MOLECULAR DYNAMICS IN THE HAIRPIN RIBOZYME: CALCULATIONAL AND EXPERIMENTAL ASPECTS By Patrick Omondi Ochieng A DISSERTATION Submitted to Michigan State University in partial fulfilment of the requirements for the degree of Biochemistry and Molecular Biology - Doctor of Philosophy 2015 ABSTRACT MOLECULAR DYNAMICS IN THE HAIRPI N RIBOZYME: CALCULATIONAL AND EXPERIMENTAL ASPECTS By Patrick Omondi Ochieng The increasing role of RNA therapy in targ eting diseases has inspired several RNA studies and especially structural RNA. Of inte rest to many scientists is how such RNA can perform their work with limited f unctional groups available to RNA. The structural versatility of RNA seems to underscore the importance of dynamics in pe rforming several functions. Ribozymes are a good example of structured R NA involved in RNA backbone cleavage with a range of strategies. Hairpin ribozyme invokes do main-domain docking to activate the cleavage process. The major loop rearrangements obs erved upon docking, as well as the kinetically unfavorable docking process, both argue for conf ormational selection by pre-organization of the catalytically-competent ac tive site of the hairpin ribozyme. In this thesis, we sought to study the behavior of loop A in sampling the docked-lik e conformation as evidence for conformational selection. We addressed three major aims whic h involved (i) understanding the dynamics in loop A using molecular dynamics simulation as a tool for assessing conformational sampling (ii) determining the right loop A construct for NMR studies and resonance as signments for structure determination and (iii) elucidation of fast a nd slow dynamics in loop A using NMR relaxation techniques. In aim 1 (Chapter 2), molecular dynamics simu lation was used to determine conformational heterogeneity in RNA based on alte rnate base-pair formation within a subset of residues in the loop region of domain A of the hairpin ribozy me. Three main conformers and several minor conformations were observed in our simulatio ns as analyzed by the Markov State model analysis. RNA base residues and backbone dynamics played a major role in the conformational heterogeneity and ensemble in loop A. Of th e conformations that were sampled, the most populated conformer, AA/CA, closely sampled conf ormational properties similar to the activated (docked) loop A conformation, suggesting the activating role indu ced by conformational heterogeneity. In aim 2 (Chapter 3), we determined the su itable loop A construct for NMR studies by NMR secondary structure analysis on various constr ucts of loop A. Using exchangeable and non- exchangeable NMR experiments we assigned certain specific proton, nitrogen and carbon resonances of loop A for structural determination and relaxation measurements. In aim 3 (Chapter 4), we asse ssed loop A dynamics using the 13C-NMR relaxation measurements. Fast internal motions in the orde r of ps - ns timescale were analyzed by Model- free approach using data from 13C R1, R1, and heteronuclear NOE of loop A. Loop A was generally found to be a rigid molecule on this timescale with internal generalized order parameters, S 2, of at least 0.9 in the helical regions. Several residues reported correlation times indicative of fast motions on the ps to ns times cale, while a few reported slow exchange in the µs-ms timescale. Our data underscores the importance of fast and slow dynamics in the formation of conformational states with structures simila r to the activated form . These conformational variability and structural transitions seem to activate RNA thereby facilitating RNA-RNA interaction via the conforma tional selection mechanism. Copyright by PATRICK OMONDI OCHIENG 2015 v I dedicate this thesis to my late parents Phile mon and Phoebe Ochieng who despite their sacrifice for me to attend school, never lived to see me begin my graduate school. My late grandpa (Ayub Oricho), who also died a few months before my thesis defense, inspired me to the end. viACKNOWLEDGEMENTS After several years of graduate school at Mich igan State University, I wish to thank the Department of Biochemistry and Molecular Bi ology for according me the opportunity to pursue my dream. My application to Michigan State Univ ersity was inspired by my former mentor and master™s thesis advisor at Western Michigan University, Dr. David Huffman, who not only provided me with an opport unity to gamble my research skills in his lab, but also mentored me methodically to be a future scientist and schola r. While on my visit tour at Michigan State University, I met Dr. Charles Hoogstraten who had been assigned to meet all the visiting prospective graduate students for breakfast. Dr. Hoogstraten was excited about his research as he shared with us, and I remember him particul arly emphasizing the us e of Nuclear Magnetic Resonance (NMR) as a tool for understanding molecu lar dynamics, not just structure. This was a new concept to me, since I had all along underst ood NMR as a tool for structure determination both in solution and solid state. When I finally jo ined Michigan State University, I was definitely interested in pursuing my graduate studies in th e NMR field. Dr. Hoogstraten accepted me in his lab and has been a pillar, a mentor, an adviso r who I will forever cherish. His insights and brilliance in biophysics, and NMR in particular, has been a useful resource in my pursuit for this project. In the course of my research, I also developed an interest in pursuing computational biochemistry alongside my experi mental work. Dr. Michael Feig readily offered to mentor me and quickly provided me a space and computer in his lab. As an amateur in Linux commands and computational biophysics, Dr. Feig encouraged me to take the n ecessary courses that prepared me for research in this complex field. He has since mentored me and offered me high quality training by always asking the hard questions that spur my thinking. I count myself fortunate to have gone through Dr. Feig™s mentorship. viiI want to express my gratitude to all the former and current members of the Hoogstraten and Feig labs. Dr. Minako Sumita and Dr. James Johnson Jr . inducted me in the H oogstraten lab, offering to train me in synthesizing and studying RNA. Dr. Shayantani Mukherjee and Dr. Sean Law trained me in Molecular Dynamics Simulation fo r which I am grateful. Dr. Monika Sharma, Dr. Beibei Wang, Dr. Parimal Kar a nd Dr. Vahid Mirjalili provided me with useful suggestions and technical support in the Molecular dynamics project. Dr. Fadh iru Kamba was a colleague in Dr. Hoogstraten™s lab whose friendship and insights I will live to ch erish. Together, we encouraged one another during good and difficu lt times. I also wish to thank Neil White for his support and friendship. I want to thank my graduate guidance committee for volunteering to mentor me as well as their suggestions that improved the quality of my resear ch and presentations. Special thanks to Dr. Bill Henry, Dr. Leslie Kuhn and Dr. John McCracken for serving in this voluntary role. I wish to thank my friends Dr. Joel Lwande and Dr. Da vid Achila for inspiring me through graduate school right from Western Mi chigan University to Michigan State University. I also thank Dr. Dan Holmes and Kermit Johnson for always providing me with NMR time, for offering solutions to my NMR challenges a nd for providing me with technical support. Last but not least, I want to thank my lovely wife Lilly Ochi eng, and son Phillip Ochieng for providing me with the desired support during my time in graduate school and for always being there for me. viiiTABLE OF CONTENTS LIST OF TABLES ................................................................................................................ .......... x LIST OF FIGURES ............................................................................................................... ........ xi KEY TO ABBREVIATIONS ...................................................................................................... xiv CHAPTER 1 ..................................................................................................................... .............. 1 INTRODUCTION TO RIBOZYMES AND TH E APPROACHES TO ASSESSING RNA DYNAMICS-FUNCTION RELATIONSHIP ................................................................................ 1 1.1 BACKGROUND .............................................................................................................. 2 1.1.1 Thesis outline and research goals ................................................................................... 2 1.1.2. RNA conformationa l transitions .................................................................................... 4 1.1.3. RNA tertiary structure ................................................................................................. .. 8 1.1.4. RNA and ribozymes .................................................................................................... 1 0 1.1.5. Mechanism of catalysis in ribozymes .......................................................................... 11 1.1.6. Therapeutic role of ribozymes ..................................................................................... 13 1.1.7. Hairpin ribozyme ....................................................................................................... .. 16 1.1.8. Structure and active site architecture of the hairpin ribozyme .................................... 17 1.1.9. Computer molecular dynamics (MD) simulations ...................................................... 19 1.2.0. Experimental assessment of dynamics ........................................................................ 23 1.2.1. Functional probing of RNA dynamics ......................................................................... 27 REFERENCES .................................................................................................................... ......... 29 CHAPTER 2 ..................................................................................................................... ............ 40 PROBING ALTERNATE BASE-PAIR REARRANGEMENTS AND CONFORMATIONAL SAMPLING IN A RIBOZYME ACTIVE-S ITE INTERNAL LOOP USING MOLECULAR DYNAMICS SIMULATIONS ..................................................................................................... 40 ABSTRACT ...................................................................................................................... ........ 41 2.1 INTRODUCTION .............................................................................................................. . 42 2.2 MATERIALS AND METHODS ........................................................................................ 47 2.2.1 Molecular dynamics (MD) simulations using NAMD and the CHARMM force field 47 2.2.2 MD simulations using sander and the Amber force field ............................................. 50 2.2.3 Targeted molecular dynamics (TMD) simulation ........................................................ 50 2.2.4 Analysis ................................................................................................................ ........ 52 2.3 RESULTS ................................................................................................................... ......... 52 2.3.1 Force field validation .................................................................................................. .. 52 2.3.2 Conformational heterogeneity of wildtype loop A ....................................................... 53 2.3.3 Conformational transitions between major states ......................................................... 65 2.4 DISCUSSION ................................................................................................................ ..... 70 2.4.1 Sampling of docked loop A structure ........................................................................... 71 2.4.2 Functional relevance .................................................................................................... . 76 2.5 CONCLUSION ................................................................................................................ ... 77 REFERENCES .................................................................................................................... ......... 78 ixCHAPTER 3 ..................................................................................................................... ............ 86 CONSTRUCT OPTIMIZATION, DOCKIN G STUDIES AND NMR RESONANCE ASSIGNMENT OF THE LOOP A DOMAIN OF THE HAIRPIN RIBOZYME ....................... 86 3.1 INTRODUCTION .............................................................................................................. . 86 3.2 MATERIALS AND METHODS ........................................................................................ 92 3.2.1 RNA preparation........................................................................................................... 92 3.2.2 Docking transitions between mutant loop A and wildtype loop B ............................... 93 3.2.3 NMR studies ............................................................................................................. .... 94 3.3 RESULTS ................................................................................................................... ......... 96 3.3.1 Docking studies ......................................................................................................... ... 96 3.3.2 Construct optimization for NMR spectroscopy ............................................................ 99 3.3.3 NMR assignments ...................................................................................................... 10 5 3.3.4 Assignment of exchangeable proton resonances ........................................................ 108 3.3.5 Non exchangeable protons .......................................................................................... 111 3.4 DISCUSSION ................................................................................................................ ... 124 ACKNOWLEDGEMENT .......................................................................................................... 127 REFERENCES .................................................................................................................... ....... 128 CHAPTER 4 ..................................................................................................................... .......... 132 INSIGHTS INTO THE INTERNAL DYNAMICS OF LOOP A USING NMR SPIN RELAXATION .................................................................................................................... ....... 132 ABSTRACT ...................................................................................................................... ...... 133 4.2 MATERIALS AND METHODS ...................................................................................... 138 4.2.1 RNA preparation......................................................................................................... 138 4.2.2 NMR studies ............................................................................................................. .. 140 4.2.3 Data analysis ........................................................................................................... .... 141 4.2.4 Model-free analysis .................................................................................................... 142 4.3 RESULTS AND DISCUSSION ....................................................................................... 142 4.3.1 13 C Relaxation measurements ..................................................................................... 142 4.3.2 Relaxation dispersion ................................................................................................. 1 44 ACKNOWLEDGMENT ......................................................................................................... 155 REFERENCES .................................................................................................................... .... 156 CHAPTER 5 ..................................................................................................................... .......... 161 SUMMARY, DISCUSSION AND FUTURE WORK ............................................................... 161 5.1 SUMMARY AND DISCUSSION .................................................................................... 162 5.2 FUTURE DIRECTION ..................................................................................................... 165 5.2.1 Base dynamics for the hairpin ribozyme .................................................................... 166 5.2.2 Functional read-out of dynamics ................................................................................ 167 REFERENCES .................................................................................................................... 170 xLIST OF TABLES Table 2-1: Summary of molecular dynamics simulations of loop A constructs ........................... 49 Table 2-2: Comparison of NOE di stance constraints with simulation averages using Amber and CHARMM .................................................................................................................................... 55 Table 2-3: Representative backbone and base-pair parameters of av erage structures derived from CHARMM36 and Amberff10 simulations respectively ............................................................... 56 Table 2-4: Characterization of the kinetic states observed during loop A simulation .................. 61 Table 2-5: Residue dynamics for selected loop residues during hpA simulations ....................... 62 Table 2-6: Conformational energies of loop A conformers calculated with MMPB/SA ............. 67 Table 2-7: Percentage population that sample sp ecific interactions observed in the docked form of loop A during isolated loop A simulation ................................................................................ 76 Table 3-1: Resonance assignments for loop A (GAAA LpAext) using NMR experiments ....... 110 Table 3-2: Non exchangeable assignments of H6/H8 aromatic pr otons in loop A .................... 118 Table 3-3: Non exchangeable assignments of adenine H2 aromatic protons in loop A ............. 119 Table 3-4: Assignments of H1'/ H2' and C1'/C2' in loop A ......................................................... 120 Table 3-5: Aromatic resonance assignment s of H1'/H2' and C1 '/C2' in loop A ......................... 122 Table 4-1: 13C R1, R1 and heteronuclear NOE measurements. .................................................. 148 Table 4-2: Model-free analysis of loop A relaxation measurements. ......................................... 149 Table 4-3: Chemical exchange parameters from T 1 dispersion analysis of loop A. .................. 154 xiLIST OF FIGURES Figure 1-1: Docked structure of the RNA hairpin ribozyme loop A (cyan) and loop B (green) domains–––––––––––––––––––––––––––––––––––5 Figure 1-2: Illustration of conformational and ch emical transition in a functional macromolecule along a reaction co-ordinate.–––––––––––––––––––––––––––6 Figure 1-3: The crystal structure of self-splicing group II intron–––––––––––––9 Figure 1-4: Secondary structure representation of the trans-acting hairpin ribozyme in the undocked form...–––––––––––––––––––––––––––––––15 Figure 1-5: Timescales and dynamics sampled by chemical and biochemical processes...––..22 Figure 2-1: Trans-acting hairpin ribozyme secondary structur e representation of metal-dependent loop-loop docking...–––––––––––––––––––––––––––––...45 Figure 2-2: Schematic representa tion of free loops A and B as well as docked loop A and B.....46 Figure 2-3: Secondary structure representation of loop A with their respective heavy atom RMSD profiles...––––––––––––––––––––––––––––––....54 Figure 2-4: Population distribution between non-methylated l oop A (green) and methylated loop A (red)––––––––––––––––––––––––––––––––––....57 Figure 2-5: Schematic representa tions (A) and structural renderings (B) of loop A simulation structures––––––––––––––––––––––––––––––––––60 Figure 2-6: Schematic representa tion of loop A conformational cl assification based on loop base-pairing and stacking–––––––––––––––––––––––––––––...61 Figure 2-7: Conformational tran sition of loop A RNA capturing top 18 macrostates derived from Markov state modeling––––––––––––––––––––––––––––..64 Figure 2-8: Targeted Molecular Dynamics (TMD ) simulation of conformer transitions from AA/UA to GA/UA and AA/CA–––––––––––––––––––––––––.68 Figure 2-9: Free energy diagrams depicting the energy profile of TMD-driven conformations..69 Figure 2-10: Loop A sampling docked conformation––––––––––––––––...73 Figure 3-1: Cartoon representation of hairpin ribozyme active site region––––––––..89 Figure 3-2: Hairpin ribozyme trans- constructs used in our studies–––––––––––..91 xiiFigure 3-3: Comparison between th e circular dichroism (CD) spectra of docking between wild type loop A and loop A(U+2C/C+3U) ......................................................................................... 97 Figure 3-4: Co(NH3)63+ titration studies to assay for metal- dependent interaction between loop A(U+2C/C+3U) and loop B––––––––––––––––––––––––––...98 Figure 3-5: ID imino NMR peaks of various loop A constructs––––––––––––.101 Figure 3-6: RNA base-pairing between A-U and G-C highlighting the involvement of imino protons of guanine and uracil–––––––––––––––––––––––––...102 Figure 3-7: 1D imino spectra of GUAA l oop A at different annealing conditions–––––.103 Figure 3-8: Temperature-de pendent 1D imino spectrum of extended GAAA loop A––––104 Figure 3-9: 2D imino NOESY spect rum of GAAA extended loop A––––––––––.106 Figure 3-10: 2D amino NOESY sp ectrum of GAAA extended loop A–––––––––..107 Figure 3-11: 2D imino 1H,15N HSQC spectrum of GAAA extended loop A–––––––..109 Figure 3-12: 2D 1H,13C-HSQC contour spectrum of the ribose region in loop A–––––...114 Figure 3-13: 2D 1H,13C-HSQC spectrum of aromatic region in loop A–––––––––..115 Figure 3-14: 2D 1H,13C CT-HSQC spectrum of aromatic region in loop A.–––––––...116 Figure 3-15: Representative 2D planes of 3D HCCH-COSY spectrum of the ribose region in loop A––––––––––––––––––––––––––––––––––...117 Figure 3-16: Representative 2D planes of 13C-edited NOESY-HSQC spectrum of loop A–...121 Figure 3-17: Schematic representation of NOE c onnectivities that correlate intra- and inter- residue H8/H8 to H1' and H2'–––––––––––––––––––––––––..122 Figure 4-1: Timescale of macromolecular internal motions relevant in biological processes–137 Figure 4-2: The hairpin ribozyme loop A construct used in NMR relaxation studies...–––.139 Figure 4-3: Representative 13C R1 and R1obs curves for A2 and A9 –..–––––––––146 Figure 4-4: Histograms of the meas ured relaxation rates in loop A...––––––––––.147 Figure 4-5: Histogram represen tation of order parameter (S 2) and exchange contribution (Rex) derived from relaxation measurements –––––––––––––––––––––...150 Figure 4-6: Simulation structur e of loop A highlighting base stacking in the GAAA tetraloop involving A2 stacked between A1 and A3 –––––––––––––––––––.....152 xiiiFigure 4-7: Representative relaxation dispersion curves of selected C8 resonances––––..153 xivKEY TO ABBREVIATIONS 1D one dimensional 2D two dimensional 2WJ two-way junction 3D three dimensional 4WJ four-way junction ATP adenosine triphosphate CD circular dichroism COSY correlated spectroscopy CPMG Carr-Purcell-Meiboom-Gill CSA chemical shift anisotropy CT constant time CTP cytosine triphosphate DC decoupler DNA Deoxyribonucleic acid DSS sodium 2,2 dimethyl-2-silapentane-5-sulfonate E. coli Escherichia coli EDTA Ethylenediaminetetraacetic acid E-S enzyme-substrate FID free induction decay FRET fluorescence resonance energy transfer GlcN6P glucosamine-6-phosphate GTP guanosine triphosphate HDV hepatitis delta virus xvHis histidine HIV human immunodeficiency virus hNOE heteronuclear nucle ar Overhauser enhancement HSQC heteronuclear single quantum coherence LNA locked nucleic acid MALDI matrix-assisted la ser desorption ionization MD molecular dynamics MMPB/SA molecular mechanic s Poisson-Boltzmann surface area mRNA messenger RNA MSM Markov state model MWCO molecular weight cutoff NADP nicotinamide adenine dinucleotide phosphate NAMD nanoscale molecular dynamics NMP nucleotide monophosphase NMR nuclear magnetic resonance NOE nuclear Overhauser enhancement NOESY nuclear Overhasuer enhancement spectroscopy NTP nucleotide triphosphate PDB protein data bank Phe phenylalanine PME particle-mesh Ewald R1 longitudinal relaxation rate R1 rotating-frame transverse relaxation rate R2 transverse relaxation rate RF radio frequency xviRMSD root mean square deviation RNA ribonucleic acid RNAse A ribonuclease A RNAP ribonucleic acid polymerase RZ ribozyme SBS substrate binding sequence T1 longitudinal relaxation time T2 transverse relaxation time T1 rotating-frame transverse relaxation time TMD targeted molecular dynamics TOCSY total correlated spectroscopy TOF time of flight tRNA transfer ribonucleic acid UTP uridine triphosphate UTR untranslated region 1 CHAPTER 1 INTRODUCTION TO RIBOZYME S AND THE APPROACHES TO ASSESSING RNA DYNAMICS-FUNCTION RELATIONSHIP 21.1 BACKGROUND 1.1.1 Thesis outline and research goals Although the factors underlying secondary structure in DNA a nd RNA are relatively well understood, the understanding of the driving forces and mechanisms of the formation of complex tertiary structures in nucleic acids has not kept pace with th e great recent advances in RNA structural biology. Due to the importance of catalytic RNAs (ribozymes) as potential therapeutics, their tertiary structure interactions in the context of binding and catalysis remain an active subject of investigation. Specifically, domain -domain interactions are an inherent feature in the active site assembly of many ribozym es, including the hairpin ribozyme. These interactions, although rate limiti ng, prepare the required residues to engage in catalysis through phosphodiester backbone cleavage. The loop-loop bi nding transition in the hairpin ribozyme therefore presents us with a suitable model fo r understanding the mech anism behind RNA-RNA interactions with possible insight s on RNA secondary interactions, tertiary structur e interactions, dynamics and function. The biological timescale for such transitions a nd interactions usually occur in the regime of picoseconds to milliseconds. The RNA-RNA interaction in hairpin ribozyme implicates two independent RNA doma ins (loop A and loop B) with each domain containing two stems interrupted by loops A an d B respectively. The interaction between these domains, often referred to as docking, is require d for hairpin ribozyme activation and subsequent cleavage of the RNA backbone. The major loop rearrangements observed upon docking, as well as the kinetically unfavorable doc king process, both argue for a tight and mechanistic role for conformational dynamics in the pre-organization of the catalytically-competent active site of the hairpin ribozyme. However, the de tails such dynamics have not b een explored in the context of 3dynamics-function relationship. Even so, conformati onal transitions and their role in activating the ground-state docking-competent forms of doma ins A and B of the hairpin ribozyme are still not well understood. Here, we have integrated Molecular Dynamics (MD) simulation and NMR to assess the possible transitions taking place in the hairpin ribozyme™s loop A domain as an initial step towards understanding hairpin ri bozyme docking mechanism. Due to the slow association rates reported for in teraction between loop A and B (1) and the extensive structural rearrangements observed upon docking ( Figure 1-1 ) (2-4), we hypothesize that the interaction between loop A and loop B is driven by a double conformational capture mechanism, in which both loops undergo dynamic fluctuations to c onformations resembling their docked forms and that only collisions between l oops that have sampled the docked forms are capable of recognition and subsequent docking. We have probed the co nformational sampling and dynamics of loop A domain of the hairpin ribozyme as an initial step to understanding the active site dynamics in RNA. A deeper understanding of d ynamics-function relationship is important in tuning the rate and mechanism of RNA-RNA interaction. Thus, this feature may be a useful tool in the design of RNA aptamers that can be used in effective gene therapy treatments which target RNA cleavage. Investigating RNA dynamics also presents us with a unique oppor tunity to study the physicochemical factors underlying RNA tertiary structure in gene ral, and has the potential to provide generalizable in sights in RNA. We have used mol ecular dynamics (MD) simulations to map out the conformational ensemble visited by loop A. In addition, we have pursued loop A NMR studies to probe dynamics and conformational sampling in this cleavage domain. Our studies present new insights in conformati onal dynamics in hairpin ribozyme loop A vis a vis its mechanistic and functional role. 41.1.2. RNA conformational transitions A critical challenge in understa nding biological processes at the molecular level is the elucidation of conformational va riations of macromolecules to wards the formation of active conformers. Indeed, these conformational change s are observed on various timescales. Of critical importance are changes that occur in timesca les necessary for biological function. Many biological processes rely on transduction of information through conformational changes in proteins and nucleic acids. Examples of such dynamics are observed in how RNA recognizes their interacting partners including proteins and small ligands ( 5) and in the utility of ribosome (6). Conformational dynamics may facilitate pre-organization of an inactive conformation of a given macromolecule (E-S) into an activated form often referred to as the pre-catalytic form (Figure 1-2 ). Both fast and slow exchange proce sses incorporate proteins and RNA into biological networks usually effected by internal molecular effectors in conjunction with external cues. Thus, RNA conformational transitions become part of biochemical pathways which overall influence biological events. RNA with various c onformational features can bind to protein which may stabilize one or multiple conformations within a subset of conformations. This feature has been observed in tyrosyl-tRNA synthetase derived from Neurospora crassa which binds to group I introns ( 7). Group 1 introns are large self-splicing ri bozymes that catalyze their own cleavage from mRNA, tRNA and rRNA precursors. In some cases, protein binding may selectively lower the surrounding energy barrier ( 8, 9 ) which activates the dynamic transitions in the RNA. An example of this behavior is typical in the U1A protein which binds to its cognate RNA and induces dynamics in parts of the RNA especia lly in direct contact with the protein. Alternate secondary structural changes also occur in RNA although they are limited by energy barriers associated with base-pair opening and fo rmation of new base-pairs. In this case, RNA gets kin econform amodulatiexempli fwells in vpaired re etically tra pations. To eing secondarfied in ham mvolving onlygions, are i nFigure 1-and loop A and B cbase-pairepped in loc aeffect these ry structural merhead rib oy variations investigated. -1: Docked sB (green) dcharacterize ed to C25 (mal conform atransitions ichanges in ozyme. In oin tertiary s tstructure o fdomains . Strthe docked fmagenta) of l5ations whic hin vivo, cellRNA by oveour work, t htructure, wi tf the RNA hructural rear rform with a loop B. Figuh limit thei rls have a v aercoming la he mechanismth limite d chhairpin rib orangement oG+1(red resure adapted ar ability to ariety of pr oarge energetims by whi changes in Wozyme loop Aof loop resid usidue) extra- and modifiedsample m uoteins capa bic barriers ( 1ch RNA expWatson-CrickA (cyan) ues in loop -helically d from ref 3.ultiple ble of 10) as plores k base . Figure 1functionsubstrate transitio nundergo btimescal eJames Jo h-2: Illustratal macrom ocomplex at n state, and Eboth local t res within ps-hnson Jr. dotion of confoolecule alo nground stateE-P is the en ansitions an -ns and µs-moctoral thesi s6ormational ng a reactio ne, E-S* is thezyme produd large confms timescale ss).and chemicn co-ordina te activated suct complex. formational es respectivelcal transiti ote. E-S is e nstate, TS rep RNA ribozyexchange at ly (Figure reon in a nzyme- resents the ymes different eprinted fro mm 7In vivo , conformational transitions occur when signa ls reaching the cell modulate the formation of multiple conformations and consequent stab ilization of such conformations. A good example of this is in riboswitch where metabolite bi nding stabilizes the riboswitch by inducing a conformational redistribution to a state that is olates RNA element into the aptamer domain ( 11). Riboswitches are a segment of RNA genetic elements usually inserted in the 5 untranslated region (5 UTR) of bacterial genes. These elements re gulate the expression of bacterial metabolic genes with changes in cellular metabolite concentration ( 12). In such processes RNA samples various conformations within th e energy landscape spanning through a series of deep local energy minima interrupted by shallow energy we lls. These deep energy minimas are usually characterized by conformations that result from equilibrium motions which are stabilized by cellular signals to effect conformational transitions ( 13, 14). Tertiary confor mational transitions are mostly seen in small ribozymes such as the hairpin and hepatitis delta virus (HDV). These transitions involve large conformational changes in the position of helical arms, usually after domain interaction to another RNA domain prom oting the conformational changes necessary for RNA catalysis. Following catalysis, another set of undocking changes promote the release of the cleavage product (15). The importance of these transitions is underscored by the robust impact they have on the overall catalytic rate constant. The Tetrahymena group I ribozyme has been shown to transition between tertiary conformati ons while reporting differe nt substrate binding affinities but with similar enzyme activities ( 16). The inter-conformationa l conversion rates were slower than the catalytic rate, implying the existence of multiple conformations. 