PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

{EB 4) '32 ’sz‘oriif
NOV 0 4 2003

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 (flammfnfl-DJS

X-RAY CRYSTALLOGRAPHIC STUDIES OF CY TOCHROME C OXIDASE
FROM RHODOBACTER SPHAEROIDES

By

Ling Qin

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirement
for the degree of

DOCTOR OF PHILOSOPHY
Department of Biochemistry and Molecular Biology

2005

ABSTRACT
X-RAY CRYSTALLOGRAPHIC STUDIES OF CYTOCHROME C OXIDASE
FROM RHODOBACT ER SPHAEROIDES
By
Ling Qin

Cytochrome c oxidase (CcO) catalyzes the final reaction of aerobic respiration
by accepting electrons from its immediate donor, cytochrome c, and passing them
onto oxygen to form water. It also translocates protons from the inside of the
mitochondrial inner membrane of eukaryotes or the periplasmic membrane of
prokaryotes to the outside. The proton gradient formed across the membrane is
utilized by ATP synthase to make ATP, a universal energy source. Efforts were
directed toward developing the methodology of protein production, purification and
crystallization in order to reproducibly obtain crystals of cytochrome c oxidase from
Rhodobacter sphaeroides (RchO), a bacterial homologue of the mammalian enzyme.
High resolution structures are needed for progress in understanding the proton
pumping mechanism.

By molecular engineering of various strain of Rhodobacter sphaeroides (R. s.)
with different forms of subunits and histidine tag positions, modifying detergent
solublization and protein purification protocols, and exploring different crystallization
and cryocooling conditions, crystals of the four subunit RchO were obtained and the
crystal structure was solved to 2.9 A resolution. A new form of the crystal containing

the catalytic core of the oxidase, subunits I and II, was also obtained with isotropic

X-ray diffraction to 2.35 A resolution from an R. 5. strain with a C-terminal histidine
tag on a shortened subunit II, and a short homogeneous version of subunit IV. The
re-dissolved I-II subunit crystals were highly active, had native spectral characteristics,
and displayed “suicide inactivation” during steady state turnover as previously
observed for cytochrome c oxidase without subunit III. In the 1-H subunit RchO
structure, the histidine tag attached to subunit II was found to chelate cadmium and
contribute to strong crystal contacts. The crystal structure clearly revealed the
presence of the previously observed covalent bond between the side chain ring atoms
of H284; and Y2881 close to the active site of the enzyme. In both crystal forms, a
cadmium binding site was found at the matrix side of the enzyme at the suggested
entrance for K-pathway for proton uptake, which could contribute to the observed
inhibitory effect of zinc/cadmium on detergent-solublized enzyme. Detergents,
detergent headgroups, and alkyl chains of detergents or lipids were also observed in
the crystal structure of 1-H subunit Rch0. The detergent sugar head groups are
associated with aromatic residues in a manner similar to phospholipid head groups.
Comparison to other CcO crystal structures reveals that alkyl chain positions of
membrane lipids and detergent substitutes are conserved, suggesting their importance
for obtaining well-ordered crystals. The method developed is readily reproducible and
should be applicable to crystallizing other forms and key site-directed mutants of
RchO, whose high resolution crystal structures will benefit our understanding of the

mechanism of this critical energy conserving process.

ACKNOWLEDGEMENTS

First and foremost, I would like to express my sincere gratitude to my graduate
advisor, Dr. Shelagh Ferguson-Miller, for her constant guidance and support. I am not
only greatly benefiting from her wealth of knowledge and experience in the field of
study, but also deeply inspired by her enthusiasm and determination in carrying out
scientific research.

I am grateful to my graduate committee members, especially Dr. R. Michael
Garavito for his constant education and guidance on everything from membrane
protein chemistry to x-ray crystallography; and to Dr. James Geiger, Dr. William
Smith and Dr. Honggao Yan, for their advice and support throughout my Ph. D.
program.

I am also indebted to my present and past lab mates for their help and
friendship. I thank Dr. Denise Mills and Dr. Carrie Hiser for their continuous help and
insightful discussions on my research and for reading through my dissertation. Dr.
Carrie Hiser also generated various strains of bacteria used in my dissertation research
and performed most of the protein gels. I thank Dr. Martyn Sharpe, Dr. Bryan
Schmidt and Dr. Yasmin Hilmi for their help and insightful discussions. Dr. Yasmin
Hilmi taught me initially to do protein purification and crystallization and she also
performed all the thin layer chromatography analysis of lipids of my samples. I thank
my fellow graduate students Jun Yang, Namjoon Kim, Jian Liu, Xi Zhang, Shujuan

Xu for their help and discussions. Xi Zhang did all the mass spectrometry analyses of

various samples from me. Jian Liu helped me with freezing crystals prior to
synchrotron trips and assisted me at synchrotron facility. I also thank all the past and
present undergraduate students who have helped me growing bacteria, making
reagents, and preparing cell membranes, particularly, Sarah House, Jamie Slater,
Robert Bauman, Matt Hawkins, Steve Kidd, and Sarah Cloutier.

I am also deeply thankful to past and present members of the Garavito Lab in
the BMB Department: Dr. Anne Mulichak for her help and education from
crystallization, freezing crystals, and data collection to structural refinement; Dr.
Rachel Powers, Christine Harman, Nicole Webb for their help and discussions on
crystallization, data collection and structural refinement.

I would also like to thank Dr. Kaillathe Padmanabhan (Pappan) for sharing his
insights on crystallography and structural refinement with me and helping me with
computer hardware and software problems.

I am also indebted to Dr. Margareta Svensson-Ek and Dr. So Iwata, both of
whom were at Uppsala University, Sweden, for their training on crystallization of
cytochrome c oxidase and for their great hospitality when I was visiting them.

I thank staff scientists at LS-CAT, DND-CAT, and SBC-CAT at Advanced
Photon Source, Argonne National Laboratory, particularly Dr. Joseph S. Brunzelle
and Dr. Zdzislaw Wawrzak for their training, support and helpful suggestions on data
collections.

And finally, I thank my parents and other family members for their support,

encouragement, and unconditioned love throughout these years.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE OF CONTENTS
LIST OF TABLES x
LIST OF FIGURES xi
ABBREVIATIONS xv
CHAPTER 1: INTRODUCTION 1
1.1 Structure and Mechanism of Cytochrome c Oxidase 1
1. 1 .1 Overall reaction 1
1.1.2 Current Understandings of the Structure/Function of Goo ----------- 4
1.1.2.1 Overall Structures of Subunits 4
1.1.2.2 Electron Transfer Pathways 12
1.1.2.3 Oxygen Chemistry 17
1.1.2.4 Proton Uptake Pathways 20
1.1.2.4.] D Pathway 2O
1.1.2.4.2 K Pathway 22
1.1.2.4.3 H Pathway in Bovine Heart Mitochondrial CcO -------------- 23
1.1.2.5 Oxygen Pathway 23
1.1.2.6 Proton and Water Exit Pathway 24
1.1.2.7 Proton Pumping Theories and Coupling of Electron Transfer and
Proton Translocation 25
1.1.2.8 Function of Subunit III of CcO 28
1.1.3 Regulation of CcO Activity and Energy Metabolism ----------------- 29
1.2 Membrane Protein Crystallography 32
1.2.1 Overview of Macromolecular Crystallography 32
1.2.2 Challenges in Membrane Protein Crystallography 33
1.2.2.1 Membrane Protein Production 33
1.2.2.2 Membrane Protein-Detergent Complexes 34
1.2.3 Progress in Membrane Protein Crystallization 37
1.2.3.1 Manipulation of Protein-Detergent Complex 38
1.2.3.2 Antibody Assisted Membrane Protein Crystallization ----------- 39
1.2.3.3 Crystallization of Membrane Protein in Lipid Cubic Phase ----- 39
1.2.4 Membrane Lipids and Membrane Protein Crystallization ------------- 40
1.2.4.1 Overview of Membrane Lipids 40
1.2.4.2 Membrane Lipids and Membrane Protein Crystallization -------- 41
1.2.4.3 Lipid Analysis of CcO Using Thin Layer Chromatography
and Mass Spectrometry 44
CHAPTER 2: METHODS 47

2.1 Molecular Engineering of Various Strains of Rhodobacter sphaeroides that

vi

2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11

Produce CcO with Various Subunit Contents

 

 

47

 

 

 

 

 

 

 

 

 

 

 

 

UV-Vrsible Spectroscopy 50
SDS-PAGE 50
Cytochrome c Oxidase Activity Assay 50
Protein Concentration Assay 51
Phosphorous Assay 51
Mass Spectrometry Analysis of Membrane Lipids 52
Growth and Harvest of R. 3. Cells 53
Preparation of R. s. Cytoplasmic Membranes 54
Detergent Solublization of R. s. Cytoplasmic Membrane 54
Column Chromatography for Purification of Enzyme
Used for Crystallization of the Four Subunit RchO 55
2.11.1 Purification of CcO with the Histidine-Tag Attached to
the C—Terminus of Subunit I for Crystallization 55
2.11.1.1 Ni-NTA Column Chromatography 55
2.11.1.2 Ion-exchange Column Chromatography Using Either
Mono Q Column or DEAE Sepharose Column 56
2.11.2 Purification of C00 with the Histidine-Tag Attached to the
Shortened C-Tenninus of Subunit II for Crystallization ----------- 57

2.12
2.13
2.14

2.15

2.16
2.17
2.18
2.19

 

2.11.2.1 Ni-NTA Column Chromatography
2.11.2.2 Ion-exchange Column Chromatography Using

 

DEAE Sepharose Column

2.11-2.3 Superdex 200 Size Exclusion Chromatography ---------------

57

59
59

 

Crystallization of the Four Subunit CcO
Flashcooling of the Four Subunit RchO Crystals

 

Soaking of Four Subunit RchO Crystals in Cadmium
Solutions Prior to Flashcooling

 

Column Chromatography for Purification of Enzyme Used

 

for Crystallization of the 1-11 Subunit RchO
Crystallization of the HI Subunit RchO

 

 

Flashcooling of the HI Subunit RchO Crystals
Data Collection and Processing

 

 

Molecular Replacement and Structural Refinement

 

 

 

 

 

CHAPTER 3: RESULTS
3.1 X-ray Crystallography of the Four Subunit RchO
3.1.1 Importance of Subunit IV in Crystallization of the Four Subunit
Rch0 and Overexpression of Subunit IV
3.1.2 Modifications of Detergent Solublization and Protein
Purification Procedures for Crystallization of CcO
3.1.3 UV/Visible Spectrum and Enzymatic Activity Assay of
Purified RchO
3.1.4 Retention of Subunit IV during the Purification and

 

Crystallization of RchO

vii

62
63
64
65
65

66

68
68

68

71

74

74

 

 

 

 

 

 

 

 

3.1.5 Factors Affecting X-Ray Diffraction Quality of RchO
Crystals and Systematic Improvements of Diffraction
Resolution Limits of RchO Crystals 78
3.1.5.1 Effect of Homogeneous Form of Subunit II 78
3.1.5.2 Effects of Different Forms of Subunit IV 81
3.1.5.3 Crystallization of Subunit II Histidine-Tagged Rch0
and its Crystal Structure 82
3.1.5.3.] Overview of the Subunit II Histidine-Tagged
Four Subunit RchO Structure 84
3.1.5.3.2 Partial Lipid Molecules 86
3.1.5.3.3 Potential Histidine-Tag Resolved at the Crystal
Contact Interface 88
3.1.5.4 Retention of Membrane Lipids in Detergent Solublization,
Protein Purification and Crystallization 91
3.1.5.4.] Identification of Lipid Species during Purification
and Crystallization 91

 

3.1.5.4.2 Quantitative Analysis of Phospholipid Content of C60
Samples by Phosphorous Assay and Inductively

 

 

 

 

 

Coupled Plasma Emission Spectroscopy 96
3.1.5.4.3 Modification of Protein Purification and Crystallization
Procedures to Retain More Bound Membrane Lipids -------- 98
3.1.5.4.3.1 Modification of Detergent Concentration used
for Membrane Solublization 100
3.1.5.432 Modification of Column Purification Methods ---------- 100
3.1.5.4.3.3 Modification of Detergent and Lipids in the
Crystallization Solution 105
3.1.5.5 Optimization of Detergent Choice, Crystallization Additives,
Crystallization and Flashcooling Procedures 106
3.1.5.5.] Optimization of Detergent(s) 106

3.1.5.5.2 Optimization of Crystallization Procedures and Additives - 110
3.1.5.5.3 Optimization of Flashcooling Procedures and

 

 

 

 

 

 

Cryoprotectants 1 10
3.1.5.6 Design of Site-Directed Mutants to Improve Crystal
Diffraction 112
3.1.5.7 Screening of New Crystallization Conditions and
New Crystal Forms 114
3.1.6 Potential Inhibition Sites of Zn2+/Ccl2+ and the Effects of ctl2+
Binding to CcO Crystals on X-ray Diffraction 116
3.1.6.1 Zn2+ / Cd2+ inhibition on CcO Activity 116
3.1.6.2 Potential Inhibition Sites of Zn2+ I ca2+ and the Effects
of ca2+ Binding to C00 Crystals on X-ray Diffraction ------- 120
3.2 X-ray Crystallography of [-11 Subunit Rch0 129
3.2.1 Engineering of an R. 6‘. Strain with a single Form of Subunit IV --- 129

3.2.2 Crystallization of I-11 Subunit Rch0 and Diffraction Quality

viii

 

of MI Subunit RchO Crystals 135

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.2.3 Biochemical Analysis of [-11 Subunit RchO Crystals -------------- 139
3.2.3.1 UV-visible Spectra and Activity Assays of Re-dissolved
I-II Subunit RchO Crystals 139
3.2.3.2 MALDI Mass Spectrometry Analysis of Lipids ---------------- 141
3.2.4 Crystal Structure of the I-11 Subunit Rch0 144
3.2.4.1 Overall Structure of I-11 Subunit Rch0 144
3.2.4.2 Crystal Packing of LI] Subunit Rch0 146
3.2.4.3 The Binuclear Center 149
3.2.4.4 Proton Uptake Pathways 151
3.2.4.5 Additional Cadmium Binding Site 155
3.2.4.6 Detergents and Lipids 157
CHAPTER 4: DISCUSSION 161
4.1 Importance of Homogeneous Subunits in the X-ray Diffraction
of RchO Crystals 161
4.2 Effects of the Histidine Tag and Metal Ions on X-ray Diffraction
of RchO Crystals 165
4.3 Effects of Membrane Lipids on the X-ray Diffraction Quality of
RchO Crystals 168
4.3.1 Importance of Lipid Retention in Obtaining High
Resolution Crystal of Rch0 168
4.3.2 Use of Mass Spectrometry in Monitoring Lipid Contents
during Protein Preparation of Rch0 for Crystallization ----------- 169
4.3.3 Conserved Lipid Binding Sites Found in the Structure of
I-11 Subunit Rch0 and Potential Substitution of Lipids
by Detergents 170
CHAPTER SzFUTUREPLANS 181
5.1 Improvement of the X-ray Diffraction of Crystals of RchO ----------------- 181
5.1.1 New Strains of Rhodobacter sphaeroides with Different
Subunit Contents 1 81
5.1.2 Screening New Conditions for Detergents, Protein Production,
Crystallization and Crystal Handling / Flashcooling ---------------- 182
5.1.3 Purification of [-11 Subunit RchO as a Crystallization Candidate - 183
5-2 Crystal Structure of RchO Complexed with Arachidonic Acid ----------- 184
5.3 Crystal Structures of Key Mutants of RchO 185
5.4 Kinetic and Crystallographic Studies of K-pathway Mutants Which
Abolish ca“ Binding 188
5.5 Summary 1 89

 

BIBLIOGRAPHY 1 90

 

Table 2.1

Table 3.1

Table 3.2

Table 3.3

Table 3.4

Table 3.5

Table 3.6

Table 3.7

Table 3.8

Table 4.1

Table 5.1

LIST OF TABLES

Different R. 6'. strains and their subunit compositions used
in this study

 

X-ray data collection and refinement statistics of RchO
crystals obtained from R. s. strain 167 and 169

49

85

 

Phosphorous contents of various RchO samples measured

 

by colorimetric method

Phosphorous contents of various RchO samples measured
using ICP

97

99

 

The effects of different secondary detergents on x-ray
diffraction resolution limit of the four subunit RchO
crystals

 

Different site-directed mutants of Rch0 to strengthen
crystal contacts and the x-ray diffraction resolution limit

 

of crystals of these mutants

X-ray data collection and refinement statistics of the four

subunit RchO which belong to a space group of P212121 ---------

X-ray data collection and refinement statistics of the four
subunit RchO crystals soaked in a solution containing
cadmium

 

X-ray data collection and refinement statistics of the HI
subunit RchO crystals

 

Summary of different R. s. strains with different types of
subunits H and IV in their expression products, their protein
expression levels and the diffraction resolution limits of

 

crystals obtained

Mutants of RchO whose crystal structure determinations will
be attempted using the established method and their observed!

 

predicted functional changes

109

115

118

122

138

162

186

Figure 1.1
Figure 1.2
Figure 1.3
Figure 1.4
Figure 1.5
Figure 1.6

Figure 1.7

Figure 1.8

Figure 1.9

Figure 1.10

Figure 2.1

Figure 2.2

Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4

Figure 3.5

LIST OF FIGURES

 

 

 

The mitochondrial respiratory chain complexes 2
Overall structure of RchO 5
Hemes and metal centers in Rch0 and their amino acid ligands ----- 7
Membrane lipids resolved in the crystal structure of RchO -------- 11
Comparison of R. s. and bovine mitochondrial CcO structures ------ 13
Electron transfer pathways in CcO 15
The oxygen chemistry cycle of reaction catalyzed by CcO ---------- 18

Two proton uptake pathways resolved in the crystal structure of
RchO 21

 

Structures of phospholipids found in Rhodobacter sphaeroides ----- 22

Structures of non-phospholipids found in Rhodobacter
Sphaeroides 23

 

--- 24

 

Amino acid sequences of RchO

Hanging drop and sitting drop vapor diffusion crystallization
set-ups 62

 

Major crystal contact regions in the crystal of the four
subunit Rch0 70

 

Crystal of the four subunit RchO which belongs to a space
group of R3 73

 

UV-visible spectra of purified Rch0 75

 

SDS-PAGE of purified enzyme and re-dissolved crystals of
RchO 77

 

The two forms of subunit II of RchO due to incomplete
proteolytic processing of the C-terminal 13 amino acids ----------- 80

xi

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Figure 3.10

Figure 3.11

Figure 3.12

Figure 3.13

Figure 3.14

Figure 3.15

Figure 3.16

Figure 3.17

Figure 3.18

Figure 3.19

The active site consisting of heme a3 and Cut; in the structure
of the four subunit RchO

 

Partial lipids resolved in the structure of the four subunit
Rch0

87

89

 

Major crystal contact regions in the crystal structure of the

 

four subunit RchO

Lipids content analyses of RchO samples at different

purification stages using MALDI mass spectrometry -----------

Elution profile of Ni—NTA affinity column purification of
Rch0 and the UV-visible spectra of the fractions under the
two peaks during the elution process

 

Crystal packing of the four subunit RchO which belongs to
a space group of R3

 

Crystals of the four subunit RchO which belong to a space

 

group of P212121

Unit cell display of the four subunit orthorhombic RchO
crystal which belongs to a space group of P212121

 

Cadmium binding sites found in the structure of the four
subunit RchO crystals soaked in a solution containing

 

cadmium

Cadmium binding sites found in the four subunit RchO

structure after the crystals were soaked in a solution containing

cadmium

 

MALDI mass spectra of the subunit IV region of purified

Rch0 and re-dissolved four subunit RchO crystals obtained

from different R. s. strains

 

Crystals of the HI subunit RchO which belong to a space group

 

Of P212121

UV-visible spectra of purified Rch0 and re-dissolved I-II
subunit RchO crystals

 

Enzymatic activity measurement of re-dissolved I-II subunit

xii

90

95

103

113

117

119

123

126

132

137

140

Figure 3.20

Figure 3.21

Figure 3.22

Figure 3.23

Figure 3.24

Figure 3.25

Figure 3.26

Figure 3.27

Figure 3.28

Figure 3.29

Figure 4.1

FigUre 4.2

Rch0 crystals under steady state conditions using an oxygen
electrode 142

 

MALDI mass spectra of the lipid regions of purified Rch0

and re-dissolved I-II subunit RchO crystals in both positive and
negative ion mode comparing the lipid species present in purified
Rch0 and in re-dissolved I-II subunit crystals 143

 

Comparison of the structure of the HI subunit Rch0 and
the four subunit RchO (PDB entry 1M56) 145

 

 

Unit cell display of MI subunit RchO 147

Major crystal contact region of LI] subunit RchO contributed
by the engineered histidine tag at the shortened C-terminus of
subunit II and the cadmium ion 148

 

The covalent linkage between the ring atoms of Y2881 and H284;
of Rch0 150

 

Comparison of the resolved waters in D proton uptake pathway
in the crystal structures of MI subunit Rch0 and the four subunit
RchO (PDB entry 1M56) 152

 

Comparison of the resolved waters in K proton uptake pathway
in the crystal structures of [-11 subunit Rch0 and the four subunit
RchO (PDB entry 1M56) 154

 

Additional cadmium binding site in the structure of I-11 subunit
Rch0 156

 

Two decyl maltoside detergent molecules resolved at the
interface of two RchO molecules (gray and wheat) 158

 

Structure of MI subunit RchO showing the resolved detergent
molecules, detergent headgroups, and alkyl tails of detergents or
membrane lipids 160

 

Superimposed surface representations of LI] subunit Rch0

and PdCcO together with the detergent molecules, maltose

headgroups and alkyl chains resolved from the structure of

Rch0 and the resolved detergent LDAOs from PdCcO ----------- 171

Structure overlay of [-1] subunit RchO, I-II subunit PdCcO

xiii

 

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

(PDB entry 1AR1), and bovine heart mitochondrial CcO
(PDB entry 1V54) 173

 

Structure overlay of RchO (PDB entry 1M56), PdCcO
(PDB entry lQLE) and bovine heart mitochondrial CcO (PDB
entry 1V54) showing the conserved phospholipid molecules ------ 175

Structure overlay of cytochrome bcr complexes from yeast

(PDB entry 1KB9), chicken (PDB entry lBCC), and bovine (PDB
entry 1PP9) showing the conserved phosphatidyl ethanolamine
molecule 176

 

Structure overlay of MI subunit RchO, I-II subunit PdCcO
(PDB entry 1AR1), and bovine heart mitochondrial CcO (PDB
entry 1V54) 178

 

Structure overlay of cytochrome bcl complexes from chicken
(PDB entry lBCC), and bovine (PDB entry 1PP9) 180

 

ATP

A. U.

BSA

CcO

CDL

CMC

DEAE

EDTA

ESI

FADHz

FPLC

ICP

MALDI

MES

NADH

Ni-NTA

Rd.

ABBREVIATIONS

amino acid

adenosine triphosphate

asymmetric unit

bovine serum albumin

cytochrome c oxidase

cardiolipin, diphosphatidylglycerol

critical micelle concentration

diethylaminoethyl

ethylenediaminetetraacetic acid

electrospray ionization

reduced form of flavin adenine dinucleotide

fast protein liquid chromatography

inductively coupled plasma emission spectroscopy
kilodalton

matrix-assisted laser desorption ionization mass spectrometry
2-(N-morpholino)ethanesulfonic acid

molecular weight

reduced form of nicotinarnide adenine dinucleotide
nickel-nitrilotriacetic acid

Paracoccus denitrificans

XV

PDC
PdCcO
PE
PEG-400
PG

PI

PS

R. s.
ROS
RchO
SDS-PAGE
SQDG
TOF

Tris

UCP

phosphatidyl choline

protein-detergent complex

cytochrome c oxidase from Paracoccus denitrificans
phosphatidyl ethanolamine

polyethylene glycol with an average molecular weight of 400
phosphatidyl glycerol

phosphatidyl inositol

phosphatidyl serine

Rhodobacter sphaeroides

reactive oxygen species

cytochrome c oxidase from Rhodobacter sphaeroides
sodium dodecyl sulfate polyacrylarnide gel electrophoresis
sulfoquinovosyl diacylglycerol

time of flight

tn's(hydroxymethyl)-aminomethane

uncoupling protein

xvi

Chapter 1. INTRODUCTION

1.1 Structure and Mechanism of Cytochrome c Oxidase
1.1.1 Overall reaction

Cytochrome c oxidase (CcO) is found in both the mitochondrial inner
membranes as well as the periplasmic membranes of many prokaryotes. It is the
terminal enzyme in the electron transfer chain in eukaryotes and many bacteria that
are capable of aerobic respiration. Figure 1.1 shows an overview picture of the
respiratory chain. Electrons from food sources are transferred to complex I (NADH
dehydrogenase) or complex H (succinate dehydrogenase) through NADH and FADHz,
respectively. These electrons are then transferred to complex III (cytochrome be]
complex) through coenzyme Q. Electrons are then transferred from complex HI to a
small soluble or membrane associated protein, cytochrome c, which in turn transfers
them to complex IV (cytochrome c oxidase). Cytochrome c oxidase provides the final
electron sink by accepting electrons from reduced cytochrome c and passing them
onto oxygen to form water (for reviews, see Ferguson-Miller and Babcock, 1996;
Michel et al., 1998). It utilizes the energy generated from this exothermic reaction to
translocate protons across the membrane against the membrane potential and pH
gradient (Wikstrom, 1977). The proton gradient formed across the inner mitochondrial
membrane or the bacterial periplasmic membrane is then utilized by ATP synthase to
generate ATP, a universal energy source.

The overall reaction that cytochrome c oxidase catalyzes is:

4 cyt C2+ + 8 H+(in) + 02 —) 4 cyt 03+ + 2 H20 + 4 H+(out)

Cytochrome c

   

succinate fumarate

  

NADH‘ NAD ‘
Complex 1 Complex 11 Complex 111
NADH Succinate Cytochrome be,
Dehydrogenase Dehydrogenase Complex Complex IV Complex V

Cytochrome c ATP Synthase
Oxidase

Figure 1.1: The mitochondrial respiratory chain complexes. The crystal structures
of Complexes 11, III, IV, V, and cytochrome c are shown (Bushnell etal., 1990; Iwata
er al., 1998; Iverson er al., 1999; Stock et al., 1999; Svensson-Ek er al., 2002).
Complex 1 is shown in cartoon based on structure obtained by electron microscopy
(Bottcher er al., 2002). Electrons from food sources are transferred to Complex I or
Complex 11 through NADH and FADI-Iz, respectively. These electrons are then
transferred to complex 111 through coenzyme Q, and subsequently to cytochrome c.
Cytochrome c then transfers the electrons to complex IV, which in turn transfers then
to dioxygen to form water. Complexes 1, III and IV are know to translocate protons
across the inner mitochondrial membrane. The proton gradient formed across the
inner mitochondrial membrane is then utilized by Complex V (ATP synthase) to
generate ATP, a universal energy source. This figure is adapted from a figure courtesy
of Dr. Denise Mills. Images in this dissertation are presented in color.

Ft'i

$16

02

f)

For each catalytic cycle, a total of eight protons are taken up from inside (N =negative
side) of the membrane. Four of them combine with oxygen to form water while four
others are pumped to the outside (P=positive side) of the protein.

CcO belongs to a superfamily of enzymes called heme/copper terminal
oxidases because members of this family all have a conserved feature of having an
iron-heme and a copper at the active site where oxygen is reduced. Besides using
cytochrome c as an immediate electron transfer donor, some bacterial heme-copper
terminal oxidases can utilize quinol as their substrate, such as cytochrome b03 quinol
oxidase from E. 6011'. Some bacteria also have different types of CcOs based on the
different heme groups involved in internal electron transfer. For example, CcO from
Thermus thermophilus has a heme b and a heme a3. Rhodobacter sphaemides (R. s.)
has several different types of CcOs including the 003 type, the Cbb3 type, and the 66103
type. The M3 type CcO, which has a heme a and a heme a3, is the one that is the most
homologous to that found in the mitochondrial respiratory chain in eukaryotes.

Because of its importance in energy metabolism, defects in CcO have been
found associated with a variety of diseased states, both neuromuscular and
non-neuromuscular in childhood and adulthood such as Leigh Syndrome, Alzheimer’s
disease, and a variety of other diseases (for reviews, see Robinson, 2000; Borisov,
2002). Extensive work has been performed over the years to understand the enzymatic
mechanism and regulation of C60 under physiological and pathological conditions.

CcO is a multi-subunit transmembrane protein complex. The number of

subunits varies from 4 in many bacteria to 13 in mammalian mitochondria. However,

omit 5

1111011;

131.121

subs

(iii

01

only subunits I, H, and 111, all being transmembrane subunits, are highly conserved
throughout the species. The rest of the subunits are likely to play regulatory roles in
the higher organisms. We have been focusing on understanding the structure/function
relationships of cytochrome c oxidase from Rhodobacter sphaeroides (RchO), a four
subunit bacterial enzyme, as a model to study its mammalian counterpart. RchO is
an excellent model to study the key function of electron transfer and proton pumping
by allowing kinetic, spectroscopic, and site-directed mutational analysis, as well as
other biochemical studies of C60 under different bacterial growth conditions and
genetic backgrounds.
1.1.2 Current Understandings of the Structure/Function of C00

The advent of several X-ray crystal structures from bovine heart and two
different bacteria, Paracoccus denitrificans (P. d.) and Rhodobacter sphaeroides,
together with extensive spectroscopic and functional analyses have greatly enhanced
our understanding of the M3 type cytochrome c oxidase (Iwata et al., 1995; Tsukihara
er al., 1995; Tsukihara et al., 1996; Osterrneier et al., 1997; Yoshikawa et al., 1998;
Svensson-Ek er al., 2002). We now have a fairly detailed picture of the structure of
this enzyme complex and also some of the functional aspects pertaining to its
structure. However, the exact mechanism of proton transfer and its coupling to
electron transfer is yet to be resolved.
1.1.2.1 Overall Structures of Subunits

Figure 1.2 shows the overall structure of the four subunit Rhodobacter

sphaeroides cytochrome c oxidase (RchO). Unless otherwise noted, all the amino

 

Figure 1.2: Overall structure of Rch0. The four subunits are labeled as shown in
the figure. Subunit I is colored in yellow, subunit II in green, subunit III in cyan, and
subunit IV is colored in gray. The heme groups and the redox active metal centers are
colored in red and purple, respectively.

and r

sut~c1

I

£111.11

rtdv

pr:

[1111

a?

”(J

 

acid residues herein are labeled by RchO numbering with the subunit number in
subscript. The membrane portion of CcO has a trapezoidal shape in a view
perpendicular to the membrane normal with an extra-membrane globular domain on
the periplasmic side. Subunits 1 (yellow) and H (green) of Rch0 are the core
catalytic subunits for both electron transfer and proton pumping activities. All the
redox active centers are located within these two subunits.

Subunit I is the largest and the most conserved subunit. It consists of 12
transmembrane helices roughly forming three 4-helix bundles in an anticlockwise
sequence viewed from the periplasmic side. These helices provide the scaffold where
prosthetic groups heme a, heme a3, and Cug reside. Also located within subunit I are
two proposed proton uptake channels. Figure 1.3 shows the metal centers in subunits I
and II. The iron centers of the two heme groups, heme a and heme a3, are buried
approximately 15 A from the periplasmic side of the membrane. The two porphyrin
planes are perpendicular to the membrane plane with a highly conserved
inter-porphyrin plane angle of approximately 104° with their propionates pointing to
the periplasmic side of the membrane. The hydroxyethylfamesyl tail of heme 0
extends downward and remains in the hydrophobic core between the transmembrane
helices. On the other hand, the hydroxyethylfamesyl tail of heme a3 extends sideways
and penetrates into the lipid bilayer, although the positions of the famesyl tails are not
highly conserved, unlike the angle of the hemes (Sharpe et al., 2005). Heme a has a
low spin iron with two conserved axial histidine ligands, HileZr and His421r. Heme

a3 has a high spin iron that is only five-coordinated, with only one conserved axial

 

Figure 1.3: Hemes and metal centers in Rch0 and their amino acid ligands. The
heme groups are colored in gray and the amino acid ligands are colored by the atom
types (C: green; 0: red; N: blue; S: yellow). Iron, copper, magnesium, and calcium
atoms are colored in red, purple, blue and bluish green, respectively. The membrane
surface of the periplasmic membrane of R. s. is shown as orange dotted line and the
interface between subunits I and II is represented by a light blue curve.

M2153II
H217II
C256“ *2 Subunit II
5
C “A

"~m~*

H411I

    

Subunit I

Figure 1.3: Hemes and metal centers in Rch0 and their amino acid ligands. The
heme groups are colored in gray and the amino acid ligands are colored by the atom
types (C: green; 0: red; N: blue; S: yellow). Iron, copper, magnesium, and calcium
atoms are colored in red, purple, blue and bluish green, respectively. The membrane
surface of the periplasmic membrane of R. s. is shown as orange dotted line and the
interface between subunits I and H is represented by a light blue curve.

add:

histidine, His419r, which is ligated to the iron on the opposite side (distal side) of the
heme a3 to the Cur; center. Due to lack of strong liganding interactions from the
proximal side, the iron in heme a3 is slightly out of the porphyrin plane toward the
axial histidine. The Cut; ion is approximately 4.9 - 5.1 A away from the heme a3 iron
in the structures of CcO from different species (Harrenga and Michel, 1999;
Svensson-Ek et al., 2002; Tsukihara et al., 2003). Together the Cut; center and heme
a3 Fe form the binuclear center active site where oxygen binds and is reduced to water.
CuB has three ligands, His333r, Hi5334r, and His284r. The latter is found to form an
unusual covalent linkage to Tyr2881 between the N82 and C82 atoms from the two
residues, respectively (not shown in Figure 1.3) (Osterrneier et al., 1997; Yoshikawa
et al., 1998; Buse et al., 1999; Tomson et al., 2002). X-ray crystallographic data also
suggests that there are more bridging ligands in between the heme a3 Fe and Gun.
Different bridging ligands have been proposed including a peroxide (Yoshikawa et al.,
1998), or the more widely accepted bridging structure with one hydroxide bound to
the Cu}; and a water ligated to the Fe ion of heme a3 (Ostermeier et al., 1997). The
exact nature of the bridging ligand(s) in the X-ray crystal structure is unclear. Besides
the heme groups and Cut; center, subunit I also has a non-redox active calcium
binding site in it and the role of the calcium ion is unknown. The ligating residues for
the calcium ion include Glu54l, Ala57r, and Gly591 (Svensson-Ek et al., 2002) (Figure
1.3).

Between subunits I and 11, right above the heme a3 propionate groups, an

additional non-redox active metal ion, magnesium, is found in the structure (Figure

Cf

In

01

Pro

1€a\

1.3). The magnesium binding site is highly conserved with D412; and H411 r as its
amino acid residue ligands and three very closely ligated water molecules in the
crystal structure of RchO (Svensson-Ek et al., 2002). However, its role is not
completely understood. It may be involved in stabilizing the interaction between
subunit I and II and may have important functions in water exit (Florens er al., 2001;
Schmidt et al., 2003).

Subunit II is composed of an N -terrninal loop, two transmembrane helices that
interact with subunit I and a C-terminal extra-membrane globular domain. X-ray
crystal structures indicate that the extrarnembrane domain is mainly composed of a
ten-stranded B-barrel in which the Cu center is located. The Cu», center (Figure 1.3)
is formed by two mixed valance copper ions (Cu"5+ - Cul'SI). The ligating residues for
one copper are: Cy3252u, Cy5256n, His2l7u, and Met263u and those for the other
copper atom are: Cys252n, Cy5256n, His260", and the carbonyl oxygen of Glu254n.
The two cysteines bridge the two copper atoms from two sides and the two sulfur
atoms lie in the same plane as the copper atoms. The Cup, center is the initial electron
acceptor from cytochrome c (Ferguson-Miller and Babcock, 1996; Malatesta er al.,
1998). In Rch0, the polypeptide chain of subunit II has been known to undergo
proteolytic cleavage events during its maturation process. As a result, the signal
peptide of the N-terminal 25 amino acids from the full length subunit H gene product
is completely removed (Steinrucke er al., 1987). Moreover, there is incomplete
proteolytic processing of the C-terminal 13 amino acids (Hiser er al., 2001). The

reason for this incomplete proteolytic cleavage is unknown and CcO containing

1110

lb
1W.
"ml

301‘)

Slim
enzi

P1211

mostly the processed or unprocessed forms of subunit II seem to be equally functional
(Hosler et al., 1992; Zhen et al., 1998).

Subunit III is also a highly conserved subunit. It is composed of seven helices
and has an overall structure of two bundles of helices separated by a V-shaped cleft. It
has no metal centers in it and its role is less well defined. Subunit III-less enzyme is
highly active but the enzyme undergoes suicide inactivation during catalytic turnovers
(Bratton et al., 1999). “Within the cleft of the V-shaped bundle, and the interface
between subunits III and I, bound membrane lipid molecules are found in the crystal
structures of bovine heart CcO, Rch0 and PdCcO (Figure 1.4).

Compared to subunit 1, II, and HI, subunit IV of CcO is not conserved among
different species. Subunit IV from RchO has only one transmembrane helix. Its
existence in the R. s. enzyme was not known until the crystal structure was solved
(Svensson-Ek et al., 2002). As shown in Figure 1.4, subunit IV lacks direct contact in
its transmembrane region with its neighboring protein subunits I and III and it
associates itself with the enzyme complex via indirect contacts mediated through
membrane lipid molecules as revealed by the crystal structure (Svensson-Ek et al.,
2002). The role of subunit IV remains unknown. The histidine-rich cytoplasmic
portion of the polypeptide chain might be part of the proton collecting antenna system.
However, genetic and biochemical studies of PdCcO, whose protein sequence and
structure are highly homologous to those of RchO, showed no difference in
enzymatic activity when the subunit IV gene was deleted (Witt and Ludwig, 1997).

