CHARACTERIZATION AND STRUCTURAL DETERMINATION OF METALLOENZYMES: DNA POLYMERASE BETA, , AND ACETYL COENZYME-A DECARBONYLASE/SYNTHASE

DISSERTATION

Presented in Partial Fulfillment of the Requirement for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

By

Joseph W. Arndt, B.S.

*****

The Ohio State University 2003

Dissertation Committee: Approved by Professor Michael K. Chan, Advisor

Professor Joseph Krzycki ______Professor Ming-Daw Tsai Advisor Department of Chemistry

ABSTRACT

My research focused on the structure determination of proteins from three metalloenzyme systems by X-ray crystallography. The first target was rat DNA polymerase β, which catalyzes the template-directed nucleotidyl transfer reaction required for DNA replication. We have determined the crystal structures of two intermediate complexes in the reaction pathway of this , (i) a pre-chemistry ternary complex containing protein, DNA, and a chromium dNTP analog and (ii) a post- chemistry complex after nucleotide incorporation. These intermediate structures have allowed us to dissect the role of the two essential magnesium ions in initiating the enzyme’s conformational change. Based on these structures, a revised mechanism for replication and fidelity is proposed.

The second part of this research involved structural studies on a carboxypeptidase

(PfuCP) from the hyperthermophilic archaeon, Pyrococcus furiosus. Like other , it catalyzes the removal of amino acids from the C-terminus of protein and peptide chains. In this project we have solved three different structures of this enzyme, an apo form and two metal-bound forms. The overall fold of this enzyme is distinct from all other known structures of carboxypeptidase. It differs significantly in sequence, however, with one important feature being a consensus HEXXH metal-binding

ii motif at its . While HEXXH motifs are common in and

, this is the first observation of this motif in a carboxypeptidase. These

structures represent the first prototype available for this growing family of

carboxypeptidase. Comparison with other metallopeptidases allowed us to propose a

catalytic mechanism of its C-terminal peptide hydrolysis reaction.

The last target of this research was the Methanosarcina barkeri acetyl coenzyme-

A decarbonylase/synthase (ACDS) complex and its five individual components. The

ACDS complex is involved in the metabolism acetyl coenzyme-A by methanogens. In

this project we have isolated ACDS and its subunits, α2ε2, β, γ, and δ, for crystallization

and subsequent structure elucidation. Several crystal forms have been identified for the

α2ε2 and β components. In addition, we have performed metal chelation and reconstitution experiments on the β component that indicate the enzyme’s acetyl

activity is dependent on nickel, and not copper.

iii

Dedicated to my wife and parents

iv

ACKNOWLEDGMENTS

I would like to express my utmost gratitude to my advisor, Dr. Michael Chan, for his generosity, encouragement, and support. The freedom for which he permitted me to perform my research allowed me to become a better scientist – thank you. I would also like to acknowledge my committee members, Drs. Tsai and Krzycki, for their wisdom, precious time, and assistance. Their passion for science is contagious. It has been an honor and my sincere privilege to work with them.

I am indebted to past post docs, Drs. Weimin Gong and Xuejun Zhong, which I have had the pleasure of working with on the polymerase β project. I would like to extend my appreciation to my lab mates, Dr. Bing Hao, Clara Isaza, Rinku Jain, Patrick

Kang, and Michelle Nauerth for their continual willingness to lend a hand, their technical insight, and numerous conversations that we shared together over the years. A special thanks goes out to Bing, Clara, and Rinku, not only for the scientific camaraderie that we shared, but also for their friendship and good times we spent together – my years at Ohio

State were made all the more enjoyable by their company.

v Particular recognition goes to Dr. Tsai and my colleagues of the Chemistry and

Biology Interface Training Program funded by the NIH from which I was honored with a fellowship. And to the Graduate Studies Committee and Miles Foundation for granting me a fellowship for the last six months of my studies, which made the writing of this dissertation possible.

I would like to acknowledge my cats, Pele-boy and Cashew, for kindly donating their whiskers for the cause of science - my crystal seeding experiments would have been unfruitful them.

Finally, I am most thankful to my wife, parents, and mother-in-law for their unwavering love and support. This degree would not have been possible without their prayers and encouragement.

vi

VITA

October 24, 1968……………………… Born, Chicago, IL

1987 - 1991…...………………………. B.S., Chemistry, Illinois State University,

Normal, IL

1991 – 1996...………………………… Research Technician,

Abbott Laboratories, North Chicago, IL

1997 - 2000…………………………… Graduate Teaching and Research Associate,

The Ohio State University

2000 – 2002…………………………… NIH Predoctoral Fellowship, Chemistry and

Biology Interface Program

2003 – present ………………………… Miles Fellowship, The Ohio State University

vii RELATED PUBLICATIONS

1. Bing Hao, Clara Isaza, Joseph W. Arndt, Michael Soltis and Michael K. Chan (2002).

Structure-based mechanism of O2 sensing and ligand discrimination by the FixL

heme domain of Bradyrhizobium japonicum, Biochemistry, 41, 12952-8.

2. Joseph W. Arndt, Bing Hao, Vijay Ramakrishnan, Timothy Cheng, Sunney I, Chan

and Michael K. Chan (2002). Crystal structure of a novel carboxypeptidase from the

hyperthermophilic Archaeon Pyrococcus furiosus, Structure, 10, 215-24.

3. Joseph W. Arndt, Weimin Gong, Xuejun Zhong, Alexander K. Showalter, Jia Liu,

Christopher Dunlap, Ming-Daw Tsai, and Michael K. Chan (2001). Insight into the

catalytic mechanism of DNA polymerase β: structure of intermediate complexes,

Biochemistry, 40, 5368-75.

