last shell) Rwork/ Rfree (%) Bond length (Å) Bond angle (°) Average B factor Number of water molecules PDB ID a Values in parentheses refer to the last resolution shell. (L1:L117E: Q108K:K40L:T53S:R58W:Y19W:L117E) 218 Table IV-6. X-ray crystallographic data and refinement statistics for hCRBPII variants L1:L117E, L1:L117D, and L1:L117D:Q4F bound with ThioFluor analogs. ThioFluor ThioFluor- ThioFluor- L1:L117E:A33W L1:L117D L1:L117D:Q4F Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per asymmetric unit Total reflection Unique reflection Completeness (%) Average I/s Rmerge C21 28.918 66.617 63.262 90 90.981 90 1 851159 32216 (3049) 99.10 (94.18) 35.5 (2.1) 0.059 (0.44) C21 29.233 66.641 63.939 90 90.474 90 1 1199572 63103 (5744) 93.09 (85.35) 44.3(1.5) 0.059 (0.727) 29.55-0.991 (1.027- 0.991) 19.17/21.46 0.005 0.88 16.73 191 TBD C21 29.196 67.337 64.025 90 89.311 90 1 1096273 16377 (1601) 99.90 (99.38) 3.04 (36.6) 0.14 (1.9) 29.8 - 1.599 (1.657- 1.599) 18.57/21.37 0.005 0.84 25.98 74 TBD Resolution (Å) (last shell) 33.31- 1.258 (1.303- 1.258) Rwork/ Rfree (%) Bond length (Å) Bond angle (°) Average B factor Number of water molecules PDB ID 16.95/18.59 0.008 0.95 16.71 179 TBD a Values in parentheses refer to the last resolution shell. 219 Table IV-7. X-ray crystallographic data and refinement statistics for hCRBPII L2, L3, and Q108K:K40D:T53A:R58L:Q38F:Q4F variants bound with ThioFluor and FR1, and ThioFluor3 analogs. ThioFluor-L2 FR1-L3 ThioFluor3 P1 36.766 54.99 69.805 109.338 97.797 102.603 4 970285 54659 (5195) 96.19 (91.67) 24.6(1.4) 0.044 (0.51) 34.95 -1.671 P1 36.687 54.515 63.931 89.996 86.63 77.686 4 950433 55438 (5085) 85.90 (79.16) 22.50 (2.12) 0.046 (0.53) 32.04 - 1.594 (1.731 -1.671) (1.651- 1.594) 18.38/22.94 19.67/23.27 0.007 0.86 29.66 405 TBD 0.008 1.18 24.08 235 TBD Q108K:K40D:T53 A:R58L:Q38F:Q4F -trimer P22121 36.932 102.41 108.478 90 90 90 3 1077503 14902 (1453) 99.89 (99.83) 18.88 (1.42) 0.097 (1.251) 34.74-2.5 (2.589-2.5) 20.76/27.97 0.008 1.24 50.39 77 TBD Space group a(Å) b(Å) c (Å) a(° ) b(° ) d(° ) Molecules per asymmetric unit Total reflection Unique reflection Completeness (%) Average I/s Rmerge Resolution (Å) (last shell) Rwork/ Rfree (%) Bond length (Å) Bond angle (°) Average B factor Number of water molecules PDB ID a Values in parentheses refer to the last resolution shell. 220 REFERENCES 221 REFERENCES Shintani, T.; Reggiori, F., Fluorescence microscopy-based assays for monitoring 1. yeast Atg protein trafficking. Methods Enzymology 2008, 451, 43-56. Botelho, H. M.; Uliyakina, I.; Awatade, N. T.; Proença, M. C.; Tischer, C.; 2. Sirianant, L.; Kunzelmann, K.; Pepperkok, R.; Amaral, M. D., Protein traffic disorders: an effective high-throughput fluorescence microscopy pipeline for drug discovery. Scientific Reports 2015, 5, 9038. Gaietta, G.; Deerinck, T. J.; Adams, S. R.; Bouwer, J.; Tour, O.; Laird, D. W.; 3. Sosinsky, G. E.; Tsien, R. Y.; Ellisman, M. H., Multicolor and electron microscopic imaging