Preliminary experiments of RchO lacking subunit IV also showed no effects on

mostly the processed or unprocessed forms of subunit II seem to be equally functional
(Hosler et al., 1992; Zhen et al., 1998).

Subunit III is also a highly conserved subunit. It is composed of seven helices
and has an overall structure of two bundles of helices separated by a V-shaped cleft. It
has no metal centers in it and its role is less well defined. Subunit III-less enzyme is
highly active but the enzyme undergoes suicide inactivation during catalytic turnovers
(Bratton et al., 1999). Within the cleft of the V-shaped bundle, and the interface
between subunits III and I, bound membrane lipid molecules are found in the crystal
structures of bovine heart CcO, Rch0 and PdCcO (Figure 1.4).

Compared to subunit 1, II, and III, subunit IV of C60 is not conserved among
different species. Subunit IV from RchO has only one transmembrane helix. Its
existence in the R. s. enzyme was not known until the crystal structure was solved
(Svensson-Ek et al., 2002). As shown in Figure 1.4, subunit IV lacks direct contact in
its transmembrane region with its neighboring protein subunits I and III and it
associates itself with the enzyme complex via indirect contacts mediated through
membrane lipid molecules as revealed by the crystal structure (Svensson-Ek et al.,
2002). The role of subunit IV remains unknown. The histidine-rich cytoplasmic
portion of the polypeptide chain might be part of the proton collecting antenna system.
However, genetic and biochemical studies of PdCcO, whose protein sequence and
structure are highly homologous to those of RchO, showed no difference in
enzymatic activity when the subunit IV gene was deleted (Witt and Ludwig, 1997).

Preliminary experiments of RchO lacking subunit IV also showed no effects on

 

Figure 1.4: Membrane lipids resolved in the crystal structure of Rch0. The two
figures represent a side view on the left and a top view on the right. A total of six
phosphatidyl ethanolamine molecules were resolved in the crystal structure of RchO.
Four of them are found surrounding subunit IV and the other two are found within the
V-shaped clefi formed by helix bundles inside subunit III and at the interface of
subunit III and subunit 1. Subunit IV associates itself with the enzyme complex via
indirect contacts mediated through the membrane lipid molecules.

enzymatic activities (Hiser et al., unpublished). However, a role in stability or
regulation cannot be ruled out.

CcO from the bovine heart mitochondria has 13 subunits (Kadenbach et al.,
1983; Tsukihara et al., 1996). The core subunits of I, II and III are encoded by the
mitochondrial genes while the other ten subunits are nuclear encoded (Capaldi, 1990).
The first three subunits from the bovine enzyme show very high structural homology
to the first three subunits of bacterial enzyme as shown in Figure 1.5. The other ten
subunits from the bovine enzyme include both transmembrane subunits and
extramembrane subunits. The specific roles of these subunits are not well defined and
they likely play roles in assembly and/or activity regulation. Subunit VIb binds a zinc
ion of unknown function and one of the cholate binding sites on subunit VIa observed
from crystal structure might indicate a possible nucleotide binding site for energy
production regulation (T sukihara er al., 1996). Multiple binding sites of ATP/ADP on
both core subunits and peripheral subunits were also suggested from biochemical
studies of yeast and bovine CcO (Kadenbach, 1986; Napiwotzki et al., 1997; Beauvoit
and Rigoulet, 2001; Ludwig et al., 2001).
1.1.2.2 Electron fiansfer Pathways

The electron transfer pathway within the enzyme is understood in some detail.
As shown in Figure 1.6, electrons are transferred from cytochrome c first to the Cu
center, and in turn to heme a, then to the binuclear center, heme a3 and Cug, where
dioxygen binds and is reduced to water (Ferguson-Miller and Babcock, 1996; Michel

er al., 1998).

 

Figure 1.5: Comparison of R. s. and bovine mitochondrial CcO structures. The
three core subunits I (yellow), 11 (green), and 111 (cyan) are highly homologous, as
well as the redox active centers (hemes a and 03 are shown in red, and coppers are
shown in purple). The non-homologous subunits, including subunit IV of Rch0 and
the additional 10 subunits from bovine CcO are shown in gray.

Cl

C)

PI

m.

Rt

red

199

There are two cytochromes c in R. s. that have been considered as substrates
for C602 cytochrome c,, which is membrane anchored and cytochrome cz, which is a
soluble, mobile protein. Both cytochromes c are capable of accepting one electron
from the cytochrome bcr complex (complex III) and donating it to CcO, but a variety
of considerations suggest that cytochrome cy is the physiological electron donor to the
aa3-type cytochrome c oxidase (Daldal et al., 2001; Daldal er al., 2003). From
molecular modeling using horse heart cytochrome c, the binding site for cytochrome c
is proposed to be on the outside of the membrane on a concave surface created by the
globular domain of subunit H and the adjacent flat surface of subunit I (Roberts and
Pique, 1999; Flock and Helms, 2002; Maneg er al., 2004). The binding of horse heart
cytochrome c to CcO is basically through electrostatic interactions between the two
docking faces involving a number of positively charged residues from cytochrome 6
including Lys8, Lysl3, Ly586 and Lys87 and negatively charged residues from
subunit H of RchO including Glu157n, Glu148u, Asp195u, and Asp214n, as
proposed by protein docking studies (Roberts and Pique, 1999), as well as chemical
modification studies of cytochrome c (Ferguson-Miller et al., 1978), and biochemical
analyses of site-directed mutants of these residues (Ferguson-Miller er al., 1978;
Roberts and Pique, 1999; Wang et al., 1999; Zhen et al., 1999).

Modeling studies and biochemical analysis suggest that a highly conserved
tryptophan residue in subunit H, Trp143n is the immediate electron acceptor from
reduced cytochrome c as shown in Figure 1.6 (Witt et al., 1998; Roberts and Pique,

1999; Wang et al., 1999; Zhen et al., 1999). In the CcO crystal structure, Trpl43n is

     
 

1e
g “‘1}\ a; 3‘.--
. F < § ,PHemeagr

 

Figure 1.6: Electron transfer pathways in 0:0. Cytochrome c is represented by a
red dotted circle. The heme c from cytochrome c and heme a and 03 from CcO are
shown in red and the copper atoms are shown in purple. The electron transfer pathway
is shown as blue dotted arrows.

111?.

do

R4

within van der Waals contact distance to the Cup, center ligands and the computer
docking model also shows its proximity to the heme ring of docked cytochrome c.
Phenylalanine and alanine mutants of the Trp143n residue reduce the electron transfer
rate by 450 and 1200 folds, respectively, while having no effects on the dissociation
constant between cytochrome c and CcO, nor do these mutants affect the Cu, site
(Wang et al., 1999; Zhen et al., 1999). These studies suggest that the indole ring of
Trp143n is most likely the conduit of electrons from the heme group in cytochrome c
to the Cup, center in the extramembrane domain of CcO.

The Cu), center is located in the extramembrane domain of subunit H and is
composed of two mixed-valence copper ions. It transports electrons to heme a in
subunit I at an extremely fast rate of approximately 2.3 x 104 s‘I in bovine heart
mitochondrial CcO (Pan et al., 1993) and 9 x 104 s'1 in Rch0 (Wang et al., 1999),
considering the distance between the Cup, center and the Fe in heme a of 19 A and the
driving force of just 50 meV (Tsukihara et al., 1995; Winkler et al., 1995). One of the
proposed major electron transfer (tunneling) pathways between the Cu». dinuclear
center and heme a is composed of 14 covalent bonds and two hydrogen bonds. It
starts with a hydrogen bond between CuA ligand His260u and a carbonyl group of the
peptide bond joining two highly conserved arginine residues, Arg4811 and Arg4821,
and then through Arg4821 to one of the propionate groups in heme a (Iwata er al.,
1995; Ramirez et al., 1995; Regan et al., 1998; Tan et al., 2004). The importance of
the residues along the pathway is supported by site-directed mutants H260nN and

R4821P, which show an approximately 2000 fold slower electron transfer rate between

16

Cu

lilt’

19C

(11.1

Under:

Figure

Cu and heme a, which cannot be accounted for by altered redox potentials of the
metal centers (Wang et al., 2002; Zhen et al., 2002; Qian et al., 2004). The result can
be interpreted to support the concept of a “through bond” pathway of electron transfer.
Although the distance between Cu); and the heme a3 is only about 1.5 A longer than
that between Cu), and heme a, evidence for significant rates of electron transfer
between Gun, and heme a3-Cu3 binuclear center has not been found (Regan et al.,
1998).

Electrons transported from heme a to the heme a3 Fe - Cur; binuclear center
could utilize either of the two different proposed pathways that connect heme a Fe
and heme a3 Fe: the closely approaching heme edges, or the histidine ligands to each
heme connected through helix VIH (Regan et al., 1998). The direct electron transfer
between the two heme porphyrin rings would occur through “edge to edge” transfer
since the closest edge to edge distance between the two heme groups is only
approximately 4.6 — 5.2 A in CcO from different organisms (Ostermeier et al., 1997;
Harrenga and Michel, 1999; Svensson-Ek et al., 2002; Tsukihara et al., 2003; Tan et
al., 2004). The center-to-center transfer via the histidine ligands involves some
through space jumps and a distance of approximately 13.5 A between the two heme
Fe atoms.
1.1.2.3 Oxygen Chemistry

Spectroscopic and fast kinetic measurements have contributed to our
understanding of the chemistry behind the dioxygen reduction reaction as shown in

Figure 1.7 (Michel et al., 1998; Zaslavsky and Gennis, 2000; Kim et al., 2004). In this

 

Oxidized (O)

9r 2:: :2: .5“\

 

 

 

 

 

 

 

 

 

 

 

Ferry-0x0 (F) 11* l e' Reduced (E)
Cu32+— OH Cu,1+ (H20)
Fea34+=02' Feaf— OH

YH Hi\lp/e-\
M +
H 2 6 Reduced (R)

Peroxy (P) C113” (H20)
Cu32*- OH ()2 Feaf‘“ (H20)
F e03“=02' ()2 Bound (A) /j/

Y‘ ‘\ CuB“ 1
F ea32*--O=O (H20)2

YH

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1.7: The oxygen chemistry cycle of reaction catalyzed by CcO. For each
catalytic cycle, four electrons are transferred to the binuclear center and combined

with dioxygen and four protons taken from the inside of the membrane to form 2
waters. The four additional protons translocate across the membrane are not shown in

the figure since the proton pumping mechanism is unknown.

18

be

61;

scheme, the oxidized form of enzyme (usually called intermediate 0) has a hydroxyl
bound to both heme a3 and Gun. The first electron that enters CcO through the
electron transfer pathway is expected to go to Cut; of the binuclear center forming
intermediate E, causing the conversion of its bound hydroxyl to water at the Cut; site.
A second electron is then transferred to heme a 3 forming intermediate R with another
water formed at heme a3. The doubly reduced binuclear center can then bind oxygen
forming intermediate A. The oxygen bond is rapidly cleaved by a four electron
transfer event using two electrons from the heme 613 to form a ferryl-oxo and one
electron from Cup, to form a cuprous hydroxide, as well as an electron and proton
from a nearby Tyr2881 (MacMillan et al., 1999). This state is referred to as the P
(peroxy) intermediate even though all evidence indicates that the peroxy form is not
present. A third electron is injected and it reduces the Tyr28& free radical back to its
stable state, forming intermediate F with concomitant uptake of a proton. Finally, a
fourth electron is injected with accompanying proton to convert the enzyme from the
Fe4+ ferryl intermediate to its Fe3+ oxidized state, with a hydroxyl on both heme a3
and Cug. How and when vectorial translocation of protons occurs during this catalytic
cycle is unclear and under debate. Proton pumping is proposed by some to occur
during the oxygen reduction cycle (Vygodina et al., 1997), involving the P -—) F and F
-—) 0 steps. Others proposed that one proton is likely to have been translocated before
the formation of the P intermediate (Michel, 1999). Still others propose that proton
pumping occurs during each one-electron step of reduction of the enzyme (Wikstrom

and Verkhovsky, 2002; Tsukihara et al., 2003). The proton pumping mechanism and

19

1b.

0b

mt]

its coupling to electron transfer remains a subject of controversy.
1.1.2.4 Proton Uptake Pathways

For each catalytic cycle, four protons (substrate protons) are needed for the
reduction of dioxygen to form water and another four protons (pumped protons) are
translocated across the membrane to form the proton gradient needed to synthesize
ATP (Wikstrom, 1977). Mutagenesis and the crystal structures of bacterial CcO reveal
two important proton uptake pathways, D and K pathways, each named after a residue
whose mutation blocks the pathway, D1321 and K3621, respectively (Fetter et al., 1995;
Iwata er al., 1995; Svensson-Ek et al., 2002). Figure 1.8 shows the two proton uptake
pathways.
1.1.2.4.1 D Pathway

A chain of crystallographically resolved water molecules were identified in
Rch0, which defined the D proton uptake pathway (Iwata et al., 1995; Svensson-Ek
et al., 2002) (Figure 1.8). The waters are also seen in the bovine CcO crystal structure
at 1.8 A resolution (Tsukihara et al., 2003). It starts with a highly conserved residue,
D1321, close to the inside of the bacterial cell membrane, and continues through
several polar residues including N1211, N1391, N2071, 8142;, Y33r, S2011, 82001,
$1971, and onto E2861. How proton transfer is achieved after E2861 is unclear and
Currently under debate since there are no more waters found in the crystal structure in
the hydrophobic cavity between E2861 and the two heme groups. However, water is
Observed around the heme propionates and a chain of waters is seen to form in a

mOIecular dynamics simulation of RchO (Cukier, 2004; Seibold et al., 2005).

20

 

Figure 1.8: Two proton uptake pathways resolved in the crystal structure of
Rch0. The D and K pathways are shown as blue dotted arrow and the
crystallographically resolved waters in the two pathways are shown in yellow.

21

’5.

C611?

1.1.

‘6

1'3

The D pathway is thought to conduct both pumped protons and substrate protons and
E2861 is suggested to be functioning as a valve through alternate conformations of its
side chain to direct proton flow to either the proton exit pathway or the binuclear
center active site (Konstantinov et al., 1997; Hofacker and Schulten, 1998 ; but
see Sharpe et al., 2005).
1.1.2.4.2 K Pathway

The K pathway is considered to start with E101" and continue through S2991,
K362], and T3591 up to the binuclear center via the hydroxyl of Y2881 that is hydrogen
bonded to the famesyl hydroxyl of heme a3 (Branden et al., 2002; Tomson et al., 2003)
(Figure 1.8); Y288; forms an unusual covalent bond with one of the Cut; ligands,
H2841, between the C82 and N82 atoms from the two residues, respectively. Such a
covalent linkage was proposed from the crystal structure (Osterrneier et al., 1997) and
verified by other high resolution crystal structure (Yoshikawa et al., 1998), and
received further experimental verification by mass spectrometry (Ostenneier et al.,
1997; Yoshikawa et al., 1998; Buse et al., 1999). The K pathway is believed to
transport substrate protons to the active site as shown by the extremely low activity of
the K3621 mutants, and relatively low activity of S299. and T3591 mutants, as well as
the observation that impairment in proton uptake could not be compensated by uptake
of protons from the outside. In contrast, the activity of the D pathway mutant D132lA
can be significantly enhanced by protons supplied from the outside in he presence of a
membrane potential (the controlled state) where the wild type enzyme activity is

inhibited (Fetter et al., 1995; Mills et al., 2000). This lack of enhancement of activity

22

subs

 

 

lnlnj
and!
und:

11.2

”11111
to b
51th

oxld

C0n5
efltt
“MM

2000

binUCI

in the K path mutants suggests that the K pathway is exclusively used for substrate
proton uptake and not connected to the external surface via any reversible proton path.
However, the observation that the K362rM mutant has a much higher activity with
H202 as substrate indicates that the D pathway might also be involved in providing
substrate protons (Konstantinov et al., 1997). Yet it remains unclear which and how
many of the four required substrate protons are supplied by each pathway, and when
and how these protons are used in the oxygen reduction reaction, a central issue in
understanding the coupling of proton pumping and electron transfer.
1.1.2.4.3 H Pathway in Bovine Heart Mitochondrial CcO

Besides the D and K pathways, an H pathway, which connects the
mitochondrial matrix side of the membrane to the outside, was defined and proposed
to be exclusively for pumped proton uptake based on the high-resolution crystal
structure of bovine heart mitochondrial CcO and an altered conformation in the
oxidized versus the reduced enzyme structure (Tsukihara et al., 1996; Yoshikawa et
al., 1998). However, some of the essential residues along this pathway are not
conserved in bacterial CcOs and mutations of the equivalent residues have little or no
effect on the activity of RchO, indicating that the H pathway is not likely to be a
universal proton uptake pathway conserved across the different species (Lee et al.,
2000).
1.1.2.5 Oxygen Pathway

One of the substrates of the enzyme, dioxygen must be accessible to the

binuclear center. Since dioxygen is nonpolar, oxygen could simply reach the active

23

Sll:

ex;

01;

(1:

$111

CO]

1.1

Pd!

Chi

and

Ca»:

site through diffusion in the hydrophobic regions of the enzyme and perhaps a defined
oxygen pathway is not completely necessary. However, three proposed specific
oxygen pathways were defined in the bovine enzyme based on the crystal structure
(Tsukihara et al., 1996). In one of them, which is supported by molecular dynamics
studies, oxygen is proposed to reach the binuclear center through a hydrophobic hole
connecting from the V-shaped cleft formed by helices of subunit HI, through subunit I
to the binuclear center active site (Hofacker and Schulten, 1998). Based on the
positions of the xenon atoms resolved in the structure of RchO crystal
pre-pressurized with xenon, an alternative oxygen pathway was proposed which
accessed the binuclear center through subunit I between two transmembrane helices
(Svensson-Ek et al., 2002).
1.1.2.6 Proton and Water Exit Pathway

One possibility for the proton exit is that protons transported through the D
pathway are picked up by E2861. Via conformation movement of E2861 and water
chain formation, protons could be moved to the vicinity of the D-ring propionate
group of heme a3 (Hofacker and Schulten, 1998; Cukier, 2004). The latter is
connected to the bulk phase through several hydrogen-bonded water networks.
However, it is not clear whether there is a specific proton exit path, nor if the route is
through the active site or via the E286r/propionate connection.

It is argued that product water can simply randomly diffuse out of the protein
and a specific water exit pathway is not necessary. However, this is not likely to be the

Case because water can serve as a pathway for proton movement, and random

24

dif

1'61

pr

(8'

$11

111'

fa.

’V

diffusion of product water would potentially short-circuit the CcO system which
requires unidirectional movement of electrons and protons. A water exit path was first
proposed in the crystal structure of bovine CcO that involved the non-redox active Mg
center right above the active site at the interface of subunit I and H (Tsukihara et al.,
1996). Evidence from ESEEM studies of the enzyme with Mn substituted at this Mg
site using 170 and H2170 indicates that the Mn site is rapidly accessed by water from
the outside and, importantly, water produced at the active site arguing for a discrete
pathway rather than exit by random diffusion (Schmidt et al., 2003).

1.1.2.7 Proton Pumping Theories and Coupling of Electron 'Il'ansfer and Proton
Translocation

Although the structure of CcO is well defined and the oxygen chemistry is
now fairly clear, this has not led to a resolution of the question of how proton
pumping occurs. A number of different proton pumping models have been proposed,
but most are difficult to test. Currently no single model is supported by compelling
experimental evidence. Proton pumping theories are generally grouped into direct
coupling and indirect coupling mechanisms. However a combination of both is
possible.

In a direct coupling mechanism, change of redox state of the metal center
facilitates proton uptake and release from one of the active site ligands. A prototype
“ligand-exchange” model proposed by Chan and colleagues involved ligand exchange
upon reduction of CuA, leading to a proton release from one of its cysteine ligands

(Larsen et al., 1992). This is an example of a redox-coupled pumping mechanism that

25

has been discounted because of the ability of CcO without Cu), to pump, and mutants
of the Cu site to retain pumping activity (Zhen et al., 2002). Another example of a
direct coupling mechanism, that involves changes at the active site, is the histidine
cycle model proposed first by Wikstrom and colleagues (Wikstrom et al., 1994).
Similar to the histidine cycle model, the histidine shuttle model was elaborated by
Michel and colleagues. It suggests that one of the Cut; ligands, His3331, undergoes
cycles of protonation states and swings between ligated and unligated conformations
(Ostcnneier et al., 1995). However, later crystal structures suggested that there was no
Cug ligand change upon reduction, lessening the interest in the histidine shuttle model
(Harrenga and Michel, 1999).

In an indirect coupling mechanism, oxygen chemistry or redox changes are
coupled to a structural change a distance away from the active site which leads to
proton movement and finally proton pumping. Based on the high resolution crystal
structures observed for oxidized and reduced bovine CcO, a residue D511 (bovine
numbering) at the top of the proposed H channel was suggested to be involved in
proton pumping (Yoshikawa et al., 1998). In the oxidized state of enzyme, D51. is
buried under the protein surface and connected to the matrix side via a hydrogen
bonded network and internal water pools. Its conformation changes upon reduction of
the enzyme, leading to exposure of its protonated carboxyl oxygen to the outer surface
to release a proton (Yoshikawa et al., 1998). Such a proton pumping mechanism
bypasses the binuclear center and is proposed to be driven by heme a redox changes

(Tsukihara et al., 2003). However, the residue D51] is not conserved in bacterial

26

enzymes. Another indirect coupling model originally proposed by Rich and
colleagues (Rich et al., 1996) states that charge neutralization is the driving force for
proton uptake from the inside, such that when an electron moves to a redox active
site, a proton moves to the vicinity to neutralize the negative charge. However, the
pumped protons are not bound directly to the metal center ligands and follow a path
that is separate from the binuclear center.

Recently the importance of E2861 at the end of the partially defined D
pathway is getting much attention. Molecular dynamics simulations suggest that
E2861 can deliver a proton via water chain formation to the binuclear center or to the
heme a3 propionate group by alternating its side chain conformation (Hofacker and
Schulten, 1998; Cukier, 2004). The flipped conformation of E2861, in which the
carboxyl oxygen is closer to the D ring propionate of heme a3, is proposed to be
stabilized by two water molecules originally located at the active site (Hofacker and
Schulten, 1998), or by other residue and water movements in the vicinity (Seibold et
al., 2005).

In a recently proposed model from our research group (Sharpe et al., 2005),
the H284r—Y288r covalent pair is suggested to go through cycles of ligation and
deligation of Cu}; in response to changes in charge and ligand state of the copper. The
rotation of its imidazole-tyrosine rings leads to alternate opening and closing of the K
pathway. Protons move into this site from the D and K pathways, depending on the
Cur; ligating state and the charge present in the binuclear center. They are transported

either to the active site to form water or to a conserved water molecule bound between

27

the

 

rel:

 

 

lthl

PM
had 1

Ulilrr

3511).;

lGllu

 

the propionates of heme 613, via the histidine ligands of Cu. The latter proton is
released to the outside of the membrane. This is a new version of the histidine cycle
model which is yet to be tested by experimental analysis.
1.1.2.8 Function of Subunit HI of C00

Although subunits I, H and HI are highly conserved throughout the species,
only subunits I and H contain the redox active centers and are required for both
electron transfer and proton pumping functions. The exact function of subunit H1 is
not completely understood and is being studied using subunit HI-less enzyme. Subunit
IH-less CcO can be prepared by Triton X-100 detergent treatment after enzyme
purification or by genetically removing the subunit IH gene (Mills et al., 2003).
Although the initial activity and stability of subunit HI-less enzyme is similar to that
of the four subunit holoenzyme, subunit HI-less CcO undergoes spontaneous and
irreversible inactivation (suicide inactivation) during steady state turnover (Bratton er
al., 1999). The mechanism of suicide inactivation is apparently nOt due to the
production of reactive oxygen species since adding catalase and superoxide dismutase
had no effect on the rate of inactivation (Bratton et al., 1999). Suicide inactivation is
ultimately associated with the loss of Cur; at the heme a3-Cu3 binuclear center as
measured by EPR spectroscopy and metal analysis, possibly due to the dissociation of
one or more of Cug’s ligands (Hosler, 2004).

It was also found that the rate of proton uptake through the D pathway, as well
as the proton exit/back flow pathway, was slowed down in the absence of subunit IH

(Gilderson et al., 2003; Mills et al., 2003). A similar inactivating effect was observed

28

for mutants that affect the D pathway, such as D1321A, but not for K the pathway
mutant T3591A. A severe inactivating effect on CcO was observed for a double
mutant D1321A/R4811K even in the presence of subunit 1H. On the other hand,
arachidonic acid, which enhances the D pathway proton uptake, slows down the rate
of inactivation of subunit HI-less CcO. Therefore, the role of subunit 1H could be to
maintain a conformation of CcO that facilitates proton flow to the active site through
the D pathway and proton backflow through the exit path to the active site. The
prolonged lifetime of the heme a3 oxoferryl (Fe4+=O) or a tyrosine radical
intermediate, or the deprotonated form of E2861 due to the slowed proton supply
might be responsible for the observed suicide inactivation (Mills and Hosler, 2005).
1.1.3 Regulation of 0.0 Activity and Energy Metabolism

Mitochondria consume 85-90% of the total oxygen transported into the cells
and generate most of the energy in the cell under aerobic conditions through the
respiratory chain complexes. Cytochrome c oxidase and the general energy
metabolism are constantly being regulated by different mechanisms. When there is an
energy excess indicated by a higher membrane potential difference across the
membrane, different uncoupling routes are triggered which lead to proton backleak
across the membrane to dissipate the high membrane potential to prevent potential
buildup of reactive oxygen species (ROS) (also see Figure 1.1 for where ROS are
formed) which have been associated with mitochondrial damage and the aging
process (Harper et al., 2004). There exists a feedback loop between H+ leak and ROS

formation (Brookes, 2005). There are two types of uncoupling mechanisms, extrinsic

29

uncoupling and intrinsic uncoupling. Extrinsic uncoupling leads to increased proton
leakiness across the membrane via various pathways, while intrinsic uncoupling
lowers the proton pumping efficiency of respiratory chain complexes.

Extrinsic uncoupling is canied out through different types of uncoupling
proteins. Uncoupling protein 1 (UCPl) is found in membranes of brown adipocytes
and it serves as a fatty acid anion canier that results in a net proton leak through the
membrane to produce heat to maintain body temperature (Green and Brand, 2004).
Other UCPs may have more complicated physiological functions including
controlling the production of ROS and pathogenesis of type-2 diabetes (Green and
Brand, 2004; Krauss et al., 2005). Besides the leak mediated by specific proteins such
as UCPs, the mitochondrial membrane itself may also be permeable for proton
backleak, although there is evidence suggesting that this basal leak of H+ is mediated
via small molecules such as free fatty acids or superoxide species (Brookes, 2005).

The activity of CcO itself can be altered by a variety of small molecules,
including nitric oxide (Antunes et al., 2004), fatty acids (Sharpe et al., 1996), thyroid
hormones (Kadenbach and Arnold, 2000), zinc (Mills et al., 2002), and by the direct
energy indicator, the ATP/ADP ratio, although the exact bindings sites for nucleotides
are still unclear (Kadenbach, 1986; Beauvoit and Rigoulet, 2001). Besides changes in
the substrate binding affinity and in oxygen reduction activity, proton pumping
efficiency as expressed by H+/e' stoichiometry is also regulated. There is evidence in
CcO for a proton backflow pathway that allows protons to be taken up from the

outside to support activity in the presence of a membrane potential, decreasing the

30

pl.

C\

of

it

proton pumping efficiency (Mills et al., 2002). This pathway could suppress
formation of ROS. The backflow pathway has been proposed to be the reverse of the
exit pathway for pumped protons. Zinc inhibition of C00 reconstituted into lipid
vesicles in the presence of a membrane potential is proposed to be due to the blockage

of the proton exit/backflow pathway (Mills et al., 2002).

Although there have been enormous amounts of effort put into the
understanding of the structure/function and regulation of CcO and great achievements
have been made over the years, our collective knowledge of the enzyme, particularly
in the underlying mechanism of the vectorial translocation of protons and how proton
pumping is coordinated with electron transfer and oxygen reduction, is still quite
limited (Mills and Ferguson-Miller, 2003). When the crystal structures of both bovine
heart mitochondrial and PdCcO came out simultaneously, it was predicted that only a
few years’ time would be needed for the proton pumping mechanisms to be solved.
However, to date there seems to be more questions than answers. To address this
central issue in bioenergetics, high resolution crystal structures are needed of various
redox states and key mutants with altered biochemical properties in the proton
pumping process. Towards this end, we have been trying to develop a reproducible
method of obtaining crystals of RchO, a bacterial homologue of the mammalian

enzyme, that diffract X-rays to high resolution for further mechanistic studies.

31

1.2 Membrane Protein Crystallography
1.2.1 Overview of Macromolecular Crystallography

Structural information on biomacromolecules, such as proteins and nucleic
acids, has been extremely helpful in our understanding of their function and
regulation and has aided in rational drug designs to fight diseases. With the
completion of various genomic sequencing projects, the need for structural
information of the macromolecules has become more pressing in functional genomic
projects. Of the different techniques, X-ray crystallography has been the most
successful technique to date in solving the three-dimensional structure of
macromolecules to atomic resolution.

X-rays have the wavelength of about 1 A, which is close to the bond lengths
in biomolecules. The diffraction of X-rays by the electron clouds of atoms from
ordered molecules in a crystal can provide invaluable information about the structure
of the biomolecule at atomic resolution. Our understanding of the structure/function
relationships of the protein of interest often takes a leap with the advent of its high
resolution crystal structure, as was the case in CcO from a few different sources
(Iwata er al., 1995; Tsukihara et al., 1995; Tsukihara et al., 1996; Ostenneier er al.,
1997; Svensson-Ek et al., 2002).

It is estimated that a third of all open reading frames in the human genome
encode for membrane proteins, including peripheral and integral membrane proteins.
These proteins perform vital cellular functions such as energy metabolism, signal

transduction, and material transport; thus, they are particularly important targets for

32

drug development. However, despite their abundance and importance in cellular
functions, a mere 75 unique structures of membrane proteins had been deposited in
the Protein Data Bank (PDB) by the end of 2003, compared with about 20,000
structures of water-soluble proteins (White, 2004). X—ray crystallography has been the
major technique of obtaining 3-D structure of membrane proteins. Currently the most
time-consuming step in successfully solving a membrane protein structure is the
growth of a well-ordered membrane protein crystal, which can often take many years
of effort.
1.2.2 Challenges in Membrane Protein Crystallography
1.2.2.1 Membrane Protein Production

The first step in membrane protein biochemistry and crystallography is to
obtain the protein of interest in sufficient amount. However, this is often difficult
because many membrane proteins are present at low levels in membranes, and
expressing recombinant proteins in host cells such as E. coli often leads to severe
problems such as protein precipitation and improper posttranslational modifications.
Those membrane proteins that are naturally abundant in biological membranes such
as enzymes involved in energy metabolic processes like photosynthesis and aerobic
respiration were among the first ones whose structures were solved successfully,
including CcO from bovine heart mitochondria and from the soil bacterium
Paracoccus denitrificans (Iwata et al., 1995; Tsukihara et al., 1995). Therefore, the
ability to produce high yields of membrane protein from the expression system is

often crucial in obtaining a crystal structure successfully. Since a membrane protein

33

may undergo active posttranslational modification, such as glycosylation and
proteolytic cleavage, efforts must be made to obtain a molecularly homogeneous form
of the protein as a suitable candidate for crystallization. There is also accumulating
evidence that the nature of the lipids that make up the membrane may be critical for
expression and correct folding of membrane proteins (Dowhan et al., 2004).

1.2.2.2 Membrane Protein-Detergent Complexes

In order to extract membrane proteins from the bilayer into aqueous solution,
detergent molecules are needed to solubilize the proteins and therefore crystallization
of a membrane protein is essentially crystallization of the membrane protein-detergent
complex (PDC). Membrane protein molecules, once out of their native membrane
bilayer, are often less stable due to loss of their native conformations. The search for a
suitable detergent that can maximally retain a stable and homogeneous conformation
of a particular membrane protein can be extremely time-consuming.

A detergent molecule is a small, amphipathic molecule with a hydrophilic
headgroup and a hydrophobic tail which can interact with the transmembrane portion
of a membrane protein (for reviews, see le Maire et al., 2000; Garavito and
Ferguson-Miller, 2001; Gohon and Popot, 2003). Detergent molecules are grouped
into three classes depending on their biochemical features of their headgroups, namely,
ionic, nonionic, and zwitteronic detergents. Detergents with nonionic sugar based
headgroups, such as alkyl glucosides or alkyl maltoside, are often found to have
relatively little denaturing effects on membrane proteins and therefore are often used

in both membrane solublization and membrane protein crystallization. Zwitterionic

34

detergents, e.g. LDAO, have been successfully used in quite a few membrane protein
crystallization experiments although they are considered to be more inactivating.
Until now, there has not been a magic detergent which leads to universal success in
membrane protein crystallization and the choice of detergent is usually by extensive
screening. Moreover, it is very common that the detergent used in membrane
solublizaton is different from what is used for crystallization trials.

Apart from the chemical composition of the detergent headgroup, the chain
length of a detergent’s hydrophobic tail is also of critical importance in membrane
protein crystallization due to the difference in the interaction of the alkyl tail with the
transmembrane portion of the protein as well as the effect on the overall size of the
PDCs. During the crystallization experiments of PdCcO, crystals were obtained in
both dodecyl (C12) maltoside and undecyl (Cu) maltoside detergents. Strikingly, the
one carbon difference in chain length was responsible for a drastic difference in the
X-ray diffraction limit of 8 A using dodecyl maltoside compared to 2.5 A using
undecyl maltoside. Additionally, when decyl (C10) maltoside was used, no crystals
were obtained (Ostenneier et al., 1997). Such strong dependence on chain length of
alkyl tails of detergents has been reported in several other membrane protein
crystallization experiments as well (Shinzawa-Itoh et al., 1995; Marone et al., 1999).

Detergent molecules are surfactants with their own particular phase behaviors.
The most important one is the critical micelle concentration (CMC), above which
detergent monomers tend to form into micelles and an increase in detergent

concentration will only increase the micelle concentration with no change in the

35

dc

$03

1111

CL.)
'IQ

ell

det

Oil:

Cr}
beh
“111

User

‘Car.

detergent monomer concentration. In a detergent-solublized membrane protein
solution, there exits a dynamic equilibrium between detergent monomers, detergent
micelles and protein-detergent complexes. It is conceivable that too many detergent
micelles in the crystallization mixture is bad for intermolecular contacts during crystal
nucleation and growth, while insufficient detergent can lead to instability and rapid
aggregation of membrane proteins. Therefore, the detergent concentration used in the
crystallization mixture is of critical importance in obtaining highly ordered membrane
protein crystals, and because the CMC of detergents is affected by protein, salts, and
other crystallization reagents as well as the presence of other detergent(s), the entire
system is very complicated and an exact optimal detergent concentration can only be
worked out by repeated trial and error procedures.

Another important characteristic of detergents is the cloud point, where
detergent molecules from a single phase solution partition into two immiscible phases,
one detergent-rich and the other detergent—depleted, possibly due to the interactions
between the detergent headgroups. The consolute boundary where such phase
behavior occurs depends on temperature and the presence and concentration of
crystallization reagents such as polyethylene glycol, salt, etc. Unwanted phase
behavior in a crystallization solution can be detrimental to the crystallization efforts
with membrane proteins, while attractive detergent micelle interactions can also be
useful in membrane protein crystallization as some membrane protein crystals are
formed under the conditions close to the consolute boundary of the detergent

(Garavito and Picot, 1990).

36

Close to half a century and approximately 25,000 structures later, successful
macromolecular crystallization to date still largely occurs through a process of trial
and error. Although progress has been made in understanding the different
theoretical aspects of crystallization in order to guide our practices, protein
crystallization remains empirical with many variables that can affect crystal formation
and X-ray diffraction quality, such as concentrations of protein sample and
precipitating reagents, pH values, temperature, and exogenous crystallization
additives, etc. When detergent is added into the system, the crystallization of
membrane proteins (membrane protein-detergent complexes) involves several-fold
more variables. Therefore, obtaining membrane protein crystals that diffract X-rays to
high resolution is often a combination of hard work and serendipity.

1.2.3 Progress in Membrane Protein Crystallization

Since the first reported membrane protein crystal structure more than 20 years
ago (Michel, 1982), a great amount of progress has been made in membrane protein
crystallography which has produced many new high resolution crystal structures. Very
high resolution (<2 A) membrane protein structures have been obtained for a number
of membrane proteins including bovine cytochrome c oxidase (T sukihara et al., 2003).
Currently the membrane protein structure with the best resolution is the structure of
the ammonia channel at 1.35 A resolution (Khademi et al., 2004). Besides extensive
screening of detergents and other crystallization conditions, methods that can help
increase the diffraction resolution of soluble protein crystals are also generally helpful

in membrane protein crystallization. Micro- and macro-seeding (Fromme and Witt,

37

1998), crystal dehydration (Kuo et al., 2003), heavy atom soak (Chang and Roth,
2001), and semi-directed mutagenesis of selected surface residues (Pautsch er al.,
1999) have all been successfully applied to obtain diffracting crystals of individual
membrane proteins or improve crystal diffraction. New reagents and concepts are also
used that target protein-detergent complexes. Besides the conventional crystallization
methods, several novel techniques have been developed that aid in membrane protein
crystallization.

1.2.3.1 Manipulation of Protein-Detergent Complex

It was proposed that small arnphiphiles such as 1,2,3 heptanetriols have the
ability to incorporate themselves into the membrane protein-detergent complex and
decrease the size of the PDC, which is helpful in membrane protein crystallization
(Trmmins et al., 1991; Cast et al., 1994; Rosenow et al., 2003). The use of
heptanetriol facilitated the crystallization of the first membrane protein, the
Rhodopseudomonas viridis photosynthetic reaction center (Michel, 1982), as well as
cytochrome c oxidase from Paracoccus denitrificans (Iwata et al., 1995) and it is
commonly used as an additive for membrane protein crystallization experiments.