FIELDS OF STUDY

Major Field: Chemistry

Studies in X-ray Crystallography of Biological Macromolecules

viii

TABLE OF CONTENTS

ABSTRACT...... ii

DEDICATION...... iv

ACKNOWLEDGMENTS ...... v

VITA...... vii

LIST OF FIGURES ...... xv

LIST OF TABLES...... xix

LIST OF PDB ID...... xx

LIST OF ABBREVIATIONS...... xxi

CHAPTER 1 ...... 1

1.1 Functional variability of metals ...... 1

1.2 Metal variability...... 2

ix 1.2.1 Magnesium-dependent ...... 3

1.2.2 Cobalt- dependent enzymes ...... 5

1.2.3 Nickel-dependent enzymes ...... 6

1.2.4 Metalloclusters in biology...... 6

1.3 Metalloproteins are just “MAD” about synchrotron radiation ...... 8

1.4 The highlight of this research ...... 8

1.4.1 Intermediate structures in catalytic pathway of DNA polymerase β...... 9

1.4.2 Pyrococcus furiosus carboxypeptidase: crystal structure of a novel class of

carboxypeptidase...... 10

1.4.3 Acetyl coenzyme-A decarbonylase synthase complex: rare organometallic

reaction intermediates require unusual metal clusters ...... 12

Chapter 1 References ...... 14

CHAPTER 2 ...... 16

2.1 Introduction...... 16

2.2 Materials and methods ...... 20

2.2.1 Purification of Pol β and preparation of the DNA and Cr(III)·dTMPPCP

substrates ...... 20

2.2.2 Preparation, crystallization, and data collection of the Pol β-DNA-

Cr(III)·dTMPPCP and Pol β-DNA-Cr(III)·PCP intermediate complexes...... 21

x 2.2.3 Structure determination and refinement...... 22

2.3 Results and discussions...... 23

2.3.1 The structure of Pol β-DNA-Cr(III)·dTMPPCP ...... 23

2.3.2 The structure of Pol β-DNA-Cr(III)·PCP ...... 24

2.3.3 Implications presented by the intermediate structures...... 26

2.4 Conclusions...... 30

Chapter 2 References ...... 40

CHAPTER 3 ...... 43

3.1 Introduction...... 43

3.2 Materials and methods ...... 45

3.2.1 Purification and Crystallization ...... 45

3.2.2 Data collection and cryoprotection ...... 46

3.2.3 Structure solution...... 46

3.2.4 Model building and refinement...... 47

3.2.5 Substrate modeling...... 48

3.3 Results and discussions...... 49

3.3.1 The overall protein fold ...... 49

3.3.2 The protein active site...... 51

3.3.3 Structural similarity with neurolysin ...... 53

xi 3.3.4 Sequence homologs of PfuCP...... 54

3.3.5 Proposed catalytic mechanism...... 55

3.4 Conclusion ...... 58

Chapter 3 References ...... 71

CHAPTER 4 ...... 74

4.1 Introduction...... 74

4.2 Materials and methods ...... 78

4.2.1 ACDS complex from M. barkeri ...... 78

4.2.1.1 Preparation of cells and cell extracts ...... 78

4.2.1.2 Purification of the M. barkeri ACDS complex...... 79

4.2.1.3 Crystallization of the ACDS complex ...... 81

4.2.2 The CODH α2ε2 component of the ACDS complex from M. barkeri...... 82

4.2.2.1 Purification of α2ε2 CODH component ...... 82

4.2.2.2 Activity assay of α2ε2 CODH component ...... 84

4.2.2.3 Crystallization and data collection of α2ε2 CODH component ...... 85

4.2.3 The acetyl-CoA synthase β component of the ACDS complex from M.

barkeri ...... 86

4.2.3.1 Purification acetyl-CoA synthase β component ...... 86

4.2.3.2 Activity assay of acetyl-CoA synthase β component ...... 88

xii 4.2.3.3 Metal dependence on acetyltransferase activity ...... 90

4.2.3.3.1 Copper reconstitution...... 90

4.2.3.3.2 Copper chelation ...... 91

4.2.3.4 Crystallization and data collection of acetyl-CoA synthase β component ... 92

4.2.4 The cobalamin methyltransferase δ component of the ACDS complex...... 92

4.2.4.1 Purification cobalamin methyltransferase δ component...... 92

4.2.4.2 Crystallization cobalamin methyltransferase δ component ...... 94

4.2.5 The iron-sulfur methyltransferase γ component of the ACDS complex from

M. barkeri ...... 94

4.2.5.1 Purification iron-sulfur methyltransferase γ component...... 94

4.2.5.2 Crystallization iron-sulfur methyltransferase γ component...... 95

4.2.6 Recombinant acetyl-CoA synthase β from M. thermoautotrophicum...... 96

4.2.6.1 Cell culturing and expression...... 96

4.2.6.2 Purification of recombinant M. thermoautotrophicum β...... 96

4.2.6.3 Characterization of recombinant M. thermoautotrophicum β ...... 98

4.2.6.4 Crystallization...... 100

4.2.7 Recombinant cobalamin methyltransferase δ from Archeoglobus fuligidus

100

4.2.7.1 Transformation and cell culturing...... 100

4.2.7.2 Isolation and protein refolding of recombinant A. fuligidus δ...... 101

4.3 Results and discussions...... 102

xiii 4.3.1 Isolation of the components of ACDS...... 102

4.3.1.1 M. barkeri α2ε2 CODH component ...... 102

4.3.1.2 The recombinant Archeoglobus fuligidus δ cobalamin-dependent

methyltransferase ...... 103

4.3.2 Oligomer structures of ACDS and its components...... 104

4.3.2.1 M. barkeri ACDS complex...... 104

4.3.2.2 Recombinant M. thermoautotrophicum β ...... 105

4.3.3 Progress of crystallization of the ACDS components...... 106

4.3.3.1 Crystallization of M. barkeri α2ε2...... 106

4.3.3.2 Crystallization of M. barkeri β ...... 110

4.3.4 Models of the subunits in lieu of the available structures...... 111

4.3.4.1 Modeling and comparison with other CODHs ...... 111

4.3.4.2 Modeling and comparison with Moorella thermoacetica ACSs ...... 112

4.3.5 Identity of metal in the A-cluster of the β component...... 113

4.4 Conclusions...... 116

Chapter 4 References ...... 162

Bibliography ...... 165

xiv

LIST OF FIGURES

Figure 2.1. Ribbons structure of Pol β-DNA-Cr(III)·dTMPPCP ternary complex ...... 32

Figure 2.2. Schematic diagram of active site of Pol β showing two metal ions...... 33

Figure 2.3. Stereoview of Fo-Fc electron density maps ...... 34

Figure 2.4. Structures of the active sites of stepwise Pol β intermediates...... 35

Figure 2.5. Stereoview of conformational change induced by MgA·dNTP binding...... 37

Figure 2.6. Proposed reaction scheme for Pol β...... 38

Figure 3.1. Ribbons representation of PfuCP homodimer...... 62

Figure 3.2. Sequence alignment of the M32 family of metallocarboxypeptidases...... 63

Figure 3.3. PfuCP monomer structrures ...... 66

Figure 3.4. Electron density of active site. A 2.3Å 2Fo-Fc electron density map

superimposed on the final active site model of PfuCP ...... 68

Figure 3.5. Structural comparison with neurolysin...... 69

Figure 3.6. Proposed catalytic mechanism...... 70

Figure 4.1. Available CODH/ACS structures...... 118

Figure 4.2. ACS/CODH active site metal clusters...... 120

Figure 4.3. Chromatographic resolution of the M. barkeri ACDS complex on Superose

6B size exclusion column ...... 121 xv Figure 4.4. Chromatographic resolution of the M. barkeri ACDS complex on Mono Q

column...... 122

Figure 4.5. The quality of the M. barkeri ACDS complex...... 123

Figure 4.6. Chromatographic resolution of the M. barkeri α2ε2 CODH and crude ACDS

complex on DEAE-FF column ...... 124

Figure 4.7. Chromatographic resolution of M. barkeri α2ε2 CODH on HAP column .. 125

Figure 4.8. Chromatographic resolution of M. barkeri α2ε2 CODH on HiLoad Q ...... 126

Figure 4.9. Chromatographic resolution of M. barkeri α2ε2 CODH on Mono Q ...... 127

Figure 4.10. The quality of the M. barkeri α2ε2 CODH component of the ACDS

complex...... 128

Figure 4.11. Chromatographic resolution of crude M. barkeri ACDS complex into its

comprising subunits on HiLoad Q column ...... 129

Figure 4.12. Chromatographic resolution of M. barkeri β on HAP column ...... 130

Figure 4.13. Chromatographic resolution of M. barkeri β on Mono Q column...... 131

Figure 4.14. The quality of M. barkeri β component ...... 132

Figure 4.15. Mass spectrum of purified M. barkeri β...... 133

Figure 4.16. Acetyl exchange activity of M. barkeri β...... 134

Figure 4.17. Quenching of BCS fluorescence of M. barkeri β...... 135

Figure 4.18. Compilation of fluorescence emission spectra of BCS titration...... 136

Figure 4.19. Chromatographic resolution of the M. barkeri δ on HAP column...... 137

Figure 4.20. Chromatographic resolution of the M. barkeri δ on Mono Q column ...... 138

Figure 4.21. The quality of the M. barkeri δ component of the ACDS complex...... 139 xvi Figure 4.22. Chromatographic resolution of the M. barkeri γ on HAP column ...... 140

Figure 4.23. Chromatographic resolution of the M. barkeri γ on Mono Q column...... 141

Figure 4.24. The quality of the M. barkeri γ component of the ACDS complex ...... 142

Figure 4.25. Chromatographic resolution of the recombinant M. thermoautotrophicum β

on Superose 6B size exclusion column...... 143

Figure 4.26. The quality of recombinant M. thermoautotrophicum β...... 144

Figure 4.27. M. barkeri α2ε2 CODH crystals form 1...... 145

Figure 4.28. M. barkeri α2ε2 CODH crystals form 2...... 146

Figure 4.29. M. barkeri α2ε2 CODH crystal form 4 ...... 147

Figure 4.30. Diffraction images of M. barkeri α2ε2 CODH crystal form 1...... 148

Figure 4.31. Diffraction images of plate-like M. barkeri α2ε2 CODH crystal form 2... 149

Figure 4.32. Diffraction images of hexagonal-like M. barkeri α2ε2 CODH crystal form 2

...... 150

Figure 4.33. Diffraction image of plate-like M. barkeri α2ε2 CODH crystal form 4. ... 151

Figure 4.34. Diffraction image of the M. barkeri β crystal...... 152

Figure 4.35. Structure-based sequence alignment of M. barkeri α CODH component and

M. thermoacetica β CODH...... 153

Figure 4.36. Sequence alignment of the C-terminus of M. barkeri α CODH component

and a portion of the N-terminus of M. thermoacetica α ACS component ...... 156

Figure 4.37. Sequence alignment of M. barkeri β vs. M. thermoacetica α...... 158

xvii Figure 4.38. Model of the M. barkeri β ACS component superimposed on M.

thermoacetica α ...... 159

Figure 4.39. ICP metal analysis of M. barkeri ACS β...... 160

Figure 4.40. Relative activities of the M. barkeri ACS β after chelation with BCS and

metal reconstitution...... 161

xviii

LIST OF TABLES

Table 1.1: Examples of the numerous metal functions observed in metalloproteins...... 2

Table 1.2: Magnesium-activated enzymes...... 4

Table 1.3: Known non-corrinoid cobalt-containing enzymes...... 5

Table 2.1: Summary of crystallographic results of the Pol β intermediate structures..... 39