81.1.3. RNA tertiary structure Most of the current knowledge about RNA tertiary structures comes from X-ray crystallography and NMR. These techniques rev eal precise snapshots of a dynamic reality, but provide little information about the dynamics at atomic resolution over time within the three- dimensional fold. Additional biophysical techniques such as single-molecule optical traps, time- resolved fluorescent resonance energy transfer (FRET) and hydroxyl radical footprinting may also provide some dynamics information. W ith these tools, it is possible to monitor conformational changes occurring within RNA molecules during their folding and assembly. Three-dimensional structures rev eal that RNAs have an ordere d organization where the primary sequence informs the architecture of secondary struct ure which also dictates the extent of tertiary folding. The secondary structural elements (he lices, loops, bulges and junctions) are networked to each other within a complex tertiary structure ( 17). RNA helices are usually A-form Watson- Crick duplexes while the loops, bulges and j unctions are mostly non-Watson-Crick regions terminated by one or more helices ( 18). Although the structural integrity of helices are maintained by Watson-Crick base-pairs and monovalent ions, the tertiary contacts are usually stabilized by non-canonical base-p airs and generally require the presence of divalent ions, especially magnesium ions. RNA architecture is weaved by a series of base-p airing and base stacking interactions leading to co-axial stacks of helical domains packed parallel or orthogonal to one another, as displayed by a recent st ructure of a group II self-splicing intron (Figure 1-3) (19). Figure 1stacking ref 3. -3:The crysand base-st astal structuracking inter a9re of self-sp lactions suppolicing grouport the terti ap II intron. ary. Figure adCo-axial dapted fromm 10Junctions promote conformational diversity that can be modulated by either RNA-RNA or RNA-ligand interactions. The hammerhead ribozyme provides a good example of this phenomenon. The hammerhead active site displays a three-way helical junction containing a central core with fifteen highly conserved nucleotides essential for catalytic activity. Its activity, however, is dependent on loop-loop tertiary inter actions in non-conserved regions far away from the active site ( 20). These long-range te rtiary interactions stabilize the active conformation of the junction positioning the relevant nucleotides in the right positions for catalysis. Understanding the formation of RNA tertiary structure is pertinent towards unravelling the mechanisms of elaborate cellular machines such as the eu karyotic spliceosome, in which numerous RNA structures and RNA protein inte ractions form and are disrupt ed at precise points along the splicing pathway (21, 22). We propose that the docking trans ition of in the hairpin ribozyme can serve a unique role as a small, tractable R NA-RNA interaction formed entirely by tertiary structure that is accompanied by a direct func tional readout in the form of self -cleavage (23, 24). 1.1.4. RNA and ribozymes The importance of RNA was suggested as early as the late 1950s when it was established that living cells harbor much more RNA than DNA. The discovery of the details of protein synthesis (25) revealed that RNA molecules are implicat ed in a variety of processes within the cells. Orgel (26) and Crick (27) independently proposed that RNA acted both as catalyst and as carrier of information. The idea of RNA as ca talysts received unquestionable proof after the discovery of natural RNA enzymes (ribozymes), by the independent groups of Sydney Altman and Thomas Cech ( 28, 29). Ribozymes are RNA molecules that catalyze reactions on themselves or other molecules. Over the years, six natura lly occurring self-cleaving ribozymes have been discovered. These are the hairpin, hammerhead, hepatitis delta virus (HDV), glmS, Varkud 11satellite (VS) and twister ribozymes ( 23, 30-34). These small ribozymes catalyze self-cleavage of the RNA backbone using a chemical strategy similar to RNase A (35). 1.1.5. Mechanism of catalysis in ribozymes As a protein enzyme that functionally mimi cs ribozymes, RNase A deploys the general acid-base catalytic mechanism with unique stra tegies to promote RNA backbone cleavage reaction. The following strategies ar e adopted by specific critical amino acids to facilitate this process. These involve (1) alignment of th e hydroxyl nucleophile, electrophilic phosphate, and the leaving group by reorienting the adjacent nucleotides to activate the nucleophile, (2) transition-state stabilization, and (3) protonation of the leaving group. In RNase A, the 5 of pyrimidine nucleotide forms hydrogen bonds with the main chain and backbone oxygen atoms of Threonine 45. Also, base-s tacking interactions be tween the nucleotide 5 of the scissile phosphate with Phe120 helps in facilitating stability ( 36). His12 is the general base responsible for deprotonating the 2-OH nucleophile while the ne gatively charged transition state is stabilized by the positively charged Lys41. Studies introducing a lysine to ar ginine mutation at position 41 does not significantly reduce RNase A activity indicating the requirement for a positively charged residue in that position (37). An intriguing question in understanding the catalytic mechanism of ribozymes is the ability of RNA to catalyze diverse chemical processes with limited se t of functional groups unlike the diversity in protein enzymes. The l ack of neutral functiona l groups in RNA with pK a close to neutrality or positively charged functional groups present a dilemma on how the chemical processes requiring these functional gr oups are achieved. However, a number of factors are thought to be essential for ca talysis. Specific nuc leotides, metals and water may be involved 12in facilitating the chemistry of different sets of ribozymes. The clear deficiencies in what was thought to be critical functiona l groups informed an incorrect assumption that most small ribozymes were metalloenzymes ( 38). The participation of divale nt metal ions, nucleobases and RNA backbone functional groups could provide an explanation for the general acid-base chemistry involving a general acid, base and electrostatic stabilization (39). RNAs mostly require positively charged ions to enhance a folded state. These ions are usually divalent. However, higher (molar) concentrati ons of monovalent cations or inert polyva lent ions are still sufficient for most ribozymes to achieve full activity ( 40, 41). Despite lacking the chemical diversity exhibite d by amino acid residues in RNase A, also a protein enzyme that cleaves RNA backbone, ribozymes are s till capable of catalyzing RNA cleavage reactions albeit 15-500 fold slow er than their RNase A counterparts ( 23, 42, 43). Using a divalent metal, the HDV ribozyme employs a specifically bound magnes ium ion and specific nucleotides to promote phosphate backbone cleavage to a rate of ~150-fold slower than the more efficient RNase A (44), while glmS ribozyme uses a glucos amine-derivative cofactor (GlcN6P) and nucleotides to attain cleavage ra tes ~100-fold slower than RNase A ( 45). Despite the similar general catalytic mechanism, there is no gene ral consensus on the origin of the catalytic proficiency showed by various ribozymes. The use of different strategies in promoting chemistry in ribozymes has sparked curiosity in the RNA world and remains an active field of study both computationally and experimentally. 131.1.6. Therapeutic role of ribozymes The attempted use of RNA molecules as therap eutic agents is relatively novel, but has received growing interest over the few decades. This curiosity has been inspired by various scientific discoveries that feature the role of R NA molecules in relaying genetic instructions in living systems as well as their natural abundance and adaptability. Interestingly, RNA molecules can assume a diverse set of conformations a nd perform a variety of cellular functions. For example some RNAs can fold into catalytic cent ers, while others may adopt conformations that facilitate specific RNAŒprotein, R NAŒDNA or RNAŒRNA interactions. The discovery of ribozymes ( 28, 29) has informed the development of a new class of trans-cleaving therapeutic RNAs with great pot ential in human medici ne. This strategy is designed around inactivating viral RNA genome by cleavage. Therapeutic trans-ribozymes are designed to specifically bind RNA substrate via base-pairing interactions. This base-pairing directs cleavage of the target RNA and subse quent release the cleavage products which may go through a recycling process to f acilitate a multi-step catalytic process. The realization that ribozymes can be directed to a specific RNA target to cleave pathoge nic viral transcripts in vitro has cultivated promise on their potential therapeutic value in vivo (46, 47). Significant progress towards assessing the potential use of trans-clea ving ribozymes has been made, with the hairpin and hammerhead ribozymes receiving more attention in this translational effort ( 48). Ribozymes can potentially cleave a range of RNA transcript s if well designed to efficiently bind and cleave these transcripts. Such transcripts may includ e oncogene transcripts, viral genomes and many other pathogenic transcripts. Th e success of a synthetic ribozyme to deactivate these transcripts via cleavage will depend on the efficiency of delivering the trans-forms of ribozymes in vivo , the concentration and duration of inhibition required to change the pathophysiology of the disease. 14For example, the hammerhead ribozyme inhibite d the replication of a murine retrovirus about 90% by co-localizing this riboz yme with its viral target in vivo (49); however, significant increases in the levels of inhibition were not achieved even in excess concentrations of the ribozyme. This suggests that the efficacy of trans-cleaving therapeutic ribozymes could be limited to cellular conditions and therefore a combination of th erapies (e.g. chemotherapy) and other therapeutics can augment the role of ri bozyme. Another major focus of research on ribozyme therapy has been on inhibiting the re plication of RNA viru ses and retroviruses, specifically HIV-1 ( 50, 51), hepatitis C virus ( 52) and hepatitis B virus ( 53). These are appropriate targets since their genomes are composed of RNA, and thus replication can be directly targeted for inhibition. FigureribozySBS anstems a e 1-4: Seco nyme in the und RZ respeare labelled ndary structundocked foctively. The stem 1-IV 15ture represerm. Substracleavage sitentation of tate and ribozte is shown wthe trans-aczyme strand swith an arrocting hairpi ns are labeled w. Helical n 161.1.7. Hairpin ribozyme The hairpin ribozyme is a small ribozyme th at catalyzes reversible, site-specific phosphodiester bond cleavage giving a 5'-OH and a 2'- 3'-cyclic phosphate termini products ( 54). Derived from the negative strand of tobacco rings pot virus satellite RNA, the hairpin ribozyme is involved in processing the produc t of rolling circle replicat ion through backbone cleavage and ligation respectively ( 55, 56 ). The cognate substrate is recogn ized by canonical base-pairing to a single-stranded region of the ribozyme referred to as the substrate binding sequence (SBS) (Figure 1-4). The substrate and the substrate binding sequence form the loop A domain while the rest of the ribozyme forms the loop B domain. Loop A and loop B together with their flanking helices constitute the catalytic core of the hairpin ribozyme ( 57). The most critical nucleotides for catalysis are found in the two loop regions ( 58). The interaction betw een loop A and loop B is a necessary step th at precedes catalysis ( 59) and is referred to as docking. The details of this process are not clearly understood. However, structural evidence infers possible conformational changes through which domain-domain interaction and subsequent catalysis is achieved. The chemical steps of the reaction proceed without the necessity of multivalent metal ions ( 59, 60) but the rate acceleration is stil l realized, a process similar to metal-dependent ribozymes ( 61). There origin of catalytic proficie ncy in hairpin ribozymes has elic ited debate with a major focus on the role of active site nucleobases with respect to general acid and base catalysis. Due to the limited diversity of the four native RNA residu es, the role of these residues in nucleophilic activation, stabilization of the charged transitio n-state and activation of the leaving group via protonation has been extensively investigated albeit with di fferent conclusions. Critical geometric restraints facilitate bond making and breaking in the trans-phosphorylation reaction promoted by nucleolytic ribozymes ( 62). For example the nucleophilic S N2 reaction aligns the 17torsion angle between the 2 -OH nucleophile, the scissile phosphate, and the 5 -O leaving group should be ~180° to facilitate an inline nucleop hilic attack on the scis sile phosphate. Hairpin ribozymes provides a unique set of stabilizing interactions to achieve this unfavorable conformation. This inline architecture was suppor ted by base-pairing interactions between residues in loop A and B as well as base-stacking interactions. Interestingly, the same trans- phosphorylation reaction is catalyzed by the protein enzyme ribonuc lease A (RNase A), with two histidine residues used as genera l acid and base respectively ( 63). Small ribozymes therefore may employ many similar RNase A tactics in promoting chemistry. However, the hairpin ribozyme can cleave in the presence of cobalt hexamine, [Co(NH 3)6]3+, a non-hydrolyzable and exchange-inert mimic of ma gnesium hexahydrate, [Mg(H 2O)6]2+ (64, 65). This strengthens the argument that metals may serve unique role s by modulating hairpi n ribozyme tertiary interactions rather than via di rect involvement in catalysis. 1.1.8. Structure and active site architecture of the hairpin ribozyme The X-ray crystal structures of the hairpin ribo zyme provide valuable insights into the active site conformation of the docked state (3, 66), transition state, and product state structures ( 67, 68) observed during ribozyme catalysis. In each stru cture, the active sites had generally similar conformations. The G+1 nucleotide is completely flippe d out of loop A and inserted in loop B pocket where it stacks between A26 and A38 residues ( 3). The flipped-out G+1 forms a tertiary Watson-Crick base-pair with C 25 of loop B, and is stabiliz ed by additional hydrogen-bonding interactions with A38 and G 36 residues. The specific contacts among G+1 and these loop B residues seem to highlight the spec ificity of G at that +1 position ( 3). Mutation of G+1 to other bases results in the loss of catalytic activity which is partially restored by compensatory mutations at C25 in loop B ( 69). Important structural informa tion is available in the docked 18structures of loop A and loop B, but additional st ructural information have also been obtained from the solution structures of the isolated loop A and loop B which have been separately determined ( 4, 70). The separate RNA pieces can be mixed together in vitro to reconstitute a functional ribozyme ( 71). The nuclear magnetic resonance (N MR) structures of loops A and B probably represent the conformatio n of these two loops before docking. In the isolated loop A structure, an anti-conformation G+1 is accommoda ted in the helical stack and forms a sheared pair with A9 ( 4). This residue is involved in a G+1:C25 tertiary base pairing with domain B in the docked structure (3), indicating a dramatic conformationa l reshuffle before or after docking. Also, the minor groove width of the isolated loop A conformation observed in NMR is considerably narrower than the docked state. The conformation of isolated loop B ( 70) also differs greatly from the conformation observed in the crystal structure (docked form). Only two of the seven non-canonical base pairs observed by NM R in isolated loop B are retained in the docked conformation. Whereas U41 is extruded into an S-turn in the undoc ked loop B, it forms base pair with A22 in the docked structure ( 3, 70). A detailed comparison between the docked crystal structure and the isolated loop structur es (loop A and loop B) shows large conformational rearrangements taking place in both RNA loops in c onsistent with the acti ve site assembly. The rearrangements involve dramatic changes in base-pairi ng schemes and signifi cant alteration of backbone geometry concomitant with the formati on of tertiary interac tions. These observations support the idea that conformational dynamics c ould at play in the process of loop-loop interaction, substrate orientation and subsequent catalysis. Our thesis is that static structures are important but insufficient to understand the m echanism of ribozyme action. Previously, an NMR structural study of the lead-dependent ribozy me (leadzyme) observed th at the highly populated conformation in solution is not well-oriented to sample its activated state, suggesting that 19conformational rearrangement was necessary for to achieve catalytic transitions ( 72). Similarly, we also hypothesize that inde pendent loops of the hairpin ribozyme may be sampling other higher energy (activated) states via specific dynamic modes and that these modes may be coupled to formation of active structures in facilitating hairpin ribozyme function. Our investigation has harnessed contemporary Molecular dynamics (MD) simulation and nuclear magnetic resonance (NMR) techniques which have in the past provided the richest information on dynamics of macromolecules. Previous work has shown that fitrans-dockingfl constructs, loops A and B, still retain catalytic activity ( 73) and that the two pieces can bind to each other with the K D in the low µM range ( 74). In this thesis, we have mainly probed loop A sampling the high energy docked form in the pro cess of conformational fluctuation. We hypothesize that the relatively slow docking rates may be due to th e isolated loop A samp ling other conformers including the docked (activated) form. In othe r words, only when structurally favorable conformers appear is loop A competent for docking. This phenomenon has precedents in several carefully-studied protein enzymes. Dihydrof olate reductase-NADPH complex, for example, samples a higher energy substate in which the empty substrate-/product-binding pocket adopts a conformation similar to that of the ligand-bound state ( 75). A similar observation has been reported for ribonuclease A (76). 1.1.9. Computer molecular dynamics (MD) simulations Molecular dynamics (MD) simulation uses the computational approach to interrogate the atomic transitions and molecular motions using classical empirical potential force fields. It describes the time evolution in a set of discrete particles by solving the Newton™s equation of classical mechanics according to equation 1-1. 20Fi = miai 1-1 where mi and ai are the mass and acceleration of atom i respectively while Fi is the force acting on the mass calculated from the gradie nt of potential energy function, U. Fi = -iU 1-2 MD simulation method calculates th e time-dependent behavior of a molecular system within the molecular force field that defines the level of in teractions among particles in their environment. MD simulations give a variety of information ranging from fast fluctuations to slower conformational changes of proteins, nucleic acid s and even lipids. Computational methods are currently attractive for use in investigating the chemical transitions and structural dynamics of many biological molecules as well as biological complexes. The robustness of this technique in assessing dynamics coupled with its cost-effectiv e sampling has provoked several studies in this field. Some of the most commonly used force fields are CHARMM ( 77-79), AMBER ( 80), GROMOS ( 81), and OPLS ( 82, 83). While the specifics of the potential energy functions used in various force fields are different, they genera lly use the general mathematical bonded and non- bonded interaction terms. The resu lts of a simulation are determin ed by the quality of the force field used, the sampled time (timescale) and the accu racy of the initial structures. Therefore, MD simulation is a tool primarily suitable to assess conformational dynamics within a wide range of time scales ( Figure 1-5) using both the all-atom simulation method ( 84) as well as enhanced sampling techniques used to explore the conformati onal space of a given molecule in the relevant timescale. Modern MD simulations de fine the potential energy function, U, as constituting the bonding terms and the non-bonding terms as a function of the Cartesian co-ordinates ( 85-87) as shown in equation 1-3. 21 EMMKbbb02bonds Kangles 02Kdihedrals 1cosn Kimpropers 04ijijijrij12ijrij6qiqjrijij The bonding terms are defined by the bond lengths ( b), bond angles ( ), dihedrals ( ), and improper dihedrals ( ) (Eqn. 1-3). Equilibrium values ar e denoted by subscript 0 whereas Kb, K, K and K are the force constants for the bondi ng terms of bond lengths, bond angles, dihedrals and improper dihedral s respectively. The dihedral a ngle term is modeled along the sinusoidal function with n and representing periodicity and phase shift respectively. The non- bonding terms are van der Waals and electrostatic interactions modele d by distances between atoms i and j respectively. Also, qi and qj are the point charges between which electrostatic interactions are calculated usi ng Coulombic potential. The van der Waals component of the equation models the repuls ion and attraction forces. 1-3 Figproproconhougure 1-5: Tiocesses. Moocesses. Impnformationa lur timescale imescales a nlecular dyn aportant biochl exchange os. nd dynamicamics simul ahemical pro coccur at diffe22s sampled bation can samcesses invol verent timesc aby chemical mple a rangeving local tr aales within pl and biochee of dynami cansitions andpicoseconds emical c d large to 23All-atom simulations have unique capabilitie s to map out the motions available to the molecule in three-dimensional detail. The use of state-of-the-art molecu lar dynamics calculations to build and complement structural and functiona l data is a key ingredient of the synergistic approach to dynamics-function relationship. The atomic-level dynamics that cannot be observed directly are easily deciphered and used to co mplement experimental techniques like NMR and X-ray crystallography. The hairpin ribozyme is a uniquely suitable model system for the pursuit of a deeper understanding of conformational sampling and dynamics, and the MD approach maximizes the possibilities of successfully ob taining this understanding in a rigorous and comprehensive fashion. Previous investigator s have focused on simulation studies on the structure and catalysis of the hairpin ribozyme ( 88-93), but the great majority of these have mostly explored the properties of the post-docking, catalytically competent state. Our exploration of ground-state loop A and the docki ng transition presents a mechanis tic detail for the activation of RNA domains as the initial step towards the form ation of the docked tertia ry structure that is a prerequisite for catalysis. This phenomenon has been comprehensively st udied and reported in chapters two and three of this thesis. 1.2.0. Experimental assessment of dynamics The first protein X-ray structure of myoglobin ( 94) opened up the field of structural biology to ask certain fundament al questions on how ligands c ould access the deeply buried heme iron center. This observation stimulated decad es of research into the dynamics of proteins and today, it is reasonably conventional that prot ein structures are in continuous motion and that such fluctuations are important for function. Early X-ray crystal structures of RNA also predicted the role of conformational dynamics especially due to the huge rearrangements observed in the 24tRNA™s helical arms upon binding tRNA synthetase ( 95). Similarly the conformational change theory in ribozymes well explains the formation of active states ( 96-98). To infer dynamics from X-ray structures, several crystal structures capturing the molecules at different stages of its function can be considered as part of the structural ensemble through which structural transitions can be d eciphered. Also, the flexible regions based on B- factors can be assumed to be dynamic. However, more definitive answ ers can be obtained by using NMR solution studies. NMR spectroscopy can be used to interrogate macromolecules motions of over a range of timescales using relaxation measurements. Relaxation in NMR is a phenomenon that is induced by field fluctua tions that occur duri ng molecular motion, and subsequently imparting variations in the loca l fields within the molecular movement. This property facilitates the NMR relaxa tion experiments to capture fluctu ations that are derived from such molecular motions. Relaxation in macromolecules probes 13C and 15N nuclei in solution which arises from dipolar interaction with covalently bound protons. The NMR techniques that are actively in use to study th ese motions are longitudinal ( T1), transverse ( T2) relaxation and heteronuclear nuclear Overhauser enhancement (h NOE) measurements. In th e early days, natural abundance 13C-NMR T1 measurements were used to study motions within ribonuclease A ( 99). This technique was successful sin ce it annotated multiple dynamics for - and -carbons. A year later, the same research gr oup analyzed detailed segmental motions of ribonuclease ( 100) to assess the robustness of T1. Over the next several years, researchers continued to obtain relaxation measurements using 13C natural abundance in DNA for ribose and base carbons. Due to the low sensitivity of the 13C natural abundance, minimal 13C enrichment was eventually used to study dynamics in DNA ( 101) and also RNA ( 102). The first attempt to quantify relaxation in uniformly 13C labeled RNA (103) with relaxation measuremen ts involving longitudinal 25relaxation, transverse relaxation and NOE parame ters was interpreted using the model-free analysis despite the challenges of dipolar couplings in the non-is olated spin systems. The model free analysis formalism ( 104-106) was developed by Lipari and Szabo to extract the fast ps-ns dynamics parameters from the measured relaxati on rates. In the fimodel-freefl approach, fast internal motion is described by a generalized order parameter, S2, which assesses the atomic spatial restriction, and an effective correlation time, e, which provides information on the rate of the motion. This formalism removes the ambiguity of internal dynamics interpretation. 13C relaxation can be complicated by crowded spin-sys tems in RNA or DNA, and this informs the use of C8/C6 and C2 resonances in simplifyi ng relaxation data. Also, magnetic interactions between neighboring 13C nuclei can significantly enhance relaxation measurements, and this may limit the accuracy of data collected especially for nucleic acids. Fortunat ely, various approaches can successfully suppress such unwanted cross-corre lation effects. The false effects of spin echo- modulation during the 13C relaxation delay usually caused by the 13C-13C dipolar coupling can be suppressed by measuring T1 experiment instead of T2 (107) which is measured as a time constant during the loss of spin magnetization under the magnetic spin-lock conditions. T1, T2, and heteronuclear NOE are Fourier transformations of the autoco rrelation function for molecular motion. Autocorrelated (T1, T2, T1, NOE) and cross-correlated relaxation measurements ( 108) are useful in monitoring fast intern al motions in the psŒns timescales. The ability of NMR to detect chemical exch ange between various conformational species usually occurs in the µsŒms timescale. The most frequently used NMR methods for probing conformational exchange the µsŒms timescale are relaxation dispersion measurements. Conformational exchange in the µsŒms timescal e results in broad NMR signals. The broadening, reflected by conformational exchange, Rex, contributes to the measured transverse relaxation rate 26(R2eff ). For a two-site dynamic process, three phys ical parameters are obtained. The parameters are; interconversion rates (k ex), the population fractions of th e exchanging species (pA and pB) and the chemical shifts between pA and pB ( ). The value of the effective transverse relaxation rates (R2(eff) or R1(eff) ) derive contributions from R2* (R2 relaxation from dipolar and chemical shift anisotropy relaxation) or R1 (relaxation rate at infinite spin-lock power) and Rex, is the relaxation attributed to conformational exch ange. Equations 1-4 and 1-5 summarize the expressions. R2(eff) = R2* + Rex 1-4 R1(eff) = R1 + Rex 1-5 There are several studies partic ularly of protein dynamics based on relaxation dispersion and off-resonance R1 (109-113). However, there are onl y a few studies focusing on conformational exchange in RNA, and even so in the of the power dependence of T1 (112, 114-116). An example of studies exploring T1 relaxation in the leadzyme indicates that the thermal fluctuation of the reactive groups are involved in transitioning hi ghly populated inactive state to the low populated active state ( 115). Also, NMR work done by Johnson et al (112), derived the transverse relaxation rates of RNA ribose during conforma tional exchange which was determined to depend on a range of factors including the nuclei exchange lifetime, relative populations of exchanging species, pulse rate, a nd the chemical shift difference between the conformational states. In this thes is, we have used hairpin ribozy me loop A as a model to further characterize similar dynamic mode s in the loop region of RNA usi ng NMR. The findings of this investigation have been presented in detail in Chapter four of this thesis. 