of connexin trafficking. Science 2002, 296 (5567), 503-507. Massoud, T. F.; Paulmurugan, R.; De, A.; Ray, P.; Gambhir, S. S., Reporter gene 4. imaging of protein-protein interactions in living subjects. Current Opinion in Biotechnology 2007, 18 (1), 31-7. Xing, S.; Wallmeroth, N.; Berendzen, K. W.; Grefen, C., Techniques for the 5. Analysis of Protein-Protein Interactions in Vivo. Plant physiology 2016, 171 (2), 727-58. Bigelow, C. E.; Vishwasrao, H. D.; Frelinger, J. G.; Foster, T. H., Imaging 6. enzyme activity with polarization-sensitive confocal fluorescence microscopy. Journal of Microscopy 2004, 215 (Pt 1), 24-33. 7. Welder, F.; McCorquodale, E. M.; Colyer, C. L., Proteinase assay by capillary electrophoresis employing fluorescence-quenched protein-dye conjugates. Electrophoresis 2002, 23 (11), 1585-90. 8. Turunen, P.; Rowan, A. E.; Blank, K., Single-enzyme kinetics with fluorogenic substrates: lessons learnt and future directions. FEBS Letters 2014, 588 (19), 3553- 3563. Svoboda, K. K. H.; Reenstra, W. R., Approaches to studying cellular signaling: a 9. primer for morphologists. The Anatomical Record 2002, 269 (2), 123-139. 10. Süel, G., Chapter thirteen - Use of Fluorescence Microscopy to Analyze Genetic Circuit Dynamics. In Methods in Enzymology, Voigt, C., Ed. Academic Press: 2011; Vol. 497, pp 275-293. 11. Raymo, F. M., Photoactivatable Synthetic Dyes for Fluorescence Imaging at the Nanoscale. The Journal of Physical Chemistry Letters 2012, 3 (17), 2379-85. 222 12. Combs, C. A., Fluorescence microscopy: a concise guide to current imaging methods. Current protocols in neuroscience 2010, Chapter 2, Unit2.1. Tsien, R. Y.; Waggoner, A., Fluorophores for Confocal Microscopy. In Handbook 13. of Biological Confocal Microscopy, Pawley, J. B., Ed. Springer US: Boston, MA, 1995; pp 267-279. Billinton, N.; Knight, A. W., Seeing the Wood through the Trees: A Review of 14. Techniques for Distinguishing Green Fluorescent Protein from Endogenous Autofluorescence. Analytical biochemistry 2001, 291 (2), 175-197. 15. Patterson, G. H.; Knobel, S. M.; Sharif, W. D.; Kain, S. R.; Piston, D. W., Use of the green fluorescent protein and its mutants in quantitative fluorescence microscopy. Biophysical Journal 1997, 73 (5), 2782-2790. 16. Ma, Y.; Sun, Q.; Zhang, H.; Peng, L.; Yu, J.-G.; Smith, S. C., The mechanism of cyclization in chromophore maturation of green fluorescent protein: a theoretical study. The Journal of Physical Chemistry B 2010, 114 (29), 9698-9705. 17. Barondeau, D. P.; Kassmann, C. J.; Tainer, J. A.; Getzoff, E. D., Understanding GFP chromophore biosynthesis:  controlling backbone cyclization and modifying post- translational chemistry. Biochemistry 2005, 44 (6), 1960-1970. 18. Chudakov, D. M.; Lukyanov, S.; Lukyanov, K. A., Fluorescent proteins as a toolkit for in vivo imaging. Trends Biotechnol 2005, 23 (12), 605-13. Jensen, E. C., Use of Fluorescent Probes: Their effect on cell biology and 19. limitations. The anatomical record 2012, 295 (12), 2031-2036. 