Table 3.1. Crystallographic data for PfuCP...... 60

Table 4.1: CODH crystallization conditions...... 107

xix

LIST OF PDB ID

1HUO DNA Polymerase β Complexed with DNA and Cr-TMPPCP

1HUZ DNA Polymerase β Complexed with DNA and Cr-PCP

1K9X Pyrococcus furiosus Carboxypeptidase Apo-Yb

1KA2 Pyrococcus furiosus Carboxypeptidase Apo-Mg

1KA4 Pyrococcus furiosus Carboxypeptidase Nat-Pb not deposited E. coli peptide deformylase complexed with actinonin not deposited E. coli peptide deformylase complexed with inhibitor GS7496 not deposited E. coli peptide deformylase complexed with inhibitor GS7428 not deposited E. coli peptide deformylase complexed with inhibitor GS7433

xx

LIST OF ABBREVIATIONS

ACDS acetyl coenzyme-A decarbonylase/synthase

ACS acetyl coenzyme-A synthase

APS Advanced Photon Source

BCS bathocuproine disulfonate

Bis-Tris bis-2(hydroxyethyl)amino-tris(hydroxymethyl)methane

CODH carbon monoxide dehydrogenase

CP carboxypeptidase ddNTP dideoxynucleotide triphosphate

DEAE diethylaminoethyl

DLS dynamic light scattering dNTP deoxynucleotide triphosphate dTMPPCP 2'-deoxythymidine-5'-β,γ-methylene triphosphate

EDTA ethylenediaminetetraacetate

EXAFS extended X-ray absorption fine structure g gravity

GSH glutathione

HAP hydroxyapatite

HEPES 4-(2-hydroxyethyl)piperazine-1-ethanesulfonate

xxi ICP-MS inductively couple plasma mass spectrometry

IMCA Industrial Macromolecular Crystallography Association

IPTG isopropyl-β-D-thiogalactopyranoside kDa kilo dalton

LB Leuria-Bertani medium

MAD multiwavelength anomalous dispersion

MALDI-TOF MS matrix-assisted laser desorption ionization time-of-flight

mass spectrometry

MDa mega Dalton

MR molecular replacement

NCS non-crystallographic symmetry

PAGE polyacrylamide gel electrophoresis

PCP methylene diphosphonate

PfuCP Pyrococcus furiosus carboxypeptidase

Pol β Rattus norvegicus DNA polymerase β

PPi pyrophosphate

RMSD root mean square deviation rpm revolutions per minute

SDS sodium dodecylsulfate

SSRL Stanford Synchrotron Radiation Laboratory

Tris tris(hydroxymethyl)aminomethane

XANES X-ray absorption near edge structure

xxii

CHAPTER 1

INTRODUCTION

Metal ions are involved in many biological processes including metabolism, respiration, development, nerve signal transmission, muscle contraction, and signal transduction (1-3). Indeed it is estimated that half of all proteins require metal ions for their biological function. Understanding how metal ions contribute to the biological functions of different classes of metalloproteins is the focus of this dissertation.

1.1 Functional variability of metals

As shown in Table 1.1, metals perform a variety of roles in proteins. Typical roles of metals include structural stabilization, small-molecule binding and transport, signaling, and electron transfer. Metals are also commonly involved in catalysis.

Chemical functions that they are associated with include one-, two-, and multi- electron oxidation-reduction chemistry, functional group transfer, elimination reactions, bond formation, hydrolytic catalysis, and catalytic rearrangements (1-3). There are also proteins involved in the sequestration, transport, and insertion of metals into their proper protein sites in proteins.

1

Classification of Examples function Structural stabilization zinc finger proteins Dioxygen transport hemoglobin, myoglobin, hemocyanin, hemerythrin Communication and calmodulin, calcyclin signaling Electron transfer ferrodoxin, high-potential Fe-S protein, cytochrome Uptake, transport, and Na and K channels, metallothionein, transferrin, ferritin storage Two electron cytochrome P-450, CO dehydrogenase, acetyl coenzyme-A decarbonylase/synthase Multi electron nitrogenase, nitric oxide synthase oxidoreductases deaminases, decarboxylases synthetases, carboxylases racemases peptidases, lipases kinases, methyltransferases, DNA polymerases

Table 1.1: Examples of the numerous metal functions observed in metalloproteins.

1.2 Metal cofactor variability

The metal cofactors used by proteins are diverse and ubiquitous. To date, 14 different inorganic elements have been found in metalloproteins (4-6). Perhaps the most abundant metals in cells are the alkali (Na, K) and alkaline earth (Mg, Ca) elements. The properties of these metals are the straightforward to understand since they are redox inert.

They also exhibit a preference for oxygen ligands. Their primary role in the cell is for triggering cellular responses, such as observed in the firing of neurons caused by the influx of sodium ions across the cell membrane.

2 Transition metals ions are also well represented. These include Mo, Cd, and W from the 2nd and 3rd rows of the periodic table, as well as most of the first row transition elements, except Sc and Ti. Transition metals are mainly used because of their redox activities, their ability to bind and exchange ligands, and their high charge density. These properties allow them to bind and activate substrates and stabilize their transition-state intermediates. While these properties are essential to the function of the metalloprotein, when unregulated, these properties catalyze reactions that are harmful to the cell, such as redox reactions that damage proteins, lipids, and nucleic acids. Thus for some metals nature utilizes an elaborate set of metal chaperones that mediate the insertion of the metal into the proper metal sites (7).

The several subclasses of metalloenzymes that are relevant to this dissertation will be reviewed in greater detail in the following sections.

1.2.1 Magnesium-dependent enzymes

Magnesium is an important component of a broad variety of enzymes (Table 1.2), with only calcium having more functions (3, 4). A search of the approximately 21,000 structures in the Protein Data Bank (PDB) yields 1820 hits for “magnesium”. Even though it can be further reduced to those in which magnesium is known to play a functional role, one is still faced an overwhelming task in its review.

The divalent magnesium ion is a hard metal and thus prefers hard ligands of low polarizability, such as carboxylates, phosphates, and water. Magnesium prefers an octahedral coordination geometry. Unlike Ca- and Zn-binding sites, which can be

3 identified by characteristic binding motifs, few such motifs have been identified for magnesium sites. These include the NADFDGD motif found in RNA polymerases, DNA polymerase I, and HIV-reverse transcriptase, and the YXDD or LXDD motifs found in reverse transcriptases (8, 9).

The high concentration of magnesium found in most organisms has likely contributed to its evolution as a functional metal for many enzymes. It can assume a number of different roles, including the stabilization of high-energy substrate conformations, the polarization of substrates for nucleophilic attack, and for orienting substrates for proper reactivity (2, 4). Within the context of enzyme function, Mg2+ is often associated with nucleotides. The chemistry of these interactions has direct implications for the role of Mg in catalysis (8, 9). This subject is explored in detail in chapter two in regards to the magnesium-dependent DNA polymerase β.

Classification of enzyme Example Mutases Phosphoenolpyruvate mutase Transferases Xanthine phosphoribosyltransferase Kinases Adenylate kinase ATPases Adenylyl cyclase Isomerases D-Xylose DNA processing enzymes DNA Polymerase β GTPases Transducin-α Carboxylase/oxygenase Ribulose-1,5-bisphosphate carboxylase oxygenase Phosphatases Fructose-1,6-bisphosphatase Synthetases Adenylosuccinate synthetase Dehydrogensases Isocitrate dehydrogenase Ribozymes Hepatitis virus genomic ribozyme

Table 1.2: Magnesium-activated enzymes (4). 4 1.2.2 Cobalt- dependent enzymes

Cobalt is a relatively rare element in the environment, and therefore it is not

surprising that it is less frequently found in metalloenzymes than other first-row transition

metals (10). It is nevertheless an important cofactor in corrinoid-dependent enzymes,

such as vitamin B12-dependent mutases (ribonucleotide reductase and aminomutase) and methyltransferases (monomethylamine methyltransferase and methionine synthase) (11).

Another example studied in our lab is the methyltransferase δ subunit of the acetyl

coenzyme-A decarbonylase/synthase complex of Methanosarcina barkeri, discussed in

chapter four. Non-corrinoid cobalt-containing enzymes are much rarer with only eight

having been identified (Table 1.3) (10). One of these unusual cobalt enzymes is the

Pyrococcus furiosus carboxypeptidase, which will be discussed in chapter three, is one of

the few non-corrinoid cobalt-containing enzymes that have been structurally

characterized (12).

Enzyme Source Postulated role of cobalt Methionine Animals, yeast, bacteria hydrolysis Proline Archae hydrolysis Nitrile hydratase Actinomycetes and hydrolysis bacteria Glucose isomerase Actinomycetes isomerization Methylmalonyl-CoA Bacteria carboxytransfer carboxytransferase Lysine-2,3-aminomutase Bacteria isomerization Bromoperoxidase Pseudomonas putida bromination Carboxypeptidase Pyrococcus furiosus hydrolysis

Table 1.3: Known non-corrinoid cobalt-containing enzymes (10).