It is proposed that interactions between detergent micelles are responsible for
the cloud point behavior observed in a detergent system when temperature or other
conditions change (Loll et al., 2001). Methods and crystallization screenings have
been developed to take advantage of this attractive force that promotes interactions
between detergent molecules in PDCs to facilitate crystallization (Wiener and Snook,

2001).

38

1.2.3.2 Antibody Assisted Membrane Protein Crystallization

One of the main reasons membrane proteins are hard to crystallize is their lack
of polar residues on the protein surface of the transmembrane portion, which are
essential in forming inter-molecular crystal contacts. The concept of
antibody-mediated crystallization is to expand the hydrophilic surface area of a
membrane protein to form more crystal contacts. It may also help to stabilize a
particular conformation of the protein. In the crystal structure of the four subunit
cytochrome c oxidase from Paracoccus denitrificans, all crystal contacts were found
between attached antibody molecules (Ostenneier et al., 1995). Such a technique was
also successful in crystallization of the two subunit form of PdCcO (Ostenneier er al.,
1997), yeast cytochrome bcr complex (Hunte et al., 2000), and KcsA K+ channel
(Zhou et al., 2001). A variation of the antibody mediated crystallization is the fusion
protein approach described in the crystallization of E. coli cytochrome b03 ubiquinol
oxidase, in which case a small peptide (peptide Z) was engineered into the membrane
protein with the hope of forming crystal contacts (Byme et al., 2000). The fusion
protein approach has not produced well-diffracting crystals yet, presumably because
of the flexible nature of the linker region.
1.2.3.3 Crystallization of Membrane Protein in Lipid Cubic Phase

Crystallization of membrane protein in lipid cubic phase was first successfully
applied to bacteriorhodopsin (Landau and Rosenbusch, 1996). In this method, a
bicontinuous membrane formed by mixing lipids and water is used as a support

matrix for nucleation and crystal growth. The exact mechanism of how membrane

39

protein crystals grow from the lipid cubic phase is unclear. Since membrane proteins
tend to be less stable when they are extracted from native membranes and
incorporated into detergent micelles, crystallizing membrane proteins in a quasisolid
membrane environment throughout the crystallization process seems to have an
advantage. This method has been particularly successful with a class of membrane
proteins with seven robust transmembrane helices including bacteriorhodopsin
(Landau and Rosenbusch, 1996; Luecke et al., 1999), halorhodopsin (Kolbe et al.,
2000), and sensory rhodopsin H (Gordeliy et al., 2002). Its applicability to general
membrane protein crystallization is being explored.

1.2.4 Membrane Lipids and Membrane Protein Crystallization

1.2.4.1 Overview of Membrane Lipids

The membrane bilayer is formed by various types of membrane lipids. The
composition of membrane lipids can vary between eukaryotes and prokaryotes and
between cell organelle membranes and plasma membranes. Besides, the growth
conditions of bacteria have a profound impact on the lipid composition of bacterial
cell membranes (Benning et al., 1995).

Phosphoglycerides are the most common membrane lipids. They all have a
glycerol backbone and fatty acids are esterified to the hydroxyls at the snl and sn2
positions. At the sn3 position, the hydroxyl is linked to a phosphate group, which is in
turn linked to different polar headgroups to form different phosphoglycerides
including phosphatidyl ethanolamine (PE), phosphatidyl choline (PC), phosphatidyl

inositol (PI), phosphatidyl serine (PS), phosphatidyl glycerol (PG) and

40

bis-phosphatidyl glycerol (cardiolipin; CDL). The structures of these phospholipids
are shown in Figure 1.9. Although PC is rarely found in bacteria, it is quite common
in R. s., while PI is not found in many bacteria including R. s. (Haselkom et al., 2001 ).
Besides phosphoglycerides, in animal cell membranes there are also
phosphosphingolipids and glycosphingolipids. Another important lipid group found in
chloroplast and bacterial membranes is glycoglycerolipids, including monogalactosyl
diglyceride (MGD), digalactosyl diglyceride (DGD) and sulfoquinovosyl
diacylglycerol (SQDG) (Figure 1.10). Sterols are also found in membranes, with
cholesterol being the most commonly found type in membranes of cells and
organelles such as lysosomes, endosomes and the Golgi complex.

In R. s. membranes, except for phospholipids and glycolipids,
omithine-containing lipids as well as betaine lipids can also be found in substantial
amounts especially under phosphate-limited conditions (Benning et al., 1995) (Figure
1.10). Monounsaturated and saturated fatty acids are the dominant fatty acid species
found in R. s. lipids with oleic acid (18: 1) being the most common (Irnhoff, 1991).
1.2.4.2 Membrane Lipids and Membrane Protein Crystallization

In the early days of membrane protein crystallography, extensive efforts were
made to obtain a clean, lipid free membrane protein sample after purification. This
concept is now being questioned as more and more evidence suggests that native
membrane lipids may be an important component in membrane protein crystallization.
Membrane lipids not only provide the matrix for membrane proteins to reside in and

serve as a diffusion barrier, they also have a profound functional impact on the

41

H

R, —c.. o - CH2 Phosphatidyl Ethanolamine (PE)
I

n .. -O-CH
2 fl 1 fi
0 uzc-O--ll’-O-Cr12--Chlz-Nfls+

0..

fl

R1"C- 0— C712

R -—0-CH
2 fl 1 l?
0 HZC—O~l|’-0—CH2— CH2-N+(CH3)3

fl

R1-C— 0- C712

n .. —o—cn
2 fl 1 fi
0 nzc- o — I" — o - an— CHOH-CH3OH

Phosphatidyl Choline (PC)

0“

Phosphatidyl Glycerol (PG)

0..

fl Phosphatidyl Serine (PS)
R1 “C— 0 - (3712

R —-J)—CH
2 fl 1 :(i +
O nzc—O~I|’—O-CH2—IC3H-NH3

COO'
Cardiolipin (CDL)
0
ll 11
R1 ~C-O— FHZ HzCl—O— C—-R3
“2 14”.“ ii i “00'. 194%
o HZC—O—r—O—CHZ—CHOH-CHZOr—O—CHZ o

O...

O‘ 0‘

Figure 1.9: Structures of phospholipids found in Rhodobacter sphaeroides.

42

 

Fl:

 

 

 

$112503- Sulfoquinovosyl Diacyl glycerol

0 (SQDG)
H
HO
0 “ 9‘2 ii
OH
HF - O - C - R1
HZC— O — fi -— R2
0
$112 OR Monogalactosyl Diacyl glycerol
(MGD)
HO O O - CH2 101
n np—o—c—nI
H2C" o - ICl ~’ R 2
OH 0

Di galactosyl Diacyl glycerol
(DGD)

P's“ ..... o.
2;. >9?" ore 1’1
H HF-—O-C-—Rl

IIZC--O—fi--R2
OH O

coo fl Omithine Lipid

np—NH—C—pn, H
(c112)3 HF—O—C—RZ

N113 n1

Figure 1.10: Structures of non-phospholipids found in Rhodobacter splitter-aides.

43

membrane protein, by maintaining lateral pressure to support the conformation of a
membrane protein and by building to specific sites. Sometimes delipidation can cause
loss of activity and dissociation of membrane protein subunits as shown in
cytochrome c oxidase (Robinson, 1982; Sedlak and Robinson, 1999) and the
cytochrome bcl complex (Yu and Yu, 1980; Schagger et al., 1990). Lipid molecules
are important in maintaining a more stable and homogeneous conformation of the
membrane protein, which is of critical importance in crystallization. In fact, lipid
molecules are resolved in many high resolution membrane protein crystal structures
including cytochrome c oxidase from a few different sources (Yoshikawa er al., 1998;
Harrenga and Michel, 1999; Svensson-Ek et al., 2002). Moreover, as the X-ray
diffraction limit of the bovine CcO crystal increases, more and more lipids are found
in the crystal structure, which may suggest a causal role for bound lipids in obtaining
well diffracting crystals (Tsukihara et al., 1996; Yoshikawa et al., 1998; Tsukihara et
al., 2003). Therefore, during detergent solublization and protein purification steps,
retention of the specifically bound lipid molecules on membrane proteins may be
critical. Moreover, in crystallization experiments for the Ca2+ pump and the
cytochrome be complex, lipids were added into the crystallization mixture in order to
get better diffracting crystals (Toyoshima et al., 2000; Zhang et al., 2003).
1.2.4.3 Lipid Analysis of 0:0 Using Thin Layer Chromatography and Mass
Spectrometry

As the importance of lipids in membrane protein crystallization is getting

more and more attention, efforts are being made to analyze the lipid components in

the CcO preparation. Results from such studies provide critical information about the
bound lipids on CcO under different bacterial growth conditions and at different
purification stages using various purification techniques. Thin layer chromatography
is a common method of separating lipid species based upon their different
hydrophobicities. Lipid extracts from different CcO samples were subjected to thin
layer chromatography and semi-quantitative results on the amount of bound lipid on
the protein samples could be obtained (Hilmi, 2002).

Matrix-assisted laser desorption-ionization time of flight mass spectrometry
(MALDI-TOP) is a sensitive and relatively quick method of obtaining lipid
component information in a membrane protein sample. It only requires picomole
amounts of sample without the need for lipid extraction, thereby reducing the chance
of lipid loss due to oxidation during lipid extraction processes. Combined with the
optimal choice of matrix, it is sometimes capable of showing molecular ions that
represent the lipid-bound protein subunits (Distler et al., 2004). Although MALDI
mass spectrometry is generally considered to be not quantitative, efforts are being
made to find the appropriate controls for quantification. Lipid detection using
MALDI-TOF mass spectrometry could be used as a routine assay during the
membrane protein purification at each step and provide valuable information to
improve the purification procedure in an effort to retain important bound lipid species.

Another ionization technique used in mass spectrometry could also be useful
for detecting lipid content in membrane protein samples. Electrospray ionization (ESI)

has the advantage of being both sensitive and quantitative. However, this method

45

requires that the lipids be extracted from the membrane protein sample, which could

lead to the oxidation and degradation of lipids.

Chapter 2. METHODS
2.1 Molecular Engineering of Various Strains of Rhodobacter sphaeroides that
Produce 0:0 with Various Subunit Contents

The molecular engineering of various strains of Rhodobacter sphaeroides was
performed by Dr. Carrie Hiser in our lab. Figure 2.1 shows the amino acid sequences
of the four subunits of R. s. cytochrome c oxidase. The positions of the engineered
histidine tag at either the C terminus of subunit I or the shortened C-terminus of
subunit H are shown in blue and red, respectively. For some of the constructs
expressing subunit 1 histidine-tagged CcO, the truncated form of the subunit H was
incorporated into the overexpression plasmid. The truncation site is indicated by a
space between Y287n and E288". Two different forms of subunit IV were cloned and
inserted into the overexpression plasmid under the control of the subunit I promoter:
the long form (complete subunit 1V sequence) and the short form (complete sequence
of subunit IV less the underlined N terminal 10 amino acid residues).

In order to obtain crystals of CcO that diffract X-rays to high resolution,
different strains of R. s. that express forms of CcO with different subunit contents
were made. All four subunits are inserted into the overexpression plasmid (Zhen et al.,
1998) in different combinations, with the histidine tag attached to either the C
terminus of subunit I or the shortened C terminus of subunit H. The overexpression
plasmid was then transformed into R. s. cells via biparental conjugation (Shapleigh
and Gennis, 1992; Hiser et al., 2001). Table 2.1 summarizes the different strains of R.

s. and the subunit contents of CcO produced.

47

.3968qu 68 was 0:3 E 58% fl = €58

can _ “833 .«o 35883.0 05 8 v2.83 ms 05252 2: mo common 23. .092 .3 82.258 Eva e£E< "fin 0.5!..—
mz¢qflqmHA¢¢>H>>deB>Em>mwmmBMMOOBHQEmw¢>mwm¢mmmmodZMUHuMZHHmZ.
"\Kn.uwd::a5m

OGBHNHmmmqm43>>o>mEBM3Hm¢m
hw>mxmm8mmmm0>quo>dqhHBO>H>mmOIm09¢EhmZ¢GwHzwdmwmddmmwmw¢0m>8mAflDQ¢H¢A¢AUS¢>Qmm
zzmm>qflmmd38¢40wquqHABZHqumBmQhBHHammmmHwQHHmmmmUmeqdmmmmmzmzflmbm2>mm2Hdem03
mdoqm>>mB$QOMQWME>>DWSBUhEBMQ>>>AOHQOZBmOmom234>¢wqu>hflu>mm2mmgHmmmqHmMQmZM¢$<Z

"HHH uncensm

Ammmmmmv
m>m>w<aemmg>mmgm whoomeoqumaxemmm>>m>eHaZwemmHooqmmoowmmmemmmm
manomqmom>mmom>wmmm>azmmH>o¢ue>o>>>emz>m>>2¢eoemqqmmommwwemquom>mmmzmzaowemmme
zwmmmmHmmoawmwwzwzowoe>m>e>ommmHmooznq>mqmm¢me>qH>Hm>Hazequamzmeamma>mzmemmmmz
>quque>nHeHmeHH>quwoqzmHoH¢>mmammoaoewomOmmeHmgmooommmmammqqummuamuuaqaammmz.

"HH onesnsm

 

Ammmmmmva + mflmmmm
gammmmAOMhHmmmmmmeg93mgBOd$MZSWZZ<B>m¢UmedfiwhH>wqmehmflmmqmflwqmm>mzzedm¢mememm
mzwoqummommmBAZ¢0>m223mmqmwfizmmwomOmZmOHBhwH0¢mme>¢wqm2>wmmm¢>>wweomwwma>m¢0mq
>HQB>UD>BMAmAmGQ¢ZAEmqumHmeBZBmHBmmHMHUEm>¢H>ZHdZZMMmOOquQU¢BWmem2>>hDA>0H¢>
2¢w>2mqumHmmmmMH¢H>mm>me¢mq>HHw>mmmmmmBquOwQ>mewwmmOhmHewmzmQBAASBHm0<A>mA¢A
AHQB¢B>mngqum>xm2926m¢m224bBBHZZHflwAHmmfiwmqm>mmquQBmwwmmmqummMQ>30meOQOUZUU
mflmqm<>¢qmb¢>wqgwmAZZmemmZQmmwqumzbwzwhwwm4¢mH>>hmZZQmeOBHZ>ZBAmOZmBozm><mm3
Am0bh0¥>qwmMAmm¢02m0>0mdzqm2mzw>9m¢>wqu>quBhAMA>UHammZHmZmszhmwmmommmwmHddfldz

" H banshee

$6.5m £5 5 18: 9828958 3553 :2: can 23.5w .u. .k 2.98.59 an." 035,—.

832.3858 33 3 3 cocoa mat—am 33:38 988m 05. 5:5 @533 8% >H EH33 ” w.
covoo wfitfim 3988a “we.“ 05 SE 9:53 28w 2 “EBB 80388 n

a “835.0. 033: no 352:8 U 05 8on we 2.. 3 c8855 338me m 8.893

a
m BEBE 033: no 33 mmrv «BEES O 30: 05 8 vow—03am ms 02653 H s.
H a

388.8“ 0 05 8 @3083 m8 0:633 5:5 H 52.53 3.53:.

822335 a so .bam 6 we: a e deem E «a: space can flange 8 absence

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I Lease ._. cease scene a cam Smog >2 2 use? ER
canoe... Ease ._. concede seen: a 8m scion >2 2 $22 $32
a toe Leanne s assess Reese u com 85% Pa me .s 88» a:
sacs Lease ... canoes 35a: meow Eamon ESE .s8mww 52
s eons sees: accesses 3 358 Sam mcfioa H2 E .s 8me m2
swag Essen: mosses a gene 2% cflmoa ESE .soomN» c2
swoon Ease message s 358 Sam ozmoa 2 ._ 2:2 a:
a coca Baa: Lease a scene 2% Rice 2 .s 02$ 5
a was 838 Lease a 538 33 85% 2 ._ 85H 02

I Lease message a Reese Sam ammo“. Be me .s 83» 78
3 com S new 2 com 2% condom 252 Season 33% scaz

modemoaeoo unseen madam change a seam Lassen anew

 

 

 

 

 

 

49

2.2 UV-Visible Spectroscopy

UV-Visible spectra were taken using a Perkin-Elmer Lambda 40F UV-visible
spectrophotometer. For a reduced minus oxidized CcO difference spectrum, sodium
dithionite and ferricyanide were used to fully reduce or oxidize the enzyme samples,
respectively. The enzyme was first diluted in a buffer containing 100 mM HEPES, pH
7.9, 1 mM EDTA, and 0.1% dodecyl maltoside prior to the spectra being taken. The
extinction coefficient used was 132606-630 = 24 cm'lmM'1 (Zhen et al., 1998). For
absolute spectra of the dithionite-reduced CcO, the extinction coefficient used was
A2606-“ = 40 cm'lmM'l (Zhen et al., 1998).
2.3 SDS-PAGE

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was
used to detect the C00 subunit composition including different forms of subunits and
the impurities contained in CcO sample. Approximately 10 pg of protein in sample
buffer containing 25 mM Tris, pH 6.5, 40% glycerol, 8% SDS and 0.08%
bromophenol blue was loaded onto an 8% acrylamide stacking gel at pH 6.8 on top of
an 18% acrylamide separating gel at pH 8.8. The gel was electrophoresed at 100 V at
room temperature (Peiffer et al., 1990). The gel was then stained with Coomassie blue
and destained with 7.5% glacial acetic acid for 3 times (Hiser et al., 2001).
Polypeptide sizes were estimated from low-molecular weight range markers from
Bio-Rad.
2.4 Cytochrome c Oxidase Activity Assay

The oxidase activity of Rch0 was measured polarographically by using a

50

Gilson model 5/6H oxygraph at 25°C in a 1.75ml reaction cell containing 50 mM
KH2P04, pH 6.5, 0.05% dodecyl maltoside, 2.8 mM ascorbate, 0.55 mM TMPD and
30 M of horse heart cytochrome c. The oxidase concentrations were in the range of 5
nM to 50 nM.
2.5 Protein Concentration Assay

To measure the total protein concentration in the membrane sample, the BCA
(bicinchoninic acid) Protein Assay Reagent Kit from Pierce was used (Smith et al.,
1985). Such an assay kit has detergent-compatible formulations. Membrane samples
and bovine serum albumin (BSA) standards were diluted in a buffer containing 10
mM Tris, pH 11.2, 40 mM KCl, and 0.25% deoxycholate. Two ml mixture of reagent
A containing 1% BCA-N32, 2% Na2CO3.HzO, 0.16% N32 tartrate, 0.4% NaOH,
0.95% NaHC03, pH 11.25 and reagent B containing 4% CuSO4-5 H20 mixed in a
50:1 ratio was added to each of the diluted sample tubes and BSA standard tubes. All
the tubes were incubated at 60°C for 30 min and the absorbance at 562 nm was
measured. Protein concentration was calculated for the membrane samples against
the standard curve generated with the absorbance measurements at 562 nm for the
BSA standards with known protein concentration.
2.6 Phosphorous Assay

Phosphate contents in CcO samples at different purification stages were
measured using an adapted standard method (Bartlett, 1959). In a typical assay, 10 u]
of 480 M CcO were mixed with H20 to a final volume of 2 ml in a glass tube tightly

sealed with glass stopper. Water alone was used as the blank control. Samples and

51

blanks were heated in an oven at ISO-160°C for at least 3 hours. Fifty pl of
phosphorous-free hydrogen peroxide was added to each sample and the tubes were
tightly sealed again and incubated for at least another 1.5 hours to decompose all the
peroxide. After that, 4.4 ml of H20 was added to each of the tubes followed by 0.2 ml
of Fiske-SubbaRow reagent (15 g anhydrous sodium bisulfite, 0.25 g
1-amino—2-naphthol-4-sulfonic acid, 0.5 g anhydrous sodium sulfite in 100 ml water,
filtered and kept in dark) and 0.2 ml of 5% ammonium molybdate. The solutions were
mixed thoroughly and heated for 10 minutes in a boiling water bath. The tubes were
cooled and the absorbance at 830 nm was measured for each tube. The absorbance
readings at 830 nm were then compared to the phosphate standards containing 0, 0.02,
0.04, 0.06, 0.08, 0.10 wnole of KH2P04 treated with the same procedure to calculate
the phosphorous content in the CcO samples.
2.7 Mass Spectrometry Analysis of Membrane Lipids

All the mass spectrometry analyses of membrane lipids in protein samples and
re-dissolved RchO crystals using the matrix-assisted laser desorption ionization
(MALDI) method were performed by Xi Zhang at the Proteomics Facility in the
Department of Biochemistry of Michigan State University using PerSeptive
Biosystems Voyager STR in linear positive and negative ion modes. Post-source
decay (PSD) negative ion mode was also used to identify lipid species. Different
matrices including 2, 5-dihydroxybenzoic acid (DHB) and a

-cyano-4-hydroxycinnamic acid (CHCA), saturated in 50% water/50% acetyl nitrile,

52

were used. Samples spots were prepared by mixing equal volumes of matrix and CcO
sample. Lipid standards were purchased from Avanti Polar Lipids.
2.8 Growth and Harvest of R. 3. Cells

R. 3. cells were grown on Sistrom's plates with 50 ug/ml spectinomycin, 50
ug/ml streptomycin and l ug/ml tetracycline pH 7.0 at 30°C for 3-5 days. Cells were
picked up from the plates and inoculated in small flasks with 100ml of Sistrom's
media with the same antibiotics and grown for two days at 30°C with vigorous
shaking at 250 rpm. Fifty ml of starter culture were used to inoculate a 2.8 L Fembach
flask with 800 m1 Sistrom's media containing 1 ug/ml tetracycline and 25 ug/ml
spectinomycin, and 25 ug/ml streptomycin. The Fembach flasks were shaken at 250
rpm at 30°C for 2-3 days until the absorbance reading at 660 nm was above 1.7 and a
pH over 8.5. R. s. cells were harvested by centrifugation in a GS-3 rotor at 14,000 x g
for 20 minutes at 4°C. The cell pellets were resuspended in buffer containing 50 mM
KI-12P04 and 1 mM EDTA, pH 6.5 and stored at -80°C (Zhen et al., 1998).

Sometimes the R. s. cells in a liquid culture undergo genetic recombination in
its overexpression plasmid. As a result, the gene(s) encoding all or part of CcO can be
cleaved off from the overexpression plasmid, leaving essentially no CcO expression.
In order to circumvent this problem, prior to inoculating liquid cultures, R. s. cells
were streaked on a Sistrom’s plate to single colonies. Small single colonies which
contained the intact overexpression plasmid were picked up and inoculated onto
another Sistrom’s plate. The plate with R. s. cells from a single small colony was then

used to start up small liquid culture. On the other hand, big colonies, which had

53

presumably cleaved overexpression plasmid and therefore grew faster on the
Sistrom’s plate, were discarded. Alternatively, R. 3. cells from an entire plate that was
grown from a single small colony were picked up and inoculated into a big flask
directly in order to minimize the chance of the occurrence of the cleavage of the
overexpression plasmid.
2.9 Preparation of R. s. Cytoplasmic Membranes

Resuspended R. 3. cells were thawed and small amounts of DNase I and
RNase were added. The cell resuspension was then homogenized and cells were
broken with two passages through the French press at 20,000 psi. Whole cells and
debris were removed by centrifugation at 30,000 x g for 30 minutes at 4°C. The
supernatant was collected and cell membranes were pelleted by using
ultracentrifugation at 200,000 x g for 1.5 hours. Depending on whether the histidine
tag was attached to the C-tenninus of subunit 1 or to the shortened C-tenninus of
subunit II, the pellet was resuspended in buffer containing 10 mM Tris, pH 8.0 with
either 40 mM KCl or 220 mM KCl, respectively. The membrane resuspensions were
quick-frozen in liquid nitrogen and stored at —80°C.
2.10 Detergent Solublization of R. s. Cytoplasmic Membrane

Before detergent solublization of the cell membranes, the protein
concentration in the membranes was analyzed by using the BCA Protein Assay
Reagent Kit as described earlier in this chapter. The R. 5. cell membrane sample was
diluted to a protein concentration of 10 mg/ml with buffer containing 10 mM Tris (pH

8.0), 40 mM or 220 mM KCl depending on the position of the histidine tag as

54

described above, and 1 mM imidazole. Dodecyl maltoside was then added to the
membrane resuspension sample at a final concentration of 1% (w/v). The solution was
stirred for 20 minutes at 4°C and unsolublized material was removed by
ultracentrifugation at 200,000 x g for 30 minutes. The supernatant containing
RchOwas collected and used immediately for further column purification.
2.11 Column Chromatography for Purification of Enzyme Used for
Crystallization of the Four Subunit RchO

In order to obtain crystals of the four subunit enzyme complex, a two-step
column purification protocol was used. The following protocols were adapted from
what was described previously (Svensson-Ek et al., 2002). To purify RchO that had
a histidine tag attached to the C-terminus of subunit 1, the first step was Ni-NTA
affinity column purification and the second step was either Mono Q ion-exchange
column purification, or DEAE Sepharose ion-exchange column purification. To purify
RchO with a histidine tag attached to the shortened C-terminus of subunit II, the first
step was Ni-NTA affinity column purification and the second step was either DEAE
Sepharose ion-exchange column purification, or Superdex 200 size-exclusion column
purification.
2.11.1 Purification of CcO with the Histidine-Tag Attached to the C-Terminus of
Subunit I for Crystallization
2.ll.ll.l Ni-NTA Column Chromatography

For enzyme with a histidine-tag attached to the C-terrninus of subunit 1, the

detergent solublized membrane sample containing typically 30-60 mg Rch0 was

55

loaded onto a home-packed Ni-NTA (Qiagen) column (typical volume: 18 ml)
connected to an AKTA FPLC system (Pharmacia) that was pre-equilibrated with 80
ml buffer A containing 10 mM Tris (pH 8.0), 40 mM KCl, 10 mM imidazole and
0.05% dodecyl maltoside. The column was washed extensively with 12 column
volumes of buffer A before a linear gradient of 0 - 100% buffer B containing 10 mM
Tris (pH8.0), 40 mM KCl, 150 mM imidazole, 0.05% dodecyl maltoside was applied.
The length of the gradient was typically 15 column volumes. The green fractions
containing CcO were pooled, and imidazole was removed by washing and
concentrating 2 times by using a Centriplus YM100 filter unit (MW cutoff 100 kDa;
maximum volume 15ml, Millipore). Sometimes, the green fractions from the Ni-NTA
column were simply pooled and loaded onto the next ion-exchange column without
any washing and concentrating, which will be described later in the Results section.
The yield of this column purification step was typically around 50%.
2.11.1.2 Ion-exchange Column Chromatography Using Either Mono Q Column
or DEAE Sepharose Column

Pooled and washed fractions of enzyme with a histidine tag attached to the
C-terminus of subunit I were loaded onto a Mono Q column (Pharmacia) (column
volume: 8 ml) pre-equilibrated with 50 ml buffer A containing 10 mM Tris, pH 8.0,
0.05% dodecyl maltoside. The column was washed with buffer A for two column
volumes before a linear gradient of 0-100% buffer B containing 10mM Tris, pH 8.0,
0.5M KCl, 0.05% dodecyl maltoside was applied. The length of the gradient was 20

column volumes. Green fractions from the linear gradient were then pooled,

56

concentrated and the buffer was exchanged to 10 mM Tris, pH 8.0, 0.16% - 0.24%
decyl maltoside by washing and concentrating 3-6 times by using a Centriplus
YM100 filter unit. The yield of this column purification step and the washing and
concentrating was typically around 50%.

When a DEAE Sepharose column was used instead of Mono Q column,
pooled and washed fractions of enzyme were loaded onto the DEAE Sepharose
column (Pharmacia) (column volume: 20 ml) pre-equilibrated with 100 ml buffer A
containing 10 mM Tris, pH 8.0, 0.05% dodecyl maltoside. The column was washed
with buffer A for three column volumes before a linear gradient of 0-100% buffer B
containing 10 mM Tris, pH 8.0, 0.5 M KCl, 0.05% dodecyl maltoside was applied.
The length of the gradient was typically 15 column volumes. Green fractions from the
linear gradient were then pooled, concentrated and the buffer was exchanged to
10mM Tris, pH 8.0, 0.16% - 0.24% decyl maltoside by washing and concentrating 3-6
times by using a Centriplus YM100 filter unit. Typical yield for this column
purification step with washing and concentrating is around 50%.

2.11.2 Purification of 0:0 with the Histidine-Tag Attached to the Shortened
C-Terminus of Subunit II for Crystallization
2.11.2.1. Ni-NTA Column Chromatography

For enzyme with a histidine-tag attached to the shortened C-terminus of
subunit II, the detergent solublized membrane sample containing typically 30 — 60 mg
Rch0 was loaded onto a home-packed Ni-NTA (Qiagen) column (typical column

volume: 18 — 20 ml) connected to an AKTA FPLC system that was pre-equilibrated

S7

with 80 ml buffer A containing 10 mM Tris (pH 8.0), 220 mM KCl, 2.5 mM
imidazole and 0.05% dodecyl maltoside. The column was washed extensively with 12
column volumes of buffer A before a linear gradient of 0 — 100% buffer B containing
10 mM Tris (pH8.0), 220 mM KCl, 150 mM imidazole, 0.05% dodecyl maltoside was
applied. The length of the gradient was 15 column volumes. Often there were two
very closely associated peaks in the elution profile. In order to completely separate the
two peaks, sometimes a two—step linear gradient was applied. First the percentage of
buffer B was increased to approximately 8-10% (approximate imidazole concentration:
15 mM) and held constant for approximately 6 column volumes to elute off the first
peak containing cytochrome c oxidase bound to a major contaminating species. After
this washing step, the concentration of buffer B was further increased to 100% (final
imidazole concentration: 150 mM) with a linear gradient (the length of the gradient:
13 column volumes) and the second peak containing purer Rch0 was eluted and the
green fractions under the second peak were pooled. Typical yield after this column
purification step is approximately 40% - 50%.

If the second purification step was a DEAE Sepharose column, the fractions
from the Ni-NTA column were pooled and diluted 4-5 times with buffer containing 10
mM Tris, pH 8.0, 0.05% dodecyl maltoside and loaded onto the DEAE Sepharose
column. If the second purification step was Superdex 200 size exclusion
chromatography, the fractions were pooled and concentrated to approximately 1 ml in
volume in a stirred ultrafiltration cell under N2 pressure by using a YM100

ultrafiltration membrane (MW cutoff 100 kDa; maximum volume 50 m1, Millipore).

58

 

 

0v

 

2.11.2.2 Ion-exchange Column Chromatography Using DEAE Sepharose Column

Pooled fractions of enzyme were diluted as described above and loaded onto
DEAE Sepharose column (column size: 20 ml) pre-equilibrated with 100 ml buffer A
containing 10 mM Tris, pH 8.0, 0.05% dodecyl maltoside. The column was washed
with buffer A for three column volumes before a linear gradient of 0-100% buffer B
containing 10 mM Tris, pH 8.0, 0.5 M KCl, 0.05% dodecyl maltoside was applied.
The length of the gradient was 15 column volumes. Green fractions from the linear
gradient were then pooled, concentrated and the buffer was exchanged to 10 mM Tris,
pH 8.0, 0.16% - 0.24% decyl maltoside by washing and concentrating 2 - 3 times by
using a Centriplus YM100 filter unit or in a stirred ultrafiltration cell as described
above. The typical yield of this column purification step with washing and
concentrating was around 50%.
2.11.2.3 Superdex 200 Size Exclusion Chromatography

Fractions of enzyme with a histidine tag attached to the shortened C-terminus
of subunit II from the Ni-NTA column in 10 mM Tris, pH 8.0, 150 mM NaCl, 0.16%
decyl maltoside were loaded onto a Superdex 200 column (Pharmacia) (column
volume: 120 ml) pre-equilibrated with the same buffer. The same buffer was then used
to wash the column. Green fractions containing CcO were pooled and concentrated by
using a Centriplus YM100 filter unit. Typical yield after this column purification step
was around 60%.
2.12 Crystallization of the Four Subunit RchO

In a typical crystallization experiment to obtain crystals of the four subunit

59

RchO, hanging drop or sitting drop crystallization set-ups were used. Figure 2.2(A)
and Figure 2.2(B) show the hanging-drop crystallization set—up and sitting-drop
crystallization set-up, respectively. Both crystallization set-ups help to achieve
supersaturation of the protein solution through vapor diffusion. However, in a
sitting-drop crystallization experiment, a much larger volume can be used in order to
obtain larger crystals. All the crystallization experiments were set up in the cold room.
In a typical hanging-drop crystallization experiment, 2 pl of enzyme solution
containing 10 mM Tris, pH 8.0, 0.16-0.24% decyl maltoside and 120-160 M CcO
was mixed with 1 pl of reservoir solution containing 100 mM sodium citrate, pH
5.6-5.8, 100 mM NaCl, 18-22.5% PEG-400 and 1 pl of crystallization additive
solution containing 5% heptanetriol, 33 mM MgC12, and 0.026% dodelcyl maltoside.
The crystallization drop was incubated at 14°C overnight to increase the rate of vapor
equilibrium, then the temperature of the incubator was lowered to 4°C at a rate of 1°C
per hour and held at 4°C for up to 2 weeks. Triangular crystals of four subunit RchO
usually appeared after 5-7 days and continued growing to up to 0.2 X 0.2 x 0.1 mm in
approximately two weeks.

If the sitting drop crystallization set-up was used instead of the hanging-drop
method, the typical volume of the drop solution increased to 10 pl of enzyme solution
containing 10 mM Tris, pH 8.0, 0.16-0.20% decyl maltoside and 80-120 pM CcO
mixed with 5 pl of reservoir solution and 5 pL of crystallization additive solution as

described above. The crystallization dr0p was incubated at 4°C and triangular crystals

60

Hanging Drop Crystallization Set-up

Cover Slide
/

_- r
U \—\ Drop

Solution

 

Reservoir
Solution

 

 

 

 

B
Sitting Drop Crystallization Set-up

Cover Slide or
_- ésealing Tape

 

Drop
Solution

 

 

——__— — _ —— Reservoir

_ —_ _ : Solution

 

 

 

 

 

Figure 2.2: Hanging drop (A) and sitting drop (B) vapor diffusion crystallization
set-ups.

61

of the four subunit RchO started to appear after approximately 10 days and
continued growing to their full size of up to 0.3 X 0.3 x 0.15 mm in approximately 4
weeks.
2.13 Flashcooling of the Four Subunit RchO Crystals

Because the crystals were grown at 4°C, and would quickly dissolve under
room temperature, the flashcooling procedures were performed in the cold room. In
order to lessen the humidity in the cold room, which could lead to ice formation
during the crystal flashcooling, Drierite or liquid nitrogen was used as a desiccant.

Crystals of the four subunit RchO were picked up from the original
crystallization drop with a cryoloop and soaked in approximately 30 pL of stabilizing
solution in a sitting drop well. The stabilizing solution was made to mimic the
assumed concentration of ingredients in the crystallization drop after equilibrium
except for a little less detergent in a sitting drop well. In more detail, the stabilizing
solution contained 91 mM sodium citrate, pH 5.6-5.8, 91 mM NaCl, 18 mM Tris, pH
8.0, 30 mM MgC12, 2.5% heptanetriol, 0.16% decyl maltoside, 0.016% dodecyl
maltoside, and 18-20% PEG-400. A small aliquot of the cryosolution which contained
the same ingredients as those in the stabilizing solution except for a higher (30-32%)
PEG-400 concentration was then added to the stabilizing solution containing the
crystals and the two solutions were carefully mixed. A small aliquot was then taken
out from the sitting drop well to return the drop volume to 30 pl. The same procedure
was repeated several times until approximately the entire solution in the sitting drop

well had been exchanged to the cryosolution. The entire procedure took

62

approximately 8-10 minutes. Adding small aliquots of the cryosolution ensured the
gradual increase in PEG-400 concentration and thus protected the crystals from being
damaged due to dramatic increase in the osmotic pressure in the surrounding
environments. The crystals were then picked up again using a cryoloop and
submerged into liquid propane prechilled with liquid nitrogen or directly into liquid
nitrogen itself. The flashcooled crystals were then stored in liquid nitrogen in
cryovials until data collection.
2.14 Soaking of Four Subunit RchO Crystals in Cadmium Solutions Prior to
Flashcooling

Crystals of the four subunit RchO were soaked in a solution containing
cadmium in order to determine the cadmium binding site(s) responsible for the
observed inhibitory effect on the enzyme. In a typical soaking experiment, crystals of
the four subunit RchO were picked up with a cryoloop and transferred to
approximately 30 pL of stabilizing solution (see the previous paragraph for the
detailed concentrations of ingredients) containing varying concentration of CdClz
from 1.5 mM to 9 mM. The crystals were soaked in the stabilizing solution containing
CdC12 for half an hour to a few days to allow binding of Cd2+ to the enzyme. After
soaking, the crystals underwent the same flashcooling procedure as described in the
previous paragraph except that the cryosolution was supplemented with the same
concentration of CdC12 as in the stabilizing solutions. The crystals were flashcooled in

liquid nitrogen directly and stored until data collection.