271.2.1. Functional probing of RNA dynamics The conformational dynamics of RNA can or iginate from the phosphate backbone, base, or ribose. However, it remains a challenge to directly probe the functional significance of a specific RNA conformational change or dynamic region. Our lab introduced a specific isotope- labeling scheme that addresses the challenges of ribose 13C NMR analysis of dynamics in RNA. Previous attempts at probing RNA dynamics by taking advantage of conf ormational preferences of nucleotides with substitutions at the 2'-carbon have been thwarted by the relatively modest magnitude of such preferences, representing a maximum of ~0.8 kcal/mol at 298 K for the most commonly used nucleotides. Our lab has also demonstrated the use of conformationally restricted locked nucleic acid (LNA) to pertur b ribose dynamics in a lead-dependent ribozyme and other systems ( 117). An LNA nucleotide features a methylene linkage between 2'-oxygen and 4'-carbon which conformationally restri cts the nucleotide to a C3'-endo pucker . Here, Julien et al demonstrated the role of modulating dynamics w ith a 20-fold increase in catalytic rate in the leadzyme upon LNA substitution at G9. This obser vation provides a precise tool for mapping conformational change at a specific ribose site that appears to be a key part of the generation of the active state of the ribozyme. LNAs have also previously be en incorporated into helical regions and recognition arms of 10-23 catalytic DNA and th e hammerhead ribozymes in an effort to improve targeting of cellular and RNA sequences resu lting in dramatic effects upon structural stability a nd target affinity ( 118, 119). This strategy can be applied to a variety of ribozymes that exhibit C3'/C2' sugar pucker transitions . Most notably, critical dynamic residues in loop A of the hairpin ribozyme including G8, U+2, C+3 and A-1 whose sugar puckers are either C2'-endo or mixed C2'/C3' ( 3, 4) can provide good probe si tes for comprehensively assessing the effects of dynamics on the function of hairpin ribozyme. 28In catalytic RNA the role of dynamics is even more prominent than most protein enzymes. It appears that the conformational fluc tuations from a stable ground state to an activated conformer are an inherent part of most ribozymes. One notable example includes the conformational rearrangem ents observed between crystal stru ctures of precu rsor and product forms of the hepatitis delta virus (HDV) with cata lytic or inhibitory metal ions binding only to the precursor form ( 120). Conformational rearrangements have also been observed between isolated and docked hairpin ribozyme loop A structures ( 3, 4). These features show that conformational dynamics is not only vital to catalytic activity but could al so be involved in directing chemistry. In this thesis, we have taken a multi-pronged approach of using Molecular dynamics (MD) simulations and NMR spectrosc opy to characterize conformational dynamics and sampling in loop A of the hairpin ribozyme. Both MD simulation and NMR spectroscopy are robust techniques which present us with a unique ability to investigate dynamic properties of molecules over a range of different timescales w ith atomic resolution, and integrate both results in understanding the mechanism behind RNA-RNA ter tiary interactions. To achieve this goal we have done extensive MD simulations and herein presented our detailed re sults in chapter two. We have also isotopically labeled the hairpin ribozyme loop A using 15N and 13C, assigned the resonances and carried our NMR relaxation studies al so presented in chapter four of this thesis. 29REFERENCES 30REFERENCES 1. Sumita, M., White, N. A., Julien, K. R., a nd Hoogstraten, C. G. (2013) Intermolecular domain docking in the hairpin ribozyme Metal dependence, binding kinetics and catalysis, RNA Biol 10, 425-435. 2. Butcher, S. E., Allain, F. H., and Feigon, J. (1999) Solution structure of the loop B domain from the hairpin ribozyme, Nat Struct Biol 6, 212-216. 3. Rupert, P. B., and Ferre-D'Amare, A. R. ( 2001) Crystal structure of a hairpin ribozyme-inhibitor complex with implications for catalysis, Nature 410, 780-786. 4. Cai, Z., and Tinoco, I., Jr. (1996) Solution st ructure of loop A from the hairpin ribozyme from tobacco ringspot virus satellite, Biochemistry 35, 6026-6036. 5. Frankel, A. D. ( 1999) If the loop fits, Nat Struct Biol 6, 1081-1083. 6. Green, R., and Noller, H. F. (1997) Ribosomes and translation, Annu Rev Biochem 66, 679-716. 7. Paukstelis, P. J., Chen, J. H., Chase, E., Lambowitz, A. M., and Golden, B. L. (2008) Structure of a tyrosyl-tRNA synthetase splicing factor bound to a group I intron RNA, Nature 451, 94-97. 8. Shajani, Z., Drobny, G., and Varani, G. (2007) Binding of U1A protein changes RNA dynamics as observed by 13C NMR relaxation studies, Biochemistry 46, 5875-5883. 9. Bokinsky, G., Nivon, L. G., Liu, S., Chai, G ., Hong, M., Weeks, K. M., and Zhuang, X. (2006) Two distinct binding modes of a pr otein cofactor with its target RNA, J Mol Biol 361, 771-784. 10. Herschlag, D., Khosla, M., Tsuchihashi, Z., and Karpel, R. L. (1994) An RNA chaperone activity of non-specific RNA binding proteins in hammerhead ribozyme catalysis, Embo J 13, 2913-2924. 11. Greenleaf, W. J., Frieda, K. L., Foster, D. A. N., Woodside, M. T., and Block, S. M. (2008) Direct observation of hierarchical folding in single riboswitch aptamers, Science 319, 630-633. 12. Winkler, W., Nahvi, A., and Breaker, R. R. (2002) Thiamine derivatives bind messenger RNAs directly to regulate bacterial gene expression, Nature 419, 952-956. 3113. Bailor, M. H., Mustoe, A. M., Brooks, C. L., 3rd, and Al-Hashimi, H. M. (2011) Topological constraints: using RNA seconda ry structure to model 3D conformation, folding pathways, and dynamic adaptation, Curr Opin Struct Biol 21, 296-305. 14. Schultes, E. A., Spasic, A., Mohanty, U., a nd Bartel, D. P. (2005) Compact and ordered collapse of randomly generated RNA sequences, Nat Struct Mol Biol 12, 1130-1136. 15. Wilson, T. J., Nahas, M., Ha, T., and Lilley, D. M. (2005) Folding and catalysis of the hairpin ribozyme, Biochem Soc Trans 33, 461-465. 16. Solomatin, S. V., Greenfeld, M., Chu, S., a nd Herschlag, D. (2010) Multiple native states reveal persistent ruggedness of an RNA folding landscape, Nature 463, 681-U117. 17. Cruz, J. A., and Westhof, E. (2009) Th e dynamic landscapes of RNA architecture, Cell 136, 604-609. 18. Tinoco, I., and Bustamante, C. (1999) How RNA folds, J Mol Biol 293, 271-281. 19. Toor, N., Keating, K. S., Taylor, S. D., and Py le, A. M. (2008) Crystal structure of a self-spliced group II intron, Science 320, 77-82. 20. Martick, M., and Scott, W. G. (2006) Tertiary contacts distant from the active site prime a ribozyme for catalysis, Cell 126, 309-320. 21. Madhani, H. D., and Guthrie, C. (1994) Dynamic RNA-RNA Interactions in the Spliceosome, Annu Rev Genet 28, 1-26. 22. Wahl, M. C., Will, C. L., and Luhrmann, R. (2009) The Spliceosome: Design Principles of a Dynamic RNP Machine, Cell 136, 701-718. 23. Fedor, M. J. (2000) Structure and function of the hairpin ribozyme, J Mol Biol 297, 269-291. 24. Fedor, M. J. (2009) Comparative enzymo logy and structural biology of RNA self-cleavage, Annu Rev Biophys 38, 271-299. 25. Crick, F. H. (1958) On protein synthesis, Symp Soc Exp Biol 12, 138-163. 26. Orgel, L. E. (1968) Evoluti on of the genetic apparatus, J Mol Biol 38, 381-393. 27. Crick, F. H. (1968) The origin of the genetic code, J Mol Biol 38, 367-379. 28. Guerriertakada, C., Gardiner, K., Marsh, T ., Pace, N., and Altman, S. (1983) The RNA Moiety of Ribonuclease-P Is the Catalytic Subunit of the Enzyme, Cell 35, 849-857. 3229. Kruger, K., Grabowski, P. J., Zaug, A. J., Sands, J., Gottschling, D. E., and Cech, T. R. (1982) Self-splicing RNA: autoexcision and autocyclization of the ribosomal RNA intervening sequence of Tetrahymena, Cell 31, 147-157. 30. Been, M. D., and Wickham, G. S. (1997) Self -cleaving ribozymes of hepatitis delta virus RNA, Eur J Biochem 247, 741-753. 31. Lilley, D. M. (2004) The Varkud satellite ribozyme, RNA 10, 151-158. 32. Lim, J., Grove, B. C., Roth, A., and Break er, R. R. (2006) Characteristics of ligand recognition by a glmS self-cleaving ribozyme, Angew Chem Int Ed Engl 45, 6689-6693. 33. McKay, D. B. (1996) Structure and function of the hammerhead ribozyme: an unfinished story, RNA 2, 395-403. 34. Roth, A., Weinberg, Z., Chen, A. G. Y., Ki m, P. B., Ames, T. D., and Breaker, R. R. (2014) A widespread self-cleaving ribozyme class is revealed by bioinformatics, Nat Chem Biol 10, 56-60. 35. Raines, R. T. (1998) Ribonuclease A, Chem Rev 98, 1045-1066. 36. Wlodawer, A., Miller, M., and Sjolin, L. ( 1983) Active site of RNase: neutron diffraction study of a complex with uridine va nadate, a transition-state analog, P Natl Acad Sci USA 80, 3628-3631. 37. Trautwein, K., Holliger, P., Stackhouse, J., and Benner, S. A. (1991) Site-directed mutagenesis of bovine pancreatic ribonuclease: lysine-41 and aspartate-121, FEBS Lett 281, 275-277. 38. Pyle, A. M. (1993) Ribozymes: a distinct class of metalloenzymes, Science 261, 709-714. 39. Sigel, R. K., and Pyle, A. M. (2007) Alterna tive roles for metal ions in enzyme catalysis and the implications for ribozyme chemistry, Chem Rev 107, 97-113. 40. Murray, J. B., Seyhan, A. A., Walter, N. G., Burke, J. M., and Scott, W. G. (1998) The hammerhead, hairpin and VS ribozymes are catal ytically proficient in monovalent cations alone, Chem Biol 5, 587-595. 41. Roth, A., Nahvi, A., Lee, M., Jona, I., and Breaker, R. R. (2006) Characteristics of the glmS ribozyme suggest only structural roles for divalent metal ions, RNA 12, 607-619. 42. Canny, M. D., Jucker, F. M., Kellogg, E., Khvo rova, A., Jayasena, S. D., and Pardi, A. (2004) Fast cleavage kinetics of a natural hammerhead ribozyme, J Am Chem Soc 126, 10848-10849. 3343. Zamel, R., Poon, A., Jaikaran, D., Andersen, A., Olive, J., De Abreu, D., and Collins, R. A. (2004) Exceptionally fast self-cleavage by a Neurospora Varkud satellite ribozyme, P Natl Acad Sci USA 101, 1467-1472. 44. Brown, T. S., Chadalavada, D. M., and Be vilacqua, P. C. (2004) Design of a highly reactive HDV ribozyme sequence uncovers faci litation of RNA folding by alternative pairings and physiological ionic strength, J Mol Biol 341, 695-712. 45. Wilkinson, S. R., and Been, M. D. (2005) A pseudoknot in the 3' non-core region of the glmS ribozyme enhances self-cleavage activity, RNA 11, 1788-1794. 46. Cech, T. R. (1988) Ribozymes and Their Medical Implications, J Am Med Assoc 260, 3030-3034. 47. Usman, N., and Blatt, L. M. (2000) Nucleas e-resistant synthetic ribozymes: developing a new class of therapeutics, J Clin Invest 106, 1197-1202. 48. Symons, R. H. (1992) Small Catalytic RNAs, Annu Rev Biochem 61, 641-671. 49. Sullenger, B. A., and Cech, T. R. (1993) Te thering Ribozymes to a Retroviral Packaging Signal for Destruction of Viral-RNA, Science 262, 1566-1569. 50. Ojwang, J. O., Hampel, A., Looney, D. J., Wongstaal, F., and Rappaport, J. (1992) Inhibition of Human-Immunodeficiency-V irus Type-1 Expression by a Hairpin Ribozyme, P Natl Acad Sci USA 89, 10802-10806. 51. Rossi, J. J. (2000) Ribozyme therapy for HIV infection, Adv Drug Deliver Rev 44, 71-78. 52. Lieber, A., He, C. Y., Polyak, S. J., Gret ch, D. R., Barr, D., and Kay, M. A. (1996) Elimination of hepatitis C virus RNA in infected human hepatocytes by adenovirus- mediated expression of ribozymes, J Virol 70, 8782-8791. 53. Pulitz, J. Z., Yu, Q., Burke, J. M., and Wa nds, J. R. (1999) Combinatorial screening and intracellular antiviral ac tivity of hairpin ribozymes directed against hepatitis B virus, J Virol 73, 5381-5387. 54. Buzayan, J. M., Gerlach, W. L., Bruening, G., Keese, P., and Gould, A. R. (1986) Nucleotide sequence of satellite tobacco ri ngspot virus RNA and its relationship to multimeric forms, Virology 151, 186-199. 55. Buzayan, J. M., Gerlach, W. L., and Brueni ng, G. (1986) Satellite tobacco ringspot virus RNA: A subset of the RNA sequence is sufficient for autolytic processing, P Natl Acad Sci USA 83, 8859-8862. 3456. Walter, N. G., and Burke, J. M. (1998) Th e hairpin ribozyme: structure, assembly and catalysis, Curr Opin Chem Biol 2, 303. 57. Hampel, A., and Tritz, R. (1989) RNA catalytic properties of the minimum (-)sTRSV sequence, Biochemistry 28, 4929-4933. 58. Anderson, P., Monforte, J., Tritz, R., Nesb itt, S., Hearst, J., and Hampel, A. (1994) Mutagenesis of the hairpin ribozyme, Nucleic Acids Res 22, 1096-1100. 59. Walter, N. G., and Burke, J. M. (1998) Th e hairpin ribozyme: structure, assembly and catalysis, Curr Opin Chem Biol 2, 24-30. 60. Lilley, D. M. (1999) Structure, folding and catalysis of the small nucleolytic ribozymes, Curr Opin Struct Biol 9, 330-338. 61. Fedor, M. J. (2000) Structure and function of the hairpin ribozyme, J Mol Biol 297, 269-291. 62. Cochrane, J. C., and Strobel, S. A. (2008) Ca talytic strategies of self-cleaving ribozymes, Acc Chem Res 41, 1027-1035. 63. Raines, R. T. (1998) Ribonuclease A, Chem Rev 98, 1045-1066. 64. Kuimelis, R. G., and McLaughlin, L. W. (1998) Mechanisms of Ribozyme-Mediated RNA Cleavage, Chem Rev 98, 1027-1044. 65. Lilley, D. M. (2003) The origins of RNA catalysis in ribozymes, Trends Biochem Sci 28, 495-501. 66. Salter, J., Krucinska, J., Alam, S., Grum-T okars, V., and Wedekind, J. E. (2006) Water in the active site of an all-RNA hairpin ribozym e and effects of Gua8 base variants on the geometry of phosphoryl transfer, Biochemistry 45, 686-700. 67. Rupert, P. B., Massey, A. P., Sigurdsson, S. T., and Ferre-D'Amare, A. R. (2002) Transition state stabilizat ion by a catalytic RNA, Science 298, 1421-1424. 68. Torelli, A. T., Krucinska, J., and Wedekind, J. E. (2007) A comparison of vanadate to a 2'-5' linkage at the active site of a small ribozyme suggests a role for water in transition- state stabilization, RNA 13, 1052-1070. 69. Pinard, R., Lambert, D., Walter, N. G., Heckman, J. E., Major, F., and Burke, J. M. (1999) Structural basis for the guanosine requirement of the hairpin ribozyme, Biochemistry 38, 16035-16039. 3570. Butcher, S. E., Allain, F. H., and Feigon, J. (1999) Solution structure of the loop B domain from the hairpin ribozyme, Nat Struct Biol 6, 212-216. 71. Butcher, S. E., Heckman, J. E., and Burk e, J. M. (1995) Reconstitution of hairpin ribozyme activity following separation of functional domains, J Biol Chem 270, 29648-29651. 72. Hoogstraten, C. G., Legault, P., and Pardi, A. (1998) NMR solution structure of the lead-dependent ribozyme: evidence for dynamics in RNA catalysis, J Mol Biol 284, 337-350. 73. Butcher, S. E., Heckman, J. E., and Burk e, J. M. (1995) Reconstitution of hairpin ribozyme activity following separation of functional domains, J Biol Chem 270, 29648-29651. 74. Hampel, K. J., Walter, N. G., and Burke, J. M. (1998) The solvent-protected core of the hairpin ribozyme-substrate complex, Biochemistry 37, 14672-14682. 75. Boehr, D. D., McElheny, D., Dyson, H. J ., and Wright, P. E. (2006) The dynamic energy landscape of dihydrofolate reductase catalysis, Science 313, 1638-1642. 76. Beach, H., Cole, R., Gill, M. L., and Loria, J. P. (2005) Conservation of mus-ms enzyme motions in the apo- and substrate-mimicked state, J Am Chem Soc 127, 9167-9176. 77. Best, R. B., Zhu, X., Shim, J., Lopes, P. E., Mittal, J., Feig, M., and Mackerell, A. D., Jr. (2012) Optimization of the additive CHARMM all-atom protein force field targeting improved sampling of the backbone phi, psi and side-chain chi(1) and chi(2) dihedral angles, J Chem Theory Comput 8, 3257-3273. 78. Vanommeslaeghe, K., Hatcher, E., Achary a, C., Kundu, S., Zhong, S., Shim, J., Darian, E., Guvench, O., Lopes, P., Vorobyov, I., and MacKerell, A. D. (2010) CHARMM General Force Field: A Force Field for Dr ug-Like Molecules Compatible with the CHARMM All-Atom Additive Biological Force Fields, J Comput Chem 31, 671-690. 79. Pastor, R. W., and MacKerell, A. D. (2011) Development of the CHARMM Force Field for Lipids, J Phys Chem Lett 2, 1526-1532. 80. Cornell, W. D., Cieplak, P., Bayly, C. I., Gould, I. R., Merz, K. M., Ferguson, D. M., Spellmeyer, D. C., Fox, T., Caldwell, J. W., and Kollman, P. A. (1996) A second generation force field for the simulation of pr oteins, nucleic acids, and organic molecules (vol 117, pg 5179, 1995), J Am Chem Soc 118, 2309-2309. 81. Oostenbrink, C., Villa, A., Mark, A. E., a nd Van Gunsteren, W. F. (2004) A biomolecular force field based on the free enthalpy of hydration and solvation: The GROMOS force-field parameter sets 53A5 and 53A6, J Comput Chem 25, 1656-1676. 3682. Jorgensen, W. L., Maxwell, D. S., and TiradoR ives, J. (1996) Development and testing of the OPLS all-atom force field on conforma tional energetics and properties of organic liquids, J Am Chem Soc 118, 11225-11236. 83. Jorgensen, W. L., and Tiradorives, J. (1988) The Opls Potential Functions for Proteins - Energy Minimizations for Crystals of Cyclic-Peptides and Crambin, J Am Chem Soc 110, 1657-1666. 84. Weiner, S. J., Kollman, P. A., Nguyen, D. T., and Case, D. A. (1986) An All Atom Force-Field for Simulations of Proteins and Nucleic-Acids, J Comput Chem 7, 230-252. 85. Karplus, M., and Petsko, G. A. (1990) Molecular-Dynamics Simulations in Biology, Nature 347, 631-639. 86. Mackerell, A. D. (2004) Empirical force fiel ds for biological macromolecules: Overview and issues, J Comput Chem 25, 1584-1604. 87. Brooks, B. R., Brooks, C. L., Mackerell, A. D., Nilsson, L., Petrella, R. J., Roux, B., Won, Y., Archontis, G., Bartels, C., Boresch, S., Caflisch, A., Caves, L., Cui, Q., Dinner, A. R., Feig, M., Fischer, S., Gao, J., Hodos cek, M., Im, W., Kuczera, K., Lazaridis, T., Ma, J., Ovchinnikov, V., Paci, E., Pastor, R. W., Post, C. B., Pu, J. Z., Schaefer, M., Tidor, B., Venable, R. M., Woodcock, H. L., Wu, X., Yang, W., York, D. M., and Karplus, M. (2009) CHARMM: The Biomolecular Simulation Program, J Comput Chem 30, 1545-1614. 88. Rhodes, M. M., Reblova, K., Sponer, J., and Walter, N. G. (2006) Trapped water molecules are essential to structural dynamics and function of a ribozyme, P Natl Acad Sci USA 103, 13380-13385. 89. McDowell, S. E., Spackova, N., Sponer, J., and Walter, N. G. (2007) Molecular dynamics simulations of RNA: An in silico single molecule approach, Biopolymers 85, 169-184. 90. Ditzler, M. A., Sponer, J., and Walter, N. G. (2009) Molecular dynamics suggest multifunctionality of an adenine imino group in acid-base catalysis of the hairpin ribozyme, RNA-A Publication of the RNA Society 15, 560-575. 91. Ditzler, M. A., Otyepka, M., Sponer, J., a nd Walter, N. G. (2010) Molecular Dynamics and Quantum Mechanics of RNA: Conformational and Chemical Change We Can Believe In, Accounts Chem Res 43, 40-47. 92. Mlynsky, V., Banas, P., Hollas, D., Reblova, K., Walter, N. G., Sponer, J., and Otyepka, M. (2010) Extensive Molecular Dynamics Si mulations Showing That Canonical G8 and Protonated A38H(+) Forms Are Most Consiste nt with Crystal Stru ctures of Hairpin Ribozyme, J Phys Chem B 114, 6642-6652. 3793. Mlynsky, V., Banas, P., Walter, N. G., S poner, J., and Otyepka, M. (2011) QM/MM Studies of Hairpin Ribozyme Self-Cleavag e Suggest the Feasibility of Multiple Competing Reaction Mechanisms, J Phys Chem B 115, 13911-13924. 94. Kendrew, J. C., Bodo, G., Dintzi s, H. M., Parrish, R. G., Wyckoff, H., and Phillips, D. C. (1958) A three-dimensional model of the myoglobin molecule obtained by X-ray analysis, Nature 181, 662-666. 95. Rould, M. A., Perona, J. J., Soll, D., and Steitz, T. A. (1989) Structure of E. coli glutaminyl-tRNA synthetase complexed with tRNA(Gln) and ATP at 2.8 Å resolution, Science 246, 1135-1142. 96. Pley, H. W., Flaherty, K. M., and McKay, D. B. (1994) Three-dimensional structure of a hammerhead ribozyme, Nature 372, 68-74. 97. Scott, W. G., Finch, J. T., and Klug, A. (1995) The crystal structure of an all-RNA hammerhead ribozyme: a proposed mechanism for RNA catalytic cleavage, Cell 81, 991-1002. 98. Wang, S., Karbstein, K., Peracchi, A., Beigelman, L., and Herschlag, D. (1999) Identification of the hammerhead ribozyme meta l ion binding site responsible for rescue of the deleterious effect of a cleavage site phosphorothioate, Biochemistry 38, 14363-14378. 99. Allerhan.A, Doddrell, D., Glushko, U., Cochra n, D. W., Wenkert, E., Lawson, P. J., and Gurd, F. R. N. (1971) Carbon-13 Fourier Tranform Nuclear Magnetic Resonance .3. Conformation and Segmenta l Motion of Native and Denatured Ribonuclease-A in Solution - Application of Natural-Abunda nce Carbon-13 Partially Relaxed Fourier Transform Nuclear Magnetic Resonance, J Am Chem Soc 93, 544-546. 100. Glushko, V., Lawson, P. J., and Gurd, F. R. N. (1972) Conformational States of Bovine Pancreatic Ribonuclease a Observed by Normal and Partially Relaxed Carbon 13 Nuclear Magnetic-Resonance, Journal of Biological Chemistry 247, 3176-3185. 101. Kojima, C., Ono, A., Kainosho, M., and James, T. L. (1998) DNA duplex dynamics: NMR relaxation studies of a decamer with uniformly 13C-labeled purine nucleotides, J Magn Reson 135, 310-333. 102. Brutscher, B., Boisbouvier, J., Pardi, A., Marion, D., and Simorre, J. P. (1998) Improved sensitivity and resolution in 1H-13C NMR experiments of RNA, J Am Chem Soc 120, 11845-11851. 103. King, G. C., Harper, J. W., and Xi, Z. (1995) Isotope labeling for 13C relaxation measurements on RNA, Methods Enzymol 261, 436-450. 38104. Lipari, G., and Szabo, A. (1981) Nuclear ma gnetic resonance relaxation in nucleic acid fragments: models for internal motion, Biochemistry 20, 6250-6256. 105. Lipari, G., and Szabo, A. (1982) Model-Free Approach to the Interpretation of Nuclear Magnetic-Resonance Relaxation in Macromol ecules .1. Theory and Range of Validity, J Am Chem Soc 104, 4546-4559. 106. Lipari, G., and Szabo, A. (1982) Model-Free Approach to the Interpretation of Nuclear Magnetic-Resonance Relaxation in Macrom olecules .2. Analysis of Experimental Results, J Am Chem Soc 104, 4559-4570. 107. Yamazaki, T., Muhandiram, R., and Kay, L. E. (1994) Nmr Experiments for the Measurement of Carbon Relaxation Propertie s in Highly Enriched, Uniformly C-13,N- 15-Labeled Proteins - Application to C-13(Alpha) Carbons, J Am Chem Soc 116, 8266-8278. 108. Palmer, A. G., Williams, J., and McDermo tt, A. (1996) Nuclear magnetic resonance studies of biopolymer dynamics, J Phys Chem 100, 13293-13310. 109. Mulder, F. A., Skrynnikov, N. R., Hon, B., Dahlquist, F. W., and Kay, L. E. (2001) Measurement of slow (micros-ms) time scale dynamics in protein side chains by (15)N relaxation dispersion NMR spectroscopy: applica tion to Asn and Gln residues in a cavity mutant of T4 lysozyme, J Am Chem Soc 123, 967-975. 110. Traaseth, N. J., and Veglia, G. (2010) Probing excited states and activation energy for the integral membrane protein phospholamb an by NMR CPMG relaxation dispersion experiments, Biochim Biophys Acta 1798, 77-81. 111. Loria, J. P., Berlow, R. B., and Watt, E. D. (2008) Characterization of enzyme motions by solution NMR relaxation dispersion, Accounts Chem Res 41, 214-221. 112. Johnson, J. E., Jr., and Hoogstraten, C. G. (2008) Extensive backbone dynamics in the GCAA RNA tetraloop analyzed using 13C NMR spin relaxation and specific isotope labeling, J Am Chem Soc 130, 16757-16769. 113. Palmer, A. G., 3rd, Kroenke, C. D., and Lori a, J. P. (2001) Nuclear magnetic resonance methods for quantifying microsecond-to -millisecond motions in biological macromolecules, Methods Enzymol 339, 204-238. 114. Shajani, Z., and Varani, G. (2005) 13C NMR relaxation studies of RNA base and ribose nuclei reveal a complex pattern of motions in the RNA binding site for human U1A protein, J Mol Biol 349, 699-715. 115. Hoogstraten, C. G., Wank, J. R., and Pardi, A. (2000) Active site dynamics in the lead- dependent ribozyme, Biochemistry 39, 9951-9958. 39 116. Blad, H., Reiter, N. J., Abildgaard, F., Ma rkley, J. L., and Butcher, S. E. (2005) Dynamics and metal ion binding in the U6 R NA intramolecular stem-loop as analyzed by NMR, J Mol Biol 353, 540-555. 117. Julien, K. R., Sumita, M., Chen, P. H., Lair d-Offringa, I. A., and Hoogstraten, C. G. (2008) Conformationally restricted nucleo tides as a probe of structure-function relationships in RNA, RNA 14, 1632-1643. 118. Christiansen, J. K., Lobedanz, S., Arar, K., Wengel, J., and Vester, B. (2007) LNA nucleotides improve cleavage efficiency of singular and binary hammerhead ribozymes, Bioorg Med Chem 15, 6135-6143. 119. Donini, S., Clerici, M., Wengel, J., Vester, B., and Peracchi, A. (2007) The advantages of being locked. Assessing the cleavage of short and long RNAs by locked nucleic acid-containing 8-17 deoxyribozymes, J Biol Chem 282, 35510-35518. 120. Ke, A., Zhou, K., Ding, F., Cate, J. H., and Doudna, J. A. (2004) A conformational switch controls hepatitis de lta virus ribozyme catalysis, Nature 429, 201-205. 40 CHAPTER 2 PROBING ALTERNATE BASE-PAIR REARRANGEMENTS AND CONFORMATIONAL SAMPLING IN A RIBOZYME ACTIVE-SITE INTERNAL LOOP USING MOLECULAR DYNAMICS SIMULATIONS 41ABSTRACT Dynamic fluctuations in RNA structure st eer changes that correlate with a broad spectrum of functions. These conformational fluc tuations yield various populations based on the free energy and the transition rates between vari ous conformers along the free energy landscape. Conformational redistribution in non-coding RNA is driven by cellular sign als that affect the rugged RNA free energy landscape. To assess conf ormational sampling in the pre-docking form of the hairpin ribozyme, we have run several explicit solv ent Molecular Dynamics (MD) simulations of the hairpin ribozyme loop A do main totaling 2.4 µs. We observed one dominant conformer and other minor states identif ied using hydrogen bonding and base stacking interactions in the ac tive site loop region of loop A. Targ eted Molecular Dynamics (TMD) was used to model inter-conformer transitions. Usin g the Markov State Mode l (MSM) analysis we constructed the kinetic pathways within each co nformer and the transition network among a set of conformers. The Molecular Mechanics Poisson-Boltzmann Surface Area (MMPB/SA) approximation was applied to predict conforma tional energies and energy barriers between conformations. The barrier height between two of the observed conformations (named AA/UA and AA/CA) is accessible based on modest energi es reported for this transition. However, transitions to a third conformer (GA/UA) appear to be energetically inaccessible. This study indicates that RNA undergoes struct ural versatility through several kinetic states and transitions, a phenomenon that underscores the formation of a rugged tertiary energy landscape with multiple accessible states for RNA, consistent w ith a role for conformational sampling in the pre- organization of loop A RNA prior to RNA-RNA interactions. 422.1 INTRODUCTION The critical role of c onformational dynamics in macromolecu lar function, allostery, catalysis, and ligand recognition is well appreciated (1-4). The dynamics-function re lationship provides the missing link in more traditional structure-function paradigms where the motional variables have been frozen out using mutation or modification ( 5, 6). While the structure-dynamics-function relationship has been explored extensively in proteins, less is known ab out how RNA structure and dynamics relates to its biological roles ( 7, 8). In the context of ribozymes, a detailed understanding of the how RNA st ructure and dynamics limit catal ytic rates involves dynamic aspects of RNA-RNA interaction processes. Two fr ameworks are often discussed to describe the formation of macromolecular inte ractions: induced fit and conformational capture. In the induced fit mechanism, binding of the mo st populated state is followed by conformational change within the bound state. In contrast, the conforma tional capture mechan ism invokes ground-state fluctuations of the binding part ners that create a state resemb ling the conformation of the bound state. Binding then occurs when the partner(s) visit the respective bind ing conformations. When the energy landscape of the binding partners is comp lex because one or both partners in isolation sample multiple conformations, the presence of a defined state or measurable population corresponding to a conformation resembling the bound state would be taken as evidence for conformational capture. Molecular Dynamics (MD) simulation and Nuclear Magnetic Reso nance (NMR) have showcased their power in assessing macromolecula r motions as well as conformational sampling (9, 10) and are thus useful for inte rrogating timescales relevant for the conformational capture process. In the case of RNA, th e energy landscape is typically mo re rugged than for proteins ( 11-13). This means that multiple alternate states, se parated by kinetic barriers, may be populated to 43a significant extent at physiological temperatures. This has hindered past efforts to fully characterize the conformational dynamics of RNA a nd in particular to e xplore to what extent conformational capture or indu ced fit processes are at play in RNA-RNA interactions. Here, we have used the hair pin ribozyme as a model to study RNA dynamics in ensemble formation and inter-conformer transitions. The hairpin ribozyme catalyzes reversible, site- specific phosphodiester bond cleavage to yield 5'-O H and a 2'- 3'-cyclic phosphate terminus ( 14, 15). The ribozyme consists of two RNA loops A and B that interact (dock) to form a multi- domain structure that ac ts to cleave the phosphodies ter bond within loop A ( Figure 2-1). Loops A and B and their flanking helices constitute the catalytic core of the hairpin ribozyme derived from tobacco ringspot virus satellite RNA ( 16). This catalytic core re veals a network of stacking and hydrogen-bonding interactions within the active si te that orient the re active phosphate in the in-line orientation for an S N2-type nucleophilic attack mechanis m. In addition, nucleotide base functional groups are oriented n ear the reactive phosphate to fac ilitate catalytic chemistry ( 17, 18). The catalytic function of th e hairpin ribozyme represents a system that can serve as a potential therapeutic agent. Previously, the hair pin ribozyme has been engineered to specifically cleave HIV-1 RNA (19, 20). NMR structures of the isolated loop A and B have previously been determined as have been the crystal structures of the numerous docked forms ( 17, 21-28). Significant structural rearrange ments of both loops are observe d upon formation of the docking complex ( 17, 21, 22, 29-31). For example, U+2 in the free form of loop A is extrahelical, while it forms a non-canonical G8-U+2 base -pair in the docked form. Similarly, U41 in the free loop B is extrahelical, but it forms the A22-U41 canonical base-pair in the docked form ( Figure 2-2). A critical residue G+1 in loop A interacts with C25 in loop B to form a G+1-C25 inter-domain 44base-pair by interrupting the G+1-A9 and C25- U37 non-canonical base-pairs in the individual loops A and B, respectively.Recently, loop A and B interaction kinetic rate s have been reported by the Hoogstraten group (32). The reported docking association rate rev eals a very slow but tight docking process between loop A and loop B (on-rate five orders of magnitude below the diffusion limit). This slow association rate as well as structural versatility between the free and docked forms, especially for the key residues involved in the interaction, has led us to the hypothesis that isolated loops A and B may both sample the docking-competent states in a possibly double conformational capture docking mechanism. The major loop rearrangements observed upon docking, as well as the kinetical ly unfavorable docking process, both argue for a tight and mechanistic role for conformational dynamics in the pre-organization of the catalytically- competent active site of the hairpin ribozyme. An alternative induced fit mechanism would be expected to be kinetically acce lerated. Here, we focus on probi ng the conformational sampling of just the loop A domain of the ha irpin ribozyme as an initial step to understanding the active site dynamics in this RNA system. As the main tool, we have used MD simulations to map out the conformational ensemble visited by loop A and we describe here the energetics and the pathway for transitions between alternate conformers identified from the simulations via transitions through kinetic states. FigurmetalDockiA regre 2-1: Tranl-dependenting of loops ion. Cleavagns-acting hat loop-loop A and B prge and ligatioairpin ribozdocking. Arecedes clea von site are sh45zyme seco nA and B d evage betwe ehown by smndary struc enote loops en A-1 and Gmall arrows inture repre sA and B rG+1 residue sn loop A. sentation o frespectively .s in the loo pf . p FigureA and (dockedRibose equilibrU42. Re 2-2: Sche mB based o d loop A an dpuckers a rium (grey) Red lines reprmatic repre sn the NM Rd B). Penta gre indicate d. The doubresent new sentation of R structures gons represe nd as follo wle-boxed re set of base-p46free loops A(free loop Ant ribose su gws: C3'-end osidues formairs in dock eA and B as wA and B) agar, while b oo (open), Cm structured ed structure.well as doc kand crystal oxes represeC2'-endo (s obinding po. ked loop structure nt bases. olid), or ocket for 472.2 MATERIALS AND METHODS 2.2.1 Molecular dynamics (MD) simulation s using NAMD and the CHARMM force field Explicit solvent Molecular Dynamics (MD) simu lations were carried out initially with loop A coordinates kindly provided by I gnacio Tinoco Jr. based on the loop A structure determined by NMR (22) to validate the force field. In additiona l simulations, stem 1 of Tinoco™s loop A (denoted as LpA) was modeled to incorpor ate a GAAA tetraloop from an RNA structure in complex with theophylline (PDB code 1O15) ( 33). Residue C-1 was mutated to A-1 using the MMTSB Tool Set ( 34) to restore the binding sequence of loop A to its native sequence. One G-C base pair was added at the end of stem II to enhance its stability during the simulation time. This construct is denoted hpA. The hpA 2'-OH of the A-1 residue was also methylated in additional simulations to explore experimental constructs. The 2'- O-methylated hpA is denoted hpAome. Experimental data from our group (unpublished) had previously show n significantly lower docking affinities for 2'- O-methylated loop A (hpAome) with loop B compared to non- methylated loop A (hpA). Constructs are illu strated in Figure 3 of the Results section. The loop A RNA structures were solvated in a cubical box with TIP3P water molecules ( 35, 36) using a minimal distance of 10 Å from any R NA atom to the edge of the simulation box. The systems were neutralized with 18 Na + ions for Tinoco™s loop A (LpA) and 26 Na + counterions for both hpA and hpAome constructs. The cubical box dimensions were (56.8 Å) 3, (62.3 Å)3, and (63.4 Å)3 for Tinoco™s loop A (LpA), hpA, and hpAom e, respectively. Peri odic boundaries were applied in all simulations. The particle-mesh Ew ald (PME) method was used to model the long- range electrostatics interactions ( 37, 38). The solute-solvent systems were minimized with an initial harmonic restraint potential applied to he avy atoms with a force constant of 5 kcal/(mol Å2). The same restraints were also applied during the initial heating phases. Finally, the systems 48were allowed to relax in several equilibration runs with a decreasing harmonic potential applied to the heavy atoms at 10 K temperature increments until a temperature of 200 K was attained. Additional equilibration wit hout restraints was then completed to allow for full relaxation of the system while slowly heating to 298 K over 100 ps. All bonds involving hydrogens were constrained holonomically by the SE TTLE algorithm, which allows us to use a time-step of 2 fs. The CHARMM36 force field was used ( 39-43) in NAMD ( 44) to run Tinoco™s loop A (LpA) construct for 100 ns. CHARMM36 was also used to simulate the loop A constructs, hpA and hpAome. Six independent simulations were run for the hpA and hpAome systems using different random seeds. Individual simulations are la beled hpA1-6 and hpAome1-6 respectively. All simulations are summarized in Table 2-1. 