5 1.2.3 Nickel-dependent enzymes

Like most other transition metals, nickel can be toxic for organisms, presumably due to its binding to DNA. Nevertheless, several biological processes utilize nickel for catalysis. Since our lab studies nickel-dependent enzymes by crystallography, I find it befitting that the first protein to be crystallized was the nickel-dependent urease in 1926 by Sumner (14) – though it was another 50 years before Zerner and coworkers established the role of nickel in its enzymatic catalysis (15). This discovery prompted efforts to shed light on the chemistry of nickel in other biological systems. Since then its presence was subsequently detected in CO dehydrogenase (CODH) and [Ni-Fe]- hydrogenase from several anaerobic organisms, in methyl coenzyme-M reductase from methanogenic bacteria, in acetyl-CoA synthase (ACS) from methanogenic and acetogenic bacteria, in superoxide dismutase from actinomycetes, and putatively in glyoxalase I and cis-trans isomerase from E. coli (16-18). The progress toward solving X-ray structures of the nickel-dependent CODH and ACS enzymes from M. barkeri is discussed in the final chapter.

1.2.4 Metalloclusters in biology

Metallocluster-containing proteins represent a unique subclass of metalloproteins, in which multiple metal ions are associated with a single cofactor. Since Fe is the fourth most abundant element in the crust of the earth and the most abundant transition metal in biology, it is not surprising that the majority of all metalloclusters are of the iron-sulfur

(FeS) variety. However, one of the most significant problems organisms face is its

6 acquisition. This paradox is due to the insolubility of iron, which for uncomplexed ferric iron in water is on the order of 10-18 M (19). To over come this problem, organisms have developed elaborate systems for iron sequestration, storage, and insertion (7).

It has been speculated that the Fe-S clusters may have been some of the earliest known cofactors, reflecting the chemical environment in the prehistoric world (20-22).

Supporting this hypothesis, Cody and coworkers reported that several simple FeS clusters spontaneously form in reductive aqueous solution of ferrous iron and sulfide (22), both of which were likely prevalent in the earliest primordial environment.

Metalloclusters are diverse in architecture and function (19-21, 23). The simplest metallocenters are the Fe2S2 and Fe4S4 clusters that are typically involved in electron transfer and serve to shuttle electrons into and out of the active sites. They have reduction potentials ranging over 400 mV to below –400 mV, a scale larger than for any other redox cofactor (20). In addition to their electron transfer function, iron-sulfur clusters act as catalytic centers (e.g., aconitase and fumarase) and sensors of oxygen and superoxide (e.g., FNR and SoxR, respectively) (19). More complex multinuclear metallocenters are responsible for redox transformations of gases such as N2 (MoFe cofactor of nitrogenase), CO (C-cluster of CODH), and H2 (H-cluster of hydrogenase).

The acetyl-CoA decarbonylase/synthase complex, discussed in chapter 4, contains not only common Fe4S4 centers (clusters B and D), but also unusual Ni-Fe-S-containing cofactors (clusters A and C).

7 1.3 Metalloproteins are just “MAD” about synchrotron radiation

The multiwavelength anomalous dispersion (MAD) method of solving the phase problem in X-ray crystallography is a perfect complement for the structure determination of metalloproteins at synchrotron radiation facilities (24-26). The tunable nature of synchrotron radiation makes it possible to take advantage of sharp changes in the atomic scattering properties of the metal ions in a protein crystal. The difference in the anomalous signals of most metals measured at different wavelengths is more pronounced than the inherent elements (C, O, N, H, S) in proteins. The effectiveness of the MAD method is evident with the increasing number of structures in the PDB solved by this technique (27). Most of the metalloproteins discussed in this dissertation have intrinsic anomalous scatters (Fe, Co, and Ni) that are appropriate for MAD phasing methods (26).

In the absence of a strong anomalous scatter, the incorporation of more amenable metal ions, such as Yb3+, into the crystal can dramatically improve the strength of the anomalous signals (28). This approach was taken in the structure determination of

Pyrococcus furiosus carboxypeptidase discussed in chapter 3.

1.4 The highlight of this research

The major focus of my graduate research is the utilization of the technique of X- ray crystallography to determine the three-dimensional structures of metalloenzymes and the subsequent use these structures to probe the mechanistic basis of their biological function. In particular, my research includes the Mg-dependent DNA polymerase β that catalyzes the template-directed nucleotidyl transfer reaction; the Co-dependent

8 Pyrococcus furiosus carboxypeptidase, a member of a novel class of metallocarboxypeptidase; and the Ni-Fe-S cluster-containing acetyl-CoA decarbonylase/synthase complex and its components, which catalyzes the two fundamental reactions associated with the synthesis and degradation of acetyl-CoA in methanogens.

1.4.1 Intermediate structures in catalytic pathway of DNA polymerase β

DNA replication is a fundamental biological process essential for cellular reproduction. DNA polymerase β (Pol β) catalyzes the template-directed nucleotidyl transfer reaction required for replication by a proposed two-metal based catalytic mechanism. In order to decipher the microscopic events along the catalytic pathway of

Pol β and to understand how DNA polymerases enhance nucleotide incorporation fidelity, we have used collaborated with the Tsai group to use the exchange inert chromium ion (Cr(III)) to experimentally distinguish the events associated with binding of each of the two metal ions. By using Cr(III) pre-bound to a nucleotide analog

(Cr(III)dTMPPCP), two functional intermediates were trapped in the enzyme’s reaction pathway in the absence of the second catalytic metal. The first intermediate structure, which is prior nucleotidyl transfer, is the Pol β/DNA/Cr(III)-dTMPPCP ternary complex.

Soaking this crystal with Mn(II) ions leads to the formation of the product complex, Pol

β/DNA/Cr(III)-PCP, which is the second intermediate structure and represents a post nucleotidyl transfer complex. In this project, I have been responsible for the data collection, structure determination, and model building and refinement of both

9 intermediate structures. These structures led us to propose that the major conformational change that the enzyme undergoes in is catalytic cycle precedes the rate-limiting step.

This result implies that fidelity is not due to this conformational change. Based on these results a revised mechanism is proposed.

1.4.2 Pyrococcus furiosus carboxypeptidase: crystal structure of a novel class of

carboxypeptidase

Pyrococcus furiosus carboxypeptidase (PfuCP) is a hyperthermophilic that catalyzes the removal of C-terminal residues of polypeptides. Most carboxypeptidases are metalloproteases that use zinc as a cofactor. PfuCP, however, belongs to a novel class of cobalt-dependent carboxypeptidase that shares no sequence homology to the other classes of carboxypeptidases. This class contains a HEXXH metal-binding motif characteristic of aminopeptidases and endopeptidases, but is unprecedented for a carboxypeptidase. Importantly, no structure of this class had been previously determined.

Several members of this new family of carboxypeptidases originate from hyperthermophilic organisms. These carboxypeptidases, including PfuCP, are of interest particularly in the biotechnological industry due to their thermostability and their potential for use in C-terminal MALDI/MS ladder sequencing. A structural investigation was initiated to classify this novel type of carboxypeptidase and to elucidate the structural elements responsible for its catalytic activity. Three PfuCP structures (an apo, a ytterbium-bound, and a lead-bound) have been determined to 2.2 Å, 2.3 Å, and 3.1 Å

10 resolution, respectively. In this project, I have been involved in the crystallization, data collection, structure determination, model building, and refinement of all the structures.

The structure of Yb-PfuCP was first solved by MAD techniques from a single crystal soaked with ytterbium. The other two structures were solved by molecular replacement method. In all the structures, PfuCP exists as a homodimer of 59 kDa subunits. PfuCP is almost entirely helical with its most predominant feature being a deep channel that transverses the entire length of the protein. The HEXXH motif lies at the bottom of this groove. The Pb-bound structure confirms the role of this motif in forming the metal- binding active site. Two histidines from this motif and a glutamate from an adjacent helix coordinate the active site Pb atom. By comparison with other metalloproteases in complex with transition-state analogs, substrate was modeled to the active site and a mechanism proposed for the C-terminal peptide bond hydrolysis.

While the PfuCP crystal structure represents the first structural characterization of this new family of carboxypeptidase, PfuCP has unexpected structural homology to the human angiotensin-converting enzyme, which is an important drug target for controlling congestive heart failure and high blood pressure, and neurolysin, the enzyme that activates neurotensin (13). Thus the fold found in PfuCP appears to have unexpected significance.

11 1.4.3 Acetyl coenzyme-A decarbonylase/synthase complex: rare organometallic

reaction intermediates require unusual metal clusters

Species of Methanosarcina are methanogens that are capable of growing on

acetate as the sole source of carbon and energy. In M. barkeri, the acetyl coenzyme-A

decarbonylase/synthase (ACDS) complex plays a central role in acetyl-CoA metabolism.