20. Sheng, W.; Nick, S. T.; Santos, E. M.; Ding, X.; Zhang, J.; Vasileiou, C.; Geiger, J. H.; Borhan, B., A near-infrared photoswitchable protein–fluorophore tag for no-wash live cell imaging. Angewandte Chemie International Edition 2018, 57 (49), 16083-16087. Yapici, I.; Lee, K. S.; Berbasova, T.; Nosrati, M.; Jia, X.; Vasileiou, C.; Wang, W.; 21. Santos, E. M.; Geiger, J. H.; Borhan, B., "Turn-on" protein fluorescence: in situ formation of cyanine dyes. Journal of the American Chemical Society 2015, 137 (3), 1073-80. Berbasova, T.; Tahmasebi Nick, S.; Nosrati, M.; Nossoni, Z.; Santos, E. M.; 22. Vasileiou, C.; Geiger, J. H.; Borhan, B., A genetically encoded ratiometric pH probe: wavelength regulation-inspired design of pH indicators. Chembiochem : a European Journal of Chemical Biology 2018, 19 (12), 1288-1295. 23. Wang, W.; Geiger, J. H.; Borhan, B., The photochemical determinants of color vision: revealing how opsins tune their chromophore's absorption wavelength. 223 BioEssays : News and Reviews in Molecular, Cellular and Developmental Biology 2014, 36 (1), 65-74. Berbasova, T.; Nosrati, M.; Vasileiou, C.; Wang, W.; Lee, K. S.; Yapici, I.; Geiger, 24. J. H.; Borhan, B., Rational design of a colorimetric pH sensor from a soluble retinoic acid chaperone. Journal of the American Chemical Society 2013, 135 (43), 16111-9. 25. Wang, W.; Nossoni, Z.; Berbasova, T.; Watson, C. T.; Yapici, I.; Lee, K. S.; Vasileiou, C.; Geiger, J. H.; Borhan, B., Tuning the electronic absorption of protein- embedded all-trans-retinal. Science 2012, 338 (6112), 1340-3. 26. Haberhauer, G.; Gleiter, R.; Burkhart, C., Planarized Intramolecular Charge Transfer: A Concept for Fluorophores with both Large Stokes Shifts and High Fluorescence Quantum Yields. Chem-Eur J 2016, 22 (3), 971-978. Sasaki, S.; Drummen, G. P. C.; Konishi, G., Recent advances in twisted 27. intramolecular charge transfer (TICT) fluorescence and related phenomena in materials chemistry. J Mater Chem C 2016, 4 (14), 2731-2743. Zhang, Z. Y.; Zhang, G. Q.; Wang, J. X.; Sun, S. S.; Zhang, Z. Z., The 28. mechanisms of Large Stokes Shift and Fluorescence Quantum Yields in anilino substituted Rhodamine analogue: TICT and PICT. Comput Theor Chem 2016, 1095, 44-53. Zhou, P.; Han, K., Unraveling the detailed mechanism of excited-state proton 29. transfer. Acc Chem Res 2018, 51 (7), 1681-1690. 30. Weinberg, D. R.; Gagliardi, C. J.; Hull, J. F.; Murphy, C. F.; Kent, C. A.; Westlake, B. C.; Paul, A.; Ess, D. H.; McCafferty, D. G.; Meyer, T. J., Proton-Coupled Electron Transfer. Chemical Reviews 2012, 112 (7), 4016-4093. Tonge, P. J.; Meech, S. R., Excited state dynamics in the green fluorescent 31. protein. Journal of Photochemistry and Photobiology A: Chemistry 2009, 205 (1), 1-11. 32. Hayashi, S.; Tajkhorshid, E.; Schulten, K., Structural changes during the formation of early intermediates in the bacteriorhodopsin photocycle. Biophys J 2002, 83 (3), 1281-97. 224