63

2.15 Column Chromatography for Purification of Enzyme Used for
Crystallization of the [-11 Subunit RchO

To purify Rch0 in order to obtain crystals of the [-11 subunit RchO, a single
step of Ni-NTA affinity column purification was used. Different CcO forms all had
histidine tags attached to the shortened C-terminus of subunit II and the purification
protocol was similar to the modified Ni-NTA purification protocol for subunit II
histidine-tagged RchO. (section 2.1 1.2. 1)

The detergent solublized membrane sample was loaded onto a home-packed
Ni-NTA column (column volume: 18-20 ml) connected to an AKTA FPLC system
that was pre-equilibrated with 80 ml buffer A containing 10 mM Tris (pH 8.0), 220
mM KCl, 2.5 mM imidazole and 0.05% dodecyl maltoside. The column was washed
extensively with 12 column volumes of buffer A. A two step gradient of increasing
concentration of buffer B containing 10 mM Tris, pH 8.0, 220 mM KCl, 150 mM
imidazole, 0.05% dodecyl maltoside was applied. First the percentage of buffer B was
increased to approximately 8-10% (approximate imidazole concentration: 15 mM)
and held constant for approximately 6 column volumes to elute off the first peak
containing cytochrome c oxidase bound to a major contaminating species. After this
washing step, the concentration of buffer B was further increased to 100% (final
imidazole concentration: 150 mM) with a linear gradient and the second peak
containing purer CcO was eluted and the green fractions under the second peak were
pooled.

The pooled fractions containing reasonably pure RchO were washed and

concentrated in a stirred ultrafiltration cell as described above. The buffer was
exchanged to 10 mM Tris, pH 8.0, 150 mM NaCl, with 0.16% decyl maltoside by
washing and concentrating 2 - 3 times. The typical yield of purified protein is
approximately 30-50%.

2.16 Crystallization of the 1-11 Subunit RchO

In a typical crystallization experiment to obtain crystals of the I-II subunit
RchO, sitting drop crystallization set-up was used as shown in Figure 2.2(B). All the
crystallization experiments were set up in the cold room at 4°C. In a typical
crystallization experiment, 6 pl of enzyme solution containing 10 mM Tris, pH 8.0, 50
mM NaCl, 0.20% decyl maltoside and 120 pM CcO was mixed with 3 pL of reservoir
solution containing 100 mM MES, pH 6.3-6.9, 24-26% PEG-400 and 3 pl of
crystallization additive solution containing 5% heptanetriol, 32 mM MgC12, 1.3 mM
CdClz and 0.026% dodecyl maltoside. Heavy protein precipitation occurred
immediately after mixing of the drop contents.

The crystallization drop was incubated at 4°C and crystal showers of tiny
triangular crystals of the four subunit RchO started to appear after approximately 3-4
days. Bigger, football-shaped crystals of MI subunit RchO crystals started to appear
after approximately 2 weeks and continued to grow to their full size of up to 0.2 x 0.2
x 0.1 nun in 3-4weeks.

2.17 Flashcooling of the [-11 Subunit RchO Crystals
The flashcooling method of the [-1] subunit crystals was the same as the

method used for flashcooling four subunit crystals (section 2.13) except that 1.3 mM

65

CdClz was supplemented in both the stabilizing solution and the cryosolution.
2.18 Data Collection and Processing

X-ray diffraction data were collected at DND-CAT, Station SID-B, and at
COM—CAT, Station 321D-B, Advanced Photon Sources, Argonne National Laboratory.
The detector used was MARCCD-M165. The data set was collected by the rotation
method. During the data collection from the four subunit CcO crystals, the crystal was
rotated about the 4) angle for a total of 120° with 0.4° rotation per frame. During the
data collection from the MI subunit CcO crystals, an 180° rotation about q» angle was
performed with 0.3° per frame. Sometimes two data sets were collected on a single
crystal, one for the high resolution range with a crystal-to-detector distance of about
180 mm and a longer X-ray exposure time, and another data set for the low resolution
range with a crystal-to-detector distance of approximately 260 mm and a shorter
X-ray exposure time. The data were processed with Denzo and scaled and merged
using Scalepack (Otwinowski and Minor, 1997).
2.19 Molecular Replacement and Structural Refinement

For the crystal structure determination of the four subunit RchO, a rigid body
refinement using the coordinates of the published structure of the four subunit RchO
(PDB entry 1M56) as the initial phasing model was performed with CNS 1.1 (Brunger
et al., 1998), followed by simulated annealing, and cycles of energy minimization,
model visualization and manual model building. A group B factor refinement was
performed at later stages of the refinement. Molecular visualization and model

building were performed by using the program CHAIN (Sack and Quiocho, 1997).

66

For the crystal structure determination of the HI subunit RchO, an initial
molecular replacement search was performed by using the program Phaser (Storoni et
al., 2004), using the coordinates of the four subunit cytochrome c oxidase (PDB entry
1M56) as the original search model. The molecular replacement search was successful
and yielded two protein molecules per asymmetric unit. However, analysis of the
crystal packing after initial rigid body refinement indicates there were severe clashes
between CcO molecules in subunit III and subunit IV. Further analysis of the electron
density maps revealed that there was no subunit III or subunit IV in the crystal
structure. The coordinates of the polypeptide chains of subunit I and H were used as
initial phasing model for structural refinement. Refinement of the structure was
performed with CNS 1.1 using cycles of simulated annealing, and. energy
minimization (Brunger et al., 1998). B factors were refined at the later stages.
Molecular visualization and model building were performed by using the program
CHAIN (Sack and Quiocho, 1997). The structure was first refined at 2.5 A resolution
and then the resolution was extended to include all data up to 2.35 A. Towards the end
of refinement, restrained refinement with TLS refinement were performed using

Refmac5 (Bailey, 1994; Murshudov et al., 1997;\V1nn et al., 2001).

67

Chapter 3. RESULTS

3.1 X-ray Crystallography of the Four Subunit RchO
3.1.1 Importance of Subunit IV in Crystallization of the Four Subunit RchO
and Overexpression of Subunit IV

Before the crystal structure of cytochrome c oxidase from Rhodobacter
sphaeroides was solved by Iwata and colleagues (Svensson-Ek et al., 2002), only
three subunits (subunits I, H, and 111) had been observed in the protein complex
(Hosler et al., 1992), although CcO from a very closely related bacterium, Paracoccus
denitrificans, has four subunits (Ostermeier et al., 1995). The crystal structure of
RchO showed that there are indeed four subunits in the complex. As shown in Figure
1.4, the single transmembrane helix of subunit IV of RchO is partially surrounded by
transmembrane helices from subunit I and subunit III. However, there is little direct
protein-protein contact between subunit IV and its neighboring subunits. Most of the
interactions are mediated through the phospholipid molecules immediately
surrounding the transmembrane helix of subunit IV as shown in the crystal structure
(Svensson-Ek et al., 2002). Due to its lack of direct protein-protein contacts with its
neighboring subunits, subunit IV could easily dissociate from the protein complex if
those lipid molecules were lost during detergent solublization and column purification
steps. As a result, subunit IV may not have been detected earlier because it was either
completely absent or present but at a low stoichiometry in the purified protein. In fact,
in the usual gels run to assess the protein purity (14% acrylamide), the ~6,000 dalton

subunit is not observed (Hosler et al., 1992). In addition, it was difficult to stain

68

subunit IV with Coomassie Blue in SDS-PAGE and when it did stain, the band
representing subunit IV was always diffuse possibly due to different levels of lipid
molecules associated with the protein.

The loss of subunit IV in the RchO purification proved to be detrimental to
the formation of well-ordered four subunit RchO crystals. Analysis of the RchO
surface crystal contacts indicates that the key crystal contact occurs between the
bottom part of one molecule, the flat surface formed by subunits I, III and IV, and the
top of another, including the globular domain of subunit H. Figure 3.1 shows the
importance of subunit IV in forming the key crystal contacts mediated through the
interaction of the cytoplasmic portion (N terminus) of subunit IV and nearby residues
from subunit 1, II and HI from a second molecule. Therefore, it is intuitive to conclude
that loss of subunit IV leads to the inability of crystal formation.

Previously our lab was engaged in active pursuit of crystallization of RchO
using a strain of R. s. which overexpressed only subunits I, H and III. The purification
procedure we applied involved two column purification steps, a Ni affinity column
followed by a DEAE-SPW ion exchange column. The latter column was thought to
extract lipid from the enzyme sample. In retrospect, the main reason for the inability
to obtain RchO crystals may have been due to substoichiometric amounts of subunit
IV in the protein complex at the beginning of purification due to lower expression of
subunit IV compared with the other three over-expressed subunits, and/or loss of
subunit IV due to delipidation during the protein purification steps, particularly during

the DEAE-SPW column purification step.

69

 

Figure 3.1: Major crystal contact regions in the crystal of the four subunit
RchO. Subunits I, II, III, and IV are colored yellow, green, cyan, and purple,
respectively. The N-terminal region of subunit IV (purple) of one RchO molecule
takes part in the major crystal contact during crystal formation.

70

In order to obtain the RchO protein product which has all four subunits in
stoichiometric amounts, the subunit IV gene was cloned, sequenced and inserted into
the overexpression plasmid under the control of the strong subunit I promoter. It was
found that there are two potential starting codons in the gene sequence of subunit IV
in the R. s. genome and therefore there are two forms of subunit IV with different N
terminus (Figure 2.1). In an earlier attempt, the complete subunit IV gene, starting
with the first starting codon, was cloned and inserted into two different
overexpression plasmids, which were then conjugated into parental R. s. strains
individually, to make two new R. s. strains, 120 and 119. Strain 120 overexpresses all
four subunits with a histidine tag attached to the C-terrninus of subunit I and the other
three subunits at their full lengths, while strain 119 is identical except for a shortened
form of subunit 11 instead of the native full length subunit II (see Table 2.1 in Chapter
2, Method section). Successful expression of the plasmid would theoretically ensure
the presence of all four subunits in Rch0 in stoichiometric amounts. In fact, RchO
obtained from strains 120 and 119 did produce four subunit crystals when modified
protein preparation procedures were applied (see below).

3.1.2 Modifications of Detergent Solublization and Protein Purification
Procedures for Crystallization of RchO

Since lipids were found to be crucial in securing subunit IV in the enzyme
complex, efforts were made to try to standardize and optimize the protein purification
procedure in order to retain more lipids. The first step in protein preparation is

detergent solublization of R. s. cell membranes. This step is crucial in that too little

71

detergent fails to extract all the protein into the aqueous solution, while too much
detergent can completely remove the membrane lipid molecules. In order to find the
optimum detergent/protein ratio for solublization, a protein concentration assay was
performed using a modified BCA method (see section 2.5). After the protein
concentration determination, the R. s. cytoplasmic membrane sample was diluted with
buffer to make solutions with different protein concentrations. Different
concentrations of dodecyl maltoside were added and the effects of the amount of
detergent on the yield of purified protein and the protein crystallization ability were
observed. It was found that a final dodecyl maltoside concentration of approximately
1.0% in a protein solution of approximately 10mg/ml was suitable. At this relatively
low detergent concentration, not all of the protein was solublized from the membrane,
as seen from the fact that only about 50% of the enzyme bound to the Ni2+-NTA
column. However, using this level of detergent concentration, apparently more lipids
were retained during the process and crystals of the enzyme could be obtained
reproducibly.

Purification of RchO involved two purification steps, a Ni affinity column
purification followed by an ion exchange column purification. In order to obtain
crystals of CcO, the second column purification step was changed from a DEAE-SPW
column, which might be more lipid-extracting due to the more hydrophobic nature of
its column matrix material, to a Mono Q column, which was used by Iwata’s group to
obtain RchO crystals (Svensson-Ek et al., 2002).

Figure 3.2 shows the picture of a single crystal of RchO obtained from R. s.

72

 

Figure 3.2: Crystal of the four subunit RchO which belongs to a space group of
R3.

strain 120 using the modified purification protocols. The exact crystallization
conditions are as listed in section 2.12. The crystal has a triangular prism shape and
belongs to the space group of R3 with a unit cell dimensions of a = b = 340.7 A, c =
90.6 A, a = B = 120°, and y = 90°.

3.1.3 UVIVisible Spectrum and Enzymatic Activity Assay of Purified RchO

Figure 3.3 (A) shows the UV/visible spectra of purified oxidized RchO (blue)
and dithionite-reduced enzyme (purple) using the modified protein purification
protocol. The purity of the enzyme preparation was judged by the ratio of Ana/Am
ratio for oxidized enzyme, which was approximately 2.0. In the case of the reduced
enzyme, the Soret to a—band ratio (A44549o/ Amgo) was approximately 5.5. Both of
these values were similar to those reported previously (Zhen et al., 1998;
Svensson-Ek et al., 2002). Figure 3.3 (B) shows the reduced — oxidized difference
spectrum of purified Rch0 for crystallization. The at peak at approximately 606 nm
appeared to be normal (Hosler er al., 1992; Zhen et al., 1998).

The turnover number of the purified wild type Rch0 was determined
polarographically to be about 1400 s'1 at room temperature by measuring the oxygen
consumption rate. The turnover number was similar to that from wild type enzyme
purified using the previous purification procedure (Zhen et al., 1998).

3.1.4 Retention of Subunit IV during the Purification and Crystallization of
RchO
One important question to ask was whether the fourth subunit was

overexpressed and successfully retained after the protein purification procedure and

74

(A) 280

 

 

 

   

 

 

 

 

0.4 ~
445
g 0.3- 421 I
a
a
:5 0.2l
E
0.1
0 J l
260 300 350 400 460 500 660 600 660
Wavelength (nm)
(3) 606
0.04 — I
8 0.03 ~
.n
g 0.02 ..
U)
.9
<1
0.01 -
o r r r i 1 J r r
500 525 550 575 600 625 650 675 700

Wavelength (nm)

Figure 3.3: UV-visible spectra of purified RchO. (A): Oxidized (blue) and
dithionite-reduced (purple) spectra of RchO. The Ana/A421 in the oxidized spectra is
approximately 2, and the Soret to a-band ratio (A445490/ W30) is approximately 5 .5.
(B): Dithionite-reduced minus ferricyanide-oxidized difference spectrum (green) of
RchO.

75

whether the crystals formed had stoichiometric amounts of subunit IV in them. In
order to answer this question, both purified enzyme samples and re-dissolved RchO
crystals were subjected to a series of SDS-PAGE gels and the subunit contents were
analyzed.

Figure 3.4 (A) shows the SDS-PAGE of purified RchO from two different R.
s. strains: 119, which expressed all four subunits of Rch0; and 25-1, which was used
for crystallization trials previously and only overexpressed subunits I, II, and III.
RchO from strain 119 (Lane 1) was purified with both a Ni column and a Mono Q
column, while RchO from strain 25-1 (Lane 2) was purified with a Ni column only.
It can be seen that the intensity of the band representing subunit IV in Lane 1 was
greater than that in Lane 2 although the latter sample underwent only Ni column
purification. Therefore, the relative amount of subunit IV in CcO from R. 3. strain 119
is much higher than that from strain 25-1. Although quantitative assessment of the
subunit IV content in the enzyme preparation was not possible, the density of the
subunit IV band in SDS-PAGE did qualitatively indicate a higher subunit IV content,
presumably due to overexpression of its gene and the modified purification
procedures. Also in Figure 3.4 (A), two different bands with different apparent
molecular weights in SDS-PAGE can be seen. The upper band represents the long
form of subunit IV, which was overexpressed in strain 119. The lower band of subunit
IV found in the CcO sample from 25-1 represents the short form of subunit IV. Since
in strain 25-1, subunit ,IV was not overexpressed in plasmid, such an observation

seems to indicate that native translation of subunit IV starts with the second starting

76

(A) (B)_
maProtem 120 120 119 119

119 251 Protein _ .
. Standards Standards Protern Cry stals Protein Crystals

       

66.4
Sub I 55.6 , I

42.7

s» n 355 . n
25.5

In
so In 20.0
14.3

uh [V

Figure 3.4: SDS-PAGE of purified enzyme and re-dissolved crystals of RchO.
(A): Purified RchO from R s. strain 119 and 25-1 with the different subunits labeled.
(B): Purified enzyme and re—dissolved crystals of RchO showing the different
subunits as labled. The overexpression plasmid of R. 3. strain 119 contains: his-tagged
subunit 1, shortened subunit II, complete subunit III, and long form of subunit IV. The
overexpression plasmid of R s. strain 120 contains: his-tagged subunit 1, complete
subunit 11, complete subunit III, and long form of subtmit IV. The overexpression
plasmid of R. s. strain 25-1 contains: his-tagged subunit 1, shortened subunit II, and
complete subunit III only.

codon, although it is also possible that the long form of subunit IV undergoes
proteolytic cleavage with the product being the short form of subunit IV.
Identification of alternative starting points of translation and proteolytic cleavage
processes of subunit IV of RchO samples from other strains of R. s. were examined
using mass spectrometry and will be discussed in later sections (see section 3.2.1).

Figure 3.4 (B) shows the SDS-PAGE of purified Rch0 together with
re-dissolved crystals from R. s. strains 120 and 119. It can be seen that subunit IV was
present in significant amounts in all four lanes of CcO samples, in both the purified
enzyme and re-dissolved crystals. Therefore, the overexpression of subunit IV by
using molecular engineering and retention of subunit IV during detergent
solublization and column purification seemed to be successful.
3.1.5 Factors Affecting X-Ray Diffraction Quality of RchO Crystals and
Systematic Improvements of Diffraction Resolution Limits of RchO Crystals
3.1.5.1 Effect of Homogeneous Form of Subunit II

Although crystals of RchO could be routinely obtained using the new R. 5.
strains and with the modified protein purification procedures, the X-ray diffraction
quality of the CcO crystals obtained from strains 120 and 119 was poor. The best CcO
crystals from strain 119 diffracted X—rays to a limit of 3.6 A resolution, while all CcO
crystals from strain 120 only diffracted to > 4 A. Since the difference between strains
120 and 119 was the overexpression of native subunit H vs. truncated subunit H, such
an observation suggested that different forms of subunit H could play an important

role in the crystal diffraction quality.

78

It was well known that subunit H of RchO undergoes incomplete proteolytic
processing of its C-terminal amino acid residues (Hosler et al., 1992). The processing
site was thought to be after the residue Ser290u based on mass spectrometry analysis
(Hiser et al., 2001). Such a processing could not be prevented by adding different
protease inhibitors to protein samples during purification, nor would it go to
completion, and FPLC purification could only partially separate these two forms
(Hosler et al., 1992; Zhen et al., 1998).

Figure 3.5 (A) shows the FPLC elution profile from Mono Q column
purification of C00 obtained from R. s. strain 120. It can be seen that there were two
very closely associated elution peaks, peak 1 and peak 2, followed by an elongated
tail, as the concentration of KCl increased. As shown in Figure 3.5 (B), SDS-PAGE of
the two peaks indicated that the first peak had mostly the long, intact form of subunit
H, while the second peak contained mostly the short, processed form of subunit H.
Although R. s. strain 120 expresses the full length subunit H sequence in its
overexpression plasmid, due to natural processing of the C terminus, mixed
populations of both intact and naturally processed forms of subunit H were present.
These two different forms of subunit H coexisted in the purified RchO complex and
Rch0 molecules with both forms of subunit H were incorporated into the crystal as
shown in Figure 3.5 (C). Such an inhomogeneity in the protein sample might be the
reason why CcO crystals obtained from strain 120 were of significantly less quality
than those from strain 119.

In order to improve the crystal diffraction resolution, two forms of the subunit

79

(A) 2"“APeak

 

 

my 1“Pe1\k/ ‘*
150-1 f \
:3; 1
E ‘\
a m \
N \
< \
50- x \\
.// \
0 I v I l. ' ' '
35 70 75 39
B
15‘ 2“‘1 ,6 _ Prot
(3) Peak Peak T3" (0C stals ein
’3‘” 3,: :“F 'i m

  

Processw';...g.,fly.fi ~"- "
Sub [I i“ V ‘

 

. 1'!

Figure 3. 5: The two forms of subunit H of RchO due to incomplete proteolytic
processing of the C-terminal 13 amino acids. (A): FPLC elution profile of Mono Q
column purification of RchO obtained fiom R. s. strain 120 showing the two closely
associated peaks. (B): SDS-PAGE of the fiaetions under the first and second peak
from the FPLC purification shown in (A). The first and second peaks represent
RchO with the intact and processed subunit H, respectively. (C): SDS-PAGE of the
purified enzyme and re-dissolved crystals of RchO from strain 120. The two forms
of subunit II coexist in both purified enzyme and in the re-dissolved crystals.

IV gene, the long form of subunit IV starting with the first starting codon, and the
short form of subunit IV starting with the second starting codon, were individually
inserted into an overexpression plasmid which contained the artificially truncated
form of subunit H. The overexpression plasmid was then conjugated into a parental R.
s. strain YZZOO that contained no subunit H or subunit HI (Zhen et al., 1998), to
eliminate any expression of native subunit H from the bacterial genome. Two new R. s.
strains, 156 and 163, overexpressed RchO containing all four subunits with the
histidine tag on the C terminus of subunit 1, the artificially truncated form of subunit
H, and the long form and short form of subunit IV, respectively (see Table 2.1). For
some reason that was not completely understood, the protein expression levels of
strains 156 and 163, as well as 119, were extremely low compared to those from other
strains, which made extensive screening of crystallization and flashcooling conditions
very difficult. However, it was clear that the diffraction limit of the CcO crystals was
improved moderately to up to 3.3 A for Rch0 crystals obtained from strain 156 and
to 3.2 A for crystals from strain 163. Such an improvement in crystal diffraction limit
indicated that it is of critical importance to have only one form of subunit H in the
C00 sample.
3.1.5.2 Effects of Different Forms of Subunit IV

Since the natural subunit IV gene contained two potential starting codons,
efforts were made to clone the long and short forms of subunit IV gene separately and
the effects of the two forms on the crystal diffraction quality were compared (see

Figure 2.1 for the amino acid sequence of subunit IV and two potential N termini).

81

Comparisons of effects of different subunit IVs on crystals were made on CcO
crystals obtained from strains 120 and 157, and from strains 156 and 163. While
crystals obtained from strain 120, which overexpressed the long form of subunit IV,
diffracted generally > 4 A, crystals obtained from strain 157, which overexpressed the
same subunits as strain 120 except for the short subunit IV, diffracted X-rays to
approximately 4 A. A slightly better diffraction limit for crystals from the R. s. strain
overexpressing the short subunit IV was also observed by comparing CcO crystals
obtained from strain 163 and 156 as shown in the previous section. Therefore, it
seems that the presence of short subunit IV was more beneficial to the RchO crystal
diffraction than the long subunit IV.

It should be noted that the above observations of diffraction limits of CcO
crystals were not based on just a few crystals from each strain of R. s. Because of the
great crystal to crystal variation, usually tens of crystals from each strain, sometimes
from different batches of bacterial growth and protein purification, were tested at the
synchrotron for diffraction limits before a generalized idea of the resolution limit
could be made.
3.1.5.3 Crystallization of Subunit H Histidine-Tagged RchO and its Crystal
Structure

Previously, subunit H histidine-tagged RchO was engineered and expressed
in our group in an effort to improve the purity of vesicles of CcO by allowing metal
affinity purification of only those reconstituted vesicles with correctly oriented CcO

molecules in them (Hiser et al., 2001). In this form of enzyme, a 6-histidine tag was

82

attached to the shortened C terminus of subunit H as shown in Figure 2.1. In order to
test if subunit H histidine-tagged Rch0 was a better candidate for crystallization,
long and short forms of subunit IV were incorporated into the overexpression plasmid
containing the histidine-tagged subunit H. The overexpression plasmids were then
conjugated into parental R. 3. strain YZZOO to make strains 167 and 169, respectively
(see Table 2.1). The histidine tag at the C-terminus of the truncated subunit H not only
facilitated Ni-affinity column purification of the enzyme, but also provided a uniform
form of subunit H, whose critical importance in crystal quality had been observed. It
was thought that since the C-terminal end of subunit H is located on the surface of the
extra-membrane domain, which is very close to the crystal contact surface as shown
in Figure 3.1, the presence of six histidine residues at this region might contribute to
new crystal contacts via polar interactions between the histidine tag and residues from
another molecule.

The protein expression levels of both strains 167 and 169 were satisfactory
and moderate improvements in the diffraction limits of CcO crystals were observed.
When combined with other modifications during protein purification, crystallization
and flashcooling of crystals (see later sections in this part), the best crystals obtained
from R. 3. strain 167 diffracted X-rays up to 3.1 A and the best crystals obtained from
strain 169 diffracted X-rays up to 2.9 A. Datasets were collected from both crystals
and structural refinement were carried out by using the original published four subunit
RchO structure (PDB entry 1M56) as the initial phasing model (Svensson-Ek et al.,

2002).

83

Table 3.1 shows the data collection and structural refinement of the two
crystals obtained from R. s. strains 167 and 169, respectively. Due to the strong
anisotropic diffraction, the completeness of both data sets was significantly lower than
100%. Such an anisotropic behavior is very common among the X-ray diffraction
patterns from membrane proteins, and the published wild type four subunit RchO
crystal structure has an overall completeness of only close to 70%. Although the
diffraction limit of the CcO crystal from strain 169 was a little further than that from
strain 167 (2.9 A vs. 3.1 A), the quality of the electron density maps and the final
R-factor and Rf... were very close to each other at about 27% and 32% for R-factor
and Rfreea respectively. The refined structure discussed here was from the crystal
obtained from R. s. strain 167. In fact, the electron density maps of the crystal from
strain 167 revealed more interesting features than that from strain 169, as described
later.
3.1.5.3.1 Overview of the Subunit II Histidine-Tagged Four Subunit RchO
Structure

The structure of the subunit H histidine-tagged CcO was highly homologous
to that of the published four subunit CcO structure with a histidine tag attached to the
C-terminus of subunit 1. The electron density maps of the main chains of the
transmembrane helices were clear, including the single transmembrane helix of the
newly-found subunit IV. The amino acid residues in transmembrane helices,
particularly those buried in the center of the protein were generally better resolved

compared with those on the surface and in the loop regions. Due to the relatively low

84

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R. s. Strain 167 169
A. Unit Cell Parameters
Space Group R3 R3
Cell Dimensions (A) a=b=339.5 a=b=339.4
c=89.1 c=89.3
No. Molecules per AU. 2 2
B. Data Collection
Resolution Range (A) 40-3.1 (3.21-3.1) 30-2.9 (3.0-2.9)
Completeness (%) 73.0 (42.8) 74.7 (33.4)
No. Unique Reflections 50,721 (2,966) 63,444 (2,841)
Overall Redundancy 6.3 (2.6) 5.1 (3.7)
Overall Rmerge (%) 6.5 (32.8) 8.2 (30.5)
C. Structural Refinement
R-factor / R free (%) 27.1 / 32.5 26.7 / 31.5
Average B-factor 101.4 113.5
Non Cryst. Symmetry restrain restrain
F / sigma Cutoff 2.0 2.5
R.m.s.d Bond Lengths (A) 0.011 0.011
R.m.s.d Bond Angles (0) 1.50 1.48
No. of Atoms 17878 17809

 

Rmerge = 2 I 1h - <Ih> l / 2 1h over all h, where 1h is the intensity of
reflection h. R-factor= 2| lFoI - chl |/ Z lFol; where F0 and Fc are
the observed and calculated structure factors, respectively. The
completeness and number of unique reflections values were
obtained from CNS refinement result file after F/sigma cutoff was
applied. Redundancy and Rmerge values were obtained from the
program Scalepack. Values in parentheses are parameters for the

highest resolution shell.

85

Table 3.1: X-ray data collection and refinement statistics of RchO crystals
obtained from R. s. strain 167 and 169.

 

resolution and quality of the dataset, as well as possibly more mobility of the side
chains, several amino acids on the surface of the protein, most of which should have
been lysines and arginines, were assigned as alanines because no side chain electron
densities were observed. Solvent molecules, such as waters, were not assigned due to
the relatively low resolution limit of the data.

The redox active metal centers and the porphyrin rings of the heme groups
were better resolved, as well as the important ligating amino acid residues. Figure 3.6
shows the structure of heme a3 - Cug binuclear center with the (2Fo- F.) difference
electron density map contoured at 10 level. The ligating residues of the heme a3 Fe
and Cu atoms are also shown in Figure 3.6, together with a nearby residue Y2881.
Although there is strong evidence for a linkage of Y288; to one of the C113 ligands,
H2841, through the side chain ring atoms from the two residues (T sukihara et al., 1996;
Ostermeier et al., 1997; Buse et al., 1999), no efforts were made to try to model in
such a covalent bond due to the limited resolution.
3.1.5.3.2 Partial Lipid Molecules

In the published four subunit CcO crystal structure, a total of 6 phosphatidyl
ethanolamine (PE) molecules per CcO molecule were resolved. Four of them were
found to surround subunit IV at the interface between subunit IV and subunits I and
III, while the other two PBS were found in the V-shaped cleft formed by
transmembrane helix bundles inside subunit HI and close to the interface of subunits I
and 1H as shown in Figure 1.4 (Svensson-Ek et al., 2002). In the current structure,

although no complete PE molecules were resolved, a total of 11 partial PE molecules

86

      

..... amt.‘ . Igldfl. m.
. 4 t A . . ... 01.1 a.
.95! mean-Dav!

     

(

S1

a
.Q

’t

V ‘V

   

H \11. fir I». flit..- ‘1
as . ex...» a fish...

flea-v inns. in»... n \ ,.-- . . -

x. .. r. l.-. .1 t _. t I. its...“ .
\~.. at». sex)... . tbs"... Int/nit. ,

.. ...\A .u.s.'l. .... ~ 7 \A’ die. I: 3'th I .
.HAMMJ a ....1..,... 2.x. 0...”- " "Dun”:n‘t‘.bl
\\. .7xx..$.../..m\w'sil. . .. r r \ or... 13‘s..

. n \04- 1.37.1! ‘V .4... A“ . pd‘.
thfiwmhdm. Iv.» 'Dflfiv... fin“\ A»... . ‘1

«Vt ) I. or a. ( t ‘ . 'J .i. A . v\‘~
.\1 . I“ i If "é ‘ 12151..
'1’ n ’1 ‘y I! t . I’ve . Ir in.

/ v...\t§:r..~1....m0\tw . > \u
m. . V5.0... 2.... w e
4 . A. s. m
H W .1. 8.x H

is shown in blue. The amino acid residues including the ligands of the heme a;-Fe

four subunit Rch0. The (2Fo — Fe) difference electron density map contoured at 10
(orange) and Cu]; (purple) are colored by atom type.

Figure 3.6: The active site consisting of heme a3 and Cunin the structure of the

were found in the two CcO molecules in an asymmetric unit. Figure 3.7 shows some
of these partially resolved PE molecules (colored by atom type) together with the
resolved PE molecules in the published structure (yellow). For some of these
molecules, only the phosphate group, the glycerol backbone, and the beginning of the
ester linkage were resolved as shown in Figure 3.7 (A). For some others, the
ethanolamine head group was additionally resolved as shown in Figure 3.7 03).
Sometimes, part of one or two fatty acid chains of the lipid molecules were also
resolved as shown in Figure 3.7 (C) and (D). It can also be seen that both the overall
positions and detailed conformations of these lipid molecules are almost identical in
the two structures in spite of the differences in the histidine—tag positions, and the
slight differences in purification and crystallization procedures. Therefore, the
presence of these partial PE molecules indicated that these lipids are an integral part
of the entire structure.
3.1.5.3.3 Potential Histidine-Tag Resolved at the Crystal Contact Interface

As shown in Figure 3.1 and Figure 3.8 (A), the key crystal contact occurs at
the interface between the bottom part of one molecule including mostly the
N-terminal cytoplasmic region of subunit IV, and the comer between the globular
domain of subunit H and the flat surface formed by subunits I and IH. The C-tenninus
of subunit H is not located within the crystal contact interface, but is in close
proximity. Part of the reason why subunit H histidine-tagged CcO was used as a
crystallization candidate was that the histidine-tag might contribute to formation of

new crystal contacts in this region by assuming a new, stable conformation different

88

 

 

 

 

 

 

r l I
Figure 3.7: Partial lipids resolved in the structure of the four subunit Rch0. The
resolved lipids are shown by sticks and colored by atom type (C: green; 0: red; N:
blue; P: purple), while those resolved in the published four subunit RchO structure
(PDB entry 1M56) are shown in yellow. Different portions of the lipids are resolved
in the current structure compared with the published crystal structure as shown in (A),
(B), (C), and (D).

 

     

   

J. .., ‘4‘ '
‘ ‘1' t "G
fife/’23“! '
'.-..»' w

'01
’1

    

 

Figure 3.8: Major crystal contact regions in the crystal structure of the four
subunit RchO. (A): Although not directly involved in forming crystal contacts, the
C-terminus of submit 11 is close to the crystal contact regions as labeled in the figure.
(B): Extra pieces of electron density are seen between the C-terminus of subunit II
(E28111) and regions of another RchO molecule (Wl7m‘ and Y53m“) in the (2Fo —
F.) difference electron density map contoured at lo (blue) and in the (F0 — F.)
difference electron density map contoured at 2.50 (orange).

from the conformation of the C-terminus of subunit H as seen in the published crystal
structure. Analysis of the CcO crystal obtained from R. s. strain 167 seemed to
suggest that it was possible that the engineered histidine-tag could be making new
crystal contacts.

Figure 3.8 (B) shows the (2Fo- Fe) difference electron density map contoured
at 10 (blue) and the (Fo- Fc) difference electron density map contoured at 2.50 level
(orange) in the area immediately adjacent to the shortened C-terminal residue of
subunit H (E28111) right before the histidine tag. A large electron density feature could
be seen which seemed to connect the C-terminus of subunit H to areas from another
molecule. This continuous piece of electron density could contain the engineered
histidine tag, which might suggest that the histidine tag had been immobilized and
contributed to crystal contacts. However, due to the limited resolution and the
relatively poor quality of the electron density, the model building of the histidine tag
was not aggressively pursued. Similar electron density was not observed in this area
in the structure of the crystal obtained from R. s. strain 169 although it also possessed
a histidine-tagged subunit H. When another refinement program, Refmac5, was used,
this piece of electron density also disappeared.
3.1.5.4 Retention of Membrane Lipids in Detergent Solublization, Protein
Purification and Crystallization
3.1.5.4.] Identification of Lipid Species during Purification and Crystallization

The importance of native membrane lipids in membrane protein crystallization

is getting more attention as more high resolution membrane protein crystal structures

91

are found to contain membrane lipids in them. Such tightly and specifically bound
lipids stabilize the native conformations of membrane proteins and contribute to the
homogeneity of membrane protein samples which is crucial to crystallization efforts.
In the published crystal structure of the four subunit CcO, a total of 6 phospholipids
were resolved, all of which were assigned as phosphatidyl ethanolamine (PE). They
reside in the crevices and interfaces between subunits (Figure 1.4). Besides their
obvious role of securing the presence of subunit IV, it is very conceivable that the
complete or partial loss of these lipids might also lead to non-uniform conformations
of the transmembrane helices and thus inhomogeneity of the protein samples.
Therefore, efforts were made to analyze the lipid content during the protein
purification processes and to modify protein preparation protocols in order to retain
more important lipid species bound to protein complexes. Thin layer chromatography
(TLC) analysis of lipid species of CcO samples at different purification stages showed
diminishing amount of lipids as the purification steps went along. These included
phospholipids PE, PC, PG PS, and non-phospholipids such as SQDG and omithine
lipids (Hilmi, 2002). As the TLC requires relatively more sample (0.2 mg), as well as
lipid extraction which could lead to its oxidation and loss, matrix-assisted laser
desorption-ionization time of flight (MALDI-TOF) mass spectrometry was applied as
an assay method to detect lipid species (Distler et al., 2004). Different lipid species
could be detected preferentially using MALDI mass spectrometry with the use of
different matrix molecules in both linear positive and negative ion modes. Post source

decay (PSD) ion mode was also used to confirm the identities of some of the lipid

92

species by matching fragmentation patterns with lipid standards.

Figure 3.9 shows the MALDI-TOF mass spectra of lipids detected in R. s.
membranes and CcO samples at various stages of purification using both positive and
negative ion modes. In the detergent solublized R. s. membrane sample, as shown in
Figure 3.9(A), a variety of phospholipid species were detected using the positive ion
mode, including PC, PE, PG and CDL, while PG, SQDG. and CDL were identified in
the negative ion mode. It should be noted that an important phospholipid, CDL, which
was not identified using the TLC system previously employed, presumably due to its
comigration with pigments and hemes, and its tendency to run off the top of the TLC
plate, was successfully detected by mass spectrometry, especially when the negative
ion mode was used.

Figure 3.9 also shows the mass spectra of the lipid region of both the two-step
column-purified CcO sample and the re-dissolved four subunit CcO crystals using
positive ion mode (B) and negative ion mode (C). It can be seen that certain lipid
species were selectively retained after the protein purification and crystallization
procedures, including PC, PE, PG, SQDG, and CDL, all of which were detected in the
mass spectra shown. Although the published crystal structure had only PBS in its
model, mass spectra of re-dissolved crystals clearly showed the presence of more lipid
species, including CDL, which had been previously suggested to play key roles in
CcO activity (Robinson, 1982). Due to the lack of an accurate standard, quantification

of each of the lipid species from the peak height was not reliable and thus no

93

 

  
 
     
 
   

  

 

    
     
 
 
    

 

 

 

 

 

 
  
  

     
     
 
 

      

 

 

 

A PC3420+2K
00 838
5 8° PE36:2+2K
.18: z: Pam m PG36.2+2K
5 50 78° hemea
,3 40 852
E 30 CDL72:4+3K
& 2o CDL7223+3K
10 1572
* A‘
710 890 1070 1250 1430 1010
no. SQDG SQDG 34
a; 10+ NegativelonMode 32 793 819 SQDG36
3 so“ 821 a:
8 1°. PG36t2 349
g 60- 773
' 5°. CDL 72:4+K
g 40. CDL 72:3+K
.1! 1494
g 30 2DDM+K
20
10-
1010
B
100 _
Punfied Enzyme heme a
80‘ 852
PG36:2+2K
00 P632.1 PC36'2+K 851
iP632.0 824
40
5
c 20
g o
g m
'fi 100
g i Re—dissolved Crystal hemea
00‘ 852
' PC36:2 . ‘
”‘Pea2z1 PEG 786 P036”?
Peazzo 7‘5 PEG
‘° 7 PE36:2+K . PE3622+2K
782 20 ,
20 ‘
0
71 0 742 774 800 030 070
ml:

Figure 3.9

94

'1‘ .