The ACDS complex catalyzes three reversible reactions: (i) acetyl-CoA synthesis via a

Ni-Fe-S-containing cluster; (ii) methyl transfer from a pterin to a corrinoid protein; and

(iii) the oxidation of CO to CO2 via a Ni-Fe-S-containing cluster. This multienzyme complex is composed of five different subunits arranged in an (αβγδε)8 structure with a

total mass of 2.1 MDa. The β component is responsible for the acetyl-CoA synthase

activity, while the methyltransferase activity is associated with the γ and δ components.

The α2ε2 components are responsible for the reversible CO oxidation or the CO dehydrogenase (CODH) activity. In order to understand the mechanisms of this massive complex and to elucidate the structures of the unusual nickel-containing clusters, we are working to determine the crystal structures of ACDS complex and its components. In this endeavor, I have developed a purification protocol under strict anaerobic conditions that has yielded high purity ACDS complex and its individual components, which are of suitable quality for crystallization screening as determined by dynamic light scattering experiments. Crystallization screening for the ACDS complex and all of its components has been performed by a wide variety of techniques including capillary batch, hanging drop, and sitting drop methods. To date, the α2ε2 component has yielded four crystal

forms while two crystal forms have been identified for the β component. Their data

12 collection and structure determination are currently in progress. Furthermore, the β component has been demonstrated to be active by means of an acetyl exchange assay in the absence of the complex. Metal chelation and reconstitution experiments performed on β indicated its activity is dependent on nickel and not on copper.

13

Chapter 1 References

1. Lippard, S. J., and Berg, J. M. (1994) Principles of Bioinorganic Chemistry.

2. Cowan, J. A. (1997) Inorganic Biochemistry: An Introduction, 2 ed.

3. Messerschmidt, A., Huber, R., Poulos, T., and Wieghardt, K. (2001) in Handbook of Metalloproteins.

4. Matte, A., and Delbaere, L. T. J. (2001) in Handbook on Metalloproteins pp 59- 91.

5. Sun, H. Z., Chen, R., and Che, C. M. (2002) Prog. Chem. 14, 257-262.

6. Huang, Z. X. (2002) Prog. Chem. 14, 318-322.

7. Frazzon, J., and Dean, D. R. (2003) Curr. Opin. Chem. Biol. 7, 166-173.

8. Dudev, T., and Lim, C. (2003) Chem. Rev. 103, 773-787.

9. Cowan, J. A. (1995) Biological Chemistry of Magnesium.

10. Kobayashi, M., and Shimizu, S. (1999) Euro. J. Biochem. 261, 1-9.

11. Pratt, J. M. (2001) in Handbook on Metalloproteins pp 603-668.

12. Cheng, T. C., Ramakrishnan, V., and Chan, S. I. (1999) Protein Sci. 8, 2474- 2486.

13. Natesh, R., Schwager, S. L. U., Sturrock, E. D., and Acharya, K. R. (2003) Nature 421, 551-554.

14. Sumner, J. B. (1926) J. Biol. Chem. 69, 435-441.

15. Dixon, N. E., Gazzola, C., Blakeley, R. L., and Zerner, J. B. (1975) J. Am. Chem. Soc. 97, 4131-4133. 14

16. Watt, R. K., and Ludden, P. W. (1999) Cell. Mol. Life Sci. 56, 604-625.

17. Walsh, C. T., and Ormejohnson, W. H. (1987) Biochemistry 26, 4901-4906.

18. Ciurli, S., and Mangani, S. (2001) in Handbook on Metalloproteins pp 669-707.

19. Bentrop, D., Capozzi, F., and Luchinat, C. (2001) in Handbook on Metalloproteins pp 357-459.

20. Rees, D. C. (2002) Annu. Rev. Biochem. 71, 221-246.

21. Rees, D. C., and Howard, J. B. (2003) Science 300, 929-931.

22. Cody, G. D., Boctor, N. Z., Filley, T. R., Hazen, R. M., Scott, J. H., Sharma, A., and Yoder, H. S. (2000) Science 289, 1337-1340.

23. Drennan, C. L., and Peters, J. W. (2003) Curr. Opin. Struct. Biol. 13, 220-226.

24. Hendrickson, W. A. (1997) FASEB J. 11, 1565.

25. Hendrickson, W. A. (1999) J. Synchrot. Radiat. 6, 845-851.

26. Hendrickson, W. A., and Ogata, C. M. (1997) Methods Enzymol. 276, 494-523.

27. Hendrickson, W. A. (2000) Trends Biochem. Sci. 25, 637-643.

28. McDonald, N. Q., Panayotatos, N., and Hendrickson, W. A. (1995) EMBO J. 14, 2689-2699.

15

CHAPTER 2

X-ray Structures of DNA Polymerase β Intermediate Complexes:

Molecular Snapshots in its Reaction Pathway

2.1 Introduction

DNA polymerases have the critical responsibility for faithfully replicating and repairing DNA as directed by a complementary template DNA strand. They are ubiquitous in all living organisms and are essential for cell survival. Since the discovery of polymerase α in 1957, dozens of polymerases have been identified that vary in amino acid sequence, structure, and biological function (1). Polymerases must select the correct nucleotide from a pool of structurally similar molecules to ensure accurate and efficient

DNA synthesis. The efficiency by which polymerases insert correct nucleotides varies widely from the highly faithful replicative polymerases with accessory proofreading domains, such as that of the well known Klenow fragment from E. coli, to the mutationally prone polymerase X from the African swine fever virus (2, 3).

Due to the imperative role of these enzymes, numerous studies have focused on elucidating their physiological function and mechanism. The enzyme that we have chosen to study is DNA polymerase β (Pol β), which was first discovered in the early

16 1970s. Decades of study have revealed that Pol β is responsible for filling in short gaps

(< 6 nucleotides) in DNA in a distributive manner. It now serves as the prototypical repair enzyme in the base excision repair pathway (4). Pol β is a fairly average polymerase in that it exhibits moderate catalytic efficiency (rate constant of 90 s-1) and fidelity (base-substitution error frequency of 1 per 1000 bases) (5).

Pol β is composed of a single 39 kDa polypeptide of 335 amino acid that contains two domains, an 8 kDa N-terminal domain (residues 1-87) that performs the 5'- deoxyribose phosphatase activity and single stranded DNA binding, and a 31 kDa domain (residues 88-335) that performs the template-directed nucleotidyl transfer reaction. A structure of a ternary complex of Pol β in complex with DNA and a deoxynucleotide (dNTP) is shown with the two domains in Figure 2.1. The 31 kDa domain can be further divided into three subdomains named fingers, thumb, and palm, as defined by Steitz (11). The 8 kDa domain leads into the thumb subdomain (residues 88-

151) that comprised of four α helices. The palm subdomain (residues 152-262) adopts a two-layered αβ sandwich formed from two α helices and five anti-parallel β strands. The

fingers subdomain (residues 263-335) consists of an αβ motif.

Due to its relatively small size and excellent expression properties, Pol β serves as an excellent model for study. To date, X-ray crystal structures of various polymerases have provided an abundance of structural information (6-10). The structures of different family of polymerases though structurally quite different, share a common overall architecture that has been likened to a right hand containing three subdomains, thumb, palm, and fingers, as can be seen in the ternary complex of Pol β (Figure 2.1). The palm

17 subdomain is responsible for the nucleotidyl transfer reaction whereas that of the fingers

subdomain interacts with the incoming dNTP. The thumb may play a role in positioning

the DNA in the processivity of replication and in translocation to the subsequent

nucleotide (11).

In addition to having a common architecture, polymerases also share a common

mechanism (Figure 2.2) that involves two Mg(II) ions (7). The first Mg, referred to as

the nucleotide-binding metal (MA), interacts with the incoming dNTP and is proposed to

facilitate the departure of the β- and γ-phosphates as pyrophosphate following nucleotide

incorporation. The second Mg, termed the catalytic metal (MB), is proposed to serve as a

Lewis acid activating the 3'-OH of the primer strand for attack on the α-phosphate of the incoming dNTP.

Two major conformations of the fingers subdomain have been observed in the crystal structures of Pol β. While the free enzyme or enzyme DNA binary complex

exhibits a more open conformation, in the presence of DNA, dNTP, and the two active

site metals, the Pol β adopts a conformation that is more closed (6-8). In this closed

conformation, the fingers subdomain moves toward the active site by an approximate 30o rotation along a hinge-like helix, helix M. This movement of the fingers subdomain encloses the dNTP and leads to formation of the proper geometry for binding the two

Mg(II) ions. In the open conformation, the enzyme active site is freely accessible to solvent with several key catalytic residues held in salt bridges. Thus this conformational change is required for chemistry and it has been suggested that it also plays an important role in the discrimination of nucleotides by Pol β (8, 12), as well as other polymerases.

18 The primary aim of our study was to probe the role of each of the active site

metals (MA and MB) in the conformational change of Pol β and to provide novel

structural intermediates in the enzyme’s reaction pathway. Our approach involved the

use of the exchange inert Cr(III) ion in association with a dNTP analog.