49Table 2-1: Summary of molecular dynami cs simulations of loop A constructs Simulation LpAa LpA a hpA b hpAome c Length of simulation (ns) 100 100 200 200 Force field CHARMM36 Amber ff10 CHARMM36 CHARMM36 Number of simulation 1 1 6 6 aLpA denotes explicit solvent simulation on Tinoco™s loop A ( Figure 3A) using CHARMM36 and Amber ff10 force fi elds respectively. bhpA represents simulation of our loop A construct ( Figure 3B ) without any methylation. The simulation replicates (six) were desi gnated hpA1, hpA2, hpA3, hpA4, hpA5 and hpA6. chpAome is our simulation construct with 2'- O-methylation at residue A-1. This simulation has also six replicates named hpAome1, hpA ome2, hpAome3, hpAome 4, hpAome5, hpAome6. 502.2.2 MD simulations using sand er and the Amber force field To compare force field performance, we al so carried out simulations with the Amber force field. Again, the starting structure was obtained from Tinoco™s loop A NMR structure (LpA). The RNA was neutralized with 18 Na + counter ions using xleap at positions of high negative electric potential and solvated in a rectangular wate r box measuring 66.4 Å x 69.4 Å x 80.4 Å with a minimum 10 Å thick layer of TI P3P water molecules around the loop A RNA. Long distance electrostatic interactions were accounted fo r again by using the particle-mesh Ewald summation ( 38). Minimization of the solute-solvent system was done in two steps. First, the RNA was restrained with a harm onic potential force of 500 kcal/(mol Å 2) in 500 steps of steep descend minimization followed by 500 steps of conjugate gradient minimization. Second, the whole system (RNA, solvent and ions) was minimized together wit hout restraints during 2,500 steps at a constant volume with periodic boundaries. After the dual-step minimization, the system was gradually equilibrated by heating from 0 K to 300 K over 100 ps using the Langevin thermostat ( 45) with 10 kcal/(mol Å 2) positional restraint using NVT followed by a 100 ps equilibration at 300 K under NPT conditions without restraints. The SHAKE algorithm ( 46) was used to constrain covalent hydrogen bonds during th e simulation and allow a 2 fs time step. The sander module of Amber 11 with the amberff10 force field was used for this simulation ( 47, 48) followed by a production phase of 100 ns. 2.2.3 Targeted molecular dynamics (TMD) simulation Targeted Molecular Dynamics (TMD) ( 49) simulations were applied to sample the transition between two sets of conformers identi fied in the unbiased simulations. In particular, TMD was used to drive one of the observed simu lation structures (denoted AA/UA, see results) 51to two different structures (denoted as GA/UA and AA/CA) in separate TMD simulations. In TMD, the following biasing potential was applied to drive one conformation to the other: 20)()(21tRMSDtRMSDNkUTMD where RMSD(t) is the root mean square deviation of heavy atom coordinates with respect to the target coordinates, RMSD 0(t) is the target RMSD value at a given time, k is the force constant and N denotes the number of atoms. Initial va lues of RMSD(t) were set to 4.01 Å for the AA/UA-GA/UA TMD simulation and to 3.79 Å for the AA/UA-AA/CA TMD simulation. We initially tried several different force constants to optimize the transition and ability to reach a final RMSD difference of less than 1 between the starting and targ et structures. After several trial runs, we decided on a forc e constant of 300 kcal/(mol Å 2) that was applied to all heavy atoms. TMD was carried out using explicit solven t with a periodic boundary box of 62 x 62 x 62 Å3 to solvate RNA neutralized by Na + counterions with the same parameters as in the unbiased CHARMM simulations described above. After initial minimization and equilibration, TMD simulations were run at 298K for 2 ns. Both sets of TMD simulations were repeated ten times each time with a different random seed. The simu lation trajectories were combined and clustered using a k-means clustering algorithm based on RMSD to obtain conformational transition pathways between the initial and target conf ormations. The observed conformations along the TMD pathway obtained via clustering were then subjected to Molecular Mechanics Poisson- Boltzmann Surface Area (MMPB/SA) analysis ( 50-59) to estimate the conformational free energies in each ensemble cluster along the TMD transition pathways. 522.2.4 Analysis The analysis of the simulations was carried out using the MMTSB (Multiscale Modeling Tools for Structural Biology) Tool Set ( 34) in conjunction with C HARMM. Markov state model (60) analysis was also used to identify and model conformational states and the kinetic transitions between those states using the MSMBuilder 2.0 software package ( 61). RMSD values were calculated for all heavy atoms excluding th e terminal bases due to fraying. Previously reported NMR nuclear Overhaus er effect (NOE) distances were measured for average CHARMM36 and Amber ff10 structures using PyMOL ( 62) and VMD (63) visualization software. 2.3 RESULTS 2.3.1 Force field validation Molecular dynamics simulations were carried out on the loop A domain of the hairpin ribozyme to study the conformational sampling of R NA, especially in the context of encounter complexes. Since stability of force fields is an ongoing problem in the RNA simulation field, , we initially carried out simulations of the loop A structure obtained from Cai and Tinoco Jr. ( 22) using the two recent force fields from th e Amber and CHARMM families (Amber ff10 and CHARMM36) to evaluate which force field results in better agreement with the experimental data. Figure 2-3A shows the heavy atom root mean squa re deviation (RMSD) with respect to Tinoco™s loop A starting structure during 100 ns of MD simulation with either the CHARMM36 or Amber ff10 force fields. Both simulations were stable as evidenced by RMSD values averaging about 3 Å from the NMR starting stru cture during the simulation. Furthermore, the agreement with NMR-derived Nuclear Overhaus er Effect (NOE) distance constraints ( 22) was evaluated. The distances between specific protons that registered strong and/or medium NOEs 53for the loop A average simulation structures were an alyzed from the simulations. The results, summarized in Table 2-2, focus on five characteristic distance s that are not present in regular A- form structures for which NOE c onstraints are available. As can be seen, in the simulation with the Amber force field, the average simulation st ructure violates all but one NOE distance constraint, while only two NOEs are missed with CHARMM36 (A9H1'-C+3H6 and A9H1'- C+3H1') and the A9H1'-C+3H6 distance in the CHARMM36 simulation is closer to the NOE range than with the Amber for ce field. We note, however, that in both of these cases, the reported solution structure has NOE distances at the upper limit of the NOE constraint range, indicating that these distance constraints may be problematic. Finally, we analyzed ribose sugar puckering and base geometries (shear, buckle et c.) and again found better agreement with the experimental data for the average structur e generated by the CHARMM36 force field ( Table 2-3) compared to the Amber ff10 aver age structure. Based on these initial simulation results, we chose the CHARMM36 force field for carrying out subsequent simulations of the wild-type loop A (Figure 2-3B) construct. 2.3.2 Conformational heterogeneity of wildtype loop A The wild-type loop A construct was modeled based on the mutant l oop A NMR structure but with C-1 mutated to A-1. While we were initia lly interested in comparing the methylated and non-methylated constructs to explain differences in experimental association rates, analysis of the simulation results does not suggest significant differences between the two forms in terms of the conformational sampling of loop A alone ( Figure 2-4 ). Using the, statistical F-test, we obtained an F-value of 1.1164 between the two population variances whic h suggests that the conformational distribution betw een methylated and non-methylat ed loop A was similar in all our simulations. FihelomreAAanCsireigure 2-3: Seavy atom Roop A (B) anmajor sequencepresents 2'- OAmber ff10 a nAmberff10 w nd C, the nonHARMM36mulation ru nepresent dif fe Secondary stRMSD profnd 2'- O-met hce variation sO-methylat end CHARMhile green ren-methylate d6 force field. ns with diff eferent runs. tructure repfiles from Tinhylated (hp As and modif ied adenosineMM36 force fepresents C Hd and meth y The RMSDerent rando m54presentationnoco™s loop Aome) loop Aications of the. In A, Tinofields respe cHARMM36 ylated loop AD profiles in Bm seed for ean of loop A A (A), non-A (C). The bhe wildtype Aoco™s loop A ctively. Red psimulations A variants weB and C repach variant. Twith their r-methylated boxes highligA-1 residuewas simula tprofile repr erespectivel yere simulat eresent inde pThe profile crespective (hpA) ght the . Am ted using esents y. In B ed using pendent colors 55Table 2-2: Comparison of NOE distance constraints with si mulation averages using Amber and CHARMMa Atom Pair NOE Distance constraint (Å) Tinoco™s structure (Å) CHARMM36 MD average (Å) Amber ff10 MD average (Å) U+2H1'- C+3H5 1.8-3.0 3.0 2.5 5.9 A9H1'- C+3H6 2.5-4.0 3.9 5.4 8.5 A9H1'- C+3H5 3.0-6.5 4.4 6.3 9.5 A9H1- C+3H1' 3.0-6.5 3.2 4.5 5.4 A9H4'- C+3H1' 1.8-4.0 4.0 7.6 7.7 a NOE distances as reported in Tinoco™s loop A structure ( 22) in comparison to distances in the average structures derived from simulation with CHARMM36 and Amber ff10 force fields repectively. Atom distances were obtai ned from 100 ns CHARMM36 and Amber ff10 simulation average structures. 56Table 2-3: Representative ba ckbone and base-pair parameters of average structures derived from CHARMM36 and Amberff10 simulations respectively The numbers in bold indicate a significant confor mational deviation of residue/base-pair with respect to Tinoco™s structure. Pseudorotation phase angle (degrees) torsion angle (degrees) Residue Tinoco™s structure CHARMM36 MD average Amberff10 MD average Tinoco™s structure CHARMM36 MD average Amberff10 MD average A4 7.2 14.8 11.5 -165.6 -161.9 -79.1 C5 6.5 13.8 19.3 -156.8 -156.8 -147.3 A6 6.7 9.0 11.9 -162.9 -152.7 -158.2 A7 59.5 10.1 14.9 -169.5 -153.6 -148.3 G8 181.3 173.7 17.3 -113.4 -107.8 -173.3 A9 19.9 8.2 12.7 -162.9 -163.8 175.6 A10 9.9 10.2 8.9 -171.6 -165.1 -169.7 G11 0.4 13.3 7.2 -175.6 -168.5 -154.7 G12 4.8 10.3 10.8 -170.4 -165.7 -166.9 C13 7.1 37.5 55.1 -161.1 -152.9 -147.5 G16 4.8 15.9 11.7 -179.2 -165.3 67.6 C17 2.6 11.8 17.5 -162.9 -160.3 -173.5 C18 16.3 14.4 17.6 -154.6 -160.9 -156.5 C19 5.2 13.4 15.1 -172.6 -156.7 -143.3 G20 186.2 175.7 169.4 -95.5 -116.8 -121.5 U21 16.8 152.6 173.3 -173.4 -109.4 -144.9 C22 40.1 11.9 24.7 -140.9 -158.8 -165.6 U23 15.42 22.5 25.4 -153.9 -152.3 -160.0 G24 2.07 13.6 11.9 -171.8 -149.2 -162.4 U25 8.8 ND 13.1 ND ND -154.8 Shear (degrees) Buckle (degrees) Base-pair Tinoco™s structure CHARMM36 MD average Amberff10 MD averageTinoco™s structure CHARMM36 MD average Amberff10 MD average A4-U25 0.2 -1.1 7.7 6.8 30.6 35.2 C5-G24 1.2 -0.4 -0.1 12.6 4.8 4.2 A6-U23 -0.4 -0.5 0.5 29.8 17.7 3.41 A7-C22 4.1 5.4 6.9 -14.1 -7.9 -11.4 G8-U21 -0.3 1.9 -4.8 163.8 38.8 -2.5 A9-G20 -4.9 -5.5 -2.6 -19.3 -19.2 -166.9 A10-C19 -2.9 -2.8 -2.6 -14.6 0.6 -167.1 G11-C18 -0.6 0.1 -2.1 -2.4 0.5 172.5 G12-C17 -0.5 -0.4 -2.1 1.9 -3.9 168.8 C13-G16 0.2 0.5 4.5 6.8 0.6 175.2 FaoxdFigure 2-4: Pand methylaof loop A. Thx 200 ns simudistributions Population ated loop A he distributioulation (six sis similar w distribution(red) based on plot captusimulations ithin 95% c o57n between non RMSD fures RMSD each). The sonfidence levnon-meth ylafrom explici tdata betweestatistical F- tvel ated loop A t solvent si mn 40-200 ns test betwee n(green) mulation of the 12 n the two 58Therefore, we combined the sampling from both sets of simulations to yield a total of 2.4 µs explicit solvent MD simulations to describe the overall conformational ensemble of loop A. As can be seen in Figure 2-3 B, there were significant RMSD variations in some of the replicas indicating conformational heterogeneity. Representa tive simulation structur es were subsequently obtained via k-means clustering based on mutual RMSD. The resulting conformations have been shown in Figure 2-5. The top and bottom three base-pairs in stems I and II maintained similar Watson-Crick base-pairing, as expected, but there we re significant variati ons in the loop region (A-1 to C+3 and A7 to A10) involving alternat e base-pairing and extra-helical bases. At the bottom of stem I, A-1 and G+1 residues were in base-pair competition with A10. In cases where A-1 was base-paired to A10, G+1 was either base-paired with A9 or stacked between other residues. G8 was always unpaired or stacked betw een other residues. At the top of stem II, U+2 and C+3 were in base-pair competition with A7. Meanwhile the other unpaired bases in the loop were either stacked or fl ipped out of the helix. Further inspection suggested grouping into three major popul ations, namely AA/CA, AA/UA and GA/UA reflecting the combination of two ma jor conformations each in stem I (AA and GA) and stem II (UA and CA). AA and GA were disti nguished by A-1:A10 (AA) vs. G+1:A10 (GA) base-pairing, whereas CA and UA were distinguished by C+3:A7 (CA) vs. U+2:A7 (UA) base- pairs. The schematic representations in Figure 2-6 highlight key variations in structural features of these populations. AA/CA (46%) was the dom inant conformer while GA/UA (22%), AA/UA (24%) and GA/CA (4%) were minor conformers ( Figure 2-6). We further characterized additional variations of these conformers within the conformers outlined above. Using Markov state model analysis ( 60, 61) we developed a network of kinetic states (macrostates) connected by transitions between and w ithin other conformers (see Figure 2-7). Six kinetically different 59macrostates (a-f) were determined from the AA/ UA ensemble simulation trajectory. The relative populations and energies of the kinetic states (macrostates) with resp ect to their primary conformations are detailed in Table 2-4. The dominant AA/UA macrostate (a, 77%) had a flipped out C+3 and G+1:A9 base-pair which undergoes kinetic transitions involving various orientations of C+3 flipping, til ting of U+2:A7 base-pair, moveme nt of C+3 into the helix and breaking of G+1:A9 base-pair towards the formati on of kinetically different macrostates. Within the AA/CA conformer, seven macrostates were id entified. The dominant kinetic state within AA/CA (53%) featured flipped out U+2 and A9 residues. Othe r macrostates had A9 either flipped out or base-paired to G+ 1, different orientations of U+ 2 flipping, stacked U+2 and C+3 flipping. Various C+3 flipping orie ntations and base stacking tr ansitions were observed within the GA/UA conformer in the observed macrosta tes. Finally, the dominant GA/UA macrostate (64%) exhibited a flipped out C+3. A unique G+1: A9 base-pair was observe d in one macrostate albeit in low (1%) population. Structural transitions among kine tic states (macrostates) and conformers were facilitated by dynamics of various key residues in the active s ite of loop A. Most notably U+2, C+3 and A9 adopted different conformations inside or outside the helix involving base-pairing, base-flipping and stacking. The kinetic transitions between the various conformations were facilitated by torsional dynamics involving particular transitions in the backbone torsion angle accompanied by transitions in the sugar pucker phase angles of U+2, C+3 and A9. Base-flipping seems to be correlated with a C2'/C3'-endo mixed sugar pucker transition in these dynamic residues. Quantitative correlation analysis between base and sugar pucker dynamics in selected loop A residues has been shown (Table 2-3). Residues A9, U+2 and C+3 reported a 1:1 C2'/C3'-endo mixed sugar pucker to base-flipping ratio while G+1 reported a 2:1 ratio respectively. FiArestradigure 2-5: SA simulation epresentatio nructures for dopts the nom Schematic r e structures.ns with 3D rethe six in depmenclature uepresentatio. The numb eenderings. Tpendent loo pused in Figu60ons (A) and ering in A anThe structurep A simulat iure 2-2. structural nd B correlates shown areions. The schrenderings te schemati ce representatihematic rep r(B) of loop c structural ive resentation 61Table 2-4: Characterization of the kinetic states observ ed during loop A simulation The kinetic transitions have been highlighted with AA/UA, GA/UA and AA/CA conformers using Markov state model analysis. The structures used in calculating RMSD are representative structures within each macrostate (a-u) population. Conformer Macrostate % Macrostate population within conformer RMSD (Å) relative to highest population G (kcal/mol) AA/UA a 77.2 0.0 0.0 b 4.6 3.2 1.7 c 12.7 3.2 1.1 d 2.3 3.2 2.1 e 2.0 4.0 2.2 f 1.0 2.8 2.5 GA/UA g 64.3 0.0 0.0 h 3.5 1.4 1.7 i 3.8 1.9 1.7 j 17.9 1.6 0.8 k 0.5 2.8 2.9 l 7.7 1.9 1.3 m 1.3 2.1 2.3 n 1.0 3.0 2.5 AA/CA o 52.9 0.0 0.0 p 1.4 1.9 2.2 q 40.7 0.7 0.2 r 2.8 0.9 1.7 s 1.0 2.2 2.6 t 1.0 2.6 2.3 u 0.5 2.6 2.8 62Table 2-5: Residue dynamics for selected loop residues during hpA simulations Residues % C2™/C3™-endo sugar pucker % Base flipped out A7 16.7 0.0 G8 16.7 0.0 A9 33.3 40.0 G+1 16.7 8.33 U+2 50.0 40.0 C+3 33.3 30.0 aSugar pucker and base flipped-out percentages for each residue were calculated as a fraction of observed mixed (C2™/C3™) pucker an d complete flip-out with resp ect to all the six hpA unbiased simulations. A cutoff percentage 25% was considered as evidence of dynamics within the residue. FbaataFigure 2-6:based on l oanalysis of aand B reprethe major c and helically Schemati coop base-p aall simulati oesent obser vonformatio ny-stacked A 9c represent aairing and ons by k-m eed conform an which de p9 in the maj o63ation of lo ostacking. Teans cluster iations AA/ Upicts equili bor conforme rop A conforThese struct uing based onUA and GAbrium betw er AA/CA. rmational cures were o n heavy-ato mA/UA while een flippe d-classificatio nbtained aftem RMSD. AC represent-out A9 bas n er A ts e FigumaccirclGA/relatmac rdepibetwure 2-7: Co ncrostates (la les represen t/UA respectitive populatirostates is recting the tr aween confor m nformationabeled a-u) dt the observ eively, while iion within eaepresented b ansition frequmational pairal transitio nderived fromed major an dinset circles ach conformy solid line, uency. The drs. 64n of loop A Rm Markov sd minor con frepresent k imer. The flux with the intdotted line r eRNA capturstate model iformers AA /inetic macr obetween indensity of ea cepresents tr aring top 18 ing. The larg/CA, AA/UAostates sized dividual ch solid line ansition poinge A and by nts 65The base-flipping with mixed puc ker correlation has been obser ved in DNA in the methylation of target cytosine by the bacterial DNA cytosine methyltransferase M.Hha ( 64). A9 adopts a broad range of conformations including base-pai ring to G+1 to intrahelical base-stacking whereas C+3 samples various extrahelical orient ations and U+2 assumes various extrahelical orientations. C+3 is mostly base -paired but can transition to extr ahelical base-flipping with the base adopting various extrahelical orientations. (Figure 2-8 , Table 2-4 ). 2.3.3 Conformational transitions between major states The MD simulations, when started from di fferent random seeds, reached different conformational states as deta iled above and illustrated in the Markov state model ( Figure 2-7). While minor transitions were observed in some of the simulations such as G+1:A10 to G+1:A9 in stem I, transitions between the major states were not obs erved, presumably because of significant kinetic barriers. In order to sample the major transitions we performed targeted MD simulation (TMD) from AA/UA to GA/UA an d AA/CA conformers respectively ( Figure 2-8) in two separate TMD simulations. In one TMD simulation (AA/UA to GA/UA), there was a transition of the A-1:A10 base-p air to G+1:A10 due to the reor ientation of A10 residue. This rearrangement led to transitional stacking of A9 between G+1 and G10. A9 moved out of position giving way for G+1 to base-pair with A10. The G+1:A10 base-pair formed is subsequently stabilized in the last 0.2 ns of the simulation ( Figure 2-8). Similarly, in another TMD simulation transition (AA/UA to AA/CA), there was a loss of U+2:A7 base-pair due to extra-helical movement of U+2. Meanwhile, the previously extrahelical C+3 moved into the helix to form a non-canonical base-pair with A7 that was previously base-paired to U+2. These U+2 and C+3 trans-directional dynamics pr opagated the formation of a new non-canonical C+3:A7 base pair observed in the AA/CA conformer ( Figure 2-8). This conformational 66transition was completed within 60% of the TMD simulation with the new C+3:A7 base-pair formed and stabilized in the remaining 40% of our simulations. Sim ilarly to the kinetic transitions, A9, U+2 and C+3 residues displayed dynamic behavior that drove the conformational transitions and subsequently conformational sampling in loop A. We calculated the energies for each confor mational ensemble and the barrier height between the two sets of conformational transitio ns in TMD simulations via Molecular Mechanics Poisson-Boltzmann Surface Area (MMPB/SA). Th e calculated conformational energies, for AA/UA and AA/CA were -2007 and -2003 kcal/mol, respectively (see Table 2-6) with the difference within the 2-4 kcal/mol error of the MMPB/SA method. The U+2:A10 to C+3:A10 base-pair transition was estimated to have an energy barrier of ~13 kcal/mol ( Figure 2-8A). The GA/UA conformer, however, reported a higher en ergy (-1990 kcal/mol) compared to AA/UA and AA/CA conformers consistent with th e low relative population (21.5%) observed during explicit solvent simulation. The A- 1:A10 to G+1:A10 base-pair barr ier height was estimated as ~30 kcal/mol ( Figure 2-8B) indicating that this transition is limited by a large kinetic barrier between AA/UA and GA/UA. This high energy barr ier for the formation of GA/UA conformer supports the dominance of A-1:A 10 over G+1:A10 base-pair in st em I effectively favoring the lower energy AA/UA and AA/CA conformations. The stem II base-pai r transition between U+2:A7 to C+3:A7 base-pair was readily ach ieved by U+2 and C+3 dynamics, especially by their ability to flip in and out of the helix to facilitate base-pair formation. These dynamics are likely responsible for the lower energetic barrier between the AA/UA and AA/CA conformations. 67Table 2-6: Conformational energies of loop A conformers calculated with MMPB/SA Loop A conformer Signature ba se pair(s) Energy (kcal/mol) AA/UA A-1:A10 and U+2:A7 -2007.1 ± 0.7 GA/UA G+1:A10 and U+2:A7 -1990.3 ± 0.6 AA/CA A-1:A10 and C+3:A7 -2002.6 ± 0.6 aConformational energy values for loop A conformers calculated by the Molecular Mechanics Poisson-Boltzmann Surface Area (MMPB/SA) for AA/UA, GA/UA and AA/CA conformers. The reported energies are averages for the clus ter ensembles representing each conformer. The statistical error was estimated based on th e standard error between block averages. Figutransmaj ofigurre 2-8: T asitions fro mr and mino res show the argeted M om AA/UA to r conformat iprogress of olecular D y GA/UA a nions observ ebase-pair lo 68ynamics ( Tnd AA/CA. Ted in simul ass and form aTMD) sim uThe target c ations of lo oation. ulation of conformationop A. Top aconformer ns represent and bottom Figurconfofree eensemGA/Uenergre 2-9: Fre ormations Aenergies wembles deriv eUA (A) is c hgy barrier bet e energy dAA/UA to G Are estimate ded from T Mharacterized btween AA/Uiagrams d eA/UA (A) a nd with the MMD simulati oby a higher UA to AA/C A69epicting th end AA/UA tMMPB/SA ons. The struenergy barriA (~13 kcal/e ener gy prto AA/CA ( Bmethod ba suctural tran sier (~30 kc a/mol) confo rrofile of T MB). The con fsed on con fsition from al/mol) com prmations (B )MD-driven formational formational AA/UA to pared to the ). 702.4 DISCUSSION The conformational dynamics of RNA are key for understanding its f unctional properties. Specifically, RNA internal loops are common motif s that constitute th e active site of many functional RNAs including ribozymes. The understa nding of RNA internal loop base-pair and stacking interactions and their conformational varia tions are therefore key f eatures in predicting RNA interaction and function. In this work, we have characterized the conformational variations of the hairpin ribozyme loop A domain containing an internal loop that is essential in forming RNA-RNA tertiary interactions with subseque nt backbone cleavage. Using unbiased explicit solvent simulation we identified a surprisingly complex conformational energy landscape with three major conformations (conformers), disti nguished by alternate base-pairing and stacking interactions within the internal loop of loop A, as well as nume rous minor kinetic states within each of the major conformers. Transitions betw een different states primarily involved the dynamics of conserved bases U+2, C+3 and A9 through backbone torsional dynamics. We find significant kinetic barriers, especially between the three major conformers that reinforce the idea of an RNA energy landscape that consists of many deep local minima separated by kinetic barriers and complex transition pathway networks ( 65, 66). For example, the high kinetic barrier between AA/UA and GA/ UA conformations ( Figure 2-8A) can be overcome by sampling several states that ultimately permit popul ation of the minor GA/UA conformation ( Figure 2-6), among other conformers. Similar mechanism may be at play in the tran sition between AA/UA to AA/CA. The kinetic states therefore provide a pathway for accessing loop A shallow and deep conformational wells that stabilize the observe d conformations. The conformational transitions implicated several active site residues in cluding U+2 and C+3. The dynamics of U+2 has previously been reported by Cai and Tinoco™s ob servation of broad U+2 NMR cross-peaks in 71NOESY and COSY experiments performed in loop A ( 22) implying that this residue indeed undergoes conformational exchange. Our results indi cate that this exchange is predominantly base-flipping facilitated by puc ker exchange. Similar NMR studi es have shown that both U+2 and C+3 sugar puckers reported mixed C2'/C3'-e ndo within the NMR structure of isolated loop A further supporting the conformational exchange of these two residues. 