 

SQDG34
819
821

 

P0362 SQDG36 CDL+K
101; 773 845 1494
1 e47 .
g .01 549 Purified Enzyme ‘
'c' SQDG32:O CDL+K+Na
g 60 I 793 cor/2:4 1516
g 1 0117223
3 w , . 1456
a J 1, 1
20; “ ‘ .
o ,
150 920 mo mo mo 1ooo

 
   
 

Re-dissolved Crystal 1494

CDL72:4
CDL7223
1456

 
 
    

 

 

mlz

Figure 3.9: Lipids content analyses of RchO samples at different purification
stages using MALDI mass spectrometry. (A): MALDI mass spectra of detergent
solublized R. 3. cell membranes using positive (blue) and negative (red) ion mode.
The inset of each spectrum shows the detailed spectra of the m/z region fi'om 710 to
850. Difi'erent lipid species are labeled as in the figure. (B): MALDI mass spectra of
purified RchO (top panel) and re-dissolved RchO crystals (bottom panel) using
positive ion mode. Different lipid species are labeled as in the figure. (C): MALDI
mass spectra of purified RchO (top panel) and re-dissolved RchO crystals (bottom
panel) using negative ion mode. Difi‘erent lipid species are labeled as in the figure.

95

comparative analyses of the amounts of the lipid species between the purified CcO
and C60 crystals were possible. However, approximate relative amounts of different
lipids could be determined and due to its relative ease and quickness, and the fact that
it only requires a very small amount of sample, mass spectrometry proved to be a
valuable routine assay for detection of lipid species during the protein purification
process.
3.1.5.4.2 Quantitative Analysis of Phospholipid Content of CcO Samples by
Phosphorous Assay and Inductively Coupled Plasma Emission Spectroscopy
One quantitative method of analyzing phospholipid content was to perform a
phosphorous assay to determine the relative content of phospholipids per CcO
molecule. Table 3.2 shows the phosphorus content of a set of CcO samples at different
stages of protein purification. It can be seen that as the purification went on, the
number of phosphorous atoms per CcO molecule decreased from 15-18 immediately
after the Ni column, to approximately 5-7 after the second column followed by
washing and detergent exchange. When only the Ni column was used to purify the
protein, the number of phosphorous per CcO molecule after washing and detergent
exchange was similar or slightly higher (z 7) than in the samples after two
purification steps. Significant reduction in phosphorous was observed with increased
number of washes (Table 3.2). Although the phosphorous assay gives important
information about the relative amount of phospholipids (including CDL which has
two phosphorous atoms per molecule), it requires relatively more sample (a few mg)

and cannot give information about non-phosphorous lipids. Therefore, its usefulness

96

 

 

 

 

 

 

 

 

 

 

 

 

Sam- Sample Descriptions [P](p.M)/
ple # [CCOKl-lM)
1 R. s. cytoplasmic membrane = 3600

2 CcO fractions after Ni Column 16.6

3 after Ni column + washing* 7

4 after Ni 4» Mono Q + washing* 6.9

5 after Ni + DEAE Sepharose + washing* 5.2 — 7.3

6 after Ni + Superdex 200 + washing* 6.8

7 after Ni + Mono Q column purification, 3-5
6x washing instead of 2-3x

 

*: washing involves first concentrating the volume of the pooled
fractions from column chromatography to under 0.5 ml through
ultrafiltration, then filling the concentrator with new solution and
concentrating the solution to under 0.5m] again. The process is
repeated a number of times to ensure complete exchange of buffer
the sample is in. The number of times refilling the concentrator

with new buffer solution is usually 2-3 times.

Table 3.2: Phosphorous contents of various RchO samples measured by

colorimetric method.

97

 

 

in routine lipid quantification is limited, particularly with limited protein samples.
Mineral analysis by inductively coupled plasma emission spectroscopy (ICP)
was also performed to further quantity the elemental contents, including metals and
phosphorous. Information on the phosphorous content could be obtained directly from
the elemental analysis. Table 3.3 shows the result of ICP analysis of different CcO
samples. In this table, the phosphorous content was calculated from the ratio of the
molar concentration of phosphorous over the concentration of enzyme derived from
the molar concentration of copper divided by three, since it is known that there are
three coppers per one CcO molecule. The Cu concentration was used as the reference
for phosphorous because it appeared most consistent with the known amount of
enzyme used for the assay and also provided an internal standard. It can be seen from
Table 3.3 that there were 6-8 phosphorous atoms per CcO molecule, which was close
to what was measured using the phosphorous assay as described above, and that there
was not a clear difference in the phosphorous contents among the samples purified
with a Ni column only, Ni plus Mono Q columns, or Ni plus DEAE sepharose
columns, as observed by phosphorous analysis. Also consistent with the phosphorous
analysis (Table 3.2) is the observed reduction in phosphorous content with more
washing (Table 3.3). Although ICP is an accurate method, it requires a lot of sample
(approximately 10 mg) and time to obtain the results and thus is not suitable as a
routine assay.
3.1.5.4.3 Modification of Protein Purification and Crystallization Procedures to

Retain More Bound Membrane Lipids

98

 

.mU— win: 3.532: 81—53 OuUufl 33.3., .«e 3:858 gene—Ewe...— "nd 033—.

 

.90 mo 3955280 H29: ofimo 32¢ 28 .55 meoHoaamosmmo :ofiabfioeoo
Boa 2: .8 88 2e :66 333%.. a 86:3 s
38.: w venue—om 803 8.5385 own—£88 Homage EB magma? St :5 .5558 O 982 28 M2 65
563898 333 £25.“ mé no.“ 3:33 Ea 5:28 $0.538

35.39 03:86 060 we :83 .8523. .o
m<mo 98 :638 MZ 5?» vegan 63:86 0.60 we :88. 3592‘ .n

863%.. M23 as 2 5? 63.55 060 .6
6.5 3 6c 336 Ba b8 =58 2 £5 aqua 0.6 .m
an .2558 we 8:83 “gamma Bob 28 £039: manouog Ba 53 Eat» a. .x Bot 983 83.53 =< A

2 5E wanna 060 .m

A3388 H32 5 $503 8.5. TN :3: .858

8:5 «rm .8» vogue.» 23 5:38

85 m-~ How way—83 28 55:8 0 982 28 5:28
“58%.:

 

 

 

 

 

 

 

 

 

 

 

we $2 wdm New wow 6m mum mam mmm SS 3% ondE
mw otm mam Em owd Em hi. 6:» 3m 8: 0:. mafia
ow 08m 9% 08 m3 Pm emm 2m we». .33 6.8 tan
3. onmm o. 2. can mow. 8m wmm Dom 8m mom v.6 1-02
m.» mmom 3cm mum awe Sm am. o? 03 83 o9. m2
. . . . x53
o _v m fl 0 mm o R o
6&3

.060. 93 came 93 9&3 921V 93 E5 945 :23 3.5
\E E E 3:: Rs: 3 6.3 . 9: F: S5. FE v.8 macaw

 

 

 

 

 

 

 

 

 

 

 

99

Since the retention of membrane lipids is of great importance, modifications
of the protein purification and crystallization were made in order to retain the
important lipid species bound to the protein complex.
3.1.5.4.3.1 Modification of Detergent Concentration used for Membrane
Solublization

The first and possibly the most important way to retain bound native lipids
was to control the amount of detergent used for membrane solublization. As described
earlier in section 3.1.2, the protein concentration in the membrane sample was
determined prior to membrane solublization and the detergent dodecyl maltoside was
added at a final concentration of 1% to the membrane sample at a protein
concentration of 10 mg/ml. This was important because too much detergent relative to
protein would likely strip off the bound lipids and possibly subunit IV as well. In fact,
when the detergent concentration was doubled, no crystals were obtained.
3.1.5.4.3.2 Modification of Column Purification Methods

Modifications during column purifications were also made in order to retain
more bound lipids. In the hope of retaining more lipids, DEAE sepharose
ion-exchange column was used instead of a Mono Q column, providing a different
matrix support which might interact differently with lipids. Although phosphorous

content measurements using colorimetric analysis and ISP analysis did not show any
improvement in the phospholipid content in the purified protein complex, the
diffraction limit of CcO crystals was improved slightly. The two best crystals obtained

from R. s. strains 167 and 169 were both purified by the DEAE sepharose column

100

 

following the Ni column. It is possible that protein purification using the DEAE
sepharose column helped retain some non~phospholipid which were specifically
bound to CcO. These lipids may also be important in maintaining an overall stable
and homogeneous conformation of CcO molecules. In fact, non-phospholipids, such
as SQDG, were routinely found in all CcO samples and crystals using TLC and
MALDI mass spectrometry, but not specifically resolved in the structure.

In other attempts to retain lipids, only a single column purification step (Ni
column) was used. When a simple linear imidazole gradient was used, there were two
very closely associated elution peaks which were very hard to separate as shown in
Figure 3.10 (A). Fractions from the first peak were brownish in color and contained
CcO bound to a major contaminating species with an absorption peak at around 560
nm (Figure 3.10 (B)), while the second peak was green and contained much purer
CcO (Figure 3.10 (C)). A second ion exchange column purification was usually able
to remove the contaminant when all fractions under the two peaks were pooled.
However, without further column purification, crystals were difficult to obtain.
Therefore, a step gradient method as described in the Methods section was developed
to separate the two closely-associated elution peaks during Ni column purification in
order to obtain a reasonably pure CcO sample without a second chromatography step.
Complete separation of the two peaks was achieved as shown in Figure 3.10 (D) by
using the step gradient. Green fractions from the second peak were pooled and
washed and detergent was exchanged. The improved Ni purification protocol required

less time and handling, and could contribute to retention of more lipids that might

lOl

 

 

 

 

 

80% . 1st Peak

70%:~

60% ' 2ml Peak
9:; 50% .
‘3
00 40%,.
0
§ 30% ~

20%:-

10%:-

M...
0% T I _: l l
250 300 350 400 450 500 550
B Sample Volume (ml)

 

60% .
1“ Peak

50% —
m40% ~
5%: 2“ Peak
m30% ~ ,
o\°

20% -

10% ~ 1

0% ‘ T r T _I" I I
230 280 330 380 430 480 530

Sample Volume (ml)

 

Figure 3.10

102

 

0.0035 -

0.0025 I

0.0015 .

0.0005 -

-0.0005 -

00015 -

 

«0.0025 fir . . . v .
0.01 - 606

Absorhance

0.008 s

0.006 +

0.004 ‘

0.002 1

-0.002 .

 

 

-0.004 I I v v I I
520 540 560 580 600 620 640

Wavelength (nm)

Figure 3.10: Elution profile of Ni-NTA affinity column purification of Rch0 and
the UV-visible spectra of the fractions under the two peaks during the elution
process. (A): Two very closely related elution peaks (blue) were observed when a
simple linear gradient of imidazole was applied (green). (B): The two peaks (blue)

were completely separated when a step gradient of imidazole (green) was applied. (C):

Comparison of the reduced minus oxidized spectra of the fractions under the first
peak in the Ni column elution profile (top panel) and the fractions under the second
peak (bottom panel). The wavelengths of the absorption peaks in each spectrum are
labeled as in the figure.

otherwise be lost during a second column purification step. CcO crystals were
routinely obtained from protein samples purified using the above method and the
phosphorous content was similar to that after a two-column purification. However, the
diffraction limit of the CcO crystals was worse compared with those obtained from
CcO samples after the two-step column purification.

Comparing the phosphorous content of CcO samples immediately after the Ni
column, and after the washing and detergent exchange, showed that the phosphorous
content decreased significantly from 15-18 to 5-7 phosphorous atoms per CcO
molecule. This result suggested that the washing and detergent exchange steps were
responsible for stripping off the majority of the phospholipids. Therefore, efforts were
made to optimize the procedure of enzyme washing and concentrating in order to
retain more bound lipids. Previously, green fractions after the second ion exchange
column purification were washed and concentrated 6 times by using a YM-lOO filter
(MW cutoff 100 kDa, maximum volume 15 ml) to ensure a thorough desalting and a
complete detergent exchange. However, this washing procedure was found to lead to
the loss of more lipids; the phosphorous content was lower (approximately 3-5
phosphorous atoms per CcO molecule) than when the sample was washed for 2 — 3
times (approximately 5-7 phosphorous atoms per CcO molecule). Thus, the number of
times of washing was reduced to 2 - 3 times. The method of washing was later
changed to the stirred ultrafiltration cell under nitrogen pressure (maximum volume
50 ml) with a YM-IOO membrane disc (MW cutoff 100 kDa). The advantages of using

this method included better efficiency of washing and detergent exchange due to a

 

larger working volume and thus fewer cycles of washing (2 x). Since the entire
enzyme solution was constantly being stirred, there was no enzyme precipitate formed,
which tended to happen when the enzyme solution was subjected to prolonged
cenuifugation. Moreover, oxidation of lipids should be lower during this process since
there is little oxygen present in the ultrafiltration cell. The phosphorous content of the
CcO sample prepared in this way was similar to before and no improvement in X-ray
diffraction quality was observed.

A different second chromatographic step, a Superdex 200 size exclusion
column, was also tested and the effect on both the retention of lipids and X-ray
diffraction quality was observed. The advantage of using this column was that the
different matrix of the Superdex 200 column might bind to lipids less and thus not
deplete lipids from the protein samples. Additionally, only a simple concentrating of
CcO sample was needed after the column since the detergent exchange was performed
when the enzyme moved through the column pre-equilibrated with the new buffer. No
improvement in crystal diffraction was observed by using CcO prepared using this
column.
3.1.5.4.3.3 Modification of Detergent and Lipids in the Crystallization Solution

In order to retain more bound lipids during crystallization, less detergent was
used in the final enzyme solution with approximately 0.20% decyl maltoside instead
of 0.24% decyl maltoside.The latter concentration was used by Iwata’s group who
published the first four subunit RchO crystal structure. Crystals obtained with

various concentrations of decyl maltoside suggested that if the decyl maltoside

105

 

 

concentration was lower than 0.16%, the crystal diffraction was worse, while there
was no clear difference in diffraction among crystals obtained with conditions having
0.20% or 0.24% decyl maltoside. Therefore, the effect of such a modification was not
clear.

In other attempts, dioleoyl-glycero—3-phosphoethanolamine (DOPE) or
dioleoyl-glycero-3-phosphocholine (DOPC) were added into the crystallization
mixture with the hope that they would bind to their native binding site(s) and
contribute to the overall stable and homogeneous conformation of the protein sample.
Unfortunately, addition of the membrane phospholipids did not lead to any crystals, or
in a few instances led to crystals with almost no diffraction. This observation was in
accordance with the observations made during the screening of secondary detergents
when a few PCs with shorter acyl chain length (9-12) were used in the crystallization
additive and no crystals were obtained. When a batch crystallization method was used,
during which enzyme solution was mixed with crystallizing reagents and
crystallization additives in a well, and the entire mixture was topped with immiscible
oil, crystals were obtained with the addition of PE, but the diffraction was worse than
without the addition of phospholipids.
3.1.5.5 Optimization of Detergent Choice, Crystallization Additives,
Crystallization and Flashcooling Procedures
3.1.5.5.] Optimization of Detergent(s)

Since crystallization of membrane proteins is essentially crystallization of

protein-detergent complexes (PDCs), detergent molecules are the most important

 

variables in a membrane protein crystallization experiment, second only to the
membrane protein itself. Crystallization of the four subunit RchO requires two
different detergents. First, dodecyl maltoside was used to solubilize the R. s.
membrane and was also present in the buffers for the two—step column purification.
After that, the detergent was exchanged to decyl maltoside by repeated washing and
concentrating using either a centriplus filter unit or the stirred untmfiltration cell
under nitrogen pressure (see Methods for details). However, during crystallization, a
small amount of dodecyl maltoside was again added back to the crystallization
mixture as an additive (1/20 of the concentration of decyl maltoside). It was difficult
to control the exact concentration of each detergent in such a system, particularly
during the detergent exchange step. The idea of having a simple, single detergent in
the system seemed attractive, since the exact concentration would be known and it
would be much easier to study the effect of the detergent concentration and repeat the
conditions. Therefore, experiments were performed using only one detergent, either
dodecyl maltoside or decyl maltoside, throughout the protein preparation and
crystallization process. However, when dodecyl maltoside was used for the entire
process of membrane solublization, protein purification and crystallization, only
needle bundles of CcO crystals were obtained, while when decyl maltoside was used
throughout the entire processes, no crystals were obtained at all. Therefore, it
appeared that the two-detergent system was necessary for successful crystallization of
RchO. Decyl maltoside should be the primary detergent during crystallization

because of the slightly smaller overall size of the PDCs. However, dodecyl

107

 

maltoside is more stabilizing than decyl maltoside for CcO as demonstrated by
previous experiments (Hilmi, 2002); thus is used during the protein preparation.

The current protocol of using dodecyl maltoside for membrane solublization
and decyl maltoside as the major detergent during crystallization was maintained and
efforts were then made to search for another secondary detergent in the crystallization
additive other than dodecyl maltoside. To do this, a group of secondary detergents,
including decyl maltoside and dodecyl maltoside, were used in the crystallization
trials at a concentration of approximately 3 x CMC and the diffraction limits of the
CcO crystals obtained from such conditions were tested at the synchrotron.

Out of the 72 detergents tried as secondary detergents in the crystallization
additives, crystals were obtained from 23 of them with varying X-ray diffraction
quality. Table 3.4 shows the diffraction limits of CcO crystals using various secondary
detergents. All of the crystals displayed strong anisotropic X-ray diffraction patterns
and only the diffraction limits along the better directions (a* and b*) are listed. Some
of the best detergents were from the series of cymals, including Cymal-4 and Cymal-S,
as well as other maltosides with shorter chains lengths, including nonyl and octyl
maltoside. The diffraction limits for CcO crystals with these secondary detergents
were all similar to or slightly better than when dodecyl maltoside was used as the
secondary detergent. When no secondary detergent was used, the diffraction limit
observed was worse than when a suitable secondary detergent was used. Therefore,

these very few secondary detergents were used for further crystallization trials.

108

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

detergent Diffraction limit“
cymal-4 3.0 A
cymal-S 3.0 A
nonyl maltoside 3.0 A
octyl thiomaltoside 3.1 A
nonyl thiomaltoside 3.1 A
deoxy bigCHAP 3.1 A
C12159 3.1 A
dodecyl maltoside 3.1 A
cymal-6 3.2 A
nonyl glucoside 3.2 A
undecyl maltoside 3.2 A
Zwittergent3-14 3.2 A
Sucrose monolaurate 3.3 A
decyl maltoside 3.3 A
decyl thiomaltoside 3.3 A
Pluronic F-68 3.3 A
nonyl thioglucoside 3.4 A
LDAO 3.5 A
prodecyl maltoside 3.5 A
tetradecyl maltoside 3.5 A
hexadecyl maltoside 3.6 A
MEGA-9 3.6 A

CTAB -

 

 

 

*: diffraction limit of the best crystal along the best direction

Table 3.4: The effects of different secondary detergents on X-ray diffraction
resolution limit of the four subunit RchO crystals.

109

3.1.5.5.2 Optimization of Crystallization Procedures and Additives

Since protein crystals generally diffract X-rays much more weakly than those
of small molecules, efforts were made to obtain larger protein crystals which could
diffract more strongly. Although in many cases protein crystals of larger size do not
necessarily have better X-ray diffraction, in the case of RchO it was indeed true that
larger crystals (> 0.3 mm on the side of the triangle) diffracted better than the smaller
ones. In an effort to increase the crystal size, the entire drop volume was increased
significantly from 4 N to 20 pl and the sitting-drop setup was used instead of the
hanging-drop method. Larger CcO crystals that diffracted X-rays better could be
obtained by the sitting-drop method that had a bigger drop size.

Metals have been known to promote protein-protein interactions during
crystallization (Trakhanov and Quiocho, 1995). In the established crystallization
protocol, the divalent cation Mg2+ (in the form of MgClz) was added to the
crystallization mixture as a crystallization additive. In an effort to find the optimal
metal ions in the crystallization mixture, a number of different chloride salts were
used as crystallization additives and the X-ray diffraction was observed at the
synchrotron. Out of the approximately 15 different metals tested, CcO crystals were
successfully obtained with 6 of them, including Mn“, Cu“, Co“, on“, Gd“, and
Sr“. However, none of these metals were found to lead to any improvements in
crystal diffraction.
3.1.5.5.3 Optimization of Flashcooling Procedures and Cryoprotectants

The flashcooling method was also optimized. The concentration of PEG-400

llO

in the cryoprotectant solution was increased to approximately 32%, which seemed to
be the minimal concentration required to suppress ice formation. Also modified was
the flashcooling procedure. Initially, the crystals were picked up from the original
drop solution and soaked in the cryosolution directly for approximately 20 minutes.
However, this procedure was found to be rather harsh on CcO crystals since many of
the crystals displayed visible cracks upon being moved into the cryosolution due to a
sudden rise in osmotic pressure. In order to decrease the pressure shock, a gradual,
step-wise method for increasing the PEG-400 concentration was used (see Method
section for details). The concentration of PEG—400 was increased slowly to 32%
within a period of 8 — 10 minutes through several gradual steps. By using the new
method, generally no crystal cracks were seen and the diffraction limit and quality of
the crystals was improved slightly.

Crystal dehydration prior to flashcooling has been applied successfully to a
number of crystals to improve the diffraction limit and decrease the crystals mosaicity
including membrane proteins (Kuo et al., 2003). Efforts were also made to try to
dehydrate CcO crystals before flashcooling by either changing the well solution to a
solution with higher PEG-400 concentration, or by soaking the crystals in a stabilizing
solution and slowly increasing the concentration of PEG-400 over a period of a few
weeks. Unfortunately, dehydration did not lead to any improvement in crystal
diffraction, but led to deterioration of CcO crystals and a compete loss of X-ray
diffraction.

Other efforts to optimize flashcooling of CcO crystals included adding small

111

molecules such as glycerol and/or ethylene glycol to the crystallization mixture to a
final concentration of 5% - 10%. Since glycerol and ethylene glycol are
cryoprotectants themselves, their presence in the crystallization drop, combined with
approximately 22% PEG-400, provided enough cryoprotection to the crystals.
Therefore, crystals were simply picked up and flashfrozen in liquid nitrogen without
any soaking or handling. Unfortunately, although ice formation was suppressed, the
crystals actually diffracted worse than when glycerol or ethylene glycol was not added,
presumably due to the disturbing effects of glycerol or ethylene glycol on the crystals.
Immiscible oil (Kwong and Liu, 1999) was also tried in flashcooling crystals of
RchO but led to worse protein X-ray diffraction and extremely strong solvent
background in the diffraction image.
3.1.5.6 Design of Site-Directed Mutants to Improve Crystal Diffraction
Site-directed mutagenesis of selected surface residues has been successfully
applied to protein crystallography to improve the crystal diffraction in a number of
different cases (Pautsch et al., 1999). Figure 3.11 (A) shows the crystal packing in the
unit cell of the four subunit RchO. There are three crystal contact regions in this
crystal form as shown in Figure 3.11 (B). The first crystal contact region was the most
extensive one, as described earlier. Besides this major contact region, two more
crystal contact regions were also found on the protein surface, which involved fewer
residues. Based on the analysis of the crystal structure, a few site-directed mutants of
surface residues were designed and generated as listed in Table 3.5. The purpose of

these surface mutants is to strengthen the crystal contacts in regions 2 and

112

 

Figure 3.11: Crystal packing of the four subunit RchO which belongs to a space
group of R3. (A): Unit cell display of the RchO crystal. There are two RchO
molecules (shown in blue and green) per asymmetric unit. (B): Detailed crystal
contacts around each Rch0 molecule. RchO molecules are colored by different
polypeptide chains. There are three regions where protein-protein contacts occur as
labeled in the figure. (C): No direct protein-protein crystal contact exists along the c“
axis of the unit cell. Molecules of Rch0 are from adjacent unit cells and they are
viewed from a different angle approximately 90° away from the view in (B).

113

3 by favoring protein-protein interactions via side chain atoms. Moreover, since the
crystal packing indicated that the crystals of the four subunit CcO were
two-dimensional in nature, with no direct protein-protein interactions along the c*
axis as shown in Figure 3.11 (C), other mutants were also designed to generate new
crystal contacts along the c* axis, either by promoting ionic interactions or by creating
new metal binding sites (See Table 3.5 for details). Unfortunately, some of these
mutant proteins did not crystallize, while others did crystallize but did not lead to any
improvements in crystal diffraction, as listed in Table 3.5.
3.1.5.7 Screening of New Crystallization Conditions and New Crystal Forms

Extensive screenings to find new crystallization conditions and possibly new
crystal forms of CcO were also performed with various commercial crystallization
screening kits. A few new crystallization conditions from the screening kits were
found to produce very small crystals of CcO. One of them contained 20% PEG-1000,
200mM NaCl, 100mM Na/K phosphate, pH 6.2; and another screening solution
contained 15% ethanol, 100 mM Tris, pH 7.0. However, further optimization of
crystallization conditions by varying pH, precipitant and salt concentration failed to
produce big enough crystals for X-ray diffraction tests despite repeated trials.

CcO crystals were also found growing from a screening solution containing
100 mM HEPES, pH 7.5, 100 mM (NI-I4)2SO4, and 18% PEG-400. The majority of
the crystals were thin tetragon plates in shape, which suggested that these crystals
might belong to a completely new crystal form. Optimization of crystallization

conditions by varying concentrations of precipitants and pH, combined with

114

.3532: 035 we manic me :8: nets—emu.—
=euofitmu APTN 2: 93 895.80 .335 eofiueohm 3 65am we 353::— gooh—YBE 2.0.—cg am 935,—.

3529595 .3 8 coma AC Hoodoeomom

 

 

I BE wfivafi 308 26: mew—«Bo 3 . magma

o: 38 so mac? 88:8 Bo: Bo 88m $2.8m “mm 2 mafia vmm
I 86 wfivfia ESE Bo: wages 3 . magma

0: mg no mac? 8880 Bo: Bo 89m hofimOm “mm? mum?» mmm

 

828885 222 950 ESQ 3
4 m.m we» was no mega 83:8 Bo: Bo 88m 03309 .Hmmz mummB mom

 

828885 082 38an .3 _
4 mm a» was to mean “03:8 so: soaom 35% .282 manna 8m

 

 

 

 

 

4 is. a; ma 8&2 83:8 ansmcoem 850a .Hmm 2 960m 8m

< 3 a.» on 8&2 “03:8 fiancee Samoa .03 2 960m com
I 8 an 8&2 33:8 anfimanem 9.05% .Hmm 2 Geog 5
i 8 a. 8&2 cease 559.0% Eamon $32 959 on:

:38 :3 Esme—Q ES 52:.

-aEE -bo 3:on ash... 352$ 853:2 SN

 

 

 

 

 

 

 

 

115

micro-seeding procedures, yielded single crystals big enough for X-ray diffraction
experiments as shown in Figure 3.12.

The new crystals of C00 were flashcooled and tested at the synchrotron for
X-ray diffraction and a dataset was collected on the best crystal which diffracted to
approximately 4.3 A with very strong anisotropicity. The crystal cell parameters and
statistics of data collection are listed in Table 3.6. The crystal belonged to the
orthorhombic crystal system with a space group of P212121. Although the crystal
diffraction and the overall completeness of the dataset were rather poor, a molecular
replacement was successfully performed using the published four subunit CcO
structure as the search model and the molecular packing of this new form of crystal
was solved as shown in Figure 3.13. It can be seen that, compared with the crystal
packing in the regular rhombohedral forms of the four subunit CcO crystal, in which
crystal contact were found in two dimensions in the a*b* plane, the crystal packing of
this form of CcO crystal, however, seemed more one-dimensional. Crystal contacts
occur primarily along the c* axis involving only a restricted number of residues as
shown in Figure 3.13. This poor packing of molecules was the likely reason why this
form of CcO crystal diffracted poorly and extensive efforts were not made to improve
the crystal diffraction of this form of crystal.
3.1.6 Potential Inhibition Sites of an’VCdz“ and the Effects of Cd2+ Binding to
CcO Crystals on X-ray Diffraction
3.1.6.1 Zn2+ / Cd“ inhibition on CcO Activity

Zn2+ has been found to have a strong inhibitory effect on a number of proton

116

 

Figure 3.12: Crystals of the four subunit 11st which belong to a space group of
P212121.

117

 

 

 

 

 

 

 

 

 

 

 

 

 

Crystal Name I 167 — tetragon
A. Unit Cell Parameters
Space Group P 21 21 21
Cell Dimensions (A) a=126.3 b=151.8 c=206.6
No. Molecules per AU. 2
B. Data Collection
Resolution Range (A) 20 — 4.3
Completeness (%) 67.2
No. Unique Reflections 18,392
Overall Redundancy 5.8
Overall Rmerge (%) 7.1

 

 

 

Rmerge = 2 l 1h - <Ih> l / 2 1h over all h, where 1h is the intensity
of reflection h. The completeness and number of unique
reflections values were obtained from CNS refinement result file
after F/sigma cutoff of 3 were applied.

Table 3.6: X-ray data collection and refinement statistics of the four subunit
RchO which belong to a space group of P212121.

118

 

Unit cell display of the four subunit orthorhombic RchO crystal

Figure 3.13

which belongs to a space group of P212121. There are two RchO molecules (shown

in blue and green) per asymmetric unit.

119

and ion channels by blocking a proton pathway, including cytochrome be; complex
(Link and von Jagow, 1995; Berry et al., 2000) and bacterial photosynthetic reaction
center (Paddock et al., 1999; Axelrod et al., 2000). Similarly, Zn2+ was also found to
bind to the outside of RchO reconstituted into lipid vesicles and inhibit its activity in
the presence of a membrane potential with a K; of S 5 11M (Mills et al., 2002). The
inhibition is reversible, pH-dependent, and not competitive with the substrate
cytochrome c (Mills et al., 2002). The exact inhibition site(s) on the outside of CcO
was unknown and was proposed to be at or close to the proton exit/backleak pathway
(Mills et al., 2002). Additional Zn“ inhibition was observed on free enzyme as well
with a much higher inhibition constant and the proposed inhibition site(s) was at the D
proton uptake pathway based on the results of single turnover experiments (Aagaard
er al., 2002). The exact binding site(s) along D pathway is unknown.

Among all the other metals tested, Cd“ was the only one found to exhibit
similar inhibitory effects on CcO (Mills et al., 2002).
3.1.6.2 Potential Inhibition Sites of Zn“ / Cd2+ and the Effects of Cd2+ Binding to
0:0 Crystals on X-ray Diffraction

In order to find out the inhibition site(s) of Zn”, ZnSO4 was added to
crystallization mixture in an effort to obtain crystals of Zn2+-bound CcO. However, it
turned out that the presence of Zn2+ caused severe precipitation of CcO in the
crystallization drop even at a very low concentration (approximately 0.5 mM), and no
crystals were obtained. Co-crystallization experiments with Cd2+ yielded crystals but

the X-ray diffraction was extremely poor (worse than 7 A). Therefore, efforts were

120

made to soak the crystals of CcO formed under normal conditions in a stabilizing
solution that contained either Zn2+ or Cd2+ as described in the Method section. The pH
value of the soaking solution was slowly raised to close to 7.0 in order to allow tighter
binding of Zn2+/Cd2+ to the enzyme, and kept at 7.0 in the cryosolution before
flashcooling. Unfortunately, the presence of Zn“ in the soaking solution was found to
be extremely damaging to the X-ray diffraction of C60 crystals as both the diffraction
resolution limit and the completeness of data were much worse than when Zn2+ was
not added. However, adding Cd2+ to the soaking solution appeared beneficial to the
diffraction of CcO crystals.

Table 3.7 shows the data collection and structural refinement statistics of a
CcO crystal, obtained from R. 5. strain 169, which was soaked in stabilizing solution
containing 6 mM CdClz for 2 hours before flashcooling. It can be seen that, although
the diffraction limit of the crystal did not improve (3.2 A), the quality of the X-ray
diffraction improved slightly as judged by the better completeness of the dataset, and
the slightly lower overall B factors and very similar final R and Rf,ee factors after
refinement, despite a smaller F/sigma cutoff.

Figure 3.14 shows the amino acid backbone of the refined structure of RchO
colored by subunits. Superimposed on the structure is the (FD-Fe) omit difference
electron density map contoured at 3.5 0. It can be seen that there are electron density
peaks (shown in purple) in three locations in the structure as labeled, which could

indicate potential binding sites of Cd2+ in the crystal structure.

121

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Crystal Name | 169 - Cd Soak 169
A. Unit Cell Parameters A
Space Group R3 R3
Cell Dimensions (A) a=b=339.5 a=b=339.4
c=89.4 c=89.3
No. Molecules per A.U. 2 2
B. Data Collection
Resolution Range (A) 3032 (331-32) 30-2.9 (3.0-2.9)
Completeness (%) 86.7 (69.0) 74.7 (33.4)
No. Unique Reflections 54,839 (4,342) 63 ,444 (2,841)
Redundancy 8.5 (3.4) 5.1 (3.7)
Rmerge (%) 10.3 (41.0) 8.2 (30.5)
C. Structural Refinement
R-factor / R free (%) 26.5 / 31.4 26.7 / 31.5
Average B-factor 96.6 113.5
Non Cryst. Symmetry restrain restrain
F / sigma Cutoff 1.01 2.5
R.m.s.d Bond Lengths (A) 0.009 0.011
R.m.s.d Bond Angles (0) 1.35 1.48
No. of Atoms 17764 17809

 

Rmerge = 2 l Ih - <Ih> / 2 lb over all h, where I h is the intensity
of reflection h. R-factor= 2 I lFol - chl l/ 2 lFoI; where F0 and PC
are the observed and calculated structure factors, respectively.
The completeness and number of unique reflections values were
obtained from CNS 1.1 refinement result file after F/sigma cutoff
was applied. Redundancy and Rmerge values were obtained from
the program Scalepack. Values in parentheses are parameters for

the highest resolution shell.

Table 3.7: X-ray data collection and refinement statistics of the four subunit
RchO crystals soaked in a solution containing cadmium. The column on the right
is taken from Table 3.1 which shows the data collection and refinement statistics of

crystals obtained from the same R. 3. strain without being soaked in cadmium.

122

 

 

Figure 3.14: Cadmium binding sites found in the structure of the four subunit
RchO crystals soaked in a solution containing cadmium. RchO molecule is
colored by subunits (1: yellow; 11: green; III: cyan; IV: gray) and the heme groups and
coppers are colored orange. Part of another RchO molecule is colored wheat. The
(F0- 6) difference electron density map contoured at 3.5 o are shown in purple. There
are three cadmium binding sites found as labeled in the figure.

123

Figure 3.15 (A) shows the first potential Cd2+ binding site, which was right at
the center of a crystal contact region. The cadmium ion was ligated to the side chains
of two residues, one of them being E533; from one CcO molecule, and the other being
the same E5331 residue from another molecule. Two H5341 residues from both CcO
molecules were also likely to take part in the metal ligation since they were in very
close proximity to the cadmium. However, the distances of the nitrogen atoms in the
imidazole rings to the cadmium were around 3.3 — 4.5 A, which were too far apart for
this ligation to occur, although there was a possibility that the current structure was
not exactly correct in details due to limited resolution, and there could be indeed a
tetrahedral ligating coordination about the cadmium involving both residues E5331
and H534; from both molecules. Both of the two residues had been suggested to form
crystal contacts through electrostatic interactions between the side chains of the
glutamic acid and histidine residues. However, in the absence of cadmium, the
electrostatic interactions were not strong enough and thus the electron densities were
not seen for the flexible side chains of residue E5331 from the two molecules, as
shown in Figure 3.15 (B). It appeared that the entering of a cadmium ion into this
region strengthened this crystal contact by mediating the protein-protein interaction
via the ligating interactions between the cadmium ion and the two E5331 residue side
chain atoms. Such a strong metal ligating interaction was responsible for the slightly
improved crystal diffraction, and this potential cadmium site was probably not the
inhibitory site that was observed experimentally.

Figure 3.15 (C) shows the second potential cadmium binding site and its

124

"I!
,3! 44 C

r" 1 I I» ' I ‘
~14 ‘fia 1‘/

"a
. ‘0‘

 

Figure 3.15

125

       

  
 
 

igl‘gé/II"
1

7
. ~a, 5"
.1.-agile",

    

A

Figure 3.15: Cadmium binding sites found in the four subunit RchO structure
after the crystals were soaked in a solution containing cadmium. (A): The first
cadmium binding site located at the intermolecular surface. One RchO molecule is
colored by atom type (C: green, 0: red, N: blue) and the other RchO molecule is
colored by a different atom type scheme (C: yellow, 0: red, N: blue). Cadmium ion is
shown as a purple sphere. The (2E, -— Fc) difference electron density map contoured at
10 is shown in blue. The ligating residues of the cadmium ion fiom the two RchO
molecules are labeled as shown in the figure. (B): The same inter-molecular contact
region without the bormd cadmium, shown in the same color scheme. (C): The second
cadmium binding site located close to the interface between subunits I and 11, together
with the ligating residues as labeled in the figure. (D): The third cadmium binding site
buried inside subimit III, together with the ligating residues as labeled in the figure.

126

t;

31...

neighboring residues together with the (2Fo- c) difference electron density map
contoured at 1 a. Two nearby amino acid residues from subunit II, E10111 and H96",
were possible ligating residues to the cadmium ion. However, due to limited
resolution and the possible presence of different conformations of the side chains due
to partial binding, the side chain electron densities of these two residues were not
clearly observed, especially for E10111 (Figure 3.15 (C)). E10111 has been suggested to
be the entrance of the K proton uptake pathway (Branden et al., 2002), and binding of
cadmium at the E101" site could lead to the blockage of the K pathway and thus
inhibit the enzymatic activity. Therefore, this site could be one of the Zn2+/Cd2+
inhibition sites.