Previously it had been shown by the Tsai group that in the absence of magnesium,

binding of Cr(III)·dNTP can induce Pol β to undergo its fast conformational change

without proceeding through the nucleotidyl transfer reaction (12). Thus, using an exchange-inert Cr(III)·dNTP complex makes it possible to capture the stable intermediate prior to the chemistry of nucleotide incorporation. Therefore the structure of the Cr(III) pre-chemistry intermediate in the absence of the catalytic metal (MB) allows us to

determine whether Pol β is in the opened or closed conformation, or somewhere in

between. Furthermore, this approach allows us to include the 3'-OH of the DNA primer

strand in the active site without chemical turnover. This critical nucleophile in the

phosphoryl transfer reaction is absent in most of the available structures of Pol β since

the primer often ends as a dideoxynucleotide (ddNTP) to prevent the chemical reaction.

In the course of our studies, we have determined the crystal structures of two

intermediate complexes, both of which represent novel intermediates in the Pol β reaction

pathway (13). The first of these structures, Pol β-DNA-Cr(III)·dTMPPCP, was shown to

exist in the closed form. This structure was used to suggest that the closing of the fingers

subdomain precedes the rate-limiting conformational (or discriminating) step ruling out

its involvement in Pol β fidelity (14). Additional evidence in support of this structure

being a functional intermediate was obtained by the observation of turnover in the crystal

19 upon soaking with Mn(II) ions, as determined by the structure of the Pol β-DNA-

Cr(III)·PCP product complex, which is also in the closed conformation. These structural results have revised our understanding of the fidelity of Pol β and potentially DNA polymerases in general.

2.2 Materials and methods

2.2.1 Purification of Pol β and preparation of the DNA and Cr(III)·dTMPPCP

substrates

The purification of Pol β and preparation of the DNA and the Cr(III)·dTMPPCP were performed in Prof. Tsai’s laboratory by Dr. Xuejun Zhong, as described previously

(13). Briefly, recombinant rat DNA Pol β was overexpressed in E. coli and purified. The purified protein sample was desalted by washing three times in a Centricon 10 filter

(Amicon) with demetalated exchange buffer (0.1 M Tris, 10 mM ammonium sulfate, pH

7.0, treated with Chelex 100 resin, Bio-Rad) and then concentrated to 30 mg/mL prior to storage at -80 °C.

The DNA template, 5'-AATAGGCGTCG-3', and primer, 5'-CGACGCC-3',

(Integrated DNA Technologies, Inc) were purified by PAGE (14). Prior to crystallization, the template and primer DNA solutions were mixed in a 1:1.1 ratio followed by heating at 80 °C for 5 min and annealing by gradually cooling.

The dNTP substrate analog of 2'-deoxythymidine-5'-β,γ-methylene triphosphate

(dTMPPCP, Amersham) was used in the crystallization. The use of this methylene- containing analog can prevent problems associated with hydrolysis. The exchange inert

20 Cr(III)·dTMPPCP complex was prepared to a concentration of 3 mM by mixing equal

volumes of 6 mM dTMPPCP and 6 mM CrCl3 at 80 °C for 10 min, as described

previously (15, 16).

2.2.2 Preparation, crystallization, and data collection of the Pol β-DNA-

Cr(III)·dTMPPCP and Pol β-DNA-Cr(III)·PCP intermediate complexes

Crystallization of Pol β-DNA-Cr(III)·dTMPPCP intermediate complex. The rat

Pol β-DNA-Cr(III)·dTMPPCP ternary complex was prepared by Dr. Zhong by mixing

100 µL of Pol β (30 mg/mL), 20 µl of the DNA solution (~5 mM), and 240 µL of the

Cr(III)·dTMPPCP solution (3 mM). Crystals were grown by the sitting drop vapor diffusion method with a reservoir solution consisting of 8% PEG 3350, 70 mM lithium sulfate, and 100 mM MES, pH 7.0. This condition is similar to the one identified previously for the Pol β-DNA-Mg2·ddCTP ternary complex containing two bound Mg(II) ions (6, 7).

Preparation and crystallization of the Pol β−DNA-Cr(III)·PCP intermediate

complex. The Pol β-DNA-Cr(III)·PCP intermediate complex was prepared by soaking a

crystal of the Pol β-DNA-Cr(III)·dTMPPCP ternary complex with a reservoir solution

containing 5 mM MnCl2 and 40% (v/v) glycerol for 30 minutes at room temperature prior

to cryo-cooling with liquid nitrogen.

Data collection. Prior to data collection, crystals were sequentially transferred to

reservoir solutions containing increasing concentrations of glycerol up to a maximum

concentration of 40% (v/v). Crystals were mounted with a nylon loop and then cooled in

21 liquid nitrogen (17). All useful datasets were collected at the synchrotron facilities of the

Brookhaven National Laboratory beamline X4A and the Stanford Synchrotron Radiation

Resource beamline 9-1. The diffraction data were processed and scaled using the

programs DENZO and SCALEPACK (18).

2.2.3 Structure determination and refinement

Both intermediate crystals belong to the space group P21 (Table 2.1) with cells that were similar to those of the rat Pol β ternary complex (PDB code: 2BPG) (6). The

structures of the two intermediate complexes could therefore be determined by molecular

replacement (MR) method. The initial R factor after rigid body refinement was 0.40

confirming the correct MR solution.

Model building was performed using the program O (19) and the refinement was

carried out using X-PLOR 3.851 without the use of sigma cutoffs (20). Eight percent of

the data were set for the free-R factor calculation (21). A bulk solvent correction and

overall anisotropic B-factor scaling were applied to the diffraction data. Strict non-

crystallographic symmetry (NCS) restraints were used in the early rounds of the

refinement, but these restraints were gradually relaxed and later entirely removed. The

Cr(III).dTMPPCP and the Cr(III).PCP plus the extended primer strand were modeled to

the active site after removal of the restraints using Fo-Fc omit maps. The final R values

(Rfree) were 22.6% (28.8%) and 22.4% (28.6%) for the pre- and post-nucleotidyl transfer

intermediates, respectively.

The average temperature factors were rather high, averaging 42 Å2 in the Pol β-

DNA-Cr(III)·dTMPPCP complex and 47 Å2 Pol β-DNA-Cr(III)·PCP complex. These 22 values, however, were similar to other published Pol β structures and have been

suggested to originate from inherent static disorder of the protein complex (8). The

geometry was checked by PROCHECK and all parameters were in the acceptable ranges

with the majority being better than acceptable (22).

2.3 Results and discussions

2.3.1 The structure of Pol β-DNA-Cr(III)·dTMPPCP

The overall fold of the Pol β-DNA-Cr(III)·dTMPPCP structure is similar to that

of rat Pol β-DNA-Mg2·ddCTP ternary complex (Cα root mean square deviation, RMSD,

0.88 Å) and human Pol β-DNA-Mg2·ddCTP ternary complex (PDB codes 2BPG and

2BPY, respectively) structures (7, 8). Examination of the active site reveals strong FO-FC electron density (6σ) at MA site in both NCS related molecules (Figure 2.3A), consistent

with binding of a single Cr(III) ion. The metal is coordinated by the carboxylate side

chains of Asp190, Asp192, and the non-bridging oxygens of the α, β, and γ phosphates of

the dTMPPCP. Confirming the absence of the catalytic metal ion, no electron density

was observed at site MB. The active site of the Pol β-DNA-Cr(III)·dTMPPCP structure

with the electron density omitted is depicted in Figure 2.4B. Furthermore, examination

of the Pol β-DNA-Cr(III)·dTMPPCP intermediate reveals that no nucleotide transfer has

taken place despite the free 3'-OH.

As mentioned previously, the fundamental question we wished to address was

whether the conformational change of Pol β is rate-limiting. If this were true, the

structure of Pol β-DNA-Cr(III)·dTMPPCP should be in the “open” form. We find, 23 however, that the Pol β-DNA-Cr(III)·dTMPPCP structure is in an essentially identical

conformation to that of the Pol β-DNA-Mg2·ddCTP complex in that they are both in the

“closed” conformation (Figure 2.5). This provides evidence that the “fingers closing”

precedes the rate-limiting step, thus ruling out its importance in enhancing fidelity.

Presumably, “fingers closing” is induced by binding of the nucleotide-metal complex

(MA·dNTP).

2.3.2 The structure of Pol β-DNA-Cr(III)·PCP

As it had been previously demonstrated that addition of metal ions that serve as

the catalytic metal results in turnover of the rate-limiting step, a crystal of the Pol β-

DNA-Cr(III)·dTMPPCP intermediate complex was soaked with Mn(II) yielding an

intermediate structure following turnover (Pol β-DNA-Cr(III)·PCP). While several

metals ions are capable of activating the enzyme for catalysis (23), Mn(II) was used

rather than the physiological Mg(II) since it is more easily detectable due to its larger

mass. Soaking times longer than 30 minutes or at concentration above 5mM MnCl2 resulted in a significant loss in the diffraction quality and the resolution of the crystals.