2.4.1 Sampling of docked loop A structure A key question is how the conformational samp ling of loop A in isolation relates to the conformation that is adopted when docked to loop B. Multi-conformational sampling of loop A results in a conformation (AA/CA) with structural similarity to the docked loop A, capable of docking with an activated loop B. By assessing th e specific conformationa l changes inherent in the docked loop A crystal structure, we tabulated some of these conformational similarities with the docked structure with th eir relative population ( Table 2-7). G+1 sampled extrahelical conformers approximately 8% of the time, mostly in our 2'- O-methylated loop A simulation, suggesting that the A-1 2'- O-methylation (also in the docked form crystal structure) may slightly stabilize the G+1 extrahelical st acking. Within the stru ctures of isolated loop A and docked loop A, the base-pair and base stacking interactions reported significant variations. For example, G+1 was base-paired with A9 in the isolated loop A, while it was extra-helically base-paired with C25 in loop B ( Figure 2-2) in the docked form. Although C+3 stacked between A7 and G8 in isolated loop A, the same residue was base-paired with A7 in the docked conformation. The dynamics of residues U+2 and C+3 in our simulations play a significant role in the formation of C+3:A7 base-pair in the AA/CA conformer, also observed in the docked structure ( 17). RNA dynamics and C+3:A7 base-pair formation suggests pre-organization from an inactive to active loop A RNA during our simulations, in which some features of the docked form are sampled. By 72structurally comparing loop A™s active site betw een our observed conformers with docked form of loop A ( Figure 2-10), we determined the relative structural similarity of the AA/CA conformer and the crystallographically-observed docked conformation of loop A with an RMSD of 2.6 Å. This conformational sampling was facil itated by reorientation of specific active site residues and RNA backbone towards the structure of docked loop A besides the formation of a unique C+3:A7 base-pair intera ction also observed in the doc ked conformation but not in isolated loop A. This is consistent with the free energy landscape of loop A RNA allowing sampling of an activated conformer resembling the docked structure. This may suggest that conformational select ion at least plays a role in activating loop A. However, other conformers (AA/UA and GA/UA) are not structurally simila r with docked loop with the heavy atom loop RMSDs of 4.2 Å and 3.2 Å respectively as referenced to the docked structure. Although the AA/CA conformer is closer to the docked form than are the other major conformers, not all base-pair and stacking configurat ions match the docked crys tal structure. It is possible that the docked structur e is visited during longer (millisecond) timescales which are not reached in our simulations. However, it is more likely that the presence of loop B is necessary to fully form the docked structure. Th is is supported by the formation of several te rtiary interactions between loop A and loop B, mostly stabilized by loop B. In the crystal structure of docked loop A, G+1 is extra-helical while ba se-pairing with C25 in loop B ( 17). In our simulations we see dynamics of G+1 with a minor G+1:A10 base -pair. This suggests a hybrid interaction mechanism for the docking of loop A to loop B where initial confor mational selection is followed by an induced fit mechanism. A simila r mechanism has recently been observed in U1A-RNA molecular recognition where Helix-C is structurally reoriented to allow RNA access in the absence of RNA ( 67). Figof cloopAAC). crysrepoloopAAreprgure 2-10: Lconformatio np A crystal sA/CA (blue) aThe critical stal structur eorted RMSDp A. CorrespA/UA (D), AAresentative sLoop A sampnal capture mstructure (cyand GA/UA C+3:A7 ba se is similarl yD values ind iponding heavA/CA (E) ansimulations. pling dockedmechanism. yan) with si m(orange) resse-pair (box ey seen in A Aicate closer Avy atom R Mnd GA/UA (73d conformaStructural comulation conspectively wed red) obseA/CA but notAA/CA strucMSD profilesF) referenc eation provideomparison bnformers A Awithin the ac terved in the dt AA/UA anctural samp ls of active si ted to dockedes initial evi between doc kA/UA (green)tive site (A, docked loop nd GA/UA .Tling for dockte residues fd loop A in thdence ked ), B, and A The ked for hree 74Most docking-competent interactions were samp led during our simulations albeit in various populations ( Table 2-7). However, we did not observe conf ormational similarity of residue A-1 between our simulation structures and the docke d loop A structure. This active site residue constitutes the cleavage site which may undergo r earrangements similar to the docked form in only the presence of loop B or at longer times cales. Intra-domain and inter-domain tertiary interactions jointly prepare loop A for docking to its partner loop B. The role of the GNRA tetraloop in stabiliz ing RNA provided a good control for assessing dynamics in loop A. It is worth noting that the GAAA tetraloop incorporated in our RNA remained quite stable over the simulation time consistent with the NMR structure ( 68) and other simulations done at similar temperatures ( 69, 70 ). This suggests that our observations were not biased by force field artifacts but that they provide a realistic behavior of this independently folded RNA. Previous studies of RNA indicat e the formation of AA and GA sheared-type non- canonical base-pairing within the RNA loop ( 71-74) suggesting a possible alternate base-pair transition. This observation is consistent with alternate A-1:A10 and G+1:A10 observed in our conformers. An earlier simulation of the docked loop A ( 75) showed limited dynamics and structural conservation with the crystal structure, suggesting that flexibility of loop A in our simulation is necessary for its pre-organization to the docked form. The observation of multiple conformations and internal loop dynamics lays down a framework for understanding RNA dynamics and especially in pre-organizing R NA for activation within a complicated rugged energy landscape. 75Table 2-7: Percentage population that sample specific interactio ns observed in the docked form of loop A during isolated loop A simulation. The cutoff for base-pair formation was set at 3.5Å Residue(s) Conformation % Population Base-pair A-1:A9 Base-pair 0.0 U+2:G8 Base-pair 2.3 C+3:A7 Base-pair 51.2 Stacking G+1 Extra-helical 8.3 Ribose pucker A-1 C2'-endo 0.0 U+2 C3'-endo 41.7 C+3 C3'-endo 75.0 A7 C3'-endo 83.2 G8 C3'-endo 41.7 762.4.2 Functional relevance In RNA, dynamics is important to facil itate the diverse stru ctural rearrangements associated with transitioning kinetic barriers to functionally competent structures. To accomplish these transitions, thermal and conformational activation provi des the energy to drive RNA transitions to their dynamic functional roles. For example, the internal motions leading to the melting of base-pairs near the internal loop of HIV-1 stem l oop 1 have been observed using NMR similar to a secondary structural tran sition that occurs during viral maturation ( 76). Moreover, cross-linking experiments have shown the formation of multiple functional folds of the hairpin ribozyme under different metal ion conditions implicating U+2 and C+3 residues (77). This suggests that under physiological c onditions, solution conditions and other factors facilitate functionally relevant RNA transitions. The assessment of loop A provides the initial step to understanding tertiary structure formation and RNA-RNA interaction mechanism and dynamics in RNA. An NMR st udy of loop B by the Feigon group reported multiple mutually- inconsistent interproton NOEs, which is also consistent with the existence multiple conformations ( 21). Some of the NOEs were eventually suggested to be consistent with the docked state. These experimental observa tions also provide support for a possible conformational sampling in loop B, the docking pa rtner of loop A. Taken together, the dynamics in loop A and subsequent formation of multiple ki netic substates and transitions present initial evidence for conformational sampling as a way of exploring the rugged free energy landscape. This process results in activated molecules capable of RNA-RNA interactions to effect functional output. Loop A domain of the hairpin ribozyme provides us with an excellent model for understanding the detailed m echanism for this process. 772.5 CONCLUSION Unbiased MD simulation was used to dete rmine conformational heterogeneity in RNA based on alternate base-pairs within a subset of residues in the loop regi on of domain A of the hairpin ribozyme. The conformers were determin ed using non-canonical ba se-pair combinations in stems I and II. The Markov state model anal ysis of our simulations determined several macrostates that undergo kinetic transitions w ithin different local minima based on subtle structural changes and dynamics within a set of residues. Base and backbone dynamics also play an important role in alternate base-pair formation and subseque ntly conformational sampling in loop A. This suggests that c onformational sampling and transi tion in loop A RNA is a key strategy to avoiding the kinetic traps that loca lize RNA in non-functional conformations. Kinetic transitions are therefore achieved through sampling a series of kine tic states which subsequently prepare the RNA to overcome the kinetic energy barriers to form functi onal (active) RNA. The most populated conformer, AA/CA , closely sampled conformationa l properties similar to the activated (docked) loop A conformation. This is significant because it highlights the role of conformational sampling in activat ing loop A (conformational selectio n) for the critical tertiary RNA-RNA interaction. This unique intrinsic base-pair rearrangement within a subset of kinetic states and transitions to conf ormations along the potential ener gy surface support the rugged but accessible free energy landscape of RNA with mechanis tic properties that direct pre-organization to the activated state. The multiple conformers and inter-conformer transitions observed lay a foundation for understanding tertiary RNA interactions in the hairpin ribozyme, as well as in functional RNA systems more generally. 78REFERENCES 79REFERENCES 1. Al-Hashimi, H. M. (2005) Dynamics-based amplification of RNA function and its characterization by using NMR spectroscopy, Chembiochem 6, 1506-1519. 2. Kern, D., and Zuiderweg, E. R. (2003) The role of dynamics in allosteric regulation, Curr Opin Struct Biol 13, 748-757. 3. Perez-Canadillas, J. M., and Varani, G. (2001) Recent advances in RNA-protein recognition, Curr Opin Struct Biol 11, 53-58. 4. Wolf-Watz, M., Thai, V., Henzler-Wildman, K ., Hadjipavlou, G., Eisenmesser, E. Z., and Kern, D. (2004) Linkage between dynamics and catalysis in a thermophilic-mesophilic enzyme pair, Nat Struct Mol Biol 11, 945-949. 5. Julien, K. R., Sumita, M., Chen, P. H., Lair d-Offringa, I. A., and Hoogstraten, C. G. (2008) Conformationally restricted nucleo tides as a probe of structure-function relationships in RNA, RNA 14, 1632-1643. 6. Zhuang, X. W., Kim, H., Pereira, M. J. B ., Babcock, H. P., Walter, N. G., and Chu, S. (2002) Correlating structural dynamics and function in single ribozyme molecules, Science 296, 1473-1476. 7. Fedor, M. J. (2000) Structure and function of the hairpin ribozyme, J Mol Biol 297, 269-291. 8. Ferre-D'amare A, R., and Rupert, P. B. (2002) The hairpin ribozyme: from crystal structure to function, Biochem Soc Trans 30, 1105-1109. 9. Dethoff, E. A., Hansen, A. L., Musselman, C., Watt, E. D., Andricioaei, I., and Al-Hashimi, H. M. (2008) Characterizing comple x dynamics in the tran sactivation response element apical loop and motional correla tions with the bulge by NMR, molecular dynamics, and mutagenesis, Biophys J 95, 3906-3915. 10. Frank, A. T., Stelzer, A. C., Al-Hashimi, H. M., and Andricioaei, I. (2009) Constructing RNA dynamical ensembles by combining MD and motionally decoupled NMR RDCs: new insights into RNA dynamics and adaptive ligand recognition, Nucleic Acids Res 37, 3670-3679. 11. Sosnick, T. R., and Pan, T. (2003) R NA folding: models and perspectives, Curr Opin Struct Biol 13, 309-316. 8012. Pljevaljcic, G., Klostermeier, D., and Millar, D. P. (2005) Th e tertiary structure of the hairpin ribozyme is formed through a slow conformational search, Biochemistry 44, 4870-4876. 13. Russell, R., Das, R., Suh, H., Traver, K. J., Laederach, A., Engelhardt, M. A., and Herschlag, D. (2006) The paradoxical behavior of a highly structured misfolded intermediate in RNA folding, J Mol Biol 363, 531-544. 14. Buzayan, J. M., Gerlach, W. L., Bruening, G., Keese, P., and Gould, A. R. (1986) Nucleotide sequence of satellite tobacco ri ngspot virus RNA and its relationship to multimeric forms, Virology 151, 186-199. 15. Walter, N. G., and Burke, J. M. (1998) Th e hairpin ribozyme: structure, assembly and catalysis, Curr Opin Chem Biol 2, 24-30. 16. Hampel, A., and Tritz, R. (1989) RNA catalytic properties of the minimum (-)sTRSV sequence, Biochemistry 28, 4929-4933. 17. Rupert, P. B., and Ferre-D'Amare, A. R. ( 2001) Crystal structure of a hairpin ribozyme-inhibitor complex with implications for catalysis, Nature 410, 780-786. 18. Pinard, R., Lambert, D., Walter, N. G., Heckman, J. E., Major, F., and Burke, J. M. (1999) Structural basis for the guanosine requirement of the hairpin ribozyme, Biochemistry 38, 16035-16039. 19. Ojwang, J. O., Hampel, A., Looney, D. J., Wong-Staal, F., and Rappaport, J. (1992) Inhibition of human immunodeficiency virus type 1 expression by a hairpin ribozyme, P Natl Acad Sci USA 89, 10802-10806. 20. Yu, M., Poeschla, E., Yamada, O., Degrandis, P., Leavitt, M. C., Heusch, M., Yees, J. K., Wong-Staal, F., and Hampel, A. (1995) In vitro and in vivo characterization of a second functional hairpin ribozyme against HIV-1, Virology 206, 381-386. 21. Butcher, S. E., Allain, F. H., and Feigon, J. (1999) Solution structure of the loop B domain from the hairpin ribozyme, Nat Struct Biol 6, 212-216. 22. Cai, Z., and Tinoco, I., Jr. (1996) Solution st ructure of loop A from the hairpin ribozyme from tobacco ringspot virus satellite, Biochemistry 35, 6026-6036. 23. Rupert, P. B., Massey, A. P., Sigurdsson, S. T., and Ferre-D'Amare, A. R. (2002) Transition state stabilizat ion by a catalytic RNA, Science 298, 1421-1424. 24. Alam, S., Grum-Tokars, V., Krucinska, J., Kundracik, M. L., and Wedekind, J. E. (2005) Conformational heterogeneity at position U 37 of an all-RNA hairpin ribozyme with implications for metal binding and the catalytic structure of the S-turn, Biochemistry 44, 14396-14408. 8125. Salter, J., Krucinska, J., Alam, S., Grum-T okars, V., and Wedekind, J. E. (2006) Water in the active site of an all-RNA hairpin ribozym e and effects of Gua8 base variants on the geometry of phosphoryl transfer, Biochemistry 45, 686-700. 26. Torelli, A. T., Spitale, R. C., Krucinska, J., and Wedekind, J. E. (2008) Shared traits on the reaction coordinates of ribonuclease and an RNA enzyme, Biochem Bioph Res Co 371, 154-158. 27. Torelli, A. T., Krucinska, J., and Wedekind, J. E. (2007) A comparison of vanadate to a 2 '-5 ' linkage at the active site of a small ribozyme suggests a role for water in transition- state stabilization, RNA-A Publication of the RNA Society 13, 1052-1070. 28. Liberman, J. A., Guo, M., Jenkins, J. L., Kr ucinska, J., Chen, Y. Y., Carey, P. R., and Wedekind, J. E. (2012) A Transition-State Interaction Shifts Nucleobase Ionization toward Neutrality To Facilitate Small Ribozyme Catalysis, J Am Chem Soc 134, 16933-16936. 29. Hampel, K. J., and Burke, J. M. (2001) A conformational change in the "loop E-like" motif of the hairpin ribozyme is coincident al with domain docking and is essential for catalysis, Biochemistry 40, 3723-3729. 30. Komatsu, Y., Kumagai, I., and Ohtsuka, E. ( 1999) Investigation of the recognition of an important uridine in an internal loop of a hairpin ribozyme prepared using post- synthetically modifi ed oligonucleotides, Nucleic Acids Res 27, 4314-4323. 31. Walter, N. G., Chan, P. A., Hampel, K. J., Millar, D. P., and Burke, J. M. (2001) A base change in the catalytic core of the hairpin ribozyme perturbs function but not domain docking, Biochemistry 40, 2580-2587. 32. Sumita, M., White, N. A., Julien, K. R., a nd Hoogstraten, C. G. (2013) Intermolecular domain docking in the hairpin ribozyme Metal dependence, binding kinetics and catalysis, RNA Biol 10, 425-435. 33. Clore, G. M., and Kuszewski, J. (2003) Improving the accuracy of NMR structures of RNA by means of conformational database po tentials of mean force as assessed by complete dipolar coupling cross-validation, J Am Chem Soc 125, 1518-1525. 34. Feig, M., Karanicolas, J., and Brooks, C. L. (2004) MMTSB Tool Set: enhanced sampling and multiscale modeling methods fo r applications in structural biology, J Mol Graph Model 22, 377-395. 35. Jorgensen, W. L. (1981) Quantum and Sta tistical Mechanical Studies of Liquids .10. Transferable Intermolecular Potential Func tions for Water, Alcohols, and Ethers - Application to Liquid Water, J Am Chem Soc 103, 335-340. 8236. Jorgensen, W. L., Chandrasekhar, J., Madur a, J. D., Impey, R. W., and Klein, M. L. (1983) Comparison of Simple Potential Functions for Simulating Liquid Water, J Chem Phys 79, 926-935. 37. Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., and Pedersen, L. G. (1995) A Smooth Particle Mesh Ewald Method, J Chem Phys 103, 8577-8593. 38. Darden, T., York, D., and Pedersen, L. ( 1993) Particle Mesh Ewald - an N.Log(N) Method for Ewald Sums in Large Systems, J Chem Phys 98, 10089-10092. 39. Foloppe, N., and MacKerell, A. D. (2000) All-atom empirical fo rce field for nucleic acids: I. Parameter optimization based on small molecule and condensed phase macromolecular target data, J Comput Chem 21, 86-104. 40. MacKerell, A. D., and Banavali, N. K. (2000) All-atom empirical force field for nucleic acids: II. Application to molecular dynamics simulations of DNA and RNA in solution, J Comput Chem 21, 105-120. 41. Best, R. B., Zhu, X., Shim, J., Lopes, P. E., Mittal, J., Feig, M., and Mackerell, A. D., Jr. (2012) Optimization of the additive CHARMM all-atom protein force field targeting improved sampling of the backbone phi, psi and side-chain chi(1) and chi(2) dihedral angles, J Chem Theory Comput 8, 3257-3273. 42. MacKerell, A. D., Bashford, D., Bellott, M ., Dunbrack, R. L., Evanseck, J. D., Field, M. J., Fischer, S., Gao, J., Guo, H., Ha, S., Jo seph-McCarthy, D., Kuchnir, L., Kuczera, K., Lau, F. T. K., Mattos, C., Michnick, S., Ngo, T., Nguyen, D. T., Prodhom, B., Reiher, W. E., Roux, B., Schlenkrich, M., Smith, J. C ., Stote, R., Straub, J., Watanabe, M., Wiorkiewicz-Kuczera, J., Yin, D., and Karplus, M. (1998) All-atom empirical potential for molecular modeling and dyna mics studies of proteins, J Phys Chem B 102, 3586-3616. 43. Huang, J., and MacKerell, A. D. (2013) CHARMM36 all-atom additive protein force field: Validation based on comparison to NMR data, J Comput Chem 34, 2135-2145. 44. Phillips, J. C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, E., Chipot, C., Skeel, R. D., Kale, L., and Schulten, K. (2005) Scalable molecular dynamics with NAMD, J Comput Chem 26, 1781-1802. 45. Adelman, S. A., and Doll, J. D. (1976) Generalized Langevin equation approach for atom/solid-surface scattering: General formulation for cla ssical scattering off harmonic solids, J Chem Phys 64, 2375-2388. 46. Ryckaert, J. P., Ciccotti, G., and Berendsen, H. J. C. (1977) Numeri cal integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n- alkanes, J Comput Phys 23, 327-341. 8347. Cornell, W. D., Cieplak, P., Bayly, C. I., Gould, I. R., Merz, K. M., Ferguson, D. M., Spellmeyer, D. C., Fox, T., Caldwell, J. W., and Kollman, P. A. (1996) A second generation force field for the simulation of pr oteins, nucleic acids, and organic molecules (vol 117, pg 5179, 1995), J Am Chem Soc 118, 2309-2309. 48. Zgarbová, M., Otyepka, M., −poner, J., Mlád ek, A., Banáı, P., Cheatham, T. E., and Jureka, P. (2011) Refinement of the Cornell et al. Nucleic Acids Force Field Based on Reference Quantum Chemical Calculati ons of Glycosidic Torsion Profiles, J Chem Theory Comput 7, 2886Œ2902. 49. Schlitter, J., Engels, M., Kruger, P., Jacoby, E., and Wollmer, A. (1993) Targeted Molecular-Dynamics Simulation of Conforma tional Change - Application to the T R Transition in Insulin, Mol Simulat 10, 291-308. 50. Floris, F., and Tomasi, J. (1989) Evalua tion of the Dispersion Contribution to the Solvation Energy - a Simple Computationa l Model in the Continuum Approximation, J Comput Chem 10, 616-627. 51. Gallicchio, E., Kubo, M. M., and Levy, R. M. (2000) Enthalpy-entropy and cavity decomposition of alkane hydration free energi es: Numerical results and implications for theories of hydrophobic solvation, J Phys Chem B 104, 6271-6285. 52. Honig, B., and Nicholls, A. (1995) Classical Electrostatics in Biology and Chemistry, Science 268, 1144-1149. 53. Kollman, P. A., Massova, I., Reyes, C., K uhn, B., Huo, S. H., Chong, L., Lee, M., Lee, T., Duan, Y., Wang, W., Donini , O., Cieplak, P., Srinivasan, J., Case, D. A., and Cheatham, T. E. (2000) Calculating structures and free energies of complex molecules: Combining molecular mechanics and continuum models, Accounts Chem Res 33, 889-897. 54. Lu, Q., and Luo, R. (2003) A Poisson- Boltzmann dynamics method with nonperiodic boundary condition, J Chem Phys 119, 11035-11047. 55. Luo, R., David, L., and Gilson, M. K. (2002) Accelerated Poisson-Boltzmann calculations for static and dynamic systems, J Comput Chem 23, 1244-1253. 56. Kar, P., and Knecht, V. (2012) Mutati on-Induced Loop Opening and Energetics for Binding of Tamiflu to Influenza N8 Neuraminidase, J Phys Chem B 116, 6137-6149. 57. Kar, P., and Knecht, V. (2012) Origin of Decrease in Potency of Darunavir and Two Related Antiviral Inhibitors against HIV-2 Compared to HIV-1 Protease, J Phys Chem B 116, 2605-2614. 58. Kar, P., and Knecht, V. (2012) Energetic basis for drug resistance of HIV-1 protease mutants against amprenavir, Journal of computer-aided molecular design 26, 215-232. 8459. Kar, P., Lipowsky, R., and Knecht, V. (2013) Importance of Polar Solvation and Configurational Entropy for Design of Antiretroviral Drugs Targeting HIV-1 Protease, J Phys Chem B 117, 5793-5805. 60. Bowman, G. R., Huang, X. H., and Pande, V. S. (2009) Using generalized ensemble simulations and Markov state models to identify conformational states, Methods 49, 197-201. 61. Beauchamp, K. A., Bowman, G. R., Lane, T. J., Maibaum, L., Haque, I. S., and Pande, V. S. (2011) MSMBuilder2: Modeling Conformational Dynamics on the Picosecond to Millisecond Scale, J Chem Theory Comput 7, 3412-3419. 62. Schrödinger, L., (Ed.) (2010) The PyMOL Molecular Graphics System, Version 1.3. Schrödinger, LLC, New York, New York. 63. Humphrey, W., Dalke, A., and Schulten, K. (1996) VMD: Visual molecular dynamics, J Mol Graph Model 14, 33-38. 64. Shieh, F. K., Youngblood, B., and Reich, N. O. (2006) The role of Arg165 towards base flipping, base stabilization and catalysis in M.HhaI, J Mol Biol 362, 516-527. 65. Bailor, M. H., Mustoe, A. M., Brooks, C. L., 3rd, and Al-Hashimi, H. M. (2011) Topological constraints: using RNA seconda ry structure to model 3D conformation, folding pathways, and dynamic adaptation, Curr Opin Struct Biol 21, 296-305. 66. Schultes, E. A., Spasic, A., Mohanty, U., a nd Bartel, D. P. (2005) Compact and ordered collapse of randomly generated RNA sequences, Nat Struct Mol Biol 12, 1130-1136. 67. Kurisaki, I., Takayanagi , M., and Nagaoka, M. (2014) Combined mechanism of conformational selection and induced fit in U1A-RNA molecular recognition, Biochemistry 53, 3646-3657. 68. Jucker, F. M., Heus, H. A., Yip, P. F., Moors, E. H. M., and Pardi, A. (1996) A network of heterogeneous hydrogen bonds in GNRA tetraloops, J Mol Biol 264, 968-980. 69. Sorin, E. J., Engelhardt, M. A., Herschlag, D., and Pande, V. S. (2002) RNA simulations: Probing hairpin unfolding and the dynamics of a GNRA tetraloop, J Mol Biol 317, 493-506. 70. Zichi, D. A. (1995) Molecular-Dynamics of RNA with the Opls Force-Field - Aqueous Simulation of a Hairpin Cont aining a Tetranucleotide Loop, J Am Chem Soc 117, 2957-2969. 71. Chen, G., Znosko, B. M., Jiao, X. Q., a nd Turner, D. H. (2004) Factors affecting thermodynamic stabilities of RNA 3x3 internal loops, Biochemistry 43, 12865-12876. 8572. Santalucia, J., Kierzek, R., and Turner, D. H. (1991) Functional-Group Substitutions as Probes of Hydrogen-Bonding between GA Mismatches in RNA Internal Loops, J Am Chem Soc 113, 4313-4322. 73. Znosko, B. M., Burkard, M. E., Schroeder, S. J., Krugh, T. R., and Turner, D. H. (2002) Sheared A(anti)center dot A(anti) base pairs in a destabilizing 2x2 internal loop: The NMR structure of 5 '(rGGCAAGCCU)(2), Biochemistry 41, 14969-14977. 74. Schroeder, S. J., and Turner, D. H. (2000) Factors affecting the thermodynamic stability of small asymmetric in ternal loops in RNA, Biochemistry 39, 9257-9274. 75. Mlynsky, V., Banas, P., Hollas, D., Reblova, K., Walter, N. G., Sponer, J., and Otyepka, M. (2010) Extensive Molecular Dynamics Si mulations Showing That Canonical G8 and Protonated A38H(+) Forms Are Most Consiste nt with Crystal Stru ctures of Hairpin Ribozyme, J Phys Chem B 114, 6642-6652. 76. Sun, X. Y., Zhang, Q., and Al-Hashimi, H. M. (2007) Resolving fast and slow motions in the internal loop containing stem-loop 1 of HIV-1 that are modulated by Mg2+ binding: role in the kissing-duplex structural transition, Nucleic Acids Res 35, 1698-1713. 77. Lambert, D., Heckman, J. E., and Burke, J. M. (2006) Cation-specific structural accommodation within a catalytic RNA, Biochemistry 45, 829-838. 86 CHAPTER 3 CONSTRUCT OPTIMIZATION, DOCK ING STUDIES AND NMR RESONANCE ASSIGNMENT OF THE LOOP A DOMAIN OF THE HAIRPIN RIBOZYME 873.1 INTRODUCTION Since globular RNAs lack a central hydrophobic core, the driving force for the interaction of multiple RNA helices and loop regions into a hi ghly ordered tertiary stru cture is still unclear in comparison to proteins. A clear understanding of the formation of RNA tertiary structure is thus vital to unravel the mechanisms behind cellular machines such as the eukaryotic spliceosome, in which several RNA structures and RNA-protein interactions form and are subsequently disrupted at specific points along the splicing pathway ( 1, 2). As discussed in Chapter 1, the hairpin ribozyme provides a m odel system for understanding RNA tertiary structure interactions correlated to function. The hairpin ribozyme™s active site constitutes two independently folding domains (loop A and loop B) ( 3). Loop A has a symmetrical internal loop interspersed between two helical stems. Docking between loop A and B pre-organizes the active site to effect catalysis (phosphodi ester bond cleavage). In nature, these two domains interact as a four-way helical junction (4WJ) ( 4) but two way junction (2WJ) a nd trans-constructs are also actively under investigation ( Figure 3-1 ). Previous i n vitro reconstitution studies have shown that a two-way helical junction (2WJ) can fo rm a functional minimal hairpin ribozyme ( 3) while independent trans-docking domains also support fu ll activity ( 5, 6). The cleavage occurs between G+1 and A-1 in loop A ( 7). Chemical cleavage by various self-cleaving RNA structural motifs is essential to the repli cation of several plant viroids ( 8), the human hepatitis virus ( 9), the mitochondrial Varkud satellite from Neurospora (10) and glmS riboswitch found in gram- positive bacteria ( 11). In the specific case of the hairpi n ribozyme, the cleavage reaction is reversible, and the ligation reaction is ~ 10 times faster than that of cleavage ( 12, 13 ). These small ribozymes are tractable systems to use in the study of the criti cal principles of RNA catalysis using molecular struct ure and functional approaches ( 14, 15). Mutational studies have 88identified residues that are critical for driving ca talysis in the hairpin ribozyme, especially in the active site ( 16-19). Burke™s group investigated bindi ng and cleavage with single base substitutions at pos itions A-1 and U+2 (19). U+2 was suggested to be involved in tertiary interactions, possibly to pre-organize the active site. The direct ro le of U+2 in catalysis, however, was not determined. Structural studies of lo op A established a conformational C2'/C3'-endo mixed sugar pucker in U+2 with a flipped out (extra-helical) base ( 20). Using time-resolved fluorescence resonance energy transfer (FRET) e xperiments to probe how U+2 mutations affect docking, cleavage rates were shown to significantly decr ease despite the fact that U+2 nucleotide substitutions did not cons iderably affect docking ( 21, 22), A uracil to cytosine mutation at position +2 (U+2C) could introduce a new Wats on-Crick base pair with G8 and limit dynamics at in this position. Such mutations may quench cer tain critical dynamics required to facilitate faster docking and catalysis. This hypothesis ha s been motivated by sequence comparisons with several residue mutations. An interesting double mutant (revertant) U+ 2C/C+3U introduced in loop A (Figure 3-2 ) completely abolishes docking under non-equilibrium, gel shift assay ( 22). U+2C/C+3U sequence stayed inactive with the native G8 in loop A, but this sequence was partially rescued by a G8U mutation. Unlike the U+2C mutant, the double mutant abrogates docking, implying a role for U+2 in docking. Th e X-ray crystal structures of the docked ribozyme confirmed the contribution of U+2 in the active site formation, since it is involved in a non-canonical base-pair interaction with G8 ( 23-25). The loop A NMR solu tion structure solved by Cai and Tinoco (20), however, contained a non-native A- 1 to C mutation previously designed to cleave HIV-1 RNA ( 26). This structure therefore may no t represent the conformation of the pre-catalytic form of loop A. FFcoFigure 3-1: Cour-way junonstructs Cartoon repnction, B) t wpresentationwo-way junct89n of hairpin tion construcribozyme act, and (c) isactive site r eolated (transegion. (A) s-) 90 To address this gap, we designed a trans-construct of native loop A sequence for our biophysical studies ( Figure 3-2 ). Studies of the trans-docking domains of hairpin ribozyme is important because of its potential to provide new insights into th e docking transition using direct measurements which assay for bimolecular equilibrium and binding kinetics using various spectroscopic techniques (27). Here, we employed spectrosc opic circular dichroism (CD) equilibrium docking assay to monitor interactio n between loop A or double mutant U+2C/C+3U loop A and loop B. This assay was developed by Dr. Minako Sumita, then a post-doctoral associate in our lab. All our trans-constructs, including loop A, loop A(U+2C/C+3U) and loop B, were designed to incorporat e a GNRA tetraloop to facilitate stem I stability and NMR assignments. Taken together, these results implic ate U+2 in aligning G8 in the active site of hairpin ribozyme to facilitate efficient cata lysis. In the U+2C/C+3U double mutant sequence, abrogated internal loop dynamics (by loop c onstriction) may limit certain key tertiary interactions between loop A and loop B in asse mbling the active site. Our CD assay tested the docking transitions between native or double mutant loop A with loop B. NMR was used to asess the RNA folding and conformation. For the pr ocess of NMR assignme nts of loop A, we evaluated several loop A constructs to determine a well-behaved loop A construct. We used in vitro transcription to incorpor ate the isotopically labeled 15N and 13C (using 15N and 13C nucleotide triphosphates) for NMR resonance as signments. We also used NMR to study loop A by first assigning the critical 1H, 13C and 15N resonances, an important step in probing loop A dynamics using the NMR relaxation measurements described in Chapter 4 of this thesis. Also, we performed several NMR experi ments (1D, 2D and 3D) to obtai n resonance assignment data for loop A. The resonance assignment data obtain ed thus far are presented here in detail. Figure (B) GAloop B box higU+2C/C3-2: Hair pAAA loop Aused in our ghlights re sC+3U doublpin riboz ymA, (C) GAAAstudies. Bluidues used e mutation wme trans-conA loop A e xue circle hig hfor stem I which intro d91nstructs usextended, (D) hlights a U1I extension duces two ad ded in our st ) loop A(U + mutation i n in C, wh iditional basetudies: (A) G+2C/C+3U) an constructs ile the red e-pairs in theGUAA loopand (E) GAA and D, g rbox highlige loop regio np A, AAA reen ghts n. 923.2 MATERIALS AND METHODS 3.2.1 RNA preparation RNA was synthesized by incorporating nucleo tides of the desired RNA constructs (see Figure 3-2) via in vitro transcription ( 28, 29) using recombinant phage T7 RNA polymerase enzyme in the presence of a synthetic DNA templa te. The synthetic template was annealed to a short complementary T7 promoter sequence. The transcription was initiated by recombinant T7 RNA polymerase following addition of DNA template, unlabeled or uniformly 15N/13C-labeled nucleotide triphosphates (ATP, UTP, CTP, and GTP) and Mg 2+. Transcription was continued for 3 hours at 37 °C then stopped by the addition of EDTA. To analyze the success of in vitro transcription, samples from the reaction cocktail were analyzed by 15% denaturing gel electrophoresis after staining using Stains All solution. Transcribed RNA was recovered by ethanol precipita tion from the transcription cocktail. The ethanol precipitate was then reconstituted in H 2O, and subjected to purification through the HiLoad 26/60 Superdex 75 column ( 30) monitored at A 260 with an isocratic flow rate of 3 mL/min sodium phosphate buffer, pH 6.40, while collecti ng fractions. Fractions co rresponding to loop A, were pooled together and syst ematically exchanged with H 2O using Amicon Ultra-4 centrifugal filter units with 3000 MWCO. R NA purity was assessed by denaturing gel and MALDI-TOF mass spectrometry. Purified RNA was then dried under vacu um and reconstituted in appropriate buffers for Circular Dichro ism (CD) and NMR studies. 