There was yet another potential cadmium binding site found inside the subunit
III of the enzyme. Figure 3.15(D) shows the third potential binding site and its
possible ligating residues, including H1521", H2371", and E2411". This potential
cadmium binding site did not seem to be one of the inhibition sites since it was found
in subunit III, which is unlikely to be involved in an‘VCd2+ inhibition because RchO
without subunit III was still inhibited by an’VCd2+ (Mills et al., 2002).

Unfortunately, potential inhibition sites for an‘VCd2+ were not found on the
outside of the protein in the crystal structure with crystals soaked with the cadmium
solution. One possible reason is that the binding of Zn2+/Cd2+ only becomes tight
when there is a membrane potential (Mills et al., 2002). Since there is no way to
maintain a membrane potential in CcO crystals, the binding of an‘VCd2+ to the

enzyme is likely too weak to be observed in the crystal structure using the above

127

method.

128

3.2 X-ray Crystallography of [-11 Subunit RchO
3.2.1 Engineering of an R. 6'. Strain with a single Form of Subunit IV

Upon realization that subunit IV is crucial in forming crystal contacts in the
crystallization of the four subunit RchO, we successfully engineered the long and
short form of subunit IV, as described in section 3.1.1 and in Methods section. The
assumption made was that the subunit IV expressed from these R. 3. strains would be
uniformly long or short, depending upon which gene sequence was inserted, and that
the RchO complex would contain only one form of subunit IV, either long or short.
However, careful analyses on both purified enzyme and re-dissolved crystals with
MALDI mass spectrometry clearly show that the gene expression and
posttranslational modification lead to more forms of subunit IV in the RchO
complex than the originally engineered forms.

Figure 3.16 shows the MALDI mass spectra of purified RchO and
re-dissolved four subunit RchO crystals in the subunit IV region. Since both positive
and negative ion modes show the protein subunits equally well and the difference in
mlz values due to different ionization modes is small, all mass spectra herein are
displayed in one color. The conditions including positive and negative ion mode are
indicated in figure legends. Figure 3.16 (A) shows the mass spectra of the subunit IV
region from purified enzyme and re-dissolved crystals of RchO obtained from R. 5.
strain 207, which has in its overexpression plasmid the gene for the long form of

subunit IV, equivalent to that in R. 3. strain 167 (see Table 2.1 and Table 3.5). It can be

129

Rel ativc Intensity

Relative Intensity

  
   

 

processed
1008 IV Purified En o
1 ‘ 5088 zym
0.8 4
long IV

0 6 — 6257
0.4 i
0. 2 ..

 

  

 

   

 
   
  
   
  

0 l T l
4500 4750 5000 5250 5500 5750 6000 6250 6500 6750
m/z “118W

1“ Re-dssdvocl 5257

1 Crystals
0'8 [recessed

longl'V
0‘6- 5088 shortIV
5270

0.4 ‘
0.2 -

 

 

0 I I I I I I I I I I

4500 4750 5000 5250 5500 5750 6000 6250 5500 6750
m/z

Figure 3.16

130

ED

.0 9

P

Relative Intensity

.0

Relative Intensity

short IV

 

 

 

 

 

 

 

 

 

 

Figure 3.16

131

1 - 5270 Purified Enzyme
3‘ modified
shortIV
5 ‘ 5312
short IV - ala
4 ‘ 5199
longIV
21 6257
L- _. ‘Zw. A.:"rv:‘h~ __._
0 r I I I I I I r I
4.500 4750 5000 5250 5500 5750 6000 6250 6500 6750
m/z
1 - shortIV Re-dssolved
5270 Crystal:
-8 ‘ modified
shortIV
6 . 5312
4
longIV
2 a 6257
f w VA v _ if
0 I I I I I I I I I
6500 4750 5000 5250 5500 5750 6000 6250 6500 6750
If:

O

 

 

Purified Enzyme
1 - shortIV-ala shortIV
5199 5270
33: 0. 8 -
‘8
g o_ 5 - modfied
H shortIV
g o. 4 - 5312 6257
d
:2

 

 

 

 

4500 4750 5000 5250 5500 5750 6000 6250 6500 6750
ml:

Figure 3.16: MALDI mass spectra of the subunit IV region of purified Rch0
and re-dissolved four subunit RchO crystals obtained from different R. 3.
strains. (A): MALDI mass spectra of purified RchO (top, negative ion mode) and
re-dissolved crystals (bottom, negative ion mode) obtained from R. s. strain 207 ,
which overexpresses the long form of subunit IV in its overexpression plasmid.
Different forms of subunit IV were observed and they are labeled as shown in the
figure with their corresponding m/z values. (See text for detailed descriptions of each
of the forms of subunit IV.) (B): MALDI mass spectra of purified RchO (top,
negative ion mode) and re-dissolved crystals (bottom, positive ion mode) obtained
fiom R. s. strain 169, which overexpresses the short form of subunit IV in its
overexpression plasmid. Different forms of subunit IV were observed and they are
labeled as shown in the figure with their corresponding m/z values. (See text for
detailed descriptions of each of the forms of subunit IV.) (C): MALDI mass spectra of
pmified RchO obtained from R. s. strain I69A4 (positive ion mode), which has the
native subunit IV DNA sequence deleted from its genome and has the short form of
subunit IV in its overexpression plasmid. Different forms of subunit IV were observed
and they are labeled as shown in the figure with their corresponding m/z values. (See
text for detailed descriptions of each of the forms of subunit IV.)

132

seen that there are several major m/z peaks in the spectra for both the purified enzyme
and re—dissolved crystals, which represents the long form of subunit IV (mlz = 6257),
the short form of subunit IV (m/z = 5270), and a third form with m/z = 5088.
(Additional peaks represents salt adducts of peptides). The third form with m/z = 5088
is the proteolytic cleaved long form of subunit IV (see later for reasoning), which
corresponds to the short form of subunit IV without either the N-terminal alanine and
aspartic acid or the C-terminal asparagine and alanine (see Figure 2.1 for complete
amino acid sequence of subunit IV). Current evidence favors the first possibility since
the electron density for the C-terminal two amino acid residues of subunit IV, N50“;
and A51w, is always clearly visible in the crystal structure. This suggests their stable
presence in stoichiometric amounts. On the other hand, no such evidence exists for
the first two residues in the subunit IV, A2“; and D3w, since the entire N-terminal 9
residues are missing in the crystal structure. By comparing the peak heights in the
spectra, it can also be seen from Figure 3.16 (A) that RchO complex with the long
form of subunit IV is selected more favorably during nucleation and/or crystal growth
process, since the relative abundance of the long form of subunit IV is much higher in
the crystal than it is in the purified enzyme. However, coexistence of different
subunits IV would be expected to negatively affect the X-ray diffraction quality of the
four subunit RchO crystals; in fact, as observed in Part 1, crystals obtained from R. s.
strains overexpressing long subunit IV generally diffracted X-rays a bit worse than
those obtained from strains overexpressing short subunit IV.

Figure 3.16 (B) shows the mass spectra of the subunit IV region from purified

133

enzyme and re-dissolved crystals of RchO from R. 3. strain 169, which has in its
overexpression plasmid the gene for the short form of subunit IV (see Table 2.1). It
can be seen from these spectra that, although there is no long subunit IV gene in the
overexpression plasmid, there is a very small amount of long form of subunit IV in
both the purified enzyme and re-dissolved crystals. This small amount of the long
subunit IV is the expression product from the R. s. genome. Moreover, compared with
the spectra in Figure 3.16 (A), there is no third form of subunit IV found in the Figure
3.16 (B), which strongly suggests that this form of subunit IV with mlz = 5088 is the
proteolytic cleavage product of the long subunit IV. Interestingly, two extra peaks
close to the peak representing short subunit IV with m/z = 5312 and m/z = 5199 are
observed. The mlz value of the first extra peak is 42 units higher than that of the short
subunit IV, which could indicate a chemical modification of the short subunit IV, such
as acetylation. However, the exact nature and position of such a chemical
modification is unclear. The mlz value of the second peak is 71 units lower than that
of the short subunit IV, which could be caused by a proteolytic cleavage of a single
alanine residue from either the N-terminus or the C-terminus of the short subunit IV.
(Again, likely the unresolved N-terrninus.)

Although RchO obtained from strain 169 is more homogeneous in terms of
subunit IV than that from strain 207, there is still a small amount of contamination of
the long subunit IV present, which could negatively affect the homogeneity of protein
product and crystal quality. In an effort to produce the ultimately homogeneous form

of subunit IV, a new R. s. strain, 169A4, was made by conjugating the overexpression

134

plasmid having the short subunit IV into a new parental R. 5. strain, AIAIV, which had
subunit I and subunit IV deleted from its genome (see Table 2.1). Figure 3.16 (C)
shows the MALDI mass spectra of the subunit IV region from purified enzyme from
R. s. strain 169A4. It can be seen that there is no long subunit IV with mlz=6257 in the
enzyme complex, as expected. The assumed acetylated form of short subunit IV and
the short subunit IV less an alanine are still present and interestingly, the relative
amount of the proteolyzed short subunit IV becomes even greater as judged from the
peak heights. There is evidence that additional peaks at mlz=573l and m/z=603l may
be lipid complexes of the short subunit IV (Distler et al., 2004). Although more than
one form of subunit IV is present, the protein product from R. .9. strain 169 is the
closest to being completely homogeneous in that there are only very small variations
among the different forms compared with the long and short forms of subunit IV.
Unfortunately, crystals of the four subunit RchO obtained from strain 169A4 showed
slightly worse X-ray diffraction (3.3 A) compared with those obtained from strains
167 or 169.
3.2.2 Crystallization of [-11 Subunit Rch0 and Diffraction Quality of LI]
Subunit RchO Crystals

RchO obtained from strain 169A4 did not lead to crystals of the four subunit
enzyme with improved X-ray diffraction, but instead a completely new and
unexpected form of the crystals was produced after several weeks (see Method
section). These crystals contained only the catalytic subunits I and II of RchO, the

enzyme was purified by using the imidazole step gradient method, with a small

135

amount of cadmium (1.3 mM) added into the crystallization mixture (see Methods
section for details of the conditions). Figure 3.17 shows a picture of the HI subunit
RchO crystals. It can be seen in the figure that there are two different crystal forms
present in the crystallization drop. The background of tiny crystals are those of the
four subunit RchO and the bigger football-shaped ones are crystals with only the two
catalytic subunits of RchO.

A new R. s. strain named 37A4 was also made by conjugating a new
overexpression plasmid which contained no subunit IV sequence into the parental
strain AIAIV (see Table 2.1). RchO obtained from strain 37A4 contained no subunit
IV at all. When the enzyme was used as a crystallization candidate, under similar
crystallization conditions, crystals of the four subunit enzyme were not formed.
However, crystals of the HI subunit RchO could be obtained with similar X-ray
diffraction quality (approximately 2.4 — 2.5 A resolution) although many of them
appeared twinned.

Due to the lack of strong crystal contacts along the direction parallel to the
native membrane, membrane protein crystals often display anisotropic diffraction
behavior which leads to difficulties in data processing and subsequent structural
refinement. The crystal of the four subunit RchO diffracts X-rays anisotropically and
the structure was solved to a resolution of 2.3A along a* and b* and 2.8A along c*
(Svensson-Ek et al., 2002). On the other hand, the HI subunit crystals displayed
isotropic X-ray diffraction to 2.35 A resolution. Table 3.8 shows the data collection

and structural refinement statistics of the best crystal obtained from R. 3. strain 169A4.

136

 

Figure 3.17: Crystals of the 1-11 subunit RchO which belong to a space group of
P212121. Tiny particles are crystal showers of the four subimit RchO immersed in a
heavy protein precipitation.

137

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Crystal Name I 169A4, [-1] subunit Crystal
A. Unit Cell Parameters 7
Space Group P212121
Cell Dimensions (A) a=125.0 b=131.3 c=176.1
No. Molecules per A.U. 2
B. Data Collection
Resolution Range (A) 30 - 2.35 (2.43 — 2.35)
Completeness (%) 99.9 (99.9)
No. Unique Reflections 112,854 (12,046)
Overall Redundancy 8.5 (6.4)
Overall Rmerge (%) 7.0 (63.5)
C. Structural Refinement
R-factor / R free (%) 19.4 / 21.4 (24.5 / 27.5)
Average B-factor 34.0
Non Cryst. Symmetry -
F / sigma Cutoff -
R.m.s.d Bond Lengths (A) 0.010
R.m.s.d Bond Angles (°) 1.16
No. of Atoms 13141

 

 

 

 

Rmerge = 2 l I h - <Ih> I / 2‘. 1h over all h, where 1h is the intensity 1
of reflection h. R-factor= 2‘. l IFol - chl I / 2 IFol; where F0 and PC
are the observed and calculated structure factors, respectively.

Values in parentheses are parameters for the highest resolution
shell.

Table 3.8: X-ray data collection and refinement statistics of the HI subunit
RchO crystals.

138

It can be seen that compared with previous four subunit RchO crystals, refinement of
the HI subunit crystals led to much improved R and Rfrec factors and lower B factors.
Compared with the four subunit RchO crystals, which were grown at 4°C and would
quickly dissolve upon a rise in temperature, the HI subunit RchO crystals were also
grown at 4°C but remained stable at room temperature, which could suggest tighter
packing of molecules in this form. Besides using cadmium as the additive,
crystallization experiments were also performed with other divalent cations such as
mercury, nickel, manganese, and strontium. However, no I-II subunit RchO crystals
were obtained under similar conditions.
3.2.3 Biochemical Analysis of MI Subunit RchO Crystals
3.2.3.1 UV-visible Spectra and Activity Assays of Re-dissolved I-II Subunit
RchO Crystals

Several crystals of LI] subunit RchO were removed from the crystallization
drop and dissolved in buffer for biochemical analysis. Figure 3.18 shows the
comparison of UV-visible absorption spectra of oxidized (blue) and reduced (purple)
spectra of purified four subunit enzyme and re-dissolved I-II subunit crystals. The
re-dissolved crystals show normal heme peaks at 445 nm and 606 nm in the reduced
form compared with the purified four subunit enzyme, but the ratio of A230 / A42. is
much lower (1.35 in Figure 3.18 (B) vs. 2.3 in Figure 3.18 (A)). The decreased A230 in
the crystals compared with the whole enzyme indicates that a significant portion of
enzyme molecules have shed subunits III and IV in the process of forming crystals.

The enzymatic activity of re-dissolved I-II subunit RchO crystals was

139

 

 

 

 

 

  

 

Absorbance

 

 

250 300 350 400 450 500 550 600 650
Wavelength (nm)

Figure 3.18: UV-visible spectra of purified RchO (A, top panel) and re-dissolved
[-1] subunit RchO crystals (B, bottom panel). The spectra of the oxidized CcO are
shown in blue, and dithionite-reduced CcO spectra are shown in purple. Normal heme
peaks are observed in the re-dissolved crystals as compared with those in the purified
enzyme as labeled in the figures. However, the A230 / A421 ratio is much lower in the
re-dissolved crystals than in the purified enzyme.

140

measured under steady state conditions using an oxygen electrode; the reaction traces
showing the rates of oxygen consumption are shown in Figure 3.19. The initial
oxygen consumption rates are listed in the inset of the figure. Note the activity
decreases as the enzyme turns over, a phenomenon not seen with the wild type RchO
but typical of subunit III-less CcO (Bratton et al., 1999; Hosler, 2004). Adding lipid
and arachidonic acid protects the enzyme from undergoing suicide inactivation, as
shown in the blue and red traces in the figure (Mills and Hosler, 2005). The initial
rates measured are similar to those reported for purified subunit III-less enzyme (Mills
and Hosler, 2005). It should be noted that these crystals had been maintained under
crystallization conditions for approximately 3 months before being dissolved and
assayed. The high activity observed strongly supports previous observations that the
[-11 oxidase is highly stable unless it is undergoing turnover. It was recently found that
subunit [II-less RchO obtained by treatment of the four subunit enzyme with Triton
X-100 actually retained subunit IV in the enzyme complex (Hosler, personal
communication). Therefore, it would be useful to compare the results of structural and
functional analyses between I-II-IV subunit RchO and the re-dissolved I-II subunit
crystals for better understanding of subunit IV, whose role is unknown to date.
3.2.3.2 MALDI Mass Spectrometry Analysis of Lipids

MALDI mass spectrometry was again used to analyze the bound lipids in both
purified enzyme and in re-dissolved [-1] subunit crystals. Figure 3.20 shows the
comparison of lipid species present in purified enzyme obtained from R. 5. strain

I69A4 and in re-dissolved I-II subunit crystals. It can be seen that in the purified

141

 

   
  

 

 

 

 

 

 

  

 

 

 

Sample Activity
25° . (e‘ / sec)
Re-dissolved crystals 570.5
200 '
+ Lipid 734.6
150 + Lipid 937.4
A ' + A racllitlonic Acid
2
:1.
‘1: 100 -
O
m u
o I I U I T T
0 100 200 300 400 500 600

Time (s)

Figure 3.19: Enzymatic activity measurement of re-dissolved I-II subunit RchO
crystals under steady state conditions using an oxygen electrode. The reaction
traces represent the oxygen consumption for re-dissolved crystals (black),
re-dissolved crystals with asolectin added to the reaction mixture (blue), and
re-dissolved crystals with asolectin and arachidonic acid added to the reaction mixture
(red). The initial oxidase activity value under each condition is listed in the inset of
the figure.

I42

$8995 «.55.: EA 328%??— 5 :5. 0.93 $05.5.— E «:30...— 832: 2&— 2: ”Egan—3 2.2: :3 9.5.32. 15

95.3.. 53 5 38...? 060.4 «.55..» :4 32835.. was 069...: 55.5.. a: 2539. 2315.... 950% 8a... 5452 Sufi 9:55
NE.

 

 

 

 

 

 

 

~>5 mam—2 mam—2
82 mm: 3". 82 8o 8&8.“ o8 95-1.- was -1- a: 112...".
{iii .3;
S A
3 i 8 egégx he... :ééx on
3
36% Ram %
3 3 M
M m.
8 . 8 mt
TV 25me 8 T; 23390 8
. «Eu:
. on! 53.50 8. N8 8.
82 83 3a? 88 8.. 2: m8 incl... 11 ham 1 i 1.311 1. 2:
jmmllll ll-, : 2.. $3,351,: .7235. i. 5J7??? 3:13.32 33.44 n.4,... n. a
on : 8
new .9; %
ace coca a. a 3.8:: 8 M
3
8. ace“: 8 m.
8 8
s: t 5 25: E 2.55 352
€3.50 A v 6:. 5 ES— 8. Sm c2

 

 

 

 

 

143

enzyme sample, two phospholipids, PE and PC, were identified using the positive ion
mode and CDL and SQDG were identified using the negative ion mode. Interestingly,
in the re-dissolved I-II subunit crystals, no PE or PC was detected using the positive
ion mode, while CDL and SQDG were identified in the spectra using negative ion
mode. The above observation could be due to the fact that PE and PC are bound to the
interfaces of subunits III and IV, and they were stripped off the enzyme complex
along with subunits III and IV. This emphasizes the likely importance of the lipid in
the structure of these two subunits.
3.2.4 Crystal Structure of the [-11 Subunit RchO
3.2.4.1 Overall Structure of LI] Subunit RchO

The overall structure of the two subunit RchO is almost identical to the
structure of the subunits I and II in the four subunit structure, with an RMSD of
approximately 0.5 A over all the Cu atoms, as revealed by DALI homology search
(Holm and Sander, 1993) (Figure 3.21). Such a su'ong similarity was also observed in
the four subunit and HI subunit structures of PdCcO (Ostermeier et al., 1997). This is
reassuring since the results and conclusions drawn from the two subunit crystal
structure can then be applied to understanding the four-subunit enzyme with
reasonable confidence. The two iron heme molecules, copper centers, and non-redox
active metal centers including Mg2+ and Ca2+ found in the four subunit structures are
also found in the two subunit structure, with their ligating residues. Unlike the case of
the four subunit crystal structure, the two molecules in the asymmetric unit of the

two-subunit crystal are different from each other with respect to the arrangement of

144

 

Figure 3.21: Comparison of the structure of the [-11 subunit Rch0 and the four
subunit RchO (PDB entry 1M56). The main chain peptide of subunits I and H fiom
the 1-11 subtmit crystal structure are shown in blue and red, respectively, while those
from the four subunit structure are shown in yellow and green, respectively. Subunits
III and IV firm the four subunit RchO structure are shown in gray. Note that the
structure of subunits I and II from the two crystal structures are almost identical.

I45

some water molecules and other bound molecules around and within them. Thus we
have two independent I-II subunit RchO crystal structures and the differences
between the two molecules are likely due to different crystallographic environments.
. 3.2.4.2 Crystal Packing of MI Subunit RchO

Figure 3.22 shows the crystal packing of a unit cell of LI] subunit RchO. It
can be seen that, compared with the crystal packing in the four subunit RchO, which
occurs primarily within the a*b* plane (Figure 3.11), crystal packing of this form of
RchO crystal is 3-dimensional. Therefore, there was no anisotropic diffraction
commonly observed with membrane proteins. As shown in Figure 3.22 and Figure
3.23, analysis of the crystal contacts reveals a new contact region involving the
subunit H extramembrane domains, mediated by two cadmium ions as well as the two
engineered histidine tags. As shown in Figure 3.23 (A), each cadmium ion is ligated
in a tetrahedral coordination geometry by the side chains of three amino acid residues,
including two histidines, H283" and H285", and one glutamic acid, E28011, from one
protein molecule and one other glutamic acid residue, E152"*, from another protein
molecule. Interestingly, the two participating histidine residues, H283“ and H285",
are from the engineered histidine tag. Another contact exactly like this exists close by,
involving the same residues from the opposite protein molecule. The two crystal
contacts display a quasi two-fold rotational symmetry as shown in Figure 3.23 (B).
Such strong crystal contact mediated by cadmium ions is likely responsible for the
superior X-ray diffraction compared with the 4 subunit protein crystals. This novel

crystal contact involving an engineered histidine tag and a cadmium ion has rarely

146

Unit cell display of [-1] subunit Rch0. The crystal belongs to a space

 

group of P212121. There are two RchO molecules per asymmetric unit and they are
colored blue and green. Two cadmium ions (purple) were found at the molecular

interface mediating key crystal contact.

Figure 3.22

147

       

o A "

-46?» .

are: 2?.» -
\I

  

2.,

 

 
      
 

'n» h 3’.” :7. -~‘- . A}; w v - .
w .. . t - «seam- --
f 4"." V a? v ,at'
h’l;§‘3"’ i: ’0'}; 1. -“‘I':.lr~. .:-
_, u , rarer"
Figure 3.23: Major crystal contact region of [-1] subunit RchO contributed by
the engineered histidine tag at the shortened C-terminus of subunit [1 and the
cadmium ion. One molecule is colored by atom type (C: green, 0: red, N: blue),
while the other molecule is colored yellow. The (2F0 — Fe) electron density map
contoured at 1.0 o is shown in blue. (A): The participating amino acid residues fi'om
the two protein molecules form a tetrahedral coordination around the cadmium ion
(purple) as labeled in the figure. (B): There are two identical crystal contact
interactions at this interface with a quasi-two-fold rotational symmetry axis at the

center.

  
  

148

been seen (see Discussion).
3.2.4.3 The Binuclear Center

The binuclear center consisting of heme a3 and Gun is the active site of the
enzyme where oxygen is reduced to water (Ferguson-Miller and Babcock, 1996). A
covalent linkage between the C82 of Y288. and the N82 of a Cug ligand, H2841, is
observed in the crystal structures of CcO from bovine mitochondria and P. d.
(Ostenneier et al., 1997; Yoshikawa et al., 1998), and its presence is supported by
mass spectrometry analysis of C00 from a variety of sources (Buse et al., 1999), as
well as by FTIR studies on the E. Coli cytochrome b03 quinol oxidase in the same
family of heme-copper oxidases (T omson et al., 2002). Such a covalent linkage was
not observed unequivocally in the wild type four subunit RchO crystal structure. It
was suggested that there was a mixture of covalently linked and non—covalently linked
species in the crystal (Svensson-Ek et al., 2002). In the current structure, however, a
covalent linkage is clearly observed. Figure 3.24 shows the simulated annealing omit
map contoured at 5.5 0’. In this figure, clear electron density is observed between the
N82 of H284; and the C82 of Y2881. Model building of this linkage did not lead to
significant distortion of the bond geometry and planarity of the two involving
residues. This observation further supports the presence of this covalent linkage and
its likely importance in chemical catalysis. Such a covalent linkage is believed to
lower the pKa of the tyrosine OH group and facilitate free radical formation of Y288I.

The tyrosine free radical is postulated to form during turnover, allowing the 2-electron

I49

-- . '__,.'
,fh."|“"”
\

' l’«‘.en‘-

. a" -\

‘Aw.J

v9.23. ,
\‘P/ \““-'-.-,‘. I
4‘ '7': 3“, ..

 

Figure 3.24: The covalent linkage between the ring atoms of Y2881 and H284; of
Rch0. The (F0— Fe) difl‘erence simulated annealing omit map contoured at 5.5 o is
shown in blue. The amino acid residues are colored by atom type (C: green; 0: red; N:
blue), the heme 03 and its Fe atom are colored light red, and the Gun atom is colored
purple. (A) and (B) are fi'om two difl‘erent views.

150

reduced enzyme [Feag2+ - CuBH] to carry out a 4—electron reduction of oxygen, and
avoid the production of a peroxy intermediate (Proshlyakov er al., 1998).

Between the Fe in the heme a3 and the Cu center, there is residual density in
the (FD-Fe) map. Previously observed density in this region has been interpreted as a
water molecule ligated to F603 and a hydroxide ion to Cu (Ostenneier et al., 1997) or
a peroxy-bridge the binuclear center (Y oshikawa et al., 1998). However, the fact that
four electrons are required to fully reduce the active site contradicts the presence of
such a peroxy form (Steffens et al., 1993). Efforts to fit two separate ligands between
F603 and Cu}; also turned out to be difficult leading to the two ligands being too close
(1.5 A) to each other. According to a recently proposed proton pumping mechanism
(Sharpe et al., 2005), there is likely to be a fixed hydroxide on the copper and a less
well defined water between the heme iron and Cu]; center in the oxidized enzyme.
Such a dynamic feature is likely why it was hard to build in models in this region.
3.2.4.4 Proton Uptake Pathways

There are two established proton pathways in the enzyme: the D and K
pathways, named after the residues D132. and K3621, respectively, whose mutational
replacement blocks each path, respectively (Ostenneier et al., 1995; Svensson-Ek et
al., 2002). In the four subunit RchO crystal structure, a clear chain of 10 water
molecules was resolved in the D pathway from D1321 leading to the vicinity of E2861,
a residue thought to be critical in conducting protons from the D path to the active site
and the external bulk phase (Hofacker and Schulten, 1998). Such a chain of waters is

not as clearly resolved in the current structure. Figure 3.25 shows the comparison of

151

0" -, \
$3

A.-
J A' ‘
o J

1‘, .1»
,.~\' .‘ Q
@554?

. t

 

     

Figure 3.25: Comparison of the resolved waters in D proton uptake pathway in
the crystal structures of [-1] subunit Rch0 and the four subunit RchO (PDB
entry 1M56). Amino acid residues from the 1-11 subunit RchO structure are colored
by atom type (C: green; 0: red; N: blue). Waters resolved in the D pathway from the
[-11 subunit 1?.ch0 are shown in red and those from the four subunit enzyme in
yellow. The (2Fo - Fe) difference electron density map contoured at 1 o is shown in
blue.

152

the resolved D channel waters in the [-11 subunit and four subunit RchO structures
colored in red and yellow, respectively. The conformations of the amino acid
residues along the pathway are very similar between the two structures and only the
residues from the current I-II subunit RchO structure are shown. It can be seen that
there are 3 fewer resolved waters near the important residues along the pathway; the
former are shown in red, and they are hydrogen bonded to D1321, N1211, N1391,
$2001, 82011, 81971, and E2861. Although plausible, it is not likely that the fewer
number of resolved water molecules are due to insufficient X-ray diffraction
resolution since this structure has better resolution in 3 dimensions than the previous
one. Interestingly, the PdCcO I-II subunit structure was also lacking in resolved
waters in the D channel (Ostenneier er al., 1997). Subunit III, although not directly
involved in either electron transfer or proton pumping, appears to play a role in the
kinetics of the D channel (Hosler, 2004). Its close proximity to the D pathway,
mediated via the phospholipid molecules, may account for this effect. A less
well-defined arrangement of water in the D pathway could contribute to a less
effective proton transfer.

Compared with the D pathway, the K pathway has very few waters resolved in
the four subunit RchO structure, although molecular dynamics studies suggest that
there are many more in the pathway (Cukier, 2005; Seibold et al., 2005). The H]
subunit RchO crystal structure shows a similar water arrangement to the four subunit

RchO structure, as shown in Figure 3.26.

153

.o‘: d
'62.! It ,1
flv'mflifl‘.

at. "— r.'.‘x?- r‘_
‘ ' Q'a-

“ .‘H I
-

-'.~.-. . i
|'“\ In?"

 

Figure 3.26: Comparison of the resolved waters in K proton uptake pathway in
the crystal structures of [-1] subunit Rch0 and the four subunit RchO (PDB
entry 1M56). Amino acid residues from the I-11 subunit RchO structure are colored
by atom type (C: green; 0: red; N: blue). Heme a3 is shown in light red. Waters
resolved in the K pathway from the HI subunit Rch0 are shown in red and those
from the four subunit enzyme in yellow. The (2F0 — Fc) difference electron density
map contoured at l o is shown in blue.

154

3.2.4.5 Additional Cadmium Binding Site

Besides the cadmium binding site found at the protein-protein interface as
shown in Figure 3.22, an additional cadmium binding site was identified which was
ligated to side chain atoms of E101" and H96“, as shown in Figure 3.27. It can be
seen from this figure that the conformations of the two ligating residues (colored by
atom type) changed significantly from the unliganded state (yellow) (published four
subunit RchO structure, PDB entry 1M56) upon binding of the cadmium ion at this
site. This cadmium binding site was also observed in the structure of the four subunit
RchO when the crystal was soaked in a CdClz solution before flashcooling (see
3.1.6.2). E101 n has been suggested to be the proton entry point for the K proton
uptake pathway in Rch0 (Branden et al., 2002; Tomson et al., 2003). It was also
found previously that an‘VCd2+ inhibits the CcO activity both in reconstituted lipid
vesicles and under steady state turnovers (Mills et al., 2002). Therefore, it is likely
that this cadmium binding site could be the inhibition site on the inside of the
membrane responsible for the lower activity under steady state turnovers. Binding
of ZnZJ'ICd2+ at this site may lead to inhibition of the enzymatic activity by blocking
the K pathway, which would contradict a previous conclusion that the Zn2+/Cd2+
inhibition site(s) on the inside of the membrane is at the D pathway (Aagaard et al.,
2002). Site-directed mutants which eliminate the binding of cadmium at this site are

currently being generated and their effects on ZnZ‘VCd2+ inhibition will be studied.

155

 

Figure 3.27: Additional cadmium binding site in the structure of [-11 subunit
Rch0. The additional cadmium binding site is located on the inside of the
membrane close to the subunit I (yellow) and 11 (green) interface, as shown in the left
panel. The (2170— Fe) difference electron density map contoured at 1.0 a (blue) is
shown in the right panel. In the current structure (colored by atom type), cadmium
(purple) is bound to E10111 and H9611 of subunit II. Binding of cadmium alters the
conformations of the ligating residues compared to that found in the unliganded four
subunit RchO crystal structure (yellow) (PDB entIy 1M56).

156

3.2.4.6 Detergents and Lipids

A number of alkyl chains and detergent head groups were also resolved in the
current structure. There are 3 complete detergent molecules resolved, two of them
decyl maltoside based on the length of the resolved alkyl chain, and the third dodecyl
maltoside. Both detergents were present in the crystallization mixture. Figure 3.28
shows the two decyl maltoside molecules which reside at the interface of two protein
molecules. Interestingly, the positions of the two detergent molecules suggest that
they may be involved in mediating contacts: their hydrophobic tails interact with the
transmembrane region of one molecule and their polar head groups interact with the
extramembrane soluble domain of another molecule.

Membrane lipids play important roles in transmembrane proteins. They not
only provide a hydrophobic matrix and diffusion barrier, but also contribute to
functional aspects of membrane proteins, a subject of ongoing investigation. In the
crystal structure of the four subunit RchO, a total of 6 phosphatidyl ethanolamines
were identified. Four of them were found surrounding subunit IV, virtually isolating it
from protein contacts, and the two others were found in the crevice within subunit III
and associated with subunit I (Svensson-Ek et al., 2002). In fact, subunit IV lacks
direct contact with its neighboring subunits I and III, and its interaction appears to be
mediated via the membrane lipid molecules (Svensson-Ek et al., 2002). Somewhat
surprisingly, in the current structure, we did not find any lipid molecules that

correspond to any of the six lipid molecules found in the four subunit structure,

despite our efforts to retain lipids by using less detergent, combined with fewer steps

157

 

Figure 3.28: No decyl maltoside detergent molecules resolved at the interface of
two RchO molecules (gray and wheat). The detergent molecules are colored by
atom type (C: green; 0: red; N: blue). Two different RchO molecules are colored
wheat and gray, respectively. The (2Fo— Fc) difference electron density map (blue)
surrounding the decyl maltosides contoured at 1.0 o is shown. Tyrosine and
nyptophan residues (yellow) were found nearby the two decyl maltosides forming
stacking interactions with the detergent sugar ring.

158

of column chromatography. It appears that when subunits III and IV dissociated from
the enzyme complex, they take the associated lipids with them. In other locations,
however, there were quite a few tube-like electron density features found in the
transmembrane portion of the enzyme (Figure 3.29). These tube-like features were
interpreted as hydrocarbon tails of either native membrane lipid molecules or
detergent molecules. The headgroups of these hydrocarbon tails were not resolved,
suggesting flexibility as in the case of the partial lipids identified even in the 1.55 A

resolution bacteriorhodopsin (Luecke et al., 1999).

159

 

Figure 3.29: Structure of [-1] subunit RchO showing the resolved detergent
molecules, detergent headgroups, and alkyl tails of detergents or membrane
lipids. The protein surface is colored by charge (blue: positive; red: negative) and the
resolved detergent molecules, detergent headgroups, and alkyl tails are shown in
yellow.

Chapter 4. DISCUSSION
4. 1 Importance of Homogeneous Subunits in the X-ray Diffraction of RchO
Crystals

X-ray crystallography of a membrane protein continues to be an extremely
difficult task despite significant progresses made during recent years (L011, 2003;
Torres et al., 2003; White, 2004). Besides the difficulty associated with working with
protein detergent complexes, which is usual for membrane proteins (Garavito and
Ferguson-Miller, 2001; Seddon et al., 2004), the efforts to successfully crystallize
RchO, a multi-subunit integral membrane protein, have also been complicated by the
fact that incomplete post-translational modification and processing of the enzyme
subunits often leaves an inhomogeneous enzyme complex, which is detrimental to
obtaining well ordered protein crystals. Such processing events include
proteocleavage of the C-terminal 13 amino acid residues of the subunit II, as well as

two different potential translation starting codons in the subunit IV gene and further
proteocleavage processing of each form of subunit IV (see Results Section and Figure

2.1). In a systematic approach to improving the X-ray diffraction quality of the four
subunit RchO crystals, these inhomogeneities in subunits II and IV were remedied

by molecular engineering of unique R. s. strains. Table 4.1 shows the different R. 3.
strains made and the different types of subunits H and IV in their expression products.

as well as the protein expression levels and the diffraction limits of crystals obtained.

It can be seen that in general, more homogeneous subunits lead to better diffraction of

the four subunit RchO crystals. This is particularly true in the case'of

l6l

 

 

 

6035:: mambo He 33.: net—:89. 533.5% 05 use 295. =£82nxo £89:— hofi .3259:—
:emmmo.axo :2: 5 a can HH mad—5.5 me 893 «first? 5.3 335m .u .2 £20,5—. He raga 0:. 03¢.

325 as? E 05.6 :8: Sesame a

8.80% e3 «5 mace 350 E59... Be 05% :5: Saopae w
.2“ u 2 35%... age a.
Ease. Easez .. Z 353 8.3 a m 3&5; .. 2 E53 was a
Z 058. «a n .v .m a 4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

an am he .. m3w§.m£ a new 25
mmd
a,“ am he use 88085 + mace anaemia a new 2%
w
2 mm 8% was + 55% Baooea + secs... Maegan m ea a:
”4 mod poem $85 + owcoH 3330.5 + vwcoH neowwSéE HH new SH
m4 Nd noon va0_ + N.touHm vommoooa + nfiosm «6385.5 H Dam moH
m4 m.m coon $85 + m.waoH @3885 + vwcoH «@8855 H saw «.2
. «38055 +

m4 0 m coca «$2 + $85 commoooa + $85 $38089 + E38: H new a:

”4 v voow ewaoH + $85. @8885 + atonm $03305 + H333 H 93m 52
m4. v A boom ewGOH $03805 + mtonm + vwcoH $03803 + 3.5ch H 93m 02
Hen—who comma

swam ocaxm 3 “Spam HH 2953 m3 Eabm

imwm .M. u“

 

 

162

subunit II, deduced by comparing the diffraction limit of crystals obtained from R. 5.
strain 120, 119, and 156, which shows significant improvements in X-ray diffraction
limit (from worse than 4 A to approximately 3.2 A) upon removing the
inhomogeneity in subunit H.