By optimizing the soaking conditions, however, we were able to obtain a crystal that had undergone turnover yet still diffracted to 2.5 Å resolution. The overall fold of the Pol β-DNA-Cr(III)·PCP intermediate structure is virtually identical to that of the

Pol β-DNA-Cr(III)·dTMPPCP structure. The active site of Pol β-DNA-Cr(III)·PCP

structure (Figure 2.3B) reveals positive 3σ Fo-Fc electron density between the incoming

nucleotide’s α phosphate and the 3'-oxygen of the primer while no density is observed

24 between the α and β phosphates, consistent with nucleotidyl transfer of the nucleotide

substrate to the primer strand. Indeed the 2Fo-Fc electron density could be fit to this

model. Thus this structure represents a post-chemistry intermediate, Pol β-DNA-

Cr(III)·PCP. The nucleotide metal Cr(III) retains its coordination. Strangely, however,

no electron density is observed at the catalytic metal site for Mn(II) perhaps indicating

that release of the catalytic metal following turnover is rapid. In addition, the Asp256-

Arg254 salt bridge, which is absent in the structure of the pre-turnover Pol β-DNA-

Cr(III)·dTMPPCP intermediate, is present in this structure (Figure 2.4D).

Turnover in polymerase crystals has also been observed for Bacillus DNA

polymerase (10). However, in that structure the PPi product is released and the enzyme has translocated to the n+1 position on the DNA. Turnover of Pol β ternary complexes containing blunt-ended DNA has also been observed in other crystals of Pol β (7). These blunt-ended structures are less biologically relevant, however, because they lack the template base. Thus our Pol β-DNA-Cr(III)·PCP structure represents the first functionally relevant post-chemistry intermediate prior to reopening of the fingers subdomain.

The point of release of catalytic metal ion in the polymerase reaction sequence has been previously unknown as has its association with the reopening of the fingers subdomain. The lack of electron density at the catalytic metal site in the Pol β-DNA-

Cr(III)·PCP structure (Figure 2.3B) indicates that release of the catalytic metal ion occurs immediately after the nucleotidyl transfer step. Furthermore, since the fingers subdomain reopening has not occurred in the Pol β-DNA-Cr(III)·PCP structure suggests that this

25 post-chemistry conformational change is not induced by release of the catalytic metal.

Comparison of this structure (Figure 2.4D) to the Pol β-nicked DNA product structure

(PDB code 1BPZ, Figure 2.4E), in which the fingers subdomain has reopened (8), would suggest that opening of the fingers subdomain is induced instead by the release of

MA·PPi.

2.3.3 Implications presented by the intermediate structures

This work provides two novel intermediates in the reaction pathway of Pol β, the pre-chemistry complex of Pol β-DNA-Cr(III)·dTMPPCP, and the post-chemistry complex of Pol β-DNA-Cr(III)·PCP in which the thymidine nucleotide has been added to the DNA primer strand. The potential significance of this work is discussed below.

There is a generally accepted theory that the catalysis of DNA polymerases involves a rate-limiting conformational change. In fact such a rate-limiting conformational change has been proposed for Pol β (8, 12). This theory is primarily based on the experimental evidence of a small thio effect and pre-chemistry fluorescence phase changes observed by stopped-flow kinetic experiments (5). However, we believe that the evidence for such a rate-limiting conformational change is insufficient. In regards to the thio effect, the small magnitude of the effect for correct base pairing (4.3 for T:A) and increased value for incorrect base pairs (9 for T:G) have been used as evidence to suggest that the chemical step is not rate-limiting in the catalysis of Pol

β (23). Nonetheless, Liu and coworkers have recently shown that great caution is needed in using the thio effect as a mechanistic tool since the common use of diastereomeric

26 mixtures of the phosphorothiolate analog of the dNTP can lead to misinterpretation of the results (13, 24). Therefore the evidence of the small thio effect supporting a rate-limiting conformational change is thus questionable, since the thio nucleotide probe was racemic

(23).

In regards to the stopped flow analysis experiments supporting a rate-limiting conformational change, Zhong and coworkers revealed two phases of fluorescence change of Pol β, a fast phase at 70 s-1 and a weaker slow phase at 6 s-1 (12). In their experiment the primer DNA terminated with a ddNTP, which prevents further turnover cycles from occurring, was monitored opposite DNA template containing the fluorophore of 2-aminopurine. Considering that both the slow phase was comparable to the pre- steady-state rate constant of Pol β and the primer strand was terminated, they concluded that the slow phase corresponded to a rate-limiting conformational change prior to chemistry. These conclusions should be also interpreted with caution, since Tsai and coworkers recently revisited these previous steady-state experiments and found contradictory results (13, 25). In their revisited experiments using a newer instrument, only the fast phase was observed in the presence of the terminated primer DNA (13).

Follow up studies with optimized placement of the 2-aminopurine in the template strand confirmed the absence of the slow phase and further revealed that the fast phase can be divided to fit two exponential phases (25). These data would suggest that there is a yet to be detected fast conformational change by fluorescence.

The insertion step has been postulated to be limited by either the chemistry or a conformational change (26). Our structural results of the Pol β-DNA-Cr(III)·dTMPPCP

27 intermediate in the closed conformation suggests that a rate-limiting conformational change has not occurred. Based on the above data we interpret these results to indicate that either there is no rate-limiting conformational change, indicating a rate-limiting chemical step, or there is a yet to be detected conformational change. The Pol β-DNA-

Cr(III)·dTMPPCP intermediate structure suggests that any rate-limiting conformational change would be expected to occur after closure of the fingers subdomain and would be more subtle, such as reorientation of side chains or local structural adjustments, since a large subdomain movement would be expected to be detected by previous stopped flow fluorescence analysis (12, 13, 25, 27). This idea is supported with recent kinetic (28-31) and modeling (32, 33) analysis of Pol β, which also indicate that the movement of the fingers subdomain is rapid relative to the chemical step.

Further support for a subtle yet rate-limiting conformation change comes from steady-state and pre-steady-state kinetic data on a valine substituted mutant of Asp276, which interacts with the base of the incoming nucleotide. These experiments suggest that the rate-limiting conformational change is not the "open" to "closed" structural transition, but instead involves rearrangement of Asp276 in the closed conformation (30). Both of our intermediate structures indicate that Asp276 is exposed to the solvent and does not hydrogen bond to the nucleotide base, thus it is unlikely to participate in a rate-limiting conformational change.

In a separate study, computational modeling with high temperature simulations suggests that Arg258 rearrangement coupled with Mg(II) ion binding is the rate-limiting factor rather than a large subdomain movement (32, 33). Inspection of our intermediate

28 structures, as well as all available closed fingers Pol β structures, discredits this particular rate-limiting rearrangement prior to chemistry since Arg258 is involved in a salt bridge with Glu295 (Figures 2.4B,C,D and 2.5) in all closed fingers structures of Pol β regardless if chemical turnover has occurred.

The binding of the catalytic metal ion may be consistent with such a subtle yet rate-limiting conformational change. Particularly since it involves the positioning of the primary ligands that bind the catalytic metal ion, the α-phosphate of the dNTP, the 3'-OH of the primer, and the side chains of residues Asp190, Asp192, and Asp256.

Examination of the Pol β-DNA-Cr(III)·dTMPPCP intermediate indicates that the side chain of Asp256 must completely flip in order to participate in the catalytic metal binding

(Figure 2.4B). It is anticipated that the position of the oxygen ligand of α-P of the dNTP is critical to the metal coordination. Since the geometry of the reacting atoms is expected to influence the rate of nucleotide insertion but also catalytic metal binding, we examined the relative distance of the C3' atom of the primer terminus and the α-P of the incoming nucleotide. We note that the distance is shorter in the Pol β-DNA-Cr(III)·dTMPPCP intermediate structure than in the Pol β-DNA-Mg2·ddCTP intermediate structure (Figure

2.4B,C) (3.95 and 4.53 Å, respectively). Therefore for the oxygen ligand of the α-P of the dNTP must shift to participate in catalytic metal binding. These subtle conformational changes induced by catalytic metal binding may be the elusive rate- limiting step prior catalysis, however it is obvious that further approaches to isolate and measure the dynamics of the catalytic metal ion are needed in order to test this hypothesis.

29 In the absence of conclusive evidence for a rate-limiting conformational change,

we present a catalytic model in which the rate-limiting step is based on chemistry.

Further implications on the reaction scheme can be inferred from the post-chemistry Pol

β-DNA-Cr(III)·PCP intermediate structure. Based on its comparison with the Pol β-

DNA binary complex (Figure 2.4E) it is likely that the conformation change of reopening

of the fingers subdomain occurs after release of the MA-PPi since it is not induced by release of the catalytic metal. Based on these structural results, we therefore suggest a modified catalytic model for Pol β (shown in Figure 2.6). In this scheme, step 1 involves binding of the MA·dNTP; step 2, closing of the fingers subdomain induced by MA·dNTP binding; step 3, catalytic metal ion binding, MB; step 4, the chemical conversion; step 5, the catalytic metal ion release; step 6, reopening of the fingers subdomain; and step 7, release of MA·PPi. Structures of Pol β have been determined previously for a number of

potential intermediates, Pol βopen-DNA (Figure 2.6, intermediate A) (7), Pol βclosed-

DNAn+1-MA·dNTP-MB (intermediate E) (6-8), and Pol βopen-DNAn+1 (intermediate H) (7).

The structures reported here represent the intermediates Pol βclosed-DNA-MA·dNTP (Pol

β-DNA-Cr(III)·dTMPPCP) and Pol βclosed-DNAn+1-MA·dNTP (Pol β-DNA-Cr(III)·PPi),

as shown in Figure 2.6 (intermediates C and F, respectively). The specific difference

between this revised model and the model proposed previously (27) is the presence of the

rate-limiting chemical step instead of a rate-limiting conformational change.

2.4 Conclusions

In this project, we have solved two unique structural intermediates of rat Pol β,

the 2.6 Å resolution structure of Pol β-DNA-Cr(III) ·dTMPPCP, which corresponds to a

30 pre-chemistry intermediate in the absence of the catalytic metal, and the post-chemistry intermediate, Pol β-DNA-Cr(III) ·PCP. These structures suggest that closing of the fingers subdomain is not the rate-limiting conformational change as previously anticipated. Furthermore, a revised model of the catalytic pathway is provided that is consistent with the available structural and kinetic data.

31

Figure 2.1. Ribbons structure of Pol β-DNA-Cr(III)·dTMPPCP ternary complex. The β strands, red; α helices, cyan; loops, green; and the chromium ion, yellow. The DNA and nucleotide are represented as stick models in blue and chartreuse, respectively. This figure was prepared using the programs MOLSCRIPT and Raster-3D (34, 35).

32

Figure 2.2. Schematic diagram of active site of Pol β showing two metal ions, the nucleotide-binding metal (site A) and the catalytic metal (site B) (6-8).

33

Figure 2.3. Stereoview of Fo-Fc electron density maps (3σ blue and 6σ red) superimposed on active site models. Carbon atoms are colored yellow; nitrogen, cyan; oxygen, red; phosphorus, green; and chromium, pink. (A) The Pol β pre-turnover intermediate showing the presence of one Cr(III) ion and the absence of catalytic metal ion. (B) The post-chemistry Pol β intermediate following Mn(II) incubation. These figures were prepared using XtalView and Raster-3D (35, 36).

34

Figure 2.4. Structures of the active sites of stepwise Pol β intermediates as stereo pairs. (A) The structure of the “opened fingers” human Pol β-DNA substrate binary complex (7). (B) “closed fingers” rat Pol β-DNA-Cr(III)·dTMPPCP intermediate complex. (C) “closed fingers” human Pol β-DNA-Mg2·ddCTP ternary complex (6, 7). (D) “closed fingers” rat Pol β-DNA-Cr(III)·PCP intermediate complex (E) “opened fingers” human Pol β-DNA product (nicked) binary complex (8). These figures were prepared using the program MOLSCRIPT and Raster-3D (34, 35)

35

Figure 2.4. Structures of the active sites of stepwise Pol β intermediates 36

Figure 2.5. Stereoview of conformational change induced by MgA·dNTP binding preceding chemistry. In blue, the open conformation based on the Pol β-DNA binary complex (1bpx); (green) closed conformation based on the Pol β-DNA-Mg2·ddCTP ternary complex (1bpy); (red) closed conformation of the Pol β-DNA-Cr(III)·dTMPPCP intermediate complex. This figure was prepared by superposition of the palm and thumb subdomains using the Cα atom of amino acids 92-273. The RMSD of the main chain atoms of helix N (amino acids 275-289) for the open Pol β-DNA binary complex (blue) and the closed Pol β-DNA-Cr(III)·dTMPPCP intermediate complex (red) is 7.06 Å; the RMSD for the two closed complexes (red and green) is 0.87 Å. These figures were prepared using the programs MOLSCRIPT and Raster-3D (34, 35).

37

Figure 2.6. Proposed reaction scheme for Pol β. Pol βopen, Pol β in the open fingers conformation; Pol βclosed, Pol β in the closed fingers conformation; MA, nucleotide metal; MB, catalytic metal; and PPi, pyrophosphate.

38

Crystal Pol β-DNA-Cr(III)·dTMPPCP Pol β-DNA-Cr(III)·PCP unit cell dimensions a (Å) 86.1 85.4 b 56.2 56.1 c 105.9 105.8 β (deg) 107.2o 106.9 o total / unique reflections 172244 / 27893 172244 / 27893 completeness (%)a 88.9 (82.4) 96.6 (91.2) dmin (Å) 2.6 2.5 a,b Rsym (%) 5.6 (22.4) 4.7 (30.5) Resolution range, Å 20 - 2.6 20 - 2.6 No. of non-hydrogen protein atoms 5198 5198 No. of non-hydrogen DNA atoms 616 656 No. of metal ions 2 2 No. of ligand atoms 58 18 No. of water molecules 119 123 RMSD bond length, Å 0.015 0.020 c R / Rfree 22.6 / 28.8 22.4 / 28.6 B overall Å2 41.91 46.86 a The number in parentheses are for the highest resolution shell. b Rsym (I)= ΣΣi |Ii - |/Σ, where Ii is the intensity of the measurements for a reflection and is the mean value for the reflection. c Rfree was calculated on 8% of the reflections randomly omitted from the refinement.

Table 2.1: Summary of crystallographic results of the Pol β intermediate structures.

39

Chapter 2 References

1. Hubscher, U., Maga, G., and Spadari, S. (2002) Annu. Rev. Biochem. 71, 133-163.

2. Beese, L. S., Derbyshire, V., and Steitz, T. A. (1993) Science 260, 352-355.

3. Showalter, A. K., Byeon, I. J. L., Su, M. I., and Tsai, M. D. (2001) Nat. Struct. Biol. 8, 942-946.

4. Wilson, S. H. (1998) Mutat. Res.-DNA Repair 407, 203-215.

5. Kunkel, T. A., and Bebenek, R. (2000) Annu. Rev. Biochem. 69, 497-529.

6. Pelletier, H., Sawaya, M. R., Kumar, A., Wilson, S. H., and Kraut, J. (1994) Science 264, 1891-1903.

7. Pelletier, H., Sawaya, M. R., Wolfle, W., Wilson, S. H., and Kraut, J. (1996) Biochemistry 35, 12742-12761.

8. Sawaya, M. R., Prasad, R., Wilson, S. H., Kraut, J., and Pelletier, H. (1997) Biochemistry 36, 11205-11215.

9. Doublie, S., Tabor, S., Long, A. M., Richardson, C. C., and Ellenberger, T. (1998) Nature 391, 251-258.

10. Kiefer, J. R., Mao, C., Braman, J. C., and Beese, L. S. (1998) Nature 391, 304- 307.

11. Steitz, T. A. (1999) J. Biol. Chem. 274, 17395-17398.

12. Zhong, X. J., Patel, S. S., Werneburg, B. G., and Tsai, M. D. (1997) Biochemistry 36, 11891-11900.

40 13. Arndt, J. W., Gong, W. M., Zhong, X. J., Showalter, A. K., Liu, J., Dunlap, C. A., Lin, Z., Paxson, C., Tsai, M. D., and Chan, M. K. (2001) Biochemistry 40, 5368- 5375.

14. Wong, I., Patel, S. S., and Johnson, K. A. (1991) Biochemistry 30, 526-537.

15. Dunaway-Mariano, D., and Cleland, W. W. (1980) Biochemistry 19, 1506-1515.

16. Cleland, W. W. (1982) Methods Enzymol. 87, 159-179.

17. Teng, T. Y. (1990) J. Appl. Crystallogr. 23, 387-391.

18. Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, 307-326.

19. Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M. (1991) Acta Crystallogr. A 47, 110-119.

20. Brünger, A. T. (1993) X-PLOR Version 3.1 manual, Yale University, New Haven, CT.

21. Brunger, A. T. (1993) Acta Crystallogr. D49, 24-36.

22. Laskowski, R. A., Macarthur, M. W., Moss, D. S., and Thornton, J. M. (1993) J. Appl. Crystallogr. 26, 283-291.

23. Werneburg, B. G., Ahn, J., Zhong, X. J., Hondal, R. J., Kraynov, V. S., and Tsai, M. D. (1996) Biochemistry 35, 7041-7050.

24. Liu, J., and Tsai, M. D. (2001) Biochemistry 40, 9014-9022.