933.2.2 Docking transitions between muta nt loop A and wildtype loop B Docking studies of the double mutant were pe rformed using the difference CD experiment previously developed in our lab ( 27). For docking studies, ten 5 µM samples of loop A(U+2C/C+3U) and loop B were prepared from each stock of loop A(U+2C/C+3U) and loop B docking in increasing [Co(NH 3)6]3+ concentration. The samples were dialyzed with different concentrations of [Co(NH 3)6]3+ and then left to equilibrate on the bench for at least 1 hour before taking the CD scans. The [Co(NH 3)6]3+concentrations used were 0 µM, 15 µM, 30 µM, 60 µM, 90 µM, 120 µM, 150 µM, 200 µM, 250 µM and 500 µM. Three sets of CD scans were acquired between 220-400 nm for the individual RNA cons tructs as well as loop A(C+3U/U+3C) and loop B at increasing concentrations of [Co(NH 3)6]3+ as outlined above. The scans were averaged and normalized with buffer cond itions. Docking was monitored by circular dichroism (CD) where the CD spectra of individual loop A(U+ 2C/C+3U) and loop B were acquired separately and mathematically added against the CD spec trum of both loop A(U+2C/C+3U) and loop B together in the same tube. The signal corre sponding to docking was th en calculated by the formula shown below: CD docking = (CD U+2C/C+3U loop A + loop B) - (CD U+2C/C+3U loop A + CD loop B) 3-1 The above docking experiment was repeated (by Dr. Mina Sumita) using native loop A instead of the double mutant loop A(C+3U/U+3C). All th e docking experiments we re done in triplicate and the data obtained exported and analyzed in MS excel. RNA was quantified by extinction coefficients at 260 nm accordin g to Beer-Lambert™s law (A= x b x c). The molar extinction coefficients ( ) were 214,000 M 1 cm 1 for wild-type loop A, 267,400 M 1 cm 1 for loop A(U+2C/C+3U) and 354,000 M1 cm1 for loop B (27). 943.2.3 NMR studies To analyze samples by NMR, purified l oop A samples were resuspended in 10 mM phosphate buffer, , 100 M EDTA pH 5.5 or pH 6.5 and loaded in a reduced-volume NMR tube (Shigemi) for NMR data acquisition and optimi zation of pH. For exchangeable proton NMR, loop A was constituted in a total volume of 250 L in 90% H 20 and 10% D 2O respectively. Sample concentrations were between 0.8-1.2 mM. Exchangeable 1D proton spectra were acquired with a 1-1 echo sequence (31) at various temperatures ranging from 5, 10, 15, 20 and 25 °C for GUAA loop A, GUAA loop A(U+2C /C+3C), GAAA loop A, and GAAA loop A extended constructs. NMR experiments were done on Varian Unity INOVA 600 MHz or Bruker 900 MHz. During acquisition, the proton radiofrequency (RF) carrier was placed on the HDO residual peak. All NMR experiments were ac quired using RNA biopack pulse sequences (Varian/Agilent), unless otherwise stated, with slight modifi cations after optimization. One- dimensional imino-optimized spectroscopy was acquire d using 1-1 echo pulse sequence ( 31) with 13106 complex points in the t1 dimension, a 3 s recycle dela y, and 96 steady-state scans. The first excitation maximum was targeted between imino and aromatic region for the second excitation maximum to fall at the center of th e imino region. The sweep width was set at 12,000 Hz. 1H-1H homonuclear NOE of exchangeable prot ons was acquired in Bruker 900 MHz with cryo probe using 1024 x 256 complex points in t 2 and t 1 dimensions, 1.5 s recycle delay, 300 ms mixing time with 22,500 Hz spectral width on both dimensions. 1H-15N heteronuclear single quantum coherence (HSQC) spectra was acqui red using 1024 × 256 complex points in the t2 and t1 dimensions, a 1.5 s recycle delay and 128 stea dy-state scans with co rresponding spectral widths of 12000 and 6078.76 Hz. The 15N carrier frequency was set at 118 ppm as determined from indirect referencing to internal refe rence standard 1 mM sodium 2,2 dimethyl-2- 95silapentane-5-sulfonate (DSS) ( 32). The non-exchangeable prot ons were acquired in 100% D 2O at 25 °C. Similarly, the proton RF carrier was positioned at the HDO residual peak while the 13C carrier at 110 ppm based on indirect referencing to internal reference standard DSS. Both ribose and aromatic 1H-13C HSQC and constant time (CT)-HSQ C spectra were acquired with 2048 × 256 complex points in the t 2 and t 1 dimensions, respectively, with corresponding spectral widths of 12000 and 6033 Hz, and a 2 s recycle delay. 13C carrier frequency was set at 85 ppm for ribose 1H-13C HSQC and 110 ppm for aromatic 1H,13C-HSQC, as determined fro m indirect referencing to internal reference standard DSS. In the CT-H SQC experiment, the consta nt time delay was set at 0.028 s while the number of points in the 13C dimension was set at 52. 1H-1H homonuclear 13C-edited nuclear Overhauser spectroscopy (NOESY -HSQC) sub-spectra we re acquired at 900 MHz NMR with 1024 × 256 complex points in the t 2 and t 1 dimensions, respectively, with corresponding spectral widths of 8992, 12589, and 7914 Hz, a 2 s recycl e delay, 256 steady-state scans and mixing time of 100 ms. A 3D 13C NOESY-HSQC experiment was done at 900 MHz NMR in 100% D 2O. The 1H carrier was set to 4.7 ppm, while the 13C carrier was set at 145 ppm. The sweep widths were 12590 ( 1), 7914 ( 2), and 8992.8 Hz ( 3). Real data poi nts acquired for the three dimensions were 144 in 1, 64 in 2, and 2048 in 3.The NOESY mixing time was 150 ms. 3D HCCH COSY was ac quired at 600 MHz with 1024 × 256 complex points in the t 2 and t 1 dimensions, respectively, with corresponding spectral widths of 6000 x 6000 x 9000 Hz, a 2 s recycle delay, and 256 steady-state scans. 3D HCCH TOCSY were acquired with 512 × 256 complex points with spectral widths of 2999.18, 2999.18 and 8295.31 Hz, a 1 s recycle delay, and 64 steady-state scans. 3D Adenine HCCH-T OCSY was acquired through sweep widths of 5997.9 ×1679.5 × 4524.8. The t3 dimension had 256 complex points with 30 and 32 data points in t 2 and t 1 dimensions respectively. All NMR da ta were processe d using FELIX 2002 96(Accelrys). Before Fourier transformation in the t 2 dimension, data was zero filled, a 20% DC offset was applied, and a 5 Hz exponential line broadening function was used. A cosine-squared apodization function was also applied in the indirect dimensions before spectral transformation 3.3 RESULTS 3.3.1 Docking studies Docking is a required step that precedes th e chemical steps of catalysis in the hairpin ribozyme reaction mechanism. In these docking st udies, we hypothesized that certain stabilizing modifications in loop A limit docking by stabil izing ground-state interactions or alternate structures and consequently eliminating function ally important interactions which drive docking and catalysis. To test our hypothesis, we starte d by analyzing the possible docking transitions in mutant U+2C/C+3U loop A together with loop B and then comp aring these results with the docking studies reported for wild type loop A and loop B ( 27). Docking studies were monitored by circular dichroism (CD) over a wavelength ra nge of 220-400 nm. Our data demonstrates the abolition of docking in the double muta nt loop A(U+2C/C+3U) at 250µM [Co(NH 3)6]3+. Similar non-docking observations were made in the absence of [Co(NH 3)6]3+ (data not shown). This suggests that loop A (U+2C/C+3U) does not under go structural transitions necessary for docking in the presence of loop B. This observation is also consistent with what Burke™s group has reported using their gel shift assay, albeit under slightly different metal ion conditions ( 22). The wild type loop A actively docked with loop B under similar conditi ons with docking transitions observed both at 230 nm and 270 nm respectively ( 27) (Figure 3-3 ). Analysis of docking as a function of the increasing concentration of [Co(NH 3)6]3+ between loop A (U+2C/C+3U) and loop B has been presented ( Figure 3-4). This data shows the lack of a specific interaction between loop A(U+2C/C+3U) and loop B between 0 - 250 µM [Co(NH3)6]3+ . Figurebetwe epresenHEPESe 3-3 : Comen wild t ypece of loop BS, 20 uM EDmparison bee loop A (A)B. The spe cDTA, pH7.5,etween th e) and loop Actra were o b, 250 µM [C97e circular dA(U+2C/C+btained in t hCo(NH3)6]3+. dichroism +3U) (B). D ohe followin g (CD) spec tocking was mg buffer co ntra of doc kmonitored innditions: 20 king n the mM Figure loop Ainteract i3-4: Co(NH(U+2C/C+ 3ion with incrH3)63+ titrat i3U) and lo oreasing [Co (ion studies top B. The (NH3)6]3+. 98to assay fo robserved scr metal-de pcatter indicapendent int eates non-speeraction be tecific RNA-tween -RNA 99Since the error bars are large within this range, we interpret this as random interaction between 2 RNA molecules. Conversely, metal-dependent dock ing transition between wild type loop A and loop B was demonstrated ( 27) in cobalt hexamine with an observed (Co) ½dockof ~ 50 µM. These results not only indicate that our trans-domain constructs were active, but also demonstrate the manifestation of structural change s in real time indicated as doc king, while the lack of docking in loop A(U+2C/C+3U) is likely due to the ove r-stabilization of loop A™s active site by an additional two Watson-Crick base pairs. This ex tra stabilization abolis hes detectable docking transitions at accessible concentrations of [Co(NH 3)6]3+ that support docking in the wild-type system. However, we cannot rule out struct ural transitions in much higher metal ion concentrations. Detailed analysis of this possi bility was not possible due to the tendency of RNA to aggregate in the presence of elevated [Co(NH 3)6]3+. 3.3.2 Construct optimization for NMR spectroscopy A critical challenge to the success of RNA structure determination is the behavior of the RNA construct to be studied. Various c onstructs of loop A (both unlabeled and 15N/13C isotopically labeled) were generated by in vitro transcription using T7 RNA polymerase and the synthesized RNA purified at high concentrations for NMR studies. The purity of the RNAs was assessed by gel electrophores is and mass spectrometry (MALDI-TOF) respectively. The constructs were subjected to NMR experime nts under various annealing conditions, buffer conditions and temperatures to optimize the s econdary structure folding of loop A RNA. We determined pH and buffer conditions within whic h loop A easily folded into a hairpin structure and these conditions were replicated in all our NMR experiments. 100Using a combination of 1D NMR, 2D 1H,15N heteronuclear single quantum coherence (HSQC) and 2D nuclear Overhauser effect sp ectroscopy (NOESY), the behavior of each construct was assessed based on its imino-region resonance (Figure 3-5). The imino protons of G and U nucleotides in RNA resonate between 10-14 ppm especially when protected from exchange ( Figure 3-6 ). Figure 3-5 depicts the results of a 1D imino spectrum for all loop A constructs tested. Overall, the number of peak s observed correspond to th e base-pairs expected except in most constructs except for the GUAA tetraloop construct which showed extra uridine imino peaks as well as lack of a prominent tetraloop G-A shear base-pair peak at ~ 10 ppm (Figure 3-5B). The extra uridine imino peak in GUAA tetraloop and the lack of G-A base-pair imino peak at ~10 ppm are indicative of the formation of a duplex structure in which the presumed tetraloop uridine and adenine are base-p aired, , explaining the a dditional uridine imino peak. Various annealing protocols that favor ha irpin formation were not successful in folding GUAA loop A into a hairpin ( Figure 3-7) as shown by the lack of the tetraloop G imino peak at ~10.5 ppm. The GUAA loop A(U+2C/C+3U) construct ( Figure 3-5A), however, clearly recorded the tetraloop G-A sheared imino peak at ~10.4 ppm, suggesting that the stabilization of base-pairing in loop A favors formation of the ha irpin as opposed to the duplex observed in the previous GUAA loop A. To address the duplex-form ation issue, we built a comparable construct with a GAAA tetraloop variant ( Figure 3-5C), which we hypothesized to be less vulnerable to duplex formation. In this constr uct, we observed two peaks at ~10 ppm (10.6 and 10.8 ppm), one of which was assigned to the G-A sheared base-pair in the GAAA tetraloop region. The observation of the G-A sheared base-pair peak indicated tetr aloop formation and the desired hairpin structure. Finally, sinc e the initial GAAA construct showed some evidence of helical fraying at moderate temperatures (data not sh own), we designed another construct with the GAAA teextendedexpecteddata ind iand dyna FigurU+2Cwith imin oetraloop intrd structure s hd Watson-Cr iicate appropamic studies .re 3-5: ID C/C+3U muextended st eo residues G roducing twohowed a 1 Dick base-pai rriate thermo. imino NMRutant (A), G Uem I highli gand U o additional D imino spe crs with stab lodynamic st aR peaks of UAA tetral oghted in re d101base-pairs i nctrum ( Figule hydrogen ability of h e various lo ooop (B), G Ad (D). Peaks n stem II ( Gure 3-5D ) cobonding upelical regio nop A const AAA tetral ohighlightedGAAA loop Aonsistent wi t to 25 oC (Fins for use intructs. GUAoop (C) and d in red bo xA extended )th the numbigure 3-8). Tn NMR stru cAA tetraloo pGAAA tetrx are the ex t). The ber of These ctural p with raloop tended Figuof ihydure 3-6: RNimino (hig hdrogen bond NA base-pai hlighted red )formation. ring betwe e) protons o102en A-U and of guanine G-C highliand uracil ighting the i; and amin oinvolvemeno protons i nnt n FigSamwegure 3-7: 1 Dmple conditere acquired D imino sp eion: 10 mM at 15oC undectra of G Uphosphate, der two anne a103UAA loop A 150 mM N aaling condit iat differen aCl, 100 µMions that fa vnt annealingM EDTA, pH vor hairpin fog conditions 6.5. Spectr aormation. . a FloµMigure 3-8: oop A. The sM EDTA, pTemperatuspectrum w aH 5.5 bufferre-dependeas acquired ur conditions.104nt 1D imi nusing 600 MH no spectrumHz NMR in m of exten d10 mM ph oded GAAA osphate, 100 1053.3.3 NMR assignments To enable further spectroscopic an alysis of the sy stem, we pursued 1H, 15N, and 13C NMR assignments in the extended-GAAA construc t of the hairpin ribozyme loop A. NMR data were primarily acquired in phosphate buffer wit hout added NaCl to limit duplex formation. Our results show that, as expected, hairpin struct ures were preferred under low salt conditions as indicated by the observation of imino peak at ~10.5 ppm ( Figure 3-8). Starting from the imino resonance in the G-A base-pair of the GAAA tetral oop (with a characteristic chemical shift of 10.5 ppm), the imino-imino region of a two-di mensional NOESY spectrum displays the sequential connectivity for the Wats onŒCrick base-pairs in stem I ( Figure 3-9) of loop A. Regions of A-form helical RNA were assigned via a sequential walk along the backbone, since neighboring imino protons are ~4 Å apart and give rise to a mode rate NOE cross-peak at longer mixing times. The G-U wobble base-pair was eas ily identified as a stro ng cross-peak between the resonances of G-H1 and U-H3 imino protons at chemical shifts of 10 to 12 ppm. A 2D NOESY experiment correlating exchangeable pr otons was acquired at 900 MHz for GAAA loop A and the GAAA extended loop A. For the non- extended GAAA loop A construct, imino-imino cross-peaks were mostly visible in stem I but not stem II. In the GAAA extended loop A construct, designed to stabilize st em II, there were cross-peaks in both helices, with very intense G-U cross-peaks observed at 11.17 and 11.5 ppm, resulting in the assignment of G-5 and U14 respectively. Imino NOE cross-peak pattern from the NOESY spectrum showed a base-pairing scheme consistent with the expected secondary structure of ( Figure 3-9). The GAAA extended loop A was the best-behaved of all the constructs we tested and was thus selected as the primary sequence for all our NMR experiments to facilitate complete resonance assignment for loop A reported herein. Figure 3-9and D2 adimension mixing ti mpH 5.5. 9: 2D imin oaxis are 1Hs respectiveme of 0.3 s. Bo NOESY s pH chemical ely. The sp eBuffer cond i106 pectrum of shift at di rectrum was itions: 10 m MGAAA exterect (D1) aacquired bM phosphatended loop and indirec tby 900 MHte, 100 µM EA. D1 t (D2) Hz at a EDTA, Figspe100gure 3-10: 2ectra were ac0 µM EDTA2D amino Ncquired usingA, pH 5.5, 10NOESY sp eg 900 MHz 0oC.107ectrum of Gat 300 ms mGAAA extenmixing time inded loop Ain 10 mM p hA. These hosphate, 1083.3.4 Assignment of exchangeable proton resonances Our assignment process was initiated by 1D imino spectrum. The peak at 10.63 ppm was easily assigned to the tetraloop G (G+7) because G imino of a G-A sheared base pair resonates at around 10.5-10.8 ppm. The G-U wobble mismatch wa s also easily assigned based on intense NOE cross-peaks ( Figure 3-9 ). The small number of uridines and their downfield 1H (13.5-14.5 ppm) 15N resonance frequencies (153-159 ppm) fac ilitated an easier assignment of imino uridines in loop A. Consequently, U-4, U+5 and U14 were assigned as shown in Table 3-1. G-5 was assigned from its intense NOE connectivity to U14, which led to the assignment of all the remaining guanine imino peaks. Both aminos and imino protons were detected by 2D exchangeable NOESY ( Figure 3-9, Figure 3-10), and 2D 1H,15N HSQC ( Figure 3-11). The proton resonance peak frequencies between 1D 1H and 2D 1H,15N-HSQC imino peaks were very similar in the proton dimension validating the frequencies obtained for each peak. The corresponding 15N resonances were also determined and tabulated alongs ide their proton resonances (Table 3-1). FDFigure 3-11Data was ac: 2D imino quired at 601H,15N HSQ0 MHz in 10109QC spectrum0 mM phos pm of GAAAphate, 100 µMA extended l M EDTA, pHoop A. H 5.5.110Table 3-1: Resonance assignments for loop A (GAAA LpAext) using NMR experiments 1H ppm 15N ppm Assignment 10.63 - G+7 10.92 159.04U+2 11.17 148.08G-5 11.51 159.20U14 12.05 145.78G4 12.40 145.06G-3 12.58 144.73G6 12.96 144.8 G15 13.04 144.63G-7 13.29 143.48G11 13.72 154.72U-4 13.97 154.65U+5 1113.3.5 Non exchangeable protons The base and sugar non-exchangeable protons were detected using ribose-optimized 2D 1H,13C-HSQC, aromatic-optimized 2D 1H,13C-HSQC, 3D HCCH-COSY, 3D HCCH-TOCSY and 3D 13C-edited NOESY-HSQC spectra. 1H-13C resonances were observed from 1H,13C-HSQC experiments acquired at 600 MHz ( Figure 3-12 , Figure 3-13) for loop A. For the ribose resonances, heavy overlap was observed for H1 /C1, H2/C2, H3/C3, H4/C4, and H5/C5 cross-peaks ( Figure 3-12 ), as is typical of RNA molecules within this size. Extensive overlap observed in the ribose resonances can be an impe diment for accurate chemical shift assignments in RNA. However, further use of three-dime nsional RNA correlation experiments can reduce overlap and resonance ambiguity by the detection of cross-peaks from multiple resonances within a spin system. For example, a 3D H CCH-COSY experiment was used to correlate covalently bound 1H-1H-13C spins usually present within the ribose ring. The HCCH-COSY pulse sequence transfers magnetizati on to adjacent protons through one-bond 13C-13C coupling within the H-C-C-H spin system. The 1H-1H HCCH-COSY sub-spectrum correlating H1 and H2 proton chemical shifts in loop A has been presented ( Figure 3-14). Twenty two well- resolved H1 -H2 cross-peaks (in both upper and lower quadrants) were observed and their chemical shift resonances ha ve also been tabulated ( Table 3-4). To provide comprehensive chemical shift assignments for ribose protons, a 3D HCCH-TOCSY data was also acquired (data not shown). HCCH-TOSCY experiments are sim ilar to HCCH-COSY experiments except that the HCCH-TOCSY pulse sequence uses a mixing period where magnetization is transferable through several 13C-13C bonds that connect all 1H-13C pairs within the ribose. This provides additional H3 , H4, and H5 correlations to H1 which facilitates assignment of the entire ribose. The aromatic optimized 1H,13C-HSQC data ( Figure 3-13) provided a fingerprint spectrum for 112loop A, especially with the peaks within the 140-150 ppm range in the 13C dimension. This region contained approximately 34 observable peaks, some of which were the paired upfield and downfield components of 13C-13C doublets for pyrimidine residues. These peaks were interpreted as C6/C8 peaks reflecting the number of nucleotides in our sequence (30 nucleotides). The interpretation of this spectrum was confirmed by a 1H,13C CT-HSQC spectrum of loop A (Figure 3-14 ), which eliminated the C5-C6 dipolar sp litting in pyrimidines and effectively reduced the number of peaks observe d to twenty-five. In RNA of this size, some resonances will not give observable peaks due to overlap, e nd fraying, or other effects, whereas end heterogeneity will sometimes give rise to spurious resonances; thus, this spectrum is roughly consistent with expectations for a 30mer RNA. It should be noted that the sensitivity of 1H,13C CT-HSQC is reduced during constant time acquisiti on, and this spectrum thus had low signal to noise and digital resolution. Th e peaks between 155 Œ 160 ppm were assigned as adenine C2 peaks indicative of the number of adenines in the construct ( Table 3-3). However, we observed seven peaks in the adenine C2 spectral range, instead of the expected nine peaks, due to resonance overlap in some residues. Using the 3D HCCH-COSY data (that provided us with H1'-H2' intra-residue resonances, Figure 3-15 ) and the aromatic 1H,13C-HSQC (which provided the H6/H8 resonances), we also obtained a 13C-edited NOESY-HSQC spectrum ( Figure 3-16) to make the intra and inter-residue NOE sequential wa lk to correlate and assign base H6/H8 protons to the ribose H1' and H2' protons respectively ( Figure 3-17 ). The detailed 13C peak assignments using the three-dimensional NOESY spectrum is still in progress. As at the time of writing this chapter, we had made good progress in loop A a ssignments to the extent of correlating the 13C-NOESY-HSQC connectivity with 3D-HCCH COSY and 13C-HSQC data to finish the aromatic residue assignments especially of the 1H,13C-HSQC resonances. We used 1H,13C-CT-HSQC 113(Figure 3-14), to distinguish between the resonan ces of purines (positively phase) and pyrimidines (negatively phased) as we continue with the NOE sequential walk ( Table 3-5). Pyrimidines were mostly clustered in one region of the spectrum while the purines were well dispersed within the 1H,13C-CT-HSQC spectrum. Several pyrimidine doublet peaks were collapsed into single peaks during 1H,13C-CT-HSQC experiment thus facilitating their easy assignment. To isolate guanine from adenine among the purines, we used a 3D-Adenine HCCH- TOCSY to record adenine resonances within the aromatic 1H,13C-HSQC region, also presented in Table 3-5. Other additional assignments for th e GAAA tetraloop and loop residues were accomplished by resonance comparison with A730 loop of the Neurospora Varkud satellite (VS) ribozyme ( 33) and the partial assignments reported for th e mutant form of hairpin ribozyme loop A (20). Tables 3-2 through 3-5 su mmarize the current state of assignments for the extended- GAAA loop A construct. Given th e good behavior of the construc t and the quality of data obtained so far, we look forward to completing a ssignments for aromatic and ribose resonances for further dynamics studies. FlbFigure 3-12loop A . Datbuffer condi2: 2D 1H,13Cta was acquiitions; 10 mMC-HSQC coired in 100 %M phosphate114ontour spect% D2O at 60 e, 100 µM Etrum of th e0 MHz N MEDTA, pH 5.e ribose re gMR in the fo l.5. gion in llowing Figure 3-13Data was a conditions; 3: 2D 1H,13cquired in 110 mM pho s3C-HSQC s 100% D 2O asphate, 100 115pectrum ofat 600 MHzµM EDTA, f aromatic z NMR in t hpH 5.5. region in lhe followin gloop A. g buffer Figure 3acquired peaks, w 3-14: 2D 1H,at 600 MH while the bla c,13C CT-HSz NMR. Th eck peaks repr116SQC spectr ue red peaks resent the p oum of aro mare negativ eositively-ph amatic re gion ely phased pased purine rin loop A pyrimidine residues Figure 3the ribosNMR in 5.5. 3-15: Represse region inbuffer condisentative 2 Dn loop A . Ditions as fol l117D planes o f ata was acqlows; 10 m Mf3D HCC Hquired in 10 0M phosphat eH-COSY spe0% D2O at e, 100 µM Eectrum of 600 MHz EDTA, pH 118Table 3-2: Non exchangeable assignments of H6/H8 aromatic protons in loop A H8/H6 (ppm) C6/C8 (ppm) Assignment 7.06 136.01 G11 7.42 136.57 G+7 7.48 139.78 Peak 14 7.48 140.28 Peak 24 7.50 140.77 Peak 23 7.53 141.07 Peak 22 7.53 141.62 Peak 33 7.56 135.89 Peak 4 7.58 137.05 Peak 5 7.61 136.43 Peak 3 7.62 139.11 Peak 13 7.63 135.95 Peak 3 7.65 140.95 Peak 21 7.65 140.22 Peak 19 7.66 138.26 G8 7.73 141.68 Peak 27 7.74 142.23 Peak 29 7.76 137.26 Peak 6 7.79 140.92 Peak 17 7.81 140.92 Peak 18 7.81 141.54 Peak 27 7.82 142.15 Peak 28 7.87 143.47 Peak 32 7.89 139.13 A2 7.91 140.45 A10 7.92 142.62 Peak 30 7.98 142.77 Peak 31 7.99 137.52 G4 8.03 139.22 Peak 9 8.13 139.75 A3 8.18 141.57 A9 8.19 139.02 Peak 8 8.20 140.48 Peak 16 8.38 142.24 Peak 26 119Table 3-3: Non exchangeable assignments of adenine H2 aromatic protons in loop A H2 (ppm) C2 (ppm) Assignment 7.35 153.08 Peak 41 7.42 153.01 Peak 38 7.86 153.12 Peak 37 7.89 153.44 Peak 36 7.85 154.31 Peak 37 7.96 154.51 Peak 34 8.07 153.55 A1 120Table 3-4: Assignments of H1'/H2' and C1'/C2' in loop A H1' (ppm) H2' (ppm) C1' (ppm) C2' (ppm) 5.35 4.22 93.20 76.10 5.49 4.41 94.00 75.60 5.49 4.53 94.00 75.60 5.49 4.59 94.00 75.70 5.60 4.42 92.80 76.10 5.61 4.36 94.00 76.00 5.71 4.69 94.40 75.60 5.78 4.44 93.00 75.70 5.78 4.84 91.00 75.20 5.79 4.55 93.20 75.70 5.81 4.57 93.20 75.70 5.82 4.01 93.20 77.80 5.82 4.69 94.00 75.60 5.82 4.75 93.80 76.10 5.85 4.13 92.40 77.40 5.89 4.63 92.20 76.50 5.89 4.87 93.00 75.20 5.92 4.21 92.40 77.30 5.95 4.62 93.20 76.10 5.96 4.69 93.60 76.00 6.05 4.11 92.00 78.20 6.06 4.67 92.60 77.10 Figloophogure 3-16: Rop A. Data osphate, 100 Representatwas acquire0 µM EDTAtive 2D pla ned at 900 MA, pH 5.5. 121nes of 13C-eMHz in the edited NOEfollowing bESY-HSQC buffer conditspectrum otion; 10 m Mof M FiganddasRepgure 3-17: Sd inter-resi shed lines wprinted with Schematic r edue H8/H 8while inte r-rpermission epresentati o8 to H1' a nresidue NO Efrom Pasca l122on of NOE d H2'. IntraEs are reprele Legault d oconnectivit ia-residue Nesented by soctoral thesi ies that cor NOEs are repsolid lines wsrelate intr apresented b ywith arrow sa- y s. 123Table 3-5: Aromatic resonance assignments of H1'/H2' and C1'/C2' in loop A H8/H6 (ppm) C6/C8 (ppm) Assignment 7.06 136.01 G11 7.42 136.57 G+7 7.48 139.78 purine 7.48 140.28 pyrimidine 7.50 140.77 pyrimidine 7.53 141.07 pyrimidine 7.53 141.62 purine 7.56 135.89 purine 7.58 137.05 purine 7.61 136.43 purine 7.62 139.11 purine 7.63 135.95 purine 7.65 140.95 pyrimidine 7.65 140.22 pyrimidine 7.66 138.26 G8 7.73 141.68 purine 7.74 142.23 pyrimidine 7.76 137.26 purine 7.79 140.92 purine 7.81 140.92 purine 7.81 141.54 purine 7.82 142.15 purine 7.87 143.47 pyrimidine 7.89 139.13 A2 7.91 140.45 A10 7.92 142.62 pyrimidine 7.98 142.77 purine 7.99 137.52 G4 8.03 139.22 purine 8.13 139.75 A3 8.18 141.57 A9 8.19 139.02 purine 8.20 140.48 purine 8.38 142.24 purine 1243.4 DISCUSSION Our docking studies of loop A and loop A( U+2C/C+3U) demonstrated that trans- constructs with wild type sequence remained active under docking condition, and that this activity was metal-driven. However, the double mu tant loop A(U+2C/C+3U) lacked activity, as expected, not because of the removal of the im portant functional groups or disruption of the structure but because of the stab ilization of the ground- state conformer that hinder transitions to the docked state. The lack of docking transition s imparted by additional base-pairing featuring U+3:A7 and C+2:G8 base-pairs, emphasizes the importance of conformational fluctuations in the formation of the docked state. In the muta nt sequence, the catalytically necessary G8 was stabilized by formation of G8-C+2 base pair. This base-pair locks loop A into its ground-state conformation leading to a stabilized form of un docked loop A. The conformational stabilization of loop A(U+2C/C+3U) is supported by NMR spectra ( Figure 3-5 ) which mostly showed similar chemical shift resonance patterns for exchangeab le protons in both mutant loop A(U+2C/C+3U) and wild type loop A implying similar global stru cture adopted by the two constructs. The use of Circular dichroism (CD) in measuring structural rearrangements that characterize docking at equilibrium conditions is a robust technique with the ability to monitor such secondary and tertiary structural changes. The docking data obtained on U+2C/C+3U loop A suggests docking perturbation based on impaired structural tran sitions visible by CD. Since, the native loop A recorded we still acknowledge the challenges that may come with interp reting the data obtained from this double mutant sequence, loop A(U+2C/C +3U). It is possible that the two additional base-pairs may have virtually rendered the se quence non-functional. Our argument is that a single G8U mutation of this inactive sequence partia lly rescues the activity of this loop to dock with its docking partner, loop B (22). 125The core of the optimization of the best-beh aved construct for NMR studies was in the design of the two stems I and II wh ile maintaining active wild-type sequence within the internal loop active site. The stabilization of both st ems was useful in limiting our observation of unwanted duplex structures. Bette r results were achieved when stem I was capped with a GAAA instead of GUAA tetraloop. GUAA ma y destabilize hairpin formati on by formation of an intra- sequence A-U base-pair which is not available for a GAAA tetraloop. Also, the stabilization of stem II with additional base-pairs generally stabilized loop A for further NMR studies. Incomplete 1H, 13C, and 15N chemical shift assignments have been determined for loop A of the hairpin ribozyme. These assignments have been realized using standard through-space and through-bond correlated spectroscop y experiments with non-labeled as well as uniformly labeled loop A. Ambiguity caused by spectral overlap pr esents challenges in th e assignment process. However, the use of two- and three-dimensional e xperiments facilitated the resolution of several overlapped peaks. Our observation of imino and amino proton peaks suggest thermally stable folded loop A RNA at least until 25 oC, while the observation of a sheared G-A imino base-pair confirmed the formation of the hairpin structur e. The observation of guanosine and uridine imino proton peaks was attributed to canonical Watson-Crick base pair s consistent in number and arrangement with our predic ted secondary structures ( Figure 3-2 ). Imino proton peaks for G8 and U+3 were observed in the mutant U+ 2C/C+3U sequence as predicted. Using 1H,15N-HSQC, imino and amino spin pairs were identified, which led to imino assignments ( Table 3-1, Figure 3-11). New imino and amino proton peaks were de tected in extended stem 1(U+2C/C+3U loop A, Figure 3-1 D) and stem II (Loop A extended, Figure 3-1 C) which facilitated the imino assignment process. These new peaks were lo calized proximal to the internal loop in U+2C/C+3U loop. The partial stability of th e loop region in the U+2C/C+3U loop A may 126support our proposal that reduced mu tant activity could impart decr eased flexibility of internal loop residues. To validate this proposition, more mutants with minimum wild type changes, especially mutants that perturb docking, need to be screened. The presence of a G-U mismatch registered strong imino NOEs which served as our starting point for im ino assignments. The incorporation of a G-U mismatch in a stem is an important strategy sinc e the G-U NOEs are very strong. We are in the process of assigning all the resonances of the wildtype extended loop A construct. The spectrum of the non-exchangeable prot ons in loop A was well dispersed in the aromatic region but not well resolved in the ri bose region due to the degeneracy of the ribose protons within their environment. Aromatic C8 /C6 and C2 resonances were clearly resolved albeit with a few overlaps especially in the pyrimidine region. This was expected especially due to the degeneracy of C6 protons in uridines and cytosine. The identification of the ribose protons was initiated by a combinat ion of 3D HCCH-TOCSY ( 34) and HCCH-COSY ( 35) experiments. The HCCH-TOCSY correlates all protons in each ribose spin system, allowing for complete assignment of ribose 1H and 13C. We identified twenty one H1'/H2' cross-peaks lying in symmetric quadrants which facilitated the assi gnment of C2' symmetric resonances based on correlation with H2' chemical shifts. The C2' and C3' resonate at approximately the same frequency, thus required the collection of the well resolved 3D-HCCH-TOCSY. A 3D 13C-edited NOESY was used to extend the 2D 1H,1H NOESY into 3D based on the chemical shift of the 13C attached to one of the protons . This method greatly improves 1H signal overlap with sufficient 13C chemical shift dispersion. Using the NOESY wa lk, significant assignments can be made to complete the process. Assignment process can be challenging especial ly due to resonance overlap and the lack of clear star ting points to facilitate the necessary connectivity. Thus far, we 127have made significant progress in loop A re sonance assignment. Th e exchangeable imino protons and nitrogen were unambig uously assigned but some work still remains to fully assign aromatic 1H and 13C resonances. The finished assignments are important for the detailed NMR spin-relaxation experiments of 13C dynamics which has been comp rehensively discussed in the next chapter (Chapter 4) of this thesis. Wher e necessary, we have reported the residue dynamics results by peak number rather th an assigned atom, since complete resonance assignments is not necessary for 13C relaxation acquisition but will be usef ul in comprehensive residue dynamics analysis. ACKNOWLEDGEMENT I wish to acknowledge Dr. Minako Sumita for doing the Circular Dichroism (CD) docking work between wildtype loop A and loop B 128REFERENCES 129REFERENCES 1. Madhani, H. D., and Guthrie, C. (1994) Dynamic RNA-RNA interactions in the spliceosome, Annu Rev Genet 28, 1-26. 