In the case of subunit IV, similar observations could also be made by pair-wise
comparison between RchO crystals obtained from strains 120 and 157, 156 and 163,
and 167 and 169, in which cases a slight improvement the in X-ray diffraction limit
(0.1 - 0.2 A) was observed when the short subunit IV gene was inserted into the
overexpression plasmid rather than the complete subunit IV gene. If the MALDI mass
spectrometry analyses of different forms of subunit IVs from R. 3. strain 169 can be
applied to other strains of R. s. which have the same subunit IV gene in their

overexpression plasmids, (see 3.2.1, Figure 3.16 (A) and (B) and Table 2.1), then in
the protein product from the R. s. strains 157, 163 and 169, the short and proteolytic
cleaved short form (difference of an alanine residue) should be the predominant
species of subunit IV. Likewise, if the MALDI mass spectrometry analyses of
different forms of subunit IVs from R. .9. strain 207 can be applied to other strains of R.
s. which have the same long subunit IV gene in their overexpression plasmids, (see
3.2.1, Figure 3.16 (A) and (B) and Table 2.1), then in the protein product from the R. 3.
strains 120, 156 and 167, the long and proteolytic cleaved long form (difference of 12
amino acid residues) were the predominant subunit IV species. In the former case, the

inhomogeneity in subunit IV is much less severe than that in the latter case, which

could lead to slightly better X-ray diffraction in the crystals.

163

 

Somewhat surprisingly, when RchO obtained from strain 169A4 was used as
a crystallization candidate, the crystal diffraction resolution limit was worse than that
obtained from strain 169 and 167 (3.3 A vs. 2.9 A and 3.1 A, respectively), although
the subunit IV in Rch0 obtained from strain 169A4 was more homogeneous than
that from strain 169 (see Figure 3.16(C)). Since RchO obtained from strain 169A4
did not contain any long form of subunit IV, whereas RchO from strain 169
contained a very small amount of subunit IV, the observed difference in diffraction
limit could be due to the long form of subunit IV being somehow beneficial for
obtaining well-ordered crystals, since the long form of subunit IV seems to be
selected more favorably during nucleation/crystal growth processes. It is also
observed that four subunit RchO crystals obtained from strain 169A4 are generally
smaller than those obtained from strains 167 and 169. In fact, no crystals similar to
the sizes of the best crystals obtained from 167 and 169 have been obtained, which
could be another reason why the diffraction was worse, since it has been found that
larger RchO crystals generally diffract further than the smaller ones. Therefore, it
seems that a homogeneous long form of subunit IV could be more helpful in obtaining
well diffracting four subunit RchO crystals. Generation of such a strain of R. 5.
would require at least a few mutations to remove the second starting codon as well as
the proteolytic cleavage site 3 amino acids downstream, which could be technically
challenging and time consuming.

Although R. 3. strain 169A4 did not lead to improvement in X-ray diffraction

of the four subunit RchO crystal, its use as the crystallization candidate did lead to

 

formation of the HI subunit crystal. The more homogeneous and long subunit IV—free
RchO complex obtained from strain 169A4 turned out to be important since RchO
obtained from strain 167 failed to produce any I-II subunit crystals under similar
crystallization conditions. It is possible that the long form of subunit IV binds to the
enzyme complex more tightly than the short form of subunit IV, and is therefore
harder to be removed from the enzyme complex under the crystallization conditions
for the HI subunit RchO crystal to form.

One puzzling observation made is that the protein expression level of R. 3.
strains 119, 156 and 163 were extremely low, presumably due to deletion of the CcO
gene(s) from its overexpression plasmid while keeping the drug resistance gene(s)
during bacterial growth. As a result, R. 3. cells that have successfully deleted the C00
gene will gain a growth advantage over other cells and thus self-replicate much faster
and consume almost all the nutrients in the growth media with no production of
RchO. A cursory examination of those R. 3. strains revealed that they all have genes
for the artificially truncated form of subunit 1] (see Figure 2.1 and Table 2.1) and
either long or short subunit IV in their overexpression plasmid, although it is not clear
if these characteristics are important. The exact reasons for the occurrence of CcO
gene deletion to occur and ways to prevent it have not been found.

4.2 Effects of the Histidine Tag and Metal Ions on X-ray Diffraction of
RchO Crystals
Table 4.1 also shows the effects of the position of the histidine tag on the

diffraction resolution limit of RchO crystals. It can be seen that, in general, crystals

165

 

of the four subunit RchO with a histidine tag attached to the shortened C-terminus of
subunit II have better X—ray diffraction than those with the histidine tag attached to
the C-terminus of subunit 1. One of the reasons for such an improvement is that the
histidine-tagged subunit H is homogeneous. Another potential reason could be that the
engineered histidine tag might have helped to form a new crystal contact. As observed
in the crystal structure obtained from strain 167, shown in Figure 3.8, extra electron
density could be seen which appeared to connect the C-terminus of subunit II to
another molecule. This continuous piece of electron density might suggest that the
histidine tag was immobilized and contributed to making new crystal contacts,
although the modeling of the histidine tag was not completed. However, similar
electron density was not observed in this area in the structure of the crystal obtained
from R. 5. strain 169, although it also possessed a histidine-tagged subunit II. A
possible explanation of this discrepancy could be that the difference in subunit IV
contents between RchO obtained from 167 and 169 leads to the observed differences
seen in this area, since the extra piece of electron density observed in the crystal
structure of RchO from strain 167 (Figure 3.8) is too large and consistent to be
purely noise. However, when a different X-ray crystallographic refinement program,
Refmac5, was used, no such large electron density was observed after a TLS
refinement followed by a restrained refinement.

The role of the engineered histidine tag at the near C-terrninus of subunit II in
forming crystal contacts in the HI subunit RchO crystallization is unequivocal. As

shown in Figure 3.23, the major crystal contacts between two adjacent Rch0

166

5.;

 

molecules are mediated through the exogenous cadmium ions. Each cadmium ion is
ligated in a tetrahedral coordination geometry by the side chains of three amino acid
residues, including two histidines and one glutamic acid, from one protein molecule
and one other glutamic acid residue from another protein molecule. Interestingly, the
two participating histidine residues, H283" and H285", are from the engineered
histidine tag.

One of the commonly used divalent cation crystallization additives, cadmium,
has the ability to coordinate with amino acid residues, water, or other solutes in the
crystallization mixture with different degrees of coordination. As a result, cadmium
ions have been resolved in a large number of crystal structures, many of them
occurring in the crystal contact regions as seen in the current structure (T rakhanov and
Quiocho, 1995; Trakhanov er al., 1998). What is more interesting about this crystal
structure is that two of the ligating amino acid residues for each cadmium are from the
engineered histidine tag. Histidine tags have rarely been resolved in crystal structures,
except for a few instances (Zhou et al., 2000; Brych et al., 2001; O'Neill et al., 2001;
Stroebel et al., 2003), because they are generally flexible and it is often times
beneficial for crystallographers to cleave off the histidine tag before crystallization
(Ma and Chang, 2004 , Harman, personal communication). However, the analysis of
the HI subunit RchO crystal structure clearly indicates the essential role of the
engineered histidine tag in forming key crystal contacts. Similar observations were
made in the crystal structure of the B1 domain of a mutant protein L from

Peptostreptococcus magnus, in which case the entire histidine tag was found

167

 

coordinated with zinc ions at the protein-protein interface (O'Neill et al., 2001), and in
the crystal structure of ArsA ATPase, in which case the histidine tag was involved in a
crystal contact via a tetrahedral coordination of cadmium with one of the ligands
being a water molecule (Zhou et al., 2000). An engineered internal, rather than N- or
C- terminal histidine tag, was successful in aiding purification and crystallization of
the membrane protein siderophore receptor FhuaA. In the crystal structure, a nickel
ion was found at the molecular interface coordinating with residues from the histidine
tag (Ferguson et al., 1998; Ferguson et al., 1998). Therefore histidine tags, through
strong interaction with other species, particularly metals, can be of general use in
protein crystallization under the right conditions.
4.3 Effects of Membrane Lipids on the X-ray Diffraction Quality of RchO
Crystals
4.3.1 Importance of Lipid Retention in Obtaining High Resolution Crystal of
RchO

It has been generally realized that the role of lipids in the crystallization of
membrane proteins is to maintain a more stable and homogeneous conformation of
membrane proteins. Lipid molecules are resolved in many high resolution membrane
protein crystal structures including that of RchO (Svensson-Ek et al., 2002). During
detergent solublization and protein purification steps, efforts were made to retain the
specifically bound lipid molecules on RchO, including measuring the protein
concentration and calculating the amount of dodecyl maltoside detergent to be added

to the membrane sample, the use of different columns for protein purification,

168

 

decreasing the number of times of washing and detergent exchange, and using a lower
concentration of decyl maltoside detergent in the buffer for crystallization. Lipids
were also added to the crystallization mixtures. Measurements of lipids by using mass
spectrometry, phosphorous analysis and ICP were also taken during different stages of
protein purification in an effort to monitor the lipid content. Although no remarkable
improvement in X-ray diffraction was achieved in the four subunit RchO crystals,
such critical experiments should be performed routinely in order to understand the
lipid content changes when different purification techniques are used and at different
stages during the protein production process.
4.3.2 Use of Mass Spectrometry in Monitoring Lipid Contents during Protein
Preparation of RchO for Crystallization

In order to study the lipid contents during different stages of protein
production of RchO using different column purifications, MALDI-TOF mass
spectrometry has been used to monitor the lipid contents directly from protein
samples during each step of purification without the need for lipid extraction since
lipid extraction using organic solvents might lead to oxidation of lipid molecules.
Because of its very high sensitivity and relative ease, MALDI-TOF mass
spectrometry could be used as an analytical tool for providing us with critical
information on lipid content. Careful analysis of the m/z peaks, together with tandem
MS/MS analysis can provide great details on the exact lipid species, including the
headgroup type, acyl chain length and its degree of unsaturation. However, because

the amount of each species desorbed from the probe may not be proportional to the

169

 

amount in the sample, MALDI-TOP analysis is not a quantitative method. Although
accurate identification on lipid species can be obtained, direct comparisons of lipid
contents in different samples cannot be made. Once a reliable standard control can be
established, the peak heights of each lipid species can then be compared to those of
standards to compare the lipid contents directly among different samples. Another
useful ionization technique used in mass spectrometry, electrospray ionization (ESI),
has the advantage of being both sensitive and quantitative. However, this method
involves lipid extraction from the sample, which could lead to the oxidation and
degradation of lipids. Extraction protocols which could lead to little or no lipid loss
are being developed.
4.3.3 Conserved Lipid Binding Sites Found in the Structure of MI Subunit
RchO and Potential Substitution of Lipids by Detergents

It was somewhat surprising that in the I-H subunit RchO crystals, no lipids
were found that corresponded to any of the 6 PEs found in the published four subunit
RchO crystal structure despite efforts to retain more lipids during the protein
purification process. However, there are a few detergent molecules, detergent
headgroups and other linear molecules which could be hydrocarbon tails of
detergents/lipids resolved at the surface of the HI subunit structure. A very interesting
observation was made when the crystal structures of the two I-II subunit CcOs from R.
s. and P. d. (PDB entry 1AR1) were overlaid, including the resolved partial

detergent/lipid molecules. Figure 4.1 (A) and (B) shows the surface representations of

HO

 

Figure 4.1: Superimposed surface representations of [-1] subunit RchO (gray)
and PdCcO (wheat) together with the detergent molecules, maltose headgroups
and alkyl chains resolved from the structure of RchO (blue) and the resolved
detergent LDAOs from PdCcO (red). The two figures represent views from two
different angles. Note that quite a few of these linear molecules from the two
structures occupy the same sites on protein surface.

171

the superimposed structures of RchO (grey) and PdCcO (salmon) together with the
detergent molecules, maltose headgroups and alkyl chains of RchO (blue) and the
resolved detergent LDAOs from PdCcO (red) (Ostermeier et al., 1997). The two
figures represent views from two different angles. The positions and orientations of
these linear molecules likely represent where membrane lipids reside. Moreover,
many of these linear molecules from the two different organisms occupy the same
positions on the surface of the protein, although the two protein structures were from
two different species and the protein underwent different purification and
crystallization procedures. This suggests that these locations could be conserved sites
for specific interactions between the membrane lipids and the transmembrane protein
surface. In fact, analyses of our structure superimposed with the 1.8 A resolution
bovine mitochondrial CcO structure revealed that one of the proposed alkyl chains in
Rch0 not only coincides with an LDAO molecule from PdCcO structure as shown
in the Figure 4.1, but also coincides with the glycerol backbone and part of the acyl
tail of a lipid molecule, triacylglcerol, resolved in the bovine CcO crystal structure, as
shown in Figure 4.2. The physiological role of triacylglycerol in mitochondrial CcO is
not completely understood; also unclear is whether triacylglycerol is present in the R.
s. cytoplasmic membrane. Direct examples of specific, conserved lipids on the
membrane protein surface also come from the published crystal structures of CcO
from R. sphaeroides (PDB entry 1M56), P. denitrrficans (PDB entry lQLE), and
bovine mitochondria (PDB entry 1V54), in which case a phospholipid molecule was

found to reside at the same position within the comer of the V-shaped cleft formed by

172

 

Figure 4.2: Structure overlay of LI! subunit RchO, I-Il subunit PdCcO (PDB
entry 1AR1), and bovine heart mitochondrial CcO (PDB entry 1V54). The
transmembrane helices of subunits I and II of these CcOs are very similar and they are
shown in gray. Subunit VIc fiom bovine CcO is shown in wheat. Resolved alkyl tail
from the [-11 subunit RchO is shown in green, the detergent LDAO in PdCcO is
shown by atom color (carbon: cyan; oxygen: red; nitrogen: blue), and the resolved
triacyl glycerol in the bovine CcO structure is shown by a difi'erent atom color scheme
(carbon: yellow; oxygen: red).

I73

 

helix bundles of subunit III in all three structures as shown in Figure 4.3. At a nearby
site at the interface of subunits I and III, a conserved lipid was also found in the
structures of both Rch0 and bovine heart mitochondria CcO (Figure 4.3). Similarly,
in the crystal structures of cytochrome be; complex from yeast (PDB entry 1KB9),
chicken (PDB entry lBCC) and bovine (PDB entry 1PP9), a conserved phosphatidyl
ethanolamine was found to reside at the same position on the protein surfaces in all
three structures as shown in Figure 4.4 (Zhang et al., 1998; Lange et al., 2001).

In our structure where detergents were resolved, tyrosine and tryptophan
residues were found nearby forming a stacking interaction between the sugar ring and
the aromatic rings as shown in Figure 3.28. Such stacking interactions are often found
involved in sugar binding in carbohydrate metabolizing enzymes (Roujeinikova er al.,
2001; Roujeinikova et al., 2002), and the same intimate interactions between the
headgroup of sugar-based detergent and aromatic residue rings have also been
observed in a number of other membrane protein structures as well (Snijder et al.,
2001; Sui et al., 2001; Okada et al., 2002; Tajkhorshid et al., 2002). Tryptophan and
tyrosine, in addition to other amino acid residues such as arginine, lysine, histidine,
and threonine, have been found to reside at the membrane interface and associate
themselves with the phosphatidyl head groups of lipids (Yau et al., 1998). Therefore,
detergent headgroups that can mimic phosphatidyl headgroups, both by polar
interactions with surrounding residues and rigid stacking interactions with aromatic
rings, could occupy the same sites. The success of alkyl glycoside detergents such as

maltosides and glucosides in both stabilizing and crystallizing

174

 

Figure 4.3: Structure overlay of RchO (PDB entry 1M56), PdCcO (PDB entry
lQLE) and bovine heart mitochondrial CcO (PDB entry 1V54) showing the
conserved phospholipid molecules. The transmembrane helices of subunits I and III
from the three crystal structures are very similar and they are shown in wheat and
gray, respectively. There are two conserved lipid sites shown in the picture. The one
on the right is found in the CcO crystals structures from R. s., P. d., and bovine. The
one on the left is found in the CcO crystal structures of R. s. and bovine. The resolved
phosphatidyl ethanolamine molecule in the RchO structure is shown by atom type
(C: green; 0: red; N: blue; P: purple), resolved phosphatidyl choline molecule in the
PdCcO structure is shown by a different atom type scheme (C: cyan; 0: red; N: blue;
P: purple) and resolved phosphatidyl glycerol molecule in the bovine CcO structure is
shown by another different atom type scheme (C: yellow; 0: red; N: blue; P: purple).

175

 

Figure 4.4: Structure overlay of cytochrome be, complexes from yeast (PDB
entry 1KB9), chicken (PDB entry lBCC), and bovine (PDB entry 1PP9) showing
the conserved phosphatidyl ethanolamine molecule. The transmembrane helices
from the three crystal structures shown are very similar and they are shown in gray.
Resolved phosphatidyl ethanolamine molecule in the crystal structure from yeast is
shown by atom type (C: cyan; 0: red; N: blue; P: purple). Resolved phosphatidyl
ethanolamine molecule in chicken is shown by a different atom type scheme (C:
yellow; 0: red; N: blue; P: purple). Resolved phosphatidyl ethanolamine molecule in
bovine is shown by another different atom type scheme (C: green; 0: red; N: blue; P:
plume)-

I76

membrane proteins may be due to their ability to interact in a manner similar to
phospholipids, and therefore partially replace those specifically bound lipids that are
lost during solublization and column purification. Consequently, when detergent
molecules are resolved in a membrane protein structure, they may be expected to
reveal the position of a specifically bound phospholipid. Careful analysis of
superimposed structures of the current I-II subunit RchO and the bovine heart
mitochondrial CcO (PDB entry 1V54) indicated that in the sites where the two decyl
maltosides are found in the RchO structure, a phosphatidyl glycerol (PG) molecule
was resolved in the bovine heart mitochondrial CcO as shown in Figure 4.5 with each
decyl maltoside occupying roughly the site of each of the two acyl tails of the
phospholipid molecule. (As shown in Figure 4.1, there is a resolved LDAO detergent
molecule in the structure of MI subunit PdCcO (PDB entry 1AR1) at this same site).
In the bovine mitochondrial CcO structure, the PG molecule is surrounded
additionally by three small subunits that are not found in bacterial enzymes as shown
in wheat color in the figure. These extra subunits probably further stabilize the
binding of the lipid through additional interactions. It is plausible that initially a
phospholipid molecule was occupying the same site in Rch0, but was lost during
detergent solublization and protein purification partially due to the absence of the
extra subunits that are found in the bovine mitochondrial enzyme. Two decyl
maltoside molecules bound to this conserved phospholipid site because of their ability
to mimic the lost lipid. Similarly, in the structure of cytochrome bcl complex from

chicken and bovine, a phosphatidyl ethanolamine (PE) from the bovine structure and

177

 

Figure 4.5: Structure overlay of [-1] subunit Rch0, [-11 sub it PdCcO (PDB
entry 1AR1), and bovine heart mitochondrial CcO (PDB entry 1V54). The
transmembrane helices of subunits I and II these CcOs are very similar and they are
shown in gray. Subunits IV, VIIb, and VIII from bovine CcO are shown in wheat.
Resolved decyl maltosides from the I-11 subunit RchO is shown by atom color
(carbon: green; oxygen: red), the detergent LDAO in PaCcO is shown by a different
atorn color scheme (carbon: cyan; oxygen: red; nitrogen: blue), and the resolved
phosphatidyl glycerol molecule in the bovine CcO structure is shown by a third atom
color scheme (carbon: yellow; oxygen: red; phosphorous: purple).

178

a detergent octyl glucoside molecule from the chicken structure were found at the
same positions with the glucoside head group superimposed onto the phosphate group
and the alkyl tail of the octyl glucoside coinciding with one of the two acyl tails of the
lipid as shown in Figure 4.6. Although lipid molecules could be assigned as detergent
molecules until the advent of structures with higher resolution (McAuley et al., 1999),
in the HI subunit RchO structure, the clear detergent head group electron density is
unequivocal and the structure overlay strongly suggest such mimicking effects of
detergents on lipid molecules.

The above findings suggest that specific interactions between membrane lipids
and membrane proteins likely do exist and in order to obtain membrane protein
crystals that diffract to high resolution, special care should be given during detergent
solublization, column purification and crystallization procedures to retain important
lipid species for better overall structural integrity and stability (Garavito and

Ferguson-Miller, 2001).

179

 

Figure 4.6: Structure overlay of cytochrome be, complexes from chicken (PDB
entry lBCC), and bovine (PDB entry 1PP9). The transmembrane helices from the
two crystal structures shown are very similar and they are shown in gray. Resolved
phosphatidyl ethanolamine molecule in the crystal structure from bovine cytochrome
bc. complex is shown by atom type (C: green; 0: red; N: blue; P: purple). At the same
site, an octyl glycoside shown by a different atom color scheme (C: yellow; 0: red; P:
purple) was resolved in the chicken cytochrome bcl complex structure. Note that the
headgroup of the detergent correspond to the headgroup of the phosphatidyl
ethanolamine molecule and the alkyl tail of the detergent corresponded to one of the
two acyl tails of the lipid.

180

Chapter 5. FUTURE PLANS
5.1 Improvement of the X-ray Diffraction of Crystals of RchO

Although current I-II RchO crystals diffract X-rays isotropically up to 2.35 A
resolution, and the structure has been refined with satisfactory R and Rm factors,
more experiments need to be performed in order to obtain crystals of RchO that
diffract beyond the current resolution to better resolve changes that occur in mutant
forms, particularly in water organization.

5.1.1 New Strains of Rhodobacter sphaeroides with Different Subunit Contents

It has been found that removing flexible loops on the surface of a protein can
help crystallize proteins that are otherwise hard to crystallize and/or improve crystal
diffraction quality (Derewenda, 2004). This can be done by limited proteolysis to
identify the flexible surface loop regions, followed by removing the DNA sequences
encoding the flexible loops. In the case of the crystallization of the ammonia channel
from E. coli, crystals of artificially truncated protein diffract X-rays to 1.35 A
resolution (Khademi er al., 2004).

In the current RchO structure, there are terminal amino acid sequences that
are not resolved, especially in subunit 1, most likely due to their being flexible.
Therefore, these sequences are good candidates for being excised for optimized
crystal packing. Efforts to make protein complexes with excised portions of subunits
are currently underway and the effects will be tested should the new protein complex
successfully crystallize.

New strains of R. s. can also be made which coexpress the membrane

18l

associated cytochrome cy with CcO. The N-terminal membrane anchor domain of
cytochrome cy (Myllykallio et al., 1997; Myllykallio er al., 1999) could be linked to
the C-terminus of subunit I of RchO, allowing stoichiometric coexpression and
copurification of the RchO-cytochrome cy complex. Crystallization experiments can
then be performed with the complex, with the aim of testing the location of proposed
binding sites of cytochrome c to CcO based on molecular modeling studies (Roberts
and Pique, 1999). It is also conceivable that the larger, more hydrophilic cytochrome
c-CcO complex will also produce higher resolution crystals.

5.1.2 Screening New Conditions for Detergents, Protein Production,
Crystallization and Crystal Handling / Flashcooling

Previous results (see Chapters 3 and 4) have indicated that many factors,
including detergent solublization, protein purification and crystallization, and crystal
handling/flashcooling, affect the crystal diffraction quality greatly. Further
modification of all these steps can be tried and their effects can be studied in order to
find the optimal conditions for maximum crystal diffraction quality.

For example, new detergents, such as 3-oxatridecyl-0t-D-mannoside, as well
as the glucoside and maltoside derivatives, which were successfully used for
crystallization of CcO from bovine heart mitochondria at 1.6 A resolution (Yoshikawa,
personal communication), can be tested for membrane solublization, protein
purification and crystallization.

A novel technique for cryocooling protein crystals has recently been

developed (Kim et al., 2005). Flashcooling protein crystals under pressure up to 200

l82

 

MPa in He gas without the use of cryoprotectant has been shown to improve the
diffraction quality for a number of protein crystals (Kim et al., 2005). This novel
technique can be tested on crystals of RchO with the hope of improving their
diffraction resolution and decreasing the crystal mosaicity.
5.1.3 Purification of LI] Subunit RchO as a Crystallization Candidate

Currently, the HI subunit crystals of Rch0 are obtained from a
crystallization drop containing purified three subunit (from R. 3. strain 37A4) or four
subunit (from R. s. strain 169A4) enzyme after several weeks, conditions that are hard
to control. Moreover, the crystals of the HI subunit RchO obtained from R. s. strain
169A4 emerged from showers of the four subunit RchO crystals. Coexistence of
these two crystal forms in the drop can lead to difficulties in crystal handling. It is
also conceivable that during crystal growth, the incorporation of three or four subunit
RchO might negatively affect the quality of growing I-II subunit crystals. Therefore,
experiments can be performed to try to purify the MI subunit Rch0 for
crystallization trials. The I-II subunit RchO, if successfully obtained and purified,
maybe a better candidate for crystallization trials due to its homogeneity.

Previously, I—II subunit PdCcO was obtained by the use of the harsh detergent
LDAO (Ostenneier et al., 1997). Similar experiments can also be performed on
Rch0 in order to obtain pure I-II subunit enzyme by using LDAO and other
detergents. The gene encoding subunit 111 can also be removed from the R. s. genome
with no detrimental effects on bacterial growth and CcO expression as previously

reported (Haltia er al., 1989; Bratton et al., 2000); the resulting subunit III-less protein

183

complex might be a good crystallization candidate, although subunit IV is still present
(Hosler et al., personal communication). Therefore, crystallization trials can be
performed on protein complexes obtained from subunit III-deleted and subunits
III/IV—deleted R. 3. strains if the HI subunit enzyme can be successfully expressed
and purified.

5.2 Crystal Structure of RchO Complexed with Arachidonic Acid

Arachidonic acid is found to increase the activity of D-pathway mutants, such
as D132A (Fetter er al., 1996), and to prevent suicide inactivation of subunit III-less
RchO (Fetter er al., 1996; Mills and Hosler, 2005). The proposed mechanism is that
binding of arachidonic acid in the vicinity of the D pathway enhances the proton
uptake through the D pathway and therefore offsets the effects of mutation and loss of
subunit III on the D pathway (Fetter er al., 1996; Mills and Hosler, 2005). However,
the exact binding site(s) for arachidonic acid remains unclear. A crystal structure of
RchO complexed with arachidonic acid could help locate the binding site(s) for
arachidonic acid and help our understanding of the D proton pathway and proton
transport along this pathway.

In order to obtain a crystal structure of RchO complexed with arachidonic
acid, two approaches will be used. First, arachidonic acid will be added to the
crystallization solution and hopefully will cocrystallize with RchO from the solution.
Alternatively, crystals of MI subunit RchO can be soaked in a solution containing

arachidonic acid for a period of time for arachidonic acid to bind before cryocooling.

184

5.3 Crystal Structures of Key Mutants of RchO

Great progress has been made in our understanding of the structure/function of
CcO during the past decade, especially after the crystal structures of CcOs from a
variety of different sources were solved (Iwata et al., 1995; Tsukihara et al., 1995;
Tsukihara er al., 1996; Ostenneier et al., 1997; Yoshikawa et al., 1998; Svensson-Ek
et al., 2002). However, these high resolution crystal structures of wild type CcOs have
not been able to help elucidate the exact proton pumping mechanism and how
electron transfer and proton translocation are coordinated in time and space. In order
to understand this central issue in bioenergetics, more high resolution crystal
structures of various forms of 0:0 are needed, with their resolved water molecules.
The latter can provide us with important information on amino acid side chain and
proton movements during the process. Thus, obtaining high resolution crystal
structures of various key mutants of RchO is a high priority.

Table 5.1 shows some of the mutant RchOs whose structure determination
will be attempted using the established method. These mutants all exhibit, or are
predicted to exhibit, pronounced functional changes as measured by kinetic and
spectroscopic methods, and it is proposed that different water arrangements may be
involved in the observed changes. It is also important to choose mutants that are
structurally stable and likely to withstand purification and crystallization.

For example, one of the most interesting mutants involves the Cup, ligand,
M263nL. The missing CuA ligand Met is envisioned to be replaced with a water

molecule. As a result, the two Cu A atoms are still in place but exhibit perturbed

185

down-Eu 3:33.5— vfiuinauotomae .52: His .552: Hafiz—£358
2.. use... usages 2. a: 83.22538 232:: Bebe ea...e ode: .e 3.5:: as 2.3.

 

.833 323:3
QBoHeoEV humane—Ho 3on H0 329 05 new See Q
263on 05 mam—Bu @353 Q sweet: .333 05 52.33 Ego

 

 

32353 2: 3 magnum 883 E moweeao 833 e .Ho 32%: Qweomm>
cargo—ace: Hmmmm 43%:
885 can cocoa»: ensues con—Ston— mso e 8 Becca—AH mammmH.
assess asses see as Sea e383 5% Ease E «680

 

@353 Bowxoeézxo cocoa co x83 38 FREE ~on +3
83% cocoa @353 Q woe—003 can LEE—Head MHHwewHFLNmHQ

 

 

 

 

H2953 Q swag—u 83% e305 cos—83 $358.9 «Lam HQ

H2958 vH awn—oh: 0x89: Exam @03on HeBfia¢H (Lama.

FBfia M 5:85 83% e893 vogue—n FBEQH¢H EemcmvH
33:83 x88 3 080: 33808

33.. Hobeoo boa—Emma c3385 92%.: «Se 080: M: mam

 

e 080: 8 82 comment 55020 330.830
«:0 Ho H3589 N88 @3385 team: 8280 <30 4:992

 

 

mow—:20 3.85955 uSfleEAHBoZoBO eeEmenH 3:332

 

 

 

186

 

spectral properties and an approximately IOOmV increase in their redox potential.
Electron transfer from Cu. to heme a is severely hindered without much effect on
proton transfer (Zhen et al., 2002). A crystal structure of this mutant will lead to
elucidation of the localized effect on the Cu center within this region and better
understanding of the coupling between proton pumping and electron transfer.

Other interesting mutant forms of Rch0 are mutants of the arginine pair
(R4811, R4821) which are associated with the propionate groups of the heme groups
either directly or via hydrogen-bonded networks of water molecules. These arginines
are along the proposed proton exit channel, as well as involved in electron transport
from CuA center to heme a (Mills et al., 2000; Qian et al., 2004). Mutants of these
two arginines lead to severely perturbed electron transfer from Cup, to heme a, as in
the case of R4821P, or very subtle changes in activity and proton pumping, as found in
R4811K (Qian et al., 2004). Based on MD calculations (Seibold et al., 2005), the
R4811K mutant was proposed to lead to conformational changes in residues in the
loop between helix III and helix IV and altered water arrangements. Because of the
central position of the arginine pairs, 3 high resolution crystal structure of the arginine
mutants with resolved water molecules can help us understand the structural basis of
the observed functional changes and understand the entire processes of electron
transfer and proton translocation.

Mutants that impair proton uptake pathways would also be desirable for
structural determination. These include mutants that impaired the proton uptake

through the K pathway, such as K3621M and T3591A (Fetter et al., 1995), as well as

187

T3591D/N (Sharpe er al., personal communication), and mutants that impaired the D
pathway, D132|A (Fetter et al., 1995) and D1321A/R4811K double mutant (Mills et al.,
2005). Besides the mutants that affect the proton uptake pathway, several other
mutants that are predicted to be involved in proton translocation have been designed
and the structure determination of them will be attempted. These include: Q2761R,
which is proposed to block the proton backflow pathway (Sharpe et al., personal
communication); T352IS, which is proposed to affect one of the Cur; ligands, H3331,
central in both the oxygen reduction reaction and proton translocation (Sharpe er al.,
2005); and V33OIS/D, which is proposed to lead to changes in the proton transfer to
the active site from the D pathway (Sharpe et al., personal communication).
5.4 Kinetic and Crystallographic Studies of K-pathway Mutants Which
Abolish Cd“ Binding

As shown in the Results section (Chapter 3), the proposed binding site for
Cd2+ on the inside of the enzyme is at the entrance of the K proton pathway. The
proposed Cd2+ binding ligands are H96“ and E101" (Figure 3.11). Mutants that are
designed to abolish the potential binding sites for Cd2+ at the entrance of the K proton
pathway are currently being made, including H96nA, EIOIHA, and H96nA/E101uA
double mutant. Kinetic studies of enzymatic activity in the presence of Zn2+lCd2+ will
be performed to test if this site is indeed the inhibitory site of Zn2+/Cd2+ and crystal
structure of these mutants will be solved to confirm that no Cd2+ can bind to this site

due to such mutations.

188

5.5 Summary
In summary, high resolution crystal structures with the resolved water
molecules will help determine the structural basis of the mechanism of energy
conservation. The currently established protein purification, crystallization and crystal
handling protocols make this goal achievable. Knowledge and experiences gained
from these studies can also help us understand how to obtain high resolution
membrane protein crystals in general, and to elucidate the biology of membrane

protein - lipid interactions.

189

BIBLIOGRAPHY

Aagaard,A., Namslauer, A., and Brzezinski, P. (2002). Inhibition of proton transfer in
cytochrome c oxidase by zinc ions: delayed proton uptake during oxygen reduction.

Biochimica et Biophysica Acta 1555, 133-139.

Antunes, F., Boveris, A., and Cadenas, E. (2004). On the mechanism and biology of
cytochrome oxidase inhibition by nitric oxide. Proceedings of the National Academy

of Sciences of the United States of America 101, 16774-16779.

Axelrod, H. L., Abresch, E. C., Paddock, M. L., Okamura, M. Y., and Feher, G. (2000).
Determination of the binding sites of the proton transfer inhibitors Cd2+ and Zn2+ in
bacterial reaction centers. Proceedings of the National Academy of Sciences of the

United States of America 97, 1542-1547.
Bailey, S. (1994). The CCP4 suite: programs for protein crystallography. Acta
Crystallographica, Section D: Biological Crystallography D50, 760-763.

Bartlett, G R. (1959). Phosphorus assay in column chromatography. Journal of

Biological Chemistry 234, 466-468.

Beauvoit, B., and Rigoulet, M. (2001). Regulation of cytochrome c oxidase by
adenylic nucleotides. Is oxidative phosphorylation feedback regulated by its

end-products? IUBMB Life 52, 143-152.

Benning, C., Huang, Z.-H., and Gage, D. A. (1995). Accumulation of a novel
glycolipid and a betaine lipid in cells of Rhodobacter sphaeroides grown under

phosphate limitation. Archives of Biochemistry and Biophysics 317, 103-111.

Berry, E. A., Zhang, Z., Bellamy, H. D., and Huang, L. (2000). Crystallographic
location of two Zn2+-binding sites in the avian cytochrome bcl complex. Biochimica

et Biophysica Acta 1459, 440-448.
Borisov, V. B. (2002). Defects in mitochondrial respiratory complexes HI and IV, and
human pathologies. Molecular Aspects of Medicine 23, 385-412.

Bottcher, B., Scheide, D., Hesterberg, M., Nagel-Steger, L., and Friedrich, T. (2002).

A novel, enzymatically active conformation of the Escherichia coli
NADH:ubiquinone oxidoreductase (complex I). Journal of Biological Chemistry 2 77,

l 7970- 1 7977.
Branden, M., Tomson, F., Gennis, R. B., and Brzezinski, P. (2002). The Entry Point of

the K-Proton-Transfer Pathway in Cytochrome c Oxidase. Biochemistry 41,
1 0794- 1 0798.

190

Bratton, M. R., Hiser, L., Antholine, W. E., Hoganson, C., and Hosler, l. P. (2000).
Identification of the Structural Subunits Required for Formation of the Metal Centers
in Subunit I of Cytochrome c Oxidase of Rhodobacter sphaeroides. Biochemistry 39,
12989-12995.

Bratton, M. R., Pressler, M. A., and Hosler, J. P. (1999). Suicide Inactivation of
Cytochrome c Oxidase: Catalytic Turnover in the Absence of Subunit III Alters the
Active Site. Biochemistry 38, 16236-16245.

Brookes, P. S. (2005). Mitochondrial H-t- leak and ROS generation: An odd couple.
Free Radical Biology & Medicine 38, 12-23.

Brunger, A. T., Adams, P. D., Clore, G M., DeLano, W. L., Gros, P., Grosse-Kunstleve,
R. W., Jiang, J.-S., Kuszewski, J., Nilges, M., Pannu, N. S., et al. (1998).
Crystallography & NMR System: a new software suite for macromolecular structure
determination. Acta Crystallographica, Section D: Biological Crystallography 054,
905-921.

Brych, S. R., Blaber, S. 1., Logan, T. M., and Blaber, M. (2001). Structure and stability
effects of mutations designed to increase the primary sequence symmetry within the
core region of a b-trefoil. Protein Science 10, 2587-2599.

Buse, G, Soulimane, T., Dewor, M., Meyer, H. B., and Bluggel, M. (1999). Evidence
for a copper-coordinated histidine-tyrosine crosslink in the active site of cytochrome
oxidase. Protein Science 8, 985-990.

Bushnell, G W., Louie, G V., and Brayer, G D. (1990). High-resolution
three-dimensional structure of horse heart cytochrome c. Journal of Molecular
Biology 214, 585-595.

Byme, B., Abramson, J., Jansson, M., Holmgren, E., and Iwata, S. (2000). Fusion
protein approach to improve the crystal quality of cytochrome bo3 ubiquinol oxidase
from Escherichia coli. Biochimica et Biophysica Acta 1459, 449-455.

Capaldi, R. A. (1990). Structure and function of cytochrome c oxidase. Annual
Review of Biochemistry 59, 569-596.

Chang, G, and Roth, C. B. (2001). Structure of MsbA from E. coli: A homolog of the
multidrug resistance ATP binding cassette (ABC) transporters. Science (Washington,
DC, United States) 293, 1793-1800.

Cukier, R. I. (2004). Quantum molecular dynamics simulation of proton transfer in
cytochrome c oxidase. Biochimica et Biophysica Acta 1656, 189-202.

Cukier, R. I. (2005). A molecular dynamics study of water chain formation in the
proton-conducting K channel of cytochrome c oxidase. Biochimica et Biophysica
Acta 1706, 134-146.