2. Wahl, M. C., Will, C. L., and Luhrmann, R. (2009) The spliceosome: design principles of a dynamic RNP machine, Cell 136, 701-718. 3. Hampel, A., and Tritz, R. (1989) RNA catal ytic properties of the minimum (-)sTRSV sequence, Biochemistry 28, 4929-4933. 4. Buzayan, J. M., Gerlach, W. L., and Brueni ng, G. (1986) Satellite tobacco ringspot virus RNA: A subset of the RNA sequence is sufficient for autolytic processing, P Natl Acad Sci USA 83, 8859-8862. 5. Butcher, S. E., Heckman, J. E., and Burke, J. M. (1995) Reconstitution of hairpin ribozyme activity following separation of functional domains, J Biol Chem 270, 29648-29651. 6. Hampel, K. J., Walter, N. G., and Burke, J. M. (1998) The solvent- protected core of the hairpin ribozyme-substrate complex, Biochemistry 37, 14672-14682. 7. Buzayan, J. M., Hampel, A., and Bruening, G. (1986) Nucleotide sequence and newly formed phosphodiester bond of spontaneously ligated satellite tobacco ringspot virus RNA, Nucleic Acids Res 14, 9729-9743. 8. Symons, R. H. (1992) Small catalytic RNAs, Annu Rev Biochem 61, 641-671. 9. Perrotta, A. T., and Been, M. D. (1992) Cleavage of oligoribonuc leotides by a ribozyme derived from the hepatitis delta virus RNA sequence, Biochemistry 31, 16-21. 10. Lilley, D. M. (2004) The Varkud satellite ribozyme, RNA 10, 151-158. 11. Winkler, W. C., Nahvi, A., Roth, A., Collins, J. A., and Breaker, R. R. (2004) Control of gene expression by a natural metabolite-responsive ribozyme, Nature 428, 281-286. 12. Esteban, J. A., Banerjee, A. R., and Burke, J. M. (1997) Kinetic mechanism of the hairpin ribozyme. Identification and characteriza tion of two nonexchangeable conformations, J Biol Chem 272, 13629-13639. 13. Fedor, M. J. (1999) Tertiary structure stabilization promotes hairpin ribozyme ligation, Biochemistry 38, 11040-11050. 13014. Bevilacqua, P. C., and Yajima, R. (2006) Nu cleobase catalysis in ribozyme mechanism, Curr Opin Chem Biol 10, 455-464. 15. Doudna, J. A., and Lorsch, J. R. (2005) Ri bozyme catalysis: not different, just worse, Nat Struct Mol Biol 12, 395-402. 16. Shippy, R., Siwkowski, A., and Hampel, A. (1998) Mutational analysis of loops 1 and 5 of the hairpin ribozyme, Biochemistry 37, 564-570. 17. Komatsu, Y., Kumagai, I., and Ohtsuka, E. ( 1999) Investigation of the recognition of an important uridine in an internal loop of a hairpin ribozyme prepared using post- synthetically modifi ed oligonucleotides, Nucleic Acids Res 27, 4314-4323. 18. Perez-Ruiz, M., Barroso-DelJesus, A., and Be rzal-Herranz, A. (1999) Specificity of the hairpin ribozyme. Sequence requirem ents surrounding the cleavage site, J Biol Chem 274, 29376-29380. 19. Chowrira, B. M., Berzal-Herranz, A., and Burke, J. M. (1991) Novel guanosine requirement for catalysis by the hairpin ribozyme, Nature 354, 320-322. 20. Cai, Z., and Tinoco, I., Jr. (1996) Solution st ructure of loop A from the hairpin ribozyme from tobacco ringspot virus satellite, Biochemistry 35, 6026-6036. 21. Walter, N. G., Chan, P. A., Hampel, K. J., M illar, D. P., and Burke, J. M. (2001) A base change in the catalytic core of the hairpin ribozyme perturbs function but not domain docking, Biochemistry 40, 2580-2587. 22. Sargueil, B., Hampel, K. J., Lambert, D., a nd Burke, J. M. (2003) In vitro selection of second site revertants analysis of the hairpin ribozyme active site, J Biol Chem 278, 52783-52791. 23. Rupert, P. B., and Ferre-D'Amare, A. R. (2001) Crystal structure of a hairpin ribozyme- inhibitor complex with implications for catalysis, Nature 410, 780-786. 24. Salter, J., Krucinska, J., Alam, S., Grum-T okars, V., and Wedekind, J. E. (2006) Water in the active site of an all-RNA hairpin ribozyme and effects of Gua8 base variants on the geometry of phosphoryl transfer, Biochemistry 45, 686-700. 25. Torelli, A. T., Krucinska, J., and Wedekind, J. E. (2007) A comparison of vanadate to a 2 '-5 ' linkage at the active site of a small ri bozyme suggests a role for water in transition- state stabilization, RNA-A Publication of the RNA Society 13, 1052-1070. 26. Ojwang, J. O., Hampel, A., Looney, D. J., Wong-Staal, F., and Rappaport, J. (1992) Inhibition of human immunodeficiency virus type 1 expression by a hairpin ribozyme, P Natl Acad Sci USA 89, 10802-10806. 13127. Sumita, M., White, N. A., Julien, K. R., and Hoogstraten, C. G. ( 2013) Intermolecular domain docking in the hairpin ribozyme Metal dependence, binding kinetics and catalysis, RNA Biol 10, 425-435. 28. Milligan, J. F., Groebe, D. R., Wither ell, G. W., and Uhlenbeck, O. C. (1987) Oligoribonucleotide synthesi s using T7 RNA polymerase and synthetic DNA templates, Nucleic Acids Res 15, 8783-8798. 29. Milligan, J. F., and Uhlenbeck, O. C. ( 1989) Synthesis of small RNAs using T7 RNA polymerase, Methods Enzymol 180, 51-62. 30. Kim, I., McKenna, S. A., Viani Puglisi, E., a nd Puglisi, J. D. (2007) Rapid purification of RNAs using fast performance liquid chromatography (FPLC), RNA 13, 289-294. 31. Plateau , P., Gueron, M. (1982) Exchangeab le proton NMR without base-line distorsion, using new strong-pulse sequences, J. Am. Chem. Soc 104, 7310-7311. 32. Rupert, P. B., Massey, A. P., Sigurdss on, S. T., and Ferre-D'Amare, A. R. (2002) Transition state stabilizat ion by a catalytic RNA, Science 298, 1421-1424. 33. Desjardins, G., Bonneau, E., Girard, N., Boisbouvier, J., and Legault, P. (2011) NMR structure of the A730 loop of the Neurospora VS ribozyme: insights into the formation of the active site, Nucleic Acids Res 39, 4427-4437. 34. Nikonowicz, E. P., and Pardi, A. (1992) Th ree-dimensional heteronuclear NMR studies of RNA, Nature 355, 184-186. 35. Bax, A., Clore, G. M., Dris coll, P. C., Gronenborn, A. M., Ikura, M., and Kay, L. E. (1990) Practical Aspects of Proton Carbon Carbon Proton 3-Dime nsional Correlation Spectroscopy of C-13-Labeled Proteins, J Magn Reson 87, 620-627. 132 CHAPTER 4 INSIGHTS INTO THE INTERNAL DYNAMIC S OF LOOP A USING NMR SPIN RELAXATION 133ABSTRACT Conformational dynamics is an im portant feature in proteins and RNA. However, there is little information outli ning the relationship between dynami cs and RNA function. Structural studies of the hairpin ribozyme have reported si gnificant structural rearrangements at the active site upon domain-domain interaction, suggesting th at dynamics may facilitate transitions from the ground to active state. In this work, we have us ed longitudinal ( 13C T1) and traverse ( 13C T1) relaxation experiments as well as heteronuclear NOE to interrogate the internal dynamics and disorder within loop A. We have also used relaxation dispersion measurements of power- dependent T 1 to understand conformational exchange of residues in the µs-ms timescale. Internal motions were observed on a wide variet y of timescales within th e non-helical regions of loop A suggesting a complex landscape of accessibl e states, and integrated correlations amongst the observed motions. These results demonstr ate the usefulness of NMR spin relaxation measurements in probing dynamics of comple x molecules exemplified in ribonucleic acids. 1344.1 INTRODUCTION Over the years, there has been an increasi ng appreciation of th e critical role of conformational dynamics in many facets of molecular function ( 1, 2). With its ability to act as a site-specific probe of the dynamic properties of biomolecules, heteronuclear NMR spin relaxation has emerged as a lead ing technique for experimental studies of molecular dynamics. Relationships between spin relaxation data and function may be drawn in several ways. If the target structure of a confor mational change is known, the ex tent and manner in which the molecule fluctuates toward that state can be assessed. More directly, it may be possible to selectively freeze out specific motional variables using mutation or modifi cation and analyze the effects on function. Currently, there is an active debate about the relationship between dynamics and function. One model suggest s a direct coupling between dynamic modes throughout the molecule to the chemical catal ytic steps using dihydrofolate re ductase (DHFR) as an example (3). Motions directly lin ked to catalysis in DHFR were inve stigated using NMR and the dynamic modes were found to be on similar timescales as chemical transformati ons. Similar correlations have also been reported in RNase A ( 4) and adenylate kinase ( 5). For RNA molecules, conformational dynamics play an equally important role in events such as RNA catalysis and ligand recognition (6-9 ). In catalytic RNA the role of dyna mics is even more prominent than most protein enzymes. It appears that the conf ormational fluctuations from a stable ground state to an activated conformer are an inherent part of most ribozym es. One notable example includes the conformational rearrangements observed between crystal structures of precursor and product states in hepatitis delt a virus (HDV) with catalytic or inhib itory metal ions binding only to the precursor form ( 10). Therefore, a detailed understanding of the structure-func tion relationship of ribozymes requires the characteri zation of both the static struct ure and conformational dynamics 135of the molecule. Conformational rearrangement s have been observed between isolated and docked hairpin ribozyme domain (loop A and B) structures ( 11-13) and the mechanism of these rearrangements underpin the focus of this stu dy. These features sugges t that conformational sampling is not only vital for conformational pr e-organization but may also be involved in directing chemistry. Recently, intermolecular loop A and B interaction kinetic rates have been reported by our group ( 14). The docking association rate re veals a very slow docking process between loop A and loop B (five orders of ma gnitude below the diffusion limit) despite a tight docking affinity. This slow association rate as we ll as structural variability between the free and docked forms, especially for th e key residues involved in the interaction, informs our hypothesis that isolated loops A and B ma y independently sample docking- competent states, raising the possibility of a double conformationa l capture docking mechanism. The relationship between conformational dynamic s and the catalytic cycle of the hairpin ribozyme is not well understood. The major l oop rearrangements observed upon docking, as well as the kinetically unfavorable doc king process, both argue for a tight and mechanistic role for conformational dynamics in the pre-organization of the catalytically-competent active site of the hairpin ribozyme. NMR spectroscopy presents us with a unique ability to investigate dynamic properties of molecules over a range of different timescales with atomic resolution ( 15-23) (Figure 4-1). The discovery and success of RNA isotope-labeling techniques ( 24, 25) permitted the use of 13C and 15N nuclei as points of investigating conformational dynamics in RNA (26, 27). Conformational transitions with links to cat alytic function in ribozymes will potentially register within the microseconds to milliseco nds time scale. Such rearrangements may be thermally-induced leading to exchange be tween two or more conformational states. Heteronuclear spin relaxation measurements ar e suitable for probing dynamics on a wide range 136of timescale especially within pi coseconds to millisec onds. For fast (ps-ns) timescale dynamics, longitudinal (T1), transverse (T 2) relaxation, and heteronuclear NOE NMR techniques are mostly utilized ( 2). Also, deuterium NMR has previously been used for the study of fast motions in proteins as well as RNA systems ( 28). For slower dynamic processes (µs to ms), tr ansverse and rotating frame relaxation rates (R 2 and R1) are ordinarily measured as a function of the effective applied power ( Figure 4-1 ). Both of these methods can be used to extract the rate of exchange (kex) between two distinct states (where k ex is the sum of the rate constants in the equilibrium between the two states) since c onformational exchange on this timescale contributes to R 2 and R1. For the simplest case of a two-site exchange model of A reversibly interconverting with B, the exchange rate constant, kex, is the sum of the forward and reverse rate constants, k1 and k-1. kex = k1 + k-1 (4-1) (4-2) The nature of the NMR exchange re gime depends on the relation between kex and , the chemical shift difference between the populated states. For a system exchanging between two states, the transverse relaxation ra te under spin-locke d conditions, R1, is described by equation 4-2 (29) where T 1 is the measured relaxation time, T 1 is the relaxation time at infinite spin- lock power, Pa and Pb are the fractional popula tions of conformations a and b (assuming two-site exchange), is the chemical shift difference between states a and b, 1= B1 is the spin-lock power expressed in radians per second, and ex is the lifetime for the exchange process calculated as 1/kex. 137 10-12 10-910-610-3100103106secHeteronuclear Spin Relaxation Model-Free Approach T/T Dispersion 12Lineshape 2D EXSY Inv. Tran. H/D Exchange DisorderConformational Exchange Global Unfolding Local Unfolding Segmental Motion Figure 4-1: Timescale of macromolecu lar internal motions relevant in biological processes . Various NMR techniques are useful for probing motions from picoseconds to milliseconds. 138In this work, we have probed the dynamics of loop A domain of the hairpin ribozyme as an initial step in understanding the active site d ynamics in RNA-RNA ter tiary interaction. The results complement the calculational work in Chapter 2 and provide insights into the role of conformational sampling of loop A in the formation of tertiary structure in the hairpin ribozyme. 4.2 MATERIALS AND METHODS 4.2.1 RNA preparation 13C and 15N labeled nucleotides were incorporat ed in the hairpin ribozyme loop A (Figure 4-2) by in vitro transcription ( 30, 31) using recombinant phage T7 RNA polymerase (T7 RNAP) enzyme in the presence of a synthetic DNA template. Transcripti on was initiated with RNAP following addition of DNA template, uniformly 15N/13C-labeled nucleotide triphosphates (ATP, UTP, CTP, and GTP) and Mg 2+. Transcription was continued for 3 hours at 37 °C, stopped by the addition of EDTA, and anal yzed by 15% denaturing gel electrophoresis. Transcription product was subjected to ethanol precipitation and RNA precipitate was then reconstituted in H 2O and subjected to purification through a HiLoad 26/60 Superdex TM 75 column ( 32) integrated with AKTA Fast Perfor mance Liquid Chromatography (FPLC) monitored at A 260 with an isocratic flow rate of 3 mL /min sodium phosphate buffer, pH 6.40. Fractions corresponding to loop A were pooled and thoroughly exchanged with H 2O using Amicon Ultra-4 centrifugal filt er units, 3000 MWCO. RNA purity was determined by denaturing gel. RNA was then dried under vacuum and reconstituted in 100% D 2O and appropriate NMR buffer conditions (10 mM phosphate buffer, 100 µM EDTA, pH 5.5) Figure 4-relaxatiowas appe -2: The hai ron studies. Rended at the erpin ribozymRed arrow is end of stem 139me loop A cthe cleavagI for stabili zconstruct use site. GNRAzation . sed in NM RA tetraloop (R (boxed) 1404.2.2 NMR studies All NMR data were acquired on a Varian UnityINOVA 600 MHz ( 13C 150 MHz) spectrometer. The 1H,13C HSQC spectrum was acquired with 1024 × 512 complex points in the t2 and t1 dimensions, respectively, w ith corresponding spectral wi dths of 6000 and 4524 Hz, and a 1 s recycle delay. The proton RF carrier was centered on the residual HDO signal, and the 13C carrier frequency was set at 145 ppm (including a 35 ppm pulse se quence adjustment) to capture the resonances in the aromatic C8/C6 and adenine C2 region. Fast dynamics T 1, T1, and 1H-13C heteronuclear NOE (hNOE) data were acquired using similar parameters as 1H,13C-HSQC with minor variations of published pulse sequences ( 33). T 1 delay times ranged from 50 -1200 ms, while the T 1 delay times were from 10-100 ms. These relaxation data were acquired over 96 time-points in the second dimension at a cons tant 2980 Hz spin-lock field. Saturated and non- saturated heteronuclear NOE experiments were acquired in interleaved fashion with a proton irradiation of 3 sec within a total recycle dela y of 5 sec in the fisaturatedfl experiment. In relaxation dispersion studies of purine C 8, pyrimidine C6 and adenine C2 carbons, R 1 experiments were acquired as a function of the applied spin lock power, 1. 13C carrier was positioned at 145 ppm with a sweep width of 6000 x 4524 Hz. A total of 106 complex t1 increments of 64 transients each were collected with a recycle delay of 2 s, in a series of 15 different spin-lock powers rangi ng from 1.8 to 6 kHz at a constant relaxation delay of 35 ms. All the dispersion data were acquired using on-resonance R 1 (33). All NMR data were processed using FELIX 2002 (Felix NMR, Inc.). Be fore Fourier transformation (FT) in t2 dimension, the FID was offset-corrected at 20%, a 5 Hz exponential line broade ning function was applied in t 2 with zero filling. A cosine-squared apodization function was used in t1. 1414.2.3 Data analysis Cross-peak integrated volumes were expor ted to Igor Pro 5.0.4 (WaveMetrics) after peak-picking and volume integration using FELIX 2002. R 1 and R 1obs relaxation rates were extracted by fitting the integrated volume inte nsities to a single exponential decay using two parameters to limit potential fitting bias ( 34). The relaxation rates were obtained from the individual fits with th e errors reported from the exponential fit. The 1H,13C hNOE was determined from the ratio of the saturate d and non-saturated peak volume intensity. For dispersion analysis, the magnit ude of the spin-lock field was c onsidered as the effective field resulting from the applied B1 field and the offset according to equation 4-3. (4-3) In addition, the observed T 1 decay process has contributi ons from both transverse and longitudinal relaxation, because of tilting the spin-l ock axis out of the xy plane. This effect was also considered using measured values of R 1 (R 1=1/T1) (35). The R 1obs rates were calculated using the equation R1obs = (1/T) × -ln (I(T)/I o), where T is the relaxation delay in seconds, I is the measured intensity, and I o is the measured intensity with no relaxation delay (reference relaxation). The data for I o was acquired three times and averaged. R 1 rates were calculated from R 1obs using the equation: R1obs = R1 cos2 + R1 sin 2 (4-4) 142where = tan -1 (1/) is the angle of the spin -lock axis from vertical, 1 is the spin lock power in Hz and is the offset in Hz. Exchange parameters k ex , ex (Pa Pb()2), and R 1 were extracted by fitting dispersion curves to R 1 equation 4-2 (21) using Igor Pro 5. 4.2.4 Model-free analysis Internal motion, quantified as a generalized order parameter S2 in the ps-ns timescale using the model-free formalism of Lipari & Szabo ( 36-38), was performed using R 1, R1, and 1H,13C hNOE measurements acquired at 600 MHz. All the model-free analysis calculations were performed by Dr. Charles Hoogstraten usi ng the computer program Modelfree 4.1. S 2, e, and Rex motional parameters were determined for atoms in the hairpin ribozyme. An estimate of the isotropic rotational correlati on time was obtained from the T 1/T1 ratio for resonances in the helical stem that showed no evidence of exchange contributions when analyzed including dipole-anisotropy relaxation mechanisms ( 39). Chemical shift anisot ropy (CSA) values used were 131, 186.5 and 167 ppm for C8, C6 and C2 atoms respectively with a rHC bond length of 1.1 Å. 4.3 RESULTS AND DISCUSSION 4.3.1 13 C Relaxation measurements Longitudinal and transverse relaxation rates (R 1 and R1) were determined for purine C8 and C2 atoms as well as pyrimidine C6 atoms of hairpin ribozyme loop A to assess the fast internal motion (ps-ns) in loop A. In Figure 4-3 we have shown representative 13C R1 and R 1obs curves for C8 atoms in the tetraloop (A2) a nd active site loop region (A9) at 25°C ( Figure 4-3). Transverse and longitudinal relaxa tion rates were obtained for all the C8 resonances of purines, the C6 resonances of pyrimidines and the C2 resona nces of adenines in in the loop A. There were 143a total of 39 probes used to assess base dynamics. Both the T 1 and T 1 data were fit to single- exponential decays and the relaxation rates extracted. Results are reported in Figure 4-3 and Table 4-1. In general, most of th e residues reported typical R1 and R 1 values of ~2.2 and ~40 s -1 as well as a heteronuclear NOE ra tio of 1.2. This is consistent w ith the relaxation rates observed in helical regions for which most of the resi dues in loop A are found. This observation suggests that most of the residues in loop A are base-paire d within the helical environment as reported in chapter 3 of this thesis. Hoogstraten et al , have also reported similar relaxation rates in the helical region of a comparable RNA system, the leadzyme ( 20). Pyrimidines mostly showed no significant differences observed fo r relaxation rates in their C6 resonances consistent with previous work done in l ead-dependent ribozyme ( 40). This was expected because most of the pyrimidines are within the helic al secondary structure of loop A except U+2 and C+3 which are in the loop region. However, one pyrimidine (assi gned to either U+2 or C+ 3) residue showed an overall increase in increase R 1 by ~31% which suggests dynamics. Th is result is consistent with our molecular dynamics simulation studies of base and sugar dynamics reported for U+2 and C+3 in Chapter 2 of this thesis. The helical C8 and C2 resonances displayed no evidence of exchange contributions from the relaxa tion data based on Model free analysis (Figure 4-4 , Table 4-2). Fast internal motion was obs erved in various residues as in dicated by increase in the R 1 values as well as a decrease in R 1 values. The R 1 for peak 4 C8 was ~19% higher than the average value of R 1 calculated for helical C8 atoms. Conversely, its R 1 was dramatically reduced by ~53% relative to the average R 1 of the helical resonances . Peak 8 reported a 33% increase in R 1 and a similar (32%) decrease in R 1. Peaks 26 and 35 repor ted an increase in R 1p but constant R 1 relative to the average helical values. An increased R 1 together with a decreased R2 (or R 1) generally indicates ps-ns disorder, while an increased R 2 (or R 1) with a constant R 1 144suggests exchange contribution. Conformational exchange is us ually validated by relaxation dispersion measurements reporting on the residue. It should be noted that peak 26 represents a purine C8 resonance while peak 35 represents an adenine C2 resonance as assigned in the previous chapter (Chapter 3) of this thesis. These dynamics may be reporting on similar residue from the C8 and C2 resonances within a sing le stranded loop region. In fact, peak 26 was assigned as A1 which resides in the GAAA tetraloop. The tetraloop A1 (peak 26) reported internal dynamics ( Figure 4-4, Table 4-2), based on the S 2 value of 0.87 indi cating rapid (ps-ns) disorder. This base was proposed to be disordered from structural data ( 41). A3 (Peak 15), similarly residing in the tetraloop, re ported a 65% enhancement in the C8 R 1 and a 68% reduction in R 1 relative to the helical regions. This is also consistent ps-ns dynamics. However, previous studies on the leadzyme showed a larger R 1 for the C8 resonance of A3 ( 40) with a follow-up study reporting a conformational exchange lifetime of 80 µs ( 20). Peak 12 recorded the highest enhancement in R 1 of > 100% ( Figure 4-4) relative to the helical rates. Also, this peak showed only modest 32% reduction of R 1. This suggests that the C8 resonance of this residue undergoes conformational exchange in the µs-ms timescale. Conformational exchange parameters determined from Model free analysis ( Figure 4-5 ) confirmed a C8 conformational exchange with R ex of ~140 s -1. However this resonance did not show ps-ns dynamics based on a higher order parameter of 0.903. This suggests that this conformational exch ange may play a role in conformational sampling. Our preliminary assi gnment suggests that this residue may be G8, which plays a catalytic role in the hairpin ribozyme. G8 is base -paired to the dynamic U+2 in the docked structure of hairpin ribozyme ( 12), suggesting that this interaction may be partly driven by conformational exchange in this residue. Peak 9 C8 rela xation data reported ps-ns dynamics based on its calculated order parameter suggesting th at this residue may be located in the loop or 145terminal region. Our preliminary assignments discusse d in Chapter 3 of this thesis indicate that peak 9 is likely G+1, located in the loop region, or a terminal G. Peak 10 (assigned as tetraloop A2), did not show any proof of dynamics, in the ps -ns timescale. This suggest s that A2 is fairly stable due to stacking interactions within the tetraloop. This is consistent with our previous observations made on A2 in the resu lts our MD simula tion of loop A ( Figure 4-6 ). Similar observations have been made on the adenine, A 2, base of a GAAA tetraloop which localizes between residues involved in stacking interacti ons while being protected from chemical modification upon RNA folding with no evidence for dynamics at the C8 site ( 40, 42). 4.3.2 Relaxation dispersion By observing an increase in the measured transverse relaxation rates as well as R ex contributions in the model-free analysis for so me residues, relaxation dispersion curves of on- resonance R1 experiments were acquired to examine c onformational exchange processes within loop A in detail. Motions on the s-ms timescales can affect the transverse relaxation. While model-free analysis can propose the presence of exchange (R ex) contributions, a more accurate determination of R ex contributions is in the analysis of relaxation di spersion curves. Relaxation dispersion curves are the dependence of the measur ed transverse relaxation rate on the effective B1 field as determined by the applied radiofrequen cy (RF) field and the resonance offset for R 1 experiments. Over the years, relaxation dispersi on curves have proved useful in investigating conformational exchange processes extensively in proteins ( 22, 23, 43-46). However, few dynamic exchange studies have been done for RNA ( 20, 47, 48). Relaxation dispersion curves were fit to equations 4-2 to determine k ex, R1, and ex, where ex = Pa Pb()2. Figure 4-3:A2 (A and Bparameter f : Represen tB) and A9 (fits to a sing ltative 13C R1(C and D). Cle exponenti1461 (A and C) Curves represal. and R1obs (sent non-line(B and D) cuear least squurves for uares two- Figure 4shown fo(C) heterand 3-3 4-4: Histogror (A) longitronuclear N Oin Chapter 3ams of the mtudinal relaxOE. The pea3. 147measured r exation, T 1 (B)ak numbers elaxation ra) transversecorrespond tates in loop relaxation, Tto those in TA are T1, and Tables 3-2 148Table 4-1: 13C R1, R1 and heteronuclear NOE measurements . R1 and R1 valued are for C8 (purines), C6 (pyrimidin e) and C2 (adenine) atoms of loop A at 600 MHz. Peaks 34-39 are adenine C2 peaks. Assignment R 1 R1 hNOE Peak 1 1.87±0.07 38.35±3.10 1.20 Peak 2 1.55±0.11 34.64±4.49 1.16 Peak 3 1.91±0.08 24.53±1.01 1.16 Peak 4 2.54±0.14 19.50±1.47 1.20 Peak 5 1.63±0.05 28.65±1.00 1.22 Peak 6 1.84±0.09 35.48±1.80 1.23 Peak 7 1.57±0.08 29.55±1.43 1.28 Peak 8 2.83±0.07 27.76±0.62 1.19 Peak 9 2.56±0.19 23.99±1.70 1.17 Peak 10 1.64±0.08 35.12±1.53 1.21 Peak 11 1.97±0.19 38.38±4.66 N/A Peak 12 1.44±0.07 167.44±45.70 1.25 Peak 13 1.78±0.05 43.06±2.46 1.21 Peak 14 2.36±0.19 41.65±4.07 1.17 Peak 15 3.50±0.68 12.74±1.40 N/A Peak 16 1.70±0.11 28.88±3.87 1.21 Peak 17 1.99±0.05 46.34±3.14 1.26 Peak 18 1.84±0.11 45.12±3.98 N/A Peak 19 2.35±0.08 46.67±2.68 1.06 Peak 20 2.41±0.11 55.14±4.34 1.11 Peak 21 2.35±0.11 45.08±2.55 1.15 Peak 22 2.32±0.12 40.36±1.23 1.17 Peak 23 2.04±0.13 45.90±3.46 1.06 Peak 24 2.26±0.11 40.05±1.94 1.16 Peak 25 2.16±0.09 47.19±3.44 1.28 Peak 26 1.86±0.07 29.11±0.94 1.34 Peak 27 2.26±0.07 41.55±1.55 1.19 Peak 28 2.57±0.01 44.57±2.90 1.21 Peak 29 2.47±0.10 40.78±2.02 1.15 Peak 30 2.45±0.14 36.32±3.25 1.22 Peak 31 2.82±0.36 55.52±6.15 1.17 Peak 32 2.83±0.29 37.06±6.37 1.23 Peak 33 2.00±0.13 39.39±4.02 1.21 Peak 34 1.73±0.09 41.10±3.15 1.25 Peak 35 1.98±0.24 29.68±4.61 1.20 Peak 36 1.90±0.07 34.91±1.22 1.36 Peak 37 1.84±0.05 44.24±2.01 1.33 Peak 38 1.89±0.13 35.90±2.52 N/A Peak 39 1.70±0.04 34.58±1.16 1.16 149Table 4-2: Model-free analysis of loop A relaxation measurements.Order parameter (S 2), correlation time ( ex) and exchange contribution to traverse relaxation (R ex) were extracted from T 1, T1 and hNOE data. Peak # S 2 ex, ps R ex, s-1 1 1.000 2 0.990 3 0.000 4 0.000 5 0.92160.20 6 0.908446.35.5 7 0.89374.00 8 0.000 9 0.688708.60 10 0.957158.13.8 12 0.9030137.9 13 0.929433.412.5 14 1.000 16 0.947317.6 17 0.859493.917.6 19 1.000 20 1.000 21 1.000 22 0.934616.3 23 1.000 24 0.926523.8 25 0.806540.319.9 26 0.870278.40 27 0.000 28 1.000 29 0.889780.40 30 0.854484.00 31 1.00000 32 0.779626.00 33 1.00000 34 0.90180.40 35 0.92400 36 0.854136.20 37 1.00006.5 39 0.90326.30 40 0.90526.80 41 0.90526.80 No satisfactory model could be fit for peak s assigned 3, 4, 8, 11, 15 and 18 using the combinations of S 2, e, and R ex and an isotropic model for overall rotation of the system. Fexanigure 4-5: Hxchange connalyzed by M Histogram rntribution (RModel free foepresentati Rex) derive dormalism. 150on of order d from rela x parameter xation meas (S 2) and surements aas 151From the R 1 dispersion experiment we have repor ted on conformational exchange on several C8/C6 resonances in loop A ( Figure 4-7 , Table 4-3). From this data, we report best-fit values of ex, as well as minimum values of , the chemical shift differen ce between the two exchanging populations, calculated by setting Pa = P b = 0.5. C8/C6 resonances represented by peaks 3, 12 and 17 reported conformational exchange as s hown in the power-depende nt dispersion curves (Figure 4-7 ). The remaining resonances, by contrast, showed transverse relaxation rates that were independent of the applied B 1 power, i.e., an absence of disp ersion, as expected from the previous results of Model free analysis. This im plies that conformational exchange in the µs-ms timescale only occurs in specific residues, presumably situated in the non-helical regions of loop A. Peak 3 reported an exchange time of 85 µs whereas peaks 13 and 17 both reported an exchange time of 667 µs suggestin g that there exchange may be correlated. Details for all the resonances exchange parameters have been tabulated ( Table 4-3). Figure 4-GAAA telimit A2 d-6: Simulatetraloop in vdynamics. ion structu rvolving A2 s152re of loop Astacked be tA hi ghlighttween A1 a nting base st and A3. Thesacking in t hse interactio he ns Figresexcgure 4-7: Resonances. Pechange (A, C epresentati veaks 3, 12 anC and D) whve relaxationd 17 show phereas peak 1153n dispersionpower-depen10 (B) does nn curves of ndent confornot undergoselected C8rmational o exchange. 8 154Table 4-3: Chemical exchange parameters from T1 dispersion analysis of loop A. was calculated from PaPb()2 using Pa = Pb = 0.5 Peak # PaPb()2 ex (µs) min (ppm) 1 2.27 x 10 6 693.0 3.20 2 4.85 x 10 5 14.2 1.48 3 3.28 x 10 5 84.5 1.22 4 6.69 x 10 4 579.7 0.55 5 - - - 6 - - - 7 1.96 x 10 5 297.4 0.94 8 4.02 x 10 5 618.4 1.35 9 - - - 10 - - - 11 - - - 12 1.40 x 10 6 666.7 2.51 13 1.40 x 10 6 667.7 2.51 14 1.66 x 10 6 676.0 2.74 15 1.64 x 10 5 6.8 0.86 16 1.45 x 10 6 426.1 2.56 17 2.62 x 10 6 667 3.44 18 9.53 x 10 5 40.1 2.07 19 1.76 x 10 6 833.5 2.82 20 - - - 21 - - 22 1.11 x 10 6 544.8 2.24 23 1.33 x 10 6 32.6 2.45 24 1.08 x 10 6 32.6 2.21 25 5.13 x 10 6 8.3 1.52 26 - - - 27 8.39 x 10 5 386.2 1.94 28 - - - 29 4.32 x 10 5 314.9 1.40 30 - - - 31 - - - 32 - - - 33 1.38 x 10 6 25.5 2.49 34 1.21 x 10 6 67.1 2.34 35 - - - 36 1.35 x 10 6 1506 2.47 37 1.76 x 10 6 595.2 2.82 38 8.46 x 10 5 284.8 1.95 39 9.49 x 10 5 318.6 2.07 155In an NMR relaxation experiment, confor mational exchange in the microsecond to millisecond timescale causes increased line-br oadening of NMR signals by an amount (R ex) that contributes to the measured overa ll transverse relaxation rate (R 2eff). From R ex three physical parameters for a two-site dynamic process ca n be obtained: rates of interconversion (k ex), the relative populations of the exchanging species ( Pa and Pb) and the chemical shifts between the exchanging species ( ). In this work, we have assesse d conformational dynamics of loop A using the structural disorder (ps-ns) parameter extracted using longitudinal relaxation, traverse relaxation and heteronuclear NOE. The dynamics reported for various some loop residues showed a wide variety of timescales ranging from picosecond to millisecond timescale. The diverse timescales reported fo r loop A residues provi de us with insights for understanding the energy landscape of RNA. ACKNOWLEDGMENT I would like to thank Max T. Rogers NMR staf f Kermit Johnson, Dr. Daniel Holmes and Dr. Li for assistance in using 600 MHz spectrometer. I would also like to thank Dr. James Johnson Jr. for programming the R 1 equation in Igor Pro (WaveMetric) software. This facilitated the analysis of all the R 1 reported in this thesis. 156REFERENCES 157REFERENCES 1. Hammes-Schiffer, S. (2002) Impact of enzyme motion on activity, Biochemistry 41, 13335-13343. 2. Al-Hashimi, H. M. (2005) Dynamics-based amplification of RNA function and its characterization by using NMR spectroscopy, Chembiochem 6, 1506-1519. 3. Boehr, D. D., McElheny, D., Dyson, H. J., and Wright, P. E. (2006) The dynamic energy landscape of dihydrofolate reductase catalysis, Science 313, 1638-1642. 4. Beach, H., Cole, R., Gill, M. L., and Loria, J. P. (2005) Conservation of mus-ms enzyme motions in the apo- and substrate-mimicked state, J Am Chem Soc 127, 9167-9176. 5. Wolf-Watz, M., Thai, V., Henzler-Wildman, K ., Hadjipavlou, G., Eisenmesser, E. Z., and Kern, D. (2004) Linkage between dynamics and catalysis in a th ermophilic-mesophilic enzyme pair, Nat Struct Mol Biol 11, 945-949. 6. Leulliot, N., and Varani, G. (2001) Current topics in RNA-protein recognition: control of specificity and biological function through induced fit and conformational capture, Biochemistry 40, 7947-7956. 7. Treiber, D. K., and Williamson, J. R. (2001) Beyond kinetic traps in RNA folding, Curr Opin Struct Biol 11, 309-314. 8. Fedor, M. J., and Williamson, J. R. ( 2005) The catalytic diversity of RNAs, Nat Rev Mol Cell Biol 6, 399-412. 9. Kuimelis, R. G., and McLaughlin, L. W. (1998) Mechanisms of Ribozyme-Mediated RNA Cleavage, Chem Rev 98, 1027-1044. 10. Ke, A., Zhou, K., Ding, F., Cate, J. H., and Doudna, J. A. (2004) A conformational switch controls hepatitis de lta virus ribozyme catalysis, Nature 429, 201-205. 11. Cai, Z., and Tinoco, I., Jr. (1996) Solution st ructure of loop A from the hairpin ribozyme from tobacco ringspot virus satellite, Biochemistry 35, 6026-6036. 12. Rupert, P. B., and Ferre-D'Amare, A. R. (2001) Crystal structure of a hairpin ribozyme- inhibitor complex with implications for catalysis, Nature 410, 780-786. 13. Butcher, S. E., Allain, F. H., and Fei gon, J. (1999) Solution st ructure of the loop B domain from the hairpin ribozyme, Nat Struct Biol 6, 212-216. 15814. Sumita, M., White, N. A., Julien, K. R., and Hoogstraten, C. G. ( 2013) Intermolecular domain docking in the hairpin ribozyme Metal dependence, binding kinetics and catalysis, RNA Biol 10, 425-435. 15. Palmer, A. G., 3rd. (2004) NMR characteriza tion of the dynamics of biomacromolecules, Chem Rev 104, 3623-3640. 16. Palmer, A. G., 3rd. (1997) Probing molecular motion by NMR, Curr Opin Struct Biol 7, 732-737. 17. Akke, M. (2002) NMR methods for characterizing microsecond to millisecond dynamics in recognition and catalysis, Curr Opin Struct Biol 12, 642-647. 18. Palmer, A. G., 3rd. (2004) NMR characteri zation of the dynamics of biomacromolecules, Chem Rev 104, 3623-3640. 19. Palmer, A. G., 3rd, and Massi, F. ( 2006) Characterization of the dynamics of biomacromolecules using rotating-frame spin relaxation NMR spectroscopy, Chem Rev 106, 1700-1719. 20. Hoogstraten, C. G., Wank, J. R., and Pardi, A. (2000) Active site dynamics in the lead- dependent ribozyme, Biochemistry 39, 9951-9958. 21. Palmer, A. G., 3rd, Kroenke, C. D., and Lori a, J. P. (2001) Nuclear magnetic resonance methods for quantifying microsecond-to-millisecond motions in biological macromolecules, Methods Enzymol 339, 204-238. 22. Korzhnev, D. M., Orekhov, V. Y., Dahlquist, F. W., and Kay, L. E. (2003) Off-resonance R1rho relaxation outside of the fast exchan ge limit: an experimental study of a cavity mutant of T4 lysozyme, J Biomol NMR 26, 39-48. 23. Kim, S., and Baum, J. (2004) An on/off resonance rotating frame relaxation experiment to monitor millisecond to microsecond timescale dynamics, J Biomol NMR 30, 195-204. 24. Nikonowicz, E. P., Sirr, A., Legault, P., Juck er, F. M., Baer, L. M., and Pardi, A. (1992) Preparation of 13C and 15N labelled RNAs for heteronuclear multi-dimensional NMR studies, Nucleic Acids Res 20, 4507-4513. 25. Batey, R. T., Inada, M., Kujawinski, E., Puglisi, J. D., and Williamson, J. R. (1992) Preparation of isotopically labeled ribonucleotides for multidimensional NMR spectroscopy of RNA, Nucleic Acids Res 20, 4515-4523. 26. Akke, M., Fiala, R., Jiang, F ., Patel, D., and Palmer, A. G ., 3rd. (1997) Base dynamics in a UUCG tetraloop RNA hairpin characterized by 15N spin relaxation: correlations with structure and stability, RNA 3, 702-709. 159 27. Hall, K. B., and Tang, C. (1998) 13C relaxati on and dynamics of the purine bases in the iron responsive element RNA hairpin, Biochemistry 37, 9323-9332. 28. Vallurupalli, P., and Kay, L. E. (2005) A su ite of 2H NMR spin relaxation experiments for the measurement of RNA dynamics, J Am Chem Soc 127, 6893-6901. 29. Blackledge, M. J., Bruschweiler, R., Griesinger, C., Schmidt, J. M., Xu, P., and Ernst, R. R. (1993) Conformational backbone dynamics of the cyclic decapeptide antamanide. Application of a new multiconformational search algorithm based on NMR data, Biochemistry 32, 10960-10974. 30. Milligan, J. F., Groebe, D. R., Wither ell, G. W., and Uhlenbeck, O. C. (1987) Oligoribonucleotide synthesi s using T7 RNA polymerase and synthetic DNA templates, Nucleic Acids Res 15, 8783-8798. 31. Milligan, J. F., and Uhlenbeck, O. C. ( 1989) Synthesis of small RNAs using T7 RNA polymerase, Methods Enzymol 180, 51-62. 32. Kim, I., McKenna, S. A., Viani Puglisi, E., a nd Puglisi, J. D. (2007) Rapid purification of RNAs using fast performance liquid chromatography (FPLC), RNA 13, 289-294. 33. Yamazaki, T., Muhandiram, R., and Kay, L. E. (1994) Nmr Experiments for the Measurement of Carbon Relaxation Proper ties in Highly Enriched, Uniformly C-13,N- 15-Labeled Proteins - Applic ation to C-13(Alpha) Carbons, J Am Chem Soc 116, 8266-8278. 34. Viles, J. H., Duggan, B. M., Zaborowski, E ., Schwarzinger, S., Hun tley, J. J., Kroon, G. J., Dyson, H. J., and Wright, P. E. (2001) Potential bias in NMR relaxation data introduced by peak intensity analysis and curve fitting methods, J Biomol NMR 21, 1-9. 35. Davis, D. G., Perlman, M. E., and London, R. E. (1994) Direct measurements of the dissociation-rate constant for inhibitor-enzyme complexes vi a the T1 rho and T2 (CPMG) methods, J Magn Reson B 104, 266-275. 36. Lipari, G., and Szabo, A. (1982) Model-Free Approach to the Inte rpretation of Nuclear Magnetic-Resonance Relaxation in Macromol ecules .1. Theory and Range of Validity, J Am Chem Soc 104, 4546-4559. 37. Lipari, G., and Szabo, A. (1982) Model-Free Approach to the Inte rpretation of Nuclear Magnetic-Resonance Relaxation in Macromol ecules .2. Analysis of Experimental Results, J Am Chem Soc 104, 4559-4570. 16038. Mandel, A. M., Akke, M., and Palmer, A. G. (1995) Backbone Dynamics of Escherichia- Coli Ribonuclease Hi - Correlati ons with Structure and Func tion in an Active Enzyme, J Mol Biol 246, 144-163. 39. Kay, L. E. (1998) Protein dynamics from NMR, Nat Struct Biol 5 Suppl, 513-517. 40. Legault, P., Hoogstraten, C. G., Metlitzky, E., and Pardi, A. (1998) Order, dynamics and metal-binding in the lead-dependent ribozyme, J Mol Biol 284, 325-335. 41. Heus, H. A., and Pardi, A. (1991) Structural Features That Give Rise to the Unusual Stability of Rna Hairpins Containing Gnra Loops, Science 253, 191-194. 42. Hoogstraten, C. G., Legault, P., and Pardi, A. (1998) NMR solution structure of the lead- dependent ribozyme: Evidence for dynamics in RNA catalysis, J Mol Biol 284, 337-350. 43. Eisenmesser, E. Z., Bosco, D. A., Akke , M., and Kern, D. (2002) Enzyme dynamics during catalysis, Science 295, 1520-1523. 44. Tollinger, M., Skrynnikov, N. R., Mulder, F. A. A., Forman-Kay, J. D., and Kay, L. E. (2001) Slow dynamics in folded and unfolded states of an SH3 domain, J Am Chem Soc 123, 11341-11352. 45. Millet, O., Loria, J. P., Kroenke, C. D., P ons, M., and Palmer, A. G. (2000) The static magnetic field dependence of chemical exchange linebroadening defines the NMR chemical shift time scale, J Am Chem Soc 122, 2867-2877. 46. Cole, R., and Loria, J. P. (2002) Evidence for flexibility in the function of ribonuclease A, Biochemistry 41, 6072-6081. 47. Blad, H., Reiter, N. J., Abildgaard, F., Ma rkley, J. L., and Butcher, S. E. (2005) Dynamics and metal ion binding in the U6 RNA intramolecular stem-loop as analyzed by NMR, J Mol Biol 353, 540-555. 48. Johnson, J. E., Jr., and Hoogstraten, C. G. (2008) Extensive backbone dynamics in the GCAA RNA tetraloop analyzed using 13C NM R spin relaxation and specific isotope labeling, J Am Chem Soc 130, 16757-16769. 161 CHAPTER 5 SUMMARY, DISCUSSION AND FUTURE WORK 162 5.1 SUMMARY AND DISCUSSION The projects pursued in this thesis were in spired by the overarching idea of investigating and correlating dynamics in RNA with possible function. Using molecular dynamics simulation and NMR spin relaxation techniques, we interroga ted residue dynamics effects in loop A of the hairpin ribozyme and reported our results herein . The combination of computational (molecular dynamics simulation) and NMR techniques provides a robust dynamics probe for macromolecules. Molecular dynamics simulations play a sign ificant role in the interpretation of experimental data bearing on the time-depende nt properties of biomolecules. MD has increasingly become common and suitable with th e improvement of force fields and computer resources. This has improved the size of tracta ble molecules and the length of time that a molecule can be simulated. Longer simulati ons can more thoroughly explore conformational sampling and thus the conformational space av ailable for the molecule. Molecular dynamics simulations could be useful in interpreting NMR experiments, in cluding for fast motions seen primarily in NMR relaxation measurements ( 1, 2). NMR spectroscopy, on the other hand, detects motions across a broad range of molecular motion timescales ( 3). While studies of molecular motions have been comprehensively done in proteins using combined computer simulations and NMR ( 4-7) , these motions have been less studied in RNA molecules despite their diverse nature. One major challenge that ha s partly hindered the co mputational studies of RNA is the force field limitations. Most of the cu rrent force fields have not been successful in supporting long simulations of RNA. An all-at om CHARMM36 force field has recently been developed based on cross-validation with NMR data ( 8). Motions have also been probed using NMR in some RNA molecules with the focus on base dynamics ( 9-12) and also ribose dynamics 163 (13). We have integrated computer simulation and NMR techniques to understand broad aspects of RNA dynamics and interaction. The evidence of active site rearrange ment in the hairpin ribozyme loops, based on structural data, suggest s mobility of certain re sidues in the activation process. Using combined analysis of simulatio ns and NMR to analyze RNA molecules, the role of dynamics can be demonstrated for RNA mo lecules during RNA folding or catalysis. In Chapter Two of this Thesis, MD simulation was used to determine conformational heterogeneity in RNA based on alte rnate base-pair formation within a subset of residues in the loop region of domain A of the hairpin ribozyme. The MD data was primarily acquired using CHARMM36 force field after force field comparison with amber ff10 force field indicated better structural overlap with CHARMM36 simulation. The observed conformers were extensively mapped and determined to form predominantly in the active site of loop A. Three main conformers and several minor conformations were observed in our simulations as analyzed by the Markov State model analysis. These three majo r conformers depict the local energy minima wells where these conformers reside within whic h there exist several minor states that undergo kinetic transitions within these local minima based on subtle structur al changes and dynamics within a set of residues. RNA base residues and backbone dynamics played a major role in alternate base-pair formation a nd conformational heterogeneity in loop A. This suggests that conformational sampling and transition in loop A R NA is a key strategy to avoiding the kinetic traps that localize RNA in non-functional conforma tions. Kinetic transitions were achieved by loop A sampling a series of kinetic states which function to populate other conformers by accessing kinetic energy barriers th at may limit the formation of f unctional (active) RNA. Of the conformations that were sampled, the most populated conformer, AA/CA, closely sampled conformational properties similar to the activated (docked) l oop A conformation, suggesting the 164 activating role induced by confor mational heterogeneity. This obser vation highlights the role of conformational sampling in pre-organizing and activating loop A (conformational selection) for tertiary RNA-RNA interactions. The slow doc king association rates reported for hairpin ribozyme interaction ( 14) implies that some structural rea rrangements occur in the independent loops (loops A and B) that initiate or stabilize the tertiary interacti ons necessary for docking together. This unique intrinsic base-pair rearrangeme nt within a subset of kinetic states along the potential energy surface supports the descri ption of a rugged but accessible free energy landscape of RNA with mechanistic properties that di rect pre-organization to the activated state. The multiple conformers and inter-conformer transitions observed lay a foundation for understanding tertiary RNA interactions in hairpin ribozyme a nd RNA. Thus far, we have made significant correlations betw een the dynamics of RNA re sidues and conformational heterogeneity that we observed in loop A in the context of active site formation. In Chapter Three, we compared the docking transitions between wild type and mutant loop A using a circular dichroism assay. Dock ing was abrogated in the U+2C/C+3U mutant, which has additional two Watson- Crick base-pairs, as expected . This suggests the role of dynamics in facilitating docking. Because the U+2C /C+3U mutant is unable to form the docked structure despite docking w ith a U+2C single mutant ( 15), it is plausible that the U+2C/C+3U mutant reduces or quenches docki ng-activating dynamics. Important cytosine and uracil residues which we observed to be dynamic in the chapter 2 of this thesis are imm obilized by base-pairing. It is also possible, that the constraint imparted on loop A by additional base-pairs could have inactivated the mutant. We also determined th e suitable loop A construct for NMR studies by a series NMR studies on various l oop A constructs. The best-behaved loop A construct was used for resonance assignments of 1H, 13C, and 15N nuclei with their chemical shifts reported in the 165 results section of this chapter. Using excha ngeable and non-exchangeable NMR experiments we have assigned significant resonances of loop A for the NMR-active nuclei. In Chapter Four, loop A dynamics were stud ied using the NMR relaxation measurements namely longitudinal relaxation (T 1), transverse relaxation (T 1) and heteronuclear NOE (hNOE) using a 13C-uniformly labeled sample. Fast internal motions in the order of ps - ns timescale were analyzed using Lipari & Sza bo™s Model-free approach by ( 16-18) with data from 13C R1, R1, and heteronuclear NOE of loop A. Loop A was ge nerally found to be a rigid molecule on this timescale with internal generalized order parameters, S 2, of at least 0.9 in th e helical regions as expected. Some atoms, however, were determined to have internal, eff ective correlation times indicative of fast motions on the ps to ns times cale, while others reporte d slow exchange in the µs-ms timescale. A wide variety of internal mo tions were observed in the non-helical portions of loop A, especially in the GAAA tetraloop and th e loop region. The loop region constitutes the cleavage site of the hairpin ribozyme, and therefore it is plausible that these dynamics may play a role in accessing various energy we lls in the complex energy lands cape of loop A, in parallel with the calculational results described above. These internal motions time scales suggest a complex landscape of accessible states and pot ential correlations be tween observed motions. 5.2 FUTURE DIRECTION In this thesis we have reported significant advancements in the analysis of RNA using integrated computational and NM R approach. However, questions still remain on some aspects of RNA dynamics and how these dynamics relate to RNA function. Below are insights into the future direction that will help us gain a d eeper understanding of dyna mics-function relationship in RNA. 166 5.2.1 Base dynamics for the hairpin ribozyme We have analyzed the base dynamics in the isolated loop A of the hairpin ribozyme using 13C uniformly-labeled sample for the side-chain C8 /C6/A-C2 sites. This represents the dynamics happening in the nucleic acid side chains. Molecular dynamics st udies suggest certain dynamics reported for loop residues that seem to be corr elated with base dynamics. To comprehensively investigate ribose dynamics in these residue s, we will adopt the ribose specific 2',4'- 13C labeling technique (19) to assess ribose backbone dynamics. Using model-free analysis of disorder on the ps-ns scale and multiple-field relaxation dispersion analysis of R 1 and CPMG data will help obtain 13C ribose relaxation rates for loop A residues that have shown chemical exchange. Our relaxation data mostly displays repressed transv erse relaxation rates in the internal loop, which supports conformational exchange in this region of the molecule. The assessment of dynamics in isolated domains can also be extended to loop B, whose structure has previously solved by NMR ( 20). Maintaining the construc t whose NMR structure is available will be important due to the availability of resonance assignments. Similarly, base and ribose dynamics should be obtained by relaxation measurements ( T1, T1 and hNOE) and relaxation dispersion to obtain dynamic rates and parameters in a variety of time regimes (ps- ms). With the ground state dynamics defined, activ e site perturbations due to docking can be probed using labeled domain A with unlabeled domain B in th e presence and absence of [Co(NH3)6]3+. To prevent cleavage, an A38U mutant ma y be used instead of wild type loop A since it supports docking but no t cleavage within the timescale of NMR experiments. Transitions relevant after during and after the docking process can be captured and analyzed in the context of available data. Such experiments also can defini tively probe timescale of docking perturbations. 167 The work detailed in this thesis investig ated dynamics in loop A RNA molecules and the results obtained reveal an integrated but complex conformational sampling and transitions throughout the RNA free energy lands cape. This complexity presen ts a clearer picture of the structural versatility of RNA in fast and intermed iate timescales relevant in the catalytic cycle of ribozymes. 5.2.2 Functional read-out of dynamics Previously, it has been difficult to probe directly the functiona l relevance of a particular RNA conformational change or dynamic region. Our lab has recently developed a specific isotope-labeling scheme that labels alternat e ribose carbons specifically to allow for ribose dynamics analysis using NMR relaxation ( 21). RNA labeled as such can be used to unambiguously probe ribose pucker C3™-endo/C2 ™-endo inter-conversions. The major question remains how we can determine the functional re levance of such interconversion? Our lab has recently demonstrated that we can preferentia lly perturb RNAs ribose pseudo-rotation using locked nucleic acids (LNA), which features a methylene linkage between 2' oxygen and 4' carbon, to completely restrict th e nucleotide to a C3'-endo, and assay for functional effects of such perturbation ( 22). This technique has been used with great success to probe three systems with distinct types of conforma tional change: the fluctuation of an inactive ground state structure of a lead-dependent ribozyme to a low-population active confor mation of an unknown structure, the unfolding of a UUCG tetraloop, and the rear rangement of the U1 snRNA hairpin II (U1hpII) from a free conformation to a protein-bound form ( 22). From these experiments Julien et al showed that the imposition of a C3'-endo conf ormation using LNA at sites U6 of the UUCG tetraloop and C5 of U1hpII (both C2™-endo) resulted in a drastic decrease in the thermodynamic stability and protein binding respectively. By contrast, LNA substitution at G7 of the leadzyme 168 and A6 of U1hpII showed only minor effects im plying that ribose conf ormational changes at these sites was not essential for the formation of the active state of leadzyme or for protein recognition by U1hpII. There was also a 20-fold incr ease in catalytic rate in the leadzyme upon LNA substitution at G9 which suggests the conformational activation by dynamics regulation. This phenomenon supports the idea of active and inactive conformational st ates observed due to dynamics in Chapter Two of this thesis. This technology presents a platform for mapping out conformational changes towards the activation of the hairpin ribozymes. LNAs have also previously been incorporated into helical regi ons and recognition arms of 10-23 catalytic DNA and the hammerhead ribozymes in an effort to improve targeting of cellular and RNA sequences resulting in dramatic effects upon struct ural stability and target affinity ( 23, 24). We can therefore employ LNA at specific dynamic residues to assess function that is coupled with ribose dynamics. This conformational change may be due to the pseudo-rotation angle inversion which couples to the RNA backbone dynamics. In the docke d structure of the hairpin ribozyme G+1 is extruded from the helical stack of loop A ( 25), and is stabilized by bu rial in its cognate loop B pocket. This constrains the ri bose of both A-1 and G+ 1 into C2™-endo pucker s. Despite the C2™- endo pucker retention in G+1 between undocked ( 26) and docked loop A ( 25) structures, the ribose pucker of the adjacent nucleotide A-1 drama tically shifted from a C3'-endo to a C2'-endo upon docking. This suggests pucker rearrangement of A-1 to ac commodate the docked state conformation. Similar pucker in terconversions were observed in G8, C+3 and U+2 residues in loop A. Our simulation studies have also observe d complementary pucker transitions in residues A7, G8, A9, G+1, U+2 and C+3, some being coupled to base-flipping. These residues provide us with initial points to begin the systematic functional probi ng of pucker dynamics using LNA substitution. These perturbations can be inve stigated by various assays including binding and 169 kinetic with relevant controls. Mutations or modi fications have the limitation of the possibility of functional effects arising from features of the probe other th an the intended perturbation. For LNAs, this possibility may be due to the lo ss of hydrogen-bonding groups on the 2'-hydroxyl or to the steric effect resulting from the bridging methylene group. A partial control for these effects should employ complementary anal ysis of 2'-deoxy and/or 2'- O-methyl modifications ( 22). In cases for which either of these modifications is poorly tolerated, the usefulness of LNA probing is reduced. Ideally, this technique would be well complemented by the possibility of locking ribose to the C2'-endo. Data from both C3' and C2' endo-locked residues would probe the ribose dynamics unambiguously. Nevertheless, the ability to lock a single residue in one conformation (C3'-endo) still presents an important step to understanding dynamics in RNA. With the available biochemical, kinetic and structural information for hairpi n ribozyme, we can ultimately relate dynamics, structure and function in hair pin ribozyme and more generally small RNA catalysts. 170 REFERENCES 171 REFERENCES 1. Lipari, G., Szabo, A., and Levy, R. M. ( 1982) Protein Dynamics and Nmr Relaxation - Comparison of Simulations with Experiment, Nature 300, 197-198. 2. Levy, R. M., Karplus, M., and Wolynes, P. G. (1981) NMR Relaxation Parameters in Molecules with Internal Motion - Exact La ngevin Trajectory Results Compared with Simplified Relaxation Models, J Am Chem Soc 103, 5998-6011. 3. Palmer, A. G., 3rd. (2004) NMR characteriza tion of the dynamics of biomacromolecules, Chem Rev 104, 3623-3640. 4. Ewald, C., Christen, M. T., Watson, R. P ., Mihajlovic, M., Zhou, T., Honegger, A., Pluckthun, A., Caflisch, A., and Zerbe, O. (2015) A Combined NM R and Computational Approach to Investigate Peptide Binding to a Designed Armadillo Repeat Protein, J Mol Biol 427, 1916-1933. 5. Liu, Q., Shi, C. W., Yu, L., Zhang, L. H., Xiong, Y., and Tian, C. L. (2015) General order parameter based correlation analysis of pr otein backbone motions between experimental NMR relaxation measurements and molecular dynamics simulations, Biochem Bioph Res Co 457, 467-472. 6. Asami, S., Porter, J. R., Lange, O. F., a nd Reif, B. (2015) Access to C alpha Backbone Dynamics of Biological Solids by C-13 T- 1 Relaxation and Molecular Dynamics Simulation, J Am Chem Soc 137, 1094-1100. 7. Allner, O., Foloppe, N., and Nilsson, L. (2015) Motions and Entropies in Proteins as Seen in NMR Relaxation Experiments and Molecular Dynamics Simulations, J Phys Chem B 119, 1114-1128. 8. Huang, J., and MacKerell, A. D. (2013) CHARMM36 all-atom additive protein force field: Validation based on comparison to NMR data, J Comput Chem 34, 2135-2145. 9. Hoogstraten, C. G., Wank, J. R., and Pardi, A. (2000) Active site dynamics in the lead- dependent ribozyme, Biochemistry 39, 9951-9958. 10. Blad, H., Reiter, N. J., Abildgaard, F., Ma rkley, J. L., and Butcher, S. E. (2005) Dynamics and metal ion binding in the U6 R NA intramolecular stem-loop as analyzed by NMR, J Mol Biol 353, 540-555. 11. Shajani, Z., and Varani, G. (2005) 13C NMR relaxation studies of RNA base and ribose nuclei reveal a complex pattern of motions in the RNA binding site for human U1A protein, J Mol Biol 349, 699-715. 172 12. Shajani, Z., and Varani, G. (2008) 13C rela xation studies of the DNA target sequence for hhai methyltransferase reveal unique motional properties, Biochemistry 47, 7617-7625. 13. Johnson, J. E., Jr., and Hoogstraten, C. G. (2008) Extensive backbone dynamics in the GCAA RNA tetraloop analyzed using 13C NMR spin relaxation and specific isotope labeling, J Am Chem Soc 130, 16757-16769. 14. Sumita, M., White, N. A., Julien, K. R., a nd Hoogstraten, C. G. (2013) Intermolecular domain docking in the hairpin ribozyme Metal dependence, binding kinetics and catalysis, RNAa Biol 10, 425-435. 15. Sargueil, B., Hampel, K. J., Lambert, D., and Burke, J. M. (2003) In vitro selection of second site revertants analysis of the hairpin ribozyme active site, J Biol Chem 278, 52783-52791. 16. Lipari, G., and Szabo, A. (1981) Nuclear ma gnetic resonance relaxation in nucleic acid fragments: models for internal motion, Biochemistry 20, 6250-6256. 17. Lipari, G., and Szabo, A. (1982) Model-Free Approach to the Interpretation of Nuclear Magnetic-Resonance Relaxation in Macromol ecules .1. Theory and Range of Validity, J Am Chem Soc 104, 4546-4559. 18. Lipari, G., and Szabo, A. (1982) Model-Free Approach to the Interpretation of Nuclear Magnetic-Resonance Relaxation in Macrom olecules .2. Analysis of Experimental Results, J Am Chem Soc 104, 4559-4570. 19. Johnson, J. E., Jr., Julien, K. R., and Hoogstr aten, C. G. (2006) Alternate-site isotopic labeling of ribonucleotides for NMR studies of ribose conformational dynamics in RNA, J Biomol NMR 35, 261-274. 20. Butcher, S. E., Allain, F. H., and Feigon, J. (1999) Solution structure of the loop B domain from the hairpin ribozyme, Nat Struct Biol 6, 212-216. 21. Johnson, J. E., Jr., Julien, K. R., and Hoogstr aten, C. G. (2006) Alternate-site isotopic labeling of ribonucleotides for NMR studies of ribose conformational dynamics in RNA, J Biomol NMR 35, 261-274. 22. Julien, K. R., Sumita, M., Chen, P. H., Lair d-Offringa, I. A., and Hoogstraten, C. G. (2008) Conformationally restricted nucleo tides as a probe of structure-function relationships in RNA, RNA 14, 1632-1643. 23. Christiansen, J. K., Lobedanz, S., Arar, K., Wengel, J., and Vester, B. (2007) LNA nucleotides improve cleavage efficiency of singular and binary hammerhead ribozymes, Bioorg Med Chem 15, 6135-6143. 173 24. Donini, S., Clerici, M., Wengel, J., Vester, B., and Peracchi, A. (2007) The advantages of being locked. Assessing the cleavage of short and long RNAs by locked nucleic acid-containing 8-17 deoxyribozymes, J Biol Chem 282, 35510-35518. 25. Rupert, P. B., and Ferre-D'Amare, A. R. ( 2001) Crystal structure of a hairpin ribozyme-inhibitor complex with implications for catalysis, Nature 410, 780-786. 26. Cai, Z., and Tinoco, I., Jr. (1996) Solution st ructure of loop A from the hairpin ribozyme from tobacco ringspot virus satellite, Biochemistry 35, 6026-6036.