191

Daldal, F., Deshmukh, M., and Prince, R. C. (2003). Membrane-anchored cytochrome
c as an electron carrier in photosynthesis and respiration: past, present and future of
an unexpected discovery. Photosynthesis Research 76, 127-134.

Daldal, F., Mandaci, S., Vtherstein, C., Myllykallio, H., Duyck, K., and Zannoni, D.
(2001). Mobile cytochrome c2 and membrane-anchored cytochrome cy are both
efficient electron donors to the cbb3- and aa3-type cytochrome c oxidases during

respiratory growth of Rhodobacter sphaeroides. Journal of Bacteriology 183,
2013-2024.

Derewenda, Z. S. (2004). The use of recombinant methods and molecular engineering
in protein crystallization. Methods (San Diego, CA, United States) 34, 354-363.

Distler, A. M., Allison, J ., Hiser, C., Qin, L., Hilmi, Y., and Ferguson-Miller, S. (2004).
Mass spectrometric detection of protein, lipid and heme components of cytochrome c
oxidase from R. sphaeroides and the stabilization of non-covalent complexes from the
enzyme. European Journal of Mass Spectrometry 10, 295-308.

Dowhan, W., Mileykovskaya, B., and Bogdanov, M. (2004). Diversity and versatility
of lipid-protein interactions revealed by molecular genetic approaches. Biochimica et
Biophysica Acta 1666, 19-39.

Ferguson-Miller, S., and Babcock, G T. (1996). Heme/Copper Terminal Oxidases.
Chemical Reviews (Washington, D C) 96, 2889-2907.

Ferguson-Miller, S., Brautigan, I). L., and Margoliash, E. (1978). Definition of
cytochrome c binding domains by chemical modification. III. Kinetics of reaction of

carboxydinitrophenyl cytochromes c with cytochrome c oxidase. Journal of Biological
Chemistry 253, 149-159.

Ferguson, A. D., Breed, J ., Diederichs, K., Welte, W., and Coulton, J. W. (1998). An
internal affinity-tag for purification and crystallization of the siderophore receptor

FhuA, integral outer membrane protein from Escherichia coli K-12. Protein Science 7,
1636-1638.

Ferguson, A. D., Hofmann, B., Coulton, J. W., Diederichs, K., and Welte, W. (1998).
Siderophore-mediated iron transport: crystal structure of FhuA with bound
lipopolysaccharide. Science (Washington, D C) 282, 2215-2220.

Fetter, J ., Sharpe, M., Qian, J., Mills, D., Ferguson-Miller, S., and Nicholls, P. (1996).
Fatty acids stimulate activity and restore respiratory control in a proton channel
mutant of cytochrome c oxidase. FEBS Letters 393, 155-160.

Fetter, J. R., Qian, J ., Shapleigh, J., Thomas, J. W., Garcia-Horsman, A., Schmidt, B.,
Hosler, J., Babcock, G T., Gennis, R. B., and Ferguson-Miller, S. (1995). Possible
proton relay pathways in cytochrome c oxidase. Proceedings of the National Academy
of Sciences of the United States of America 92, 1604-1608.

192

Flock, D., and Helms, V. (2002). Protein-protein docking of electron transfer
complexes: cytochrome c oxidase and cytochrome c. Proteins: Structure, Function,
and Genetics 47, 75-85.

Florens, L., Schmidt, B., McCracken, J ., and Ferguson-Miller, S. (2001). Fast
Deuterium Access to the Buried Magnesium/Manganese Site in Cytochrome c
Oxidase. Biochemistry 40, 7491-7497.

Fromme, P., and Witt, H. T. (1998). Improved isolation and crystallization of
Photosystem I for structural analysis. Biochimica et Biophysica Acta 1365, 175-184.

Garavito, R. M., and Ferguson-Miller, S. (2001). Detergents as tools in membrane
biochemistry. Journal of Biological Chemistry 276, 32403-32406.

Garavito, R. M., and Picot, D. (1990). The art of crystallizing membrane proteins.
Methods (San Diego, CA, United States) 1, 57-69.

Gast, P., Hemelrijk, P., and Hoff, A. J. (1994). Determination of the number of
detergent molecules associated with the reaction center protein isolated from the
photosynthetic bacterium Rhodopseudomonas viridis. Effects of the amphiphilic
molecule 1,2,3-heptanetriol. FEBS Letters 337, 39-42.

Gilderson, G, Salomonsson, L., Aagaard, A., Gray, J ., Brzezinski, P., and Hosler, J.
(2003). Subunit III of cytochrome c oxidase of Rhodobacter sphaeroides is required to
maintain rapid proton uptake through the D pathway at physiologic pH. Biochemistry
42, 7400-7409.

Gohon, Y., and Popot, J .-L. (2003). Membrane protein-surfactant complexes. Current
Opinion in Colloid & Interface Science 8, 15-22.

Gordeliy, V. 1., Labahn, J., Moukhametzianov, R., Efremov, R., Granzin, J .,
Schlesinger, R., Bueldt, G, Savopol, T., Scheidig, A. J., Klare, J. P., and Engelhard, M.
(2002). Molecular basis of transmembrane signalling by sensory rhodopsin
II-transducer complex. Nature (London, United Kingdom) 419, 484-487.

Green, K., and Brand, M. D. (2004). The significance of uncoupling proteins in
diabetes. Canadian Journal of Diabetes 28, 50-58.

Haltia, T., Finel, M., Harms, N ., Nakari, T., Raitio, M., \Vlkstrom, M., and Saraste, M.
(1989). Deletion of the gene for subunit III leads to defective assembly of bacterial
cytochrome oxidase. EMBO Journal 8, 3571-3579.

Harper, M. E., Bevilacqua, L., Hagopian, K., Weindruch, R., and Ramsey, J. J. (2004).
Ageing, oxidative stress, and mitochondrial uncoupling. Acta Physiologica
Scandinavica 182, 321-331.

Harrenga, A., and Michel, H. (1999). The cytochrome c oxidase from Paracoccus

193

 

denitrificans does not change the metal center ligation upon reduction. Journal of
Biological Chemistry 274, 33296-33299.

Haselkom, R., Lapidus, A., Kogan, Y., Vlcek, C., Paces, J., Paces, V., Ulbrich, P.,
Pecenkova, T., Rebrekov, D., Milgram, A., et al. (2001). The Rhodobacter capsulatus
genome. Photosynthesis Research 70, 43-52.

Hilmi, Y. (2002) Use of mutagenesis, novel detergents, and lipid analysis to optimize
conditions for achieving high resolution 3-dimensional crystal structure of
Rhodobacter sphaeroides cytochrome c oxidase.

Hiser, C., Mills, D. A., Schall, M., and Ferguson-Miller, S. (2001). C-Terminal
Truncation and Histidine-Tagging of Cytochrome c Oxidase Subunit II Reveals the
Native Processing Site, Shows Involvement of the C-Terminus in Cytochrome c
Binding, and Improves the Assay for Proton Pumping. Biochemistry 40, 1606-1615.

Hofacker, 1., and Schulten, K. (1998). Oxygen and proton pathways in cytochrome c
oxidase. Proteins: Structure, Function, and Genetics 30, 100-107.

Holm, L., and Sander, C. (1993). Protein structure comparison by alignment of
distance matrixes. Journal of Molecular Biology 233, 123-138.

Hosler, J. P. (2004). The influence of subunit HI of cytochrome c oxidase on the D
pathway, the proton exit pathway and mechanism-based inactivation in subunit 1.
Biochimica et Biophysica Acta 1655, 332-339.

Hosler, J. P., Fetter, J ., Tecklenburg, M. M. J., Espe, M., Lerma, C., and
Ferguson-Miller, S. (1992). Cytochrome aa3 of Rhodobacter sphaeroides as a model
for mitochondrial cytochrome c-oxidase. Purification, kinetics, proton pumping, and
spectral analysis. Journal of Biological Chemistry 267, 24264-24272.

Hunte, C., Koepke, J., Lange, C., Rossmanith, T., and Michel, H. (2000). Structure at
2.3.ANG resolution of the cytochrome bcl complex from the yeast Saccharomyces

cerevisiae co-crystallized with an antibody Fv fragment. Structure (London) 8,
669-684.

Imhoff, J. F. (1991). Polar lipids and fatty acids in the genus Rhodobacter. Systematic
and Applied Microbiology 14, 228-234.

Iverson, T. M., Luna-Chavez, C., Cecchini, G, and Rees, D. C. (1999). Structure of
the Escherichia coli fumarate reductase respiratory complex. Science (Washington, D
C) 284, 1961-1966.

Iwata, S., Lee, J. W., Okada, K., Lee, J. K., Iwata, M., Rasmussen, B., Link, T. A.,
Ramaswamy, S., and Jap, B. K. (1998). Complete structure of the ll-subunit bovine
mitochondrial cytochrome bcl complex. Science (Washington, D C) 281, 64-71.

194

Iwata, S., Ostenneier, C., Ludwig, B., and Michel, H. (1995). Structure at 2.8.ANG
resolution of cytochrome c oxidase from Paracoccus denitrificans. Nature (London)
376, 660-669.

Kadenbach, B. (1986). Regulation of respiration and ATP synthesis in higher
organisms: hypothesis. Journal of Bioenergetics and Biomembranes 18, 39-54.

Kadenbach, B., and Arnold, S. (2000). 3,5-Diiodothyronine binds to subunit Va of
cytochrome c oxidase: possible mechanism of short-term effects of thyroid hormones.
Membrane Structure in Disease and Drug Therapy, 271-281.

Kadenbach, B., J arausch, J ., Hartmann, R., and Merle, P. (1983). Separation of
mammalian cytochrome c oxidase into 13 polypeptides by a sodium dodecyl
sulfate-gel electrophoretic procedure. Analytical Biochemistry 129, 517-521.

F

Khademi, S., O'Connell, J ., HI, Remis, J ., Robles-Colmenares, Y., Miercke, L. J. W.,
and Stroud, R. M. (2004). Mechanism of Ammonia Transport by Amt/MEP/Rh:
Structure of AmtB at 1.35.ANG Science (Washington, DC, United States) 305,
1587-1594.

Kim, C. U., Kapfer, R., and Gruner, S. M. (2005). High-pressure cooling of protein
crystals without cryoprotectants. Acta Crystallographica, Section D: Biological
Crystallography D61, 881-890.

Kim, B., Chufan, E. B., Karnaraj, K., and Karlin, K. D. (2004). Synthetic Models for
Heme-Copper Oxidases. Chemical Reviews (Washington, DC, United States) 104,
1077-1133.

Kolbe, M., Besir, H., Essen, L.-O., and Oesterhelt, D. (2000). Structure of the
light-driven chloride pump halorhodopsin at 1.8.ANG resolution. Science
(Washington, D C) 288, 1390-1396.

Konstantinov, A. A., Siletsky, S., Mitchell, D., Kaulen, A., and Gennis, R. B. (1997).
The roles of the two proton input channels in cytochrome c oxidase from Rhodobacter
sphaeroides probed by the effects of site-directed mutations on time-resolved
electrogenic intraprotein proton transfer. Proceedings of the National Academy of

Sciences of the United States of America 94, 9085-9090.

Krauss, S., Zhang, C.-Y., and Lowell, B. B. (2005). The mitochondrial
uncoupling-protein homologues. Nature Reviews Molecular Cell Biology 6, 248-261.

Kuo, A., Bowler, M. W., Zimmer, J ., Antcliff, J. F., and Doyle, D. A. (2003).
Increasing the diffraction limit and internal order of a membrane protein crystal by
dehydration. Journal of Structural Biology 141, 97-102.

Kwong, P. D., and Liu, Y. (1999). Use of cryoprotectants in combination with
immiscible oils for flash cooling macromolecular crystals. Journal of Applied

195

Crystallography 32, 102-105.

Landau, E. M., and Rosenbusch, J. P. (1996). Lipidic cubic phases: a novel concept
for the crystallization of membrane proteins. Proceedings of the National Academy of
Sciences of the United States of America 93, 14532-14535.

Lange, C., Nett, J. H., Trumpower, B. L., and Hunte, C. (2001). Specific roles of
protein-phospholipid interactions in the yeast cytochrome bcl complex structure.
EMBO Journal 20, 6591-6600.

Larsen, R. W., Pan, L. P., Musser, S. M., Li, Z., and Chan, S. I. (1992). Could CuB be
the site of redox linkage in cytochrome c oxidase? Proceedings of the National
Academy of Sciences of the United States of America 89, 723-727.

le Maire, M., Champeil, P., and Moller, J. V. (2000). Interaction of membrane proteins
and lipids with solubilizing detergents. Biochimica et Biophysica Acta 1508, 86—111.

Lee, H.-m., Das, T. K., Rousseau, D. L., Mills, D., Ferguson-Miller, S., and Gennis, R.
B. (2000). Mutations in the Putative H-Channel in the Cytochrome c Oxidase from
Rhodobacter sphaeroides Show That This Channel Is Not Important for Proton
Conduction but Reveal Modulation of the Properties of Heme a. Biochemistry 39,
2989-2996.

Link, T. A., and von Jagow, G (1995). Zinc ions inhibit the QP center of bovine heart
mitochondrial bcl complex by blocking a protonatable group. Journal of Biological
Chemistry 270, 25001-25006.

Loll, P. J. (2003). Membrane protein structural biology: the high throughput challenge.
Journal of Structural Biology 142, 144-153.

Loll, P. J ., Allaman, M., and \Vlencek, J. (2001). Assessing the role of
detergent-detergent interactions in membrane protein crystallization. Journal of
Crystal Growth 232, 432-438.

Ludwig, B., Bender, E., Arnold, S., Huttemann, M., Lee, I., and Kadenbach, B. (2001).
Cytochrome C oxidase and the regulation of oxidative phosphorylation.
Chembiochem: a European journal of chemical biology 2, 392-403.

Luecke, H., Schobert, B., Richter, H.-T., Cartailler, J.-P., and Lanyi, J. K. (1999).
Structure of bacteriorhodopsin at 1.55.AN G resolution. Journal of Molecular Biology
291, 899-911.

Ma, Q, and Chang, G (2004). Crystallography of the integral membrane protein
EmrE from Escherichia coli. Acta Crystallographica, Section D: Biological
Crystallography D60, 2399-2402.

MacMillan, F., Kannt, A., Behr, J ., Prisner, T., and Michel, H. (1999). Direct Evidence

196

for a Tyrosine Radical in the Reaction of Cytochrome c Oxidase with Hydrogen
Peroxide. Biochemistry 38, 9179-9184.

Malatesta, F., Nicoletti, F., Zickerrnann, V., Ludwig, B., and Brunori, M. (1998).
Electron entry in a CuA mutant of cytochrome c oxidase from Paracoccus

denitrificans. Conclusive evidence on the initial electron entry metal center. FEBS
Letters 434, 322-324.

Maneg, O., Malatesta, F., Ludwig, B., and Drosou, V. (2004). Interaction of
cytochrome c with cytochrome oxidase: two different docking scenarios. Biochimica
et Biophysica Acta 1655, 274-281.

Marone, P. A., Thiyagarajan, P., Wagner, A. M., and Tiede, D. M. (1999). Effect of
detergent alkyl chain length on crystallization of a detergent-solubilized membrane
protein: correlation of protein-detergent particle size and particle-particle interaction
with crystallization of the photosynthetic reaction center from Rhodobacter
sphaeroides. Journal of Crystal Growth 207, 214-225.

McAuley, K. E., Fyfe, P. K., Ridge, J. P., Isaacs, N. W., Cogdell, R. J., and Jones, M.
R. (1999). Structural details of an interaction between cardiolipin and an integral

membrane protein. Proceedings of the National Academy of Sciences of the United
States of America 96, 14706-14711.

Michel, H. (1982). Three-dimensional crystals of a membrane protein complex. The
photosynthetic reaction center from Rhodopseudomonas viridis. Journal of Molecular
Biology 158, 567-572.

Michel, H. (1999). Cytochrome c Oxidase: Catalytic Cycle and Mechanisms of Proton
Pumping-A Discussion. Biochemistry 38, 15129-15140.

Michel, H., Behr, J ., Harrenga, A., and Kannt, A. (1998). Cytochrome c oxidase:
structure and spectroscopy. Annual Review of Biophysics and Biomolecular Structure
27, 329-356.

Mills, D. A., and Ferguson-Miller, S. (2003). Understanding the mechanism of proton
movement linked to oxygen reduction in cytochrome c oxidase: lessons from other
proteins. FEBS Letters 545, 47-51.

Mills, D. A., Florens, L., Hiser, C., Qian, J ., and Ferguson-Miller, S. (2000). Where is
'outside' in cytochrome c oxidase and how and when do protons get there? Biochimica
et Biophysica Acta 1458, 180-187.

Mills, D. A., Geren, L., Hiser, C., Schmidt, B., Durham, B., Millett, F., and
Ferguson-Miller, S. (2005). An arginine to lysine mutation in the vicinity of the heme
propionates affects the redox potentials of the hemes and associated electron and
proton transfer in cytochrome c oxidase. Biochemistry 44, 10457-10465.

197

Mills, D. A., and Hosler, J. P. (2005). Slow Proton Transfer through the Pathways for
Pumped Protons in Cytochrome c Oxidase Induces Suicide Inactivation of the
Enzyme. Biochemistry 44, 4656-4666.

Mills, D. A., Schmidt, B., Hiser, C., Westley, E., and Ferguson-Miller, S. (2002).
Membrane potential-controlled inhibition of cytochrome c oxidase by zinc. Journal of
Biological Chemistry 277, 14894-14901.

Mills, D. A., Tan, 2., Ferguson-Miller, S., and Hosler, J. (2003). A role for subunit III

in proton uptake into the D pathway and a possible proton exit pathway in
Rhodobacter sphaeroides cytochrome c oxidase. Biochemistry 42, 7410-7417.

Murshudov, G N ., Vagin, A. A., and Dodson, E. J. (1997). Refinement of

macromolecular structures by the maximum-likelihood method. Acta
Crystallographica, Section D: Biological Crystallography D53, 240-255.

Myllykallio, H., Jenney, F. E., Jr., Moomaw, C. R., Slaughter, C. A., and Daldal, F.
(1997). Cytochrome cy of Rhodobacter capsulatus is attached to the cytoplasmic

membrane by an uncleaved signal sequence-like anchor. Journal of Bacteriology 179,
2623-2631.

Myllykallio, H., Zannoni, D., and Daldal, F. (1999). The membrane—attached electron
carrier cytochrome cy from Rhodobacter sphaeroides is functional in respiratory but
not in photosynthetic electron transfer. Proceedings of the National Academy of
Sciences of the United States of America 96, 4348-4353.

Napiwotzki, J ., Shinzawa-Itoh, K., Yoshikawa, S., and Kadenbach, B. (1997). ATP
and ADP bind to cytochrome c oxidase and regulate its activity. Biological Chemistry
378, 1013-1021.

O'Neill, J. W., Kim, D. E., Baker, D., and Zhang, K. Y. J. (2001). Structures of the B1
domain of protein L from Peptostreptococcus magnus with a tyrosine to tryptophan
substitution. Acta Crystallographica, Section D: Biological Crystallography 057,
480-487.

Okada, T., Fujiyoshi, Y., Silow, M., Navarro, J., Landau, E. M., and Shichida, Y.
(2002). Functional role of internal water molecules in rhodopsin revealed by x-ray

crystallography. Proceedings of the National Academy of Sciences of the United
States of America 99, 5982-5987.

Ostermeier, C., Harrenga, A., Errnler, U., and Michel, H. (1997). Structure at 2.7.ANG
resolution of the Paracoccus denitrificans two-subunit cytochrome c oxidase
complexed with an antibody Fv fragment. Proceedings of the National Academy of
Sciences of the United States of America 94, 10547-10553.

Ostermeier, C., Iwata, S., Ludwig, B., and Michel, H. (1995). Fv fragment-mediated
crystallization of the membrane protein bacterial cytochrome c oxidase. Nature

198

Structural Biology 2, 842-846.

Otwinowski, Z., and Minor, W. (1997). Processing of x-ray diffraction data collected
in oscillation mode. Methods in Enzymology 276, 307-326.

Paddock, M. L., Graige, M. S., Feher, G, and Okamura, M. Y. (1999). Identification
of the proton pathway in bacterial reaction centers: inhibition of proton transfer by
binding of Zn2+ or Cd2+. Proceedings of the National Academy of Sciences of the
United States of America 96, 6183-6188.

Pan, L. P., Hibdon, S., Liu, R. Q., Durham, B., and Millett, F. (1993). Intracomplex
elecn'on transfer between ruthenium-cytochrome c derivatives and cytochrome c
oxidase. Biochemistry 32, 8492-8498.

Pautsch, A., Vogt, J ., Model, K., Siebold, C., and Schulz, G E. (1999). Strategy for
membrane protein crystallization exemplified with OmpA and Ome. Proteins:
Structure, Function, and Genetics 34, 167-172.

Peiffer, W. B., Ingle, R. T., and Ferguson-Miller, S. (1990). Structurally unique plant
cytochrome c oxidase isolated from wheat germ, a rich source of plant mitochondrial
enzymes. Biochemistry 29, 8696-8701.

Proshlyakov, D. A., Pressler, M. A., and Babcock, G T. (1998). Dioxygen activation
and bond cleavage by mixed-valence cytochrome c oxidase. Proceedings of the
National Academy of Sciences of the United States of America 95, 8020-8025.

Qian, J ., Mills, D. A., Geren, L., Wang, K., Hoganson, C. W., Schmidt, B., Hiser, C.,
Babcock, G T., Durham, B., Millett, F., and Ferguson-Miller, S. (2004). Role of the
Conserved Arginine Pair in Proton and Electron Transfer in Cytochrome c Oxidase.
Biochemistry 43, 5748-5756.

Ramirez, B. E., Malmstrom, B. G, Winkler, J. R., and Gray, H. B. (1995). The
currents of life: the terminal electron-transfer complex of respiration. Proceedings of
the National Academy of Sciences of the United States of America 92, 11949-11951.

Regan, J. J., Ramirez, B. E., Winkler, J. R., Gray, H. B., and Malmstrom, B. G (1998).
Pathways for electron tunneling in cytochrome c oxidase. Journal of Bioenergetics
and Biomembranes 30, 35-39.

Rich, P. R., Meunier, B., Mitchell, R., and Moody, A. J. (1996). Coupling of charge
and proton movement in cytochrome c oxidase. Biochimica et Biophysica Acta 12 75,
91-95.

Roberts, V. A., and Pique, M. E. (1999). Definition of the interaction domain for
cytochrome c on cytochrome c oxidase: III. Prediction of the docked complex by a
complete, systematic search. Journal of Biological Chemistry 274, 38051-38060.

199

 

Robinson, B. H. (2000). Human cytochrome oxidase deficiency. Pediatric Research
48, 5 8 1 -5 85.

Robinson, N. C. (1982). Specificity and binding affinity of phospholipids to the
high-affinity cardiolipin sites of beef heart cytochrome c oxidase. Biochemistry 21,
184-188.

Rosenow, M. A., Brune, D., and Allen, J. P. (2003). The influence of detergents and
amphiphiles on the solubility of the light-harvesting I complex. Acta
Crystallographica, Section D: Biological Crystallography D59, 1422-1428.

Roujeinikova, A., Raasch, C., Burke, J ., Baker, P. J ., Liebl, W., and Rice, D. W. (2001).
The Crystal Structure of Therrnotoga maritima Maltosyltransferase and its
Implications for the Molecular Basis of the Novel Transfer Specificity. Journal of
Molecular Biology 312, 119-131.

Roujeinikova, A., Raasch, C., Sedelnikova, S., Liebl, W., and Rice, D. W. (2002).
Crystal structure of Therrnotoga maritima 4-a-glucanotransferase and its acarbose

complex: implications for substrate specificity and catalysis. Journal of Molecular
Biology 321, 149-162.

Sack, J. S., and Quiocho, F. A. (1997). CHAIN: A crystallographic modeling program.
Methods in Enzymology 2 77, 158-173.

Schagger, H., Hagen, T., Roth, B., Brandt, U., Link, T. A., and von Jagow, G (1990).
Phospholipid specificity of bovine heart be] complex. European journal of
biochemistry IFEBS 190, 123-130.

Schmidt, B., McCracken, J ., and Ferguson-Miller, S. (2003). A discrete water exit
pathway in the membrane protein cytochrome c oxidase. Proceedings of the National
Academy of Sciences of the United States of America 100, 15539-15542.

Seddon, A. M., Curnow, P., and Booth, P. J. (2004). Membrane proteins, lipids and
detergents: not just a soap opera. Biochimica et Biophysica Acta 1666, 105-117.

Sedlak, B., and Robinson, N. C. (1999). Phospholipase A2 Digestion of Cardiolipin
Bound to Bovine Cytochrome c Oxidase Alters Both Activity and Quaternary
Structure. Biochemistry 38, 14966-14972.

Seibold, S. A., Mills, D. A., Ferguson-Miller, S., and Cukier, R. I. (2005). Water Chain
Formation and Possible Proton Pumping Routes in Rhodobacter sphaeroides
Cytochrome c Oxidase: A Molecular Dynamics Comparison of the Wild Type and
R481K Mutant Biochemistry 44, 10475-10485.

Shapleigh, J. P., and Gennis, R. B. (1992). Cloning, sequencing and deletion from the
chromosome of the gene encoding subunit I of the aa3-type cytochrome c oxidase of
Rhodobacter sphaeroides. Molecular Microbiology 6, 635-642.

200

Sharpe, M., Perin, I., Wrigglesworth, J ., and Nicholls, P. (1996). Fatty acids as
modulators of cytochrome c oxidase in proteoliposomes. Biochemical Journal 320,
557-561.

Sharpe, M., Qin, L., and Ferguson-Miller, S. (2005). Proton entry, exit and pathways
in cytochrome oxidase: Insight from "Conserved" Water, In Biophysical and
Structural Aspects of Bioenergetics, M. Wikstroem, ed. (Cambridge, accepted: Royal
Society of Chemistry).

Shinzawa-Itoh, K., Ueda, H., Yoshikawa, S., Aoyama, H., Yamashita, E., and
Tsukihara, T. (1995). Effects of ethyleneglycol chain length of dodecyl
polyethyleneglycol monoether on the crystallization of bovine heart cytochrome c
oxidase. Journal of Molecular Biology 246, 572-575.

Smith, P. K., Krohn, R. 1., Hermanson, G T., Mallia, A. K., Gartner, F. H., Provenzano,
M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J., and Klenk, D. C. (1985).
Measurement of protein using bicinchoninic acid. Analytical Biochemistry 150,

76-85.

Snijder, H. J., Van Eerde, J. H., Kingrna, R. L., Kalk, K. H., Dekker, N., Egmond, M.
R., and Dijkstra, B. W. (2001). Structural investigations of the active-site mutant
Asn156Ala of outer membrane phospholipase A: Function of the Asn-His interaction
in the catalytic triad. Protein Science 10, 1962-1969.

Steffens, G C. M., Soulimane, T., Wolff, G, and Buse, G (1993). Stoichiometry and
redox behavior of metals in cytochrome-c oxidase. European Jounlal of Biochemistry
213, 1149-1157.

Steinrucke, P., Steffens, G C., Panskus, G, Buse, G, and Ludwig, B. (1987). Subunit
II of cytochrome c oxidase from Paracoccus denitrificans. DNA sequence, gene
expression and the protein. European journal of biochemistry / FEBS 167, 431-439.

Stock, D., Leslie, A. G W., and Walker, J. E. (1999). Molecular architecture of the
rotary motor in ATP synthase. Science (Washington, D C) 286, 1700-1705.

Storoni, L. C., McCoy, A. J., and Read, R. J. (2004). Likelihood-enhanced fast
rotation functions. Acta Crystallographica, Section D: Biological Crystallography D60,
432-438.

Stroebel, D., Choquet, Y., Popot, J .-L., and Picot, D. (2003). An atypical haem in the
cytochrome b6f complex. Nature (London, United Kingdom) 426, 413-418.

Sui, H., Han, B.-G, Lee, J. K., Walian, P., and Jap, B. K. (2001). Structural basis of
water-specific transport through the aqpl water channel. Nature (London, United
Kingdom) 414, 872-878.

Svensson-Ek, M., Abramson, J ., Larsson, G, Tomroth, S., Brzezinski, P., and Iwata, S.

201

(2002). The X-ray Crystal Structures of Wild-type and EQ(I-286) Mutant Cytochrome
c Oxidases from Rhodobacter sphaeroides. Journal of Molecular Biology 321,
329-339.

Tajkhorshid, E., Nollert, P., Jensen, M. O., Miercke, L. J. W., O'Connell, J ., Stroud, R.
M., and Schulten, K. (2002). Control of the selectivity of the aquaporin water channel
family by global orientational tuning. Science (Washington, DC, United States) 296,
525-530.

Tan, M.-L., Balabin, I., and Onuchic, J. N. (2004). Dynamics of electron transfer
pathways in cytochrome c oxidase. Biophysical Journal 86, 1813-1819.

Timmins, P. A., Hauk, J., Wacker, T., and Welte, W. (1991). The influence of
heptane-1,2,3-triol on the size and shape of LDAO micelles. Implications for the
crystallization of membrane proteins. FEBS Letters 280, 115-120.

Tomson, E, Bailey, J. A., Gennis, R. B., Unkefer, C. J., Li, Z., Silks, L. A., Martinez,
R. A., Donohoe, R. J ., Dyer, R. B., and Woodruff, W. H. (2002). Direct Infrared
Detection of the Covalently Ring Linked His-Tyr Structure in the Active Site of the
Heme-Copper Oxidases. Biochemistry 41, 14383-14390.

Tomson, F. L., Morgan, J ., Gu, G, Barquera, B., Vygodina, T. V., and Gennis, R. B.
(2003). Substitutions for Glutamate 101 in Subunit II of Cytochrome c Oxidase from
Rhodobacter sphaeroides Result in Blocking the Proton-Conducting K—Channel.
Biochemistry 42, 1711-1717.

Torres, J ., Stevens, T. J ., and Samso, M. (2003). Membrane proteins: the \"Wild
West\" of structural biology. Trends in Biochemical Sciences 28, 137-144.

Toyoshima, C., Nakasako, M., Nomura, H., and Ogawa, H. (2000). Crystal structure
of the calcium pump of sarcoplasmic reticulum at 2.6.ANG resolution. Nature
(London) 405, 647-655.

Trakhanov, S., Kreimer, D. 1., Parkin, S., Ames, G F.-L., and Rupp, B. (1998).
Cadmium-induced crystallization of proteins: 11. Crystallization of the Salmonella
typhimurium histidine-binding protein in complex with L—histidine, L-arginine, or
L-lysine. Protein Science 7, 600-604.

Trakhanov, S., and Quiocho, F. A. (1995). Influence of divalent cations in protein
crystallization. Protein Science 4, 1914-1919.

Tsukihara, T., Aoyama, H., Yamashita, E., Tomizaki, T., Yamaguchi, H.,
Shinzawa-Itoh, K., Nakashima, R., Yaono, R., and Yoshikawa, S. (1995). Structures of
metal sites of oxidized bovine heart cytochrome c oxidase at 2.8.ANG Science
(Washington, D C) 269, 1069-1074.

Tsukihara, T., Aoyama, H., Yamashita, B., Tomizaki, T., Yamaguchi, H.,

202

Shinzawa-Itoh, K., Nakashima, R., Yaono, R., and Yoshikawa, S. (1996). The whole
structure of the 13-subunit oxidized cytochrome c oxidase at 2.8.ANG Science
(Washington, D C) 272, 1136-1144.

Tsukihara, T., Shimokata, K., Katayama, Y., Shimada, H., Muramoto, K., Aoyama, H.,
Mochizuki, M., Shinzawa-itoh, K., Yamashita, E., Yao, M., et al. (2003). The low-spin
heme of cytochrome c oxidase as the driving element of the proton-pumping process.
Proceedings of the National Academy of Sciences of the United States of America
100, 15304-15309.

Vygodina, T. V., Capitanio, N ., Papa, S., and Konstantinov, A. A. (1997). Proton
pumping by cytochrome c oxidase is coupled to peroxidase half of its catalytic cycle.
FEBS Letters 412, 405-409.

Wang, K., Geren, L., Zhen, Y., Ma, L., Ferguson-Miller, 8., Durham, B., and Millett, F.
(2002). Mutants of the CuA site in cytochrome c oxidase of Rhodobacter sphaeroides:
II. Rapid kinetic analysis of electron transfer. Biochemistry 41, 2298-2304.

Wang, K., Zhen, Y., Sadoski, R., Grinnell, 8., Geren, L., Ferguson-Miller, S., Durham,
B., and Millett, F. (1999). Definition of the interaction domain for cytochrome c on
cytochrome c oxidase: II. Rapid kinetic analysis of electron transfer from cytochrome

c to Rhodobacter sphaeroides cytochrome oxidase surface mutants. Journal of
Biological Chemistry 274, 38042-38050.

White, S. H. (2004). The progress of membrane protein structure determination.
Protein Science 13, 1948-1949.

Wiener, M. C., and Snook, C. F. (2001). The development of membrane protein
crystallization screens based upon detergent solution properties. Journal of Crystal
Growth 232, 426-431.

Wikstrom, M., Bogachev, A., Fine], M., Morgan, J. E., Puustinen, A., Raitio, M.,
Verkhovskaya, M., and Verkhovsky, M. I. (1994). Mechanism of proton translocation

by the respiratory oxidases. The histidine cycle. Biochimica et Biophysica Acta 1187,
106-111.

Wikstrom, M., and Verkhovsky, M. I. (2002). Proton translocation by cytochrome c
oxidase in different phases of the catalytic cycle. Biochimica et Biophysica Acta 1555,
1 28-1 32.

Wikstrom, M. K. (1977). Proton pump coupled to cytochrome c oxidase in
mitochondria. Nature 266, 271-273.

Winkler, J. R., Malmstroem, B. G, and Gray, H. B. (1995). Rapid electron injection
into multisite metalloproteins: intramolecular electron transfer in cytochrome oxidase.
Biophysical Chemistry 54, 199-209.

203

Winn, M. D., Isupov, M. N., and Murshudov, G N. (2001). Use of TLS parameters to
model anisotropic displacements in macromolecular refinement. Acta
Crystallographica, Section D: Biological Crystallography D57, 122-133.

Witt, H., and Ludwig, B. (1997). Isolation, analysis, and deletion of the gene coding
for subunit IV of cytochrome c oxidase in Paracoccus denitrificans. Journal of
Biological Chemistry 272, 5514-5517.

Witt, H., Malatesta, F., Nicoletti, F., Brunori, M., and Ludwig, B. (1998). Tryptophan
121 of subunit II is the electron entry site to cytochrome-c oxidase in Paracoccus
denitrificans. Involvement of a hydrophobic patch in the docking reaction. Journal of

Biological Chemistry 273, 5132-5136.

Yau, W.-M., Wimley, W. C., Gawrisch, K., and White, S. H. (1998). The Preference of
Tryptophan for Membrane Interfaces. Biochemistry 37, 14713-14718.

Yoshikawa, S., Shinzawa-itoh, K., Nakashima, R., Yaono, R., Yamashita, B., Inoue, N.,
Yao, M., Fei, M. J ., Libeu, C. P., Mizushima, T., et al. (1998). Redox-coupled crystal
structural changes in bovine heart cytochrome c oxidase. Science (Washington, D C)
280, 1723-1729.

Yu, C.-A., and Yu, L. (1980). Structural role of phospholipids in
ubiquinol-cytochrome c reductase. Biochemistry 19, 5715-5720.

Zaslavsky, D., and Gennis, R. B. (2000). Proton pumping by cytochrome oxidase:
progress, problems and postulates. Biochimica et Biophysica Acta 1458, 164-179.

Zhang, H., Kurisu, G, Smith, J. L., and Cramer, W. A. (2003). A defined
protein-detergent-lipid complex for crystallization of integral membrane proteins: The
cytochrome b6f complex of oxygenic photosynthesis. Proceedings of the National
Academy of Sciences of the United States of America 100, 5160-5163.

Zhang, Z., Huang, L., Shulmeister, V. M., Chi, Y.-I., Kim, K. K., Hung, L.-W., Crofts,
A. R., Berry, E. A., and Kim, S.-H. (1998). Electron transfer by domain movement in
cytochrome bcl. Nature (London) 392, 677-684.

Zhen, Y., Hoganson, C. W., Babcock, G T., and Ferguson-Miller, S. (1999). Definition
of the interaction domain for cytochrome c on cytochrome c oxidase: 1. Biochemical,
spectral, and kinetic characterization of surface mutants in subunit H of Rhodobacter
sphaeroides cytochrome aa3. Journal of Biological Chemistry 274, 38032-38041.

Zhen, Y., Qian, J ., Follmann, K., Hayward, T., Nilsson, T., Dahn, M., Hilmi, Y.,
Hamer, A. G, Hosler, J. P., and Ferguson-Miller, S. (1998). Overexpression and
purification of cytochrome c oxidase from Rhodobacter sphaeroides. Protein
Expression and Purification 13, 326-336.

Zhen, Y., Schmidt, B., Kang, U. G, Antholine, W., and Ferguson-Miller, S. (2002).

204

Mutants of the CuA site in cytochrome c oxidase of Rhodobacter sphaeroides: 1.
Spectral and functional properties. Biochemistry 41 , 2288-2297.

Zhou, T., Radaev, S., Rosen, B. P., and Gatti, D. L. (2000). Structure of the ArsA
ATPase: the catalytic subunit of a heavy metal resistance pump. EMBO Journal 19,
4838-4845.

Zhou, Y., Morals-Cabral, J. H., Kaufman, A., and MacKinnon, R. (2001). Chemistry
of ion coordination and hydration revealed by a K+ channel-Fab complex at 2.0.ANG
resolution. Nature (London, United Kingdom) 414, 43-48.

205

                  
    
       

A

(11111

BRS

1111'

4

11111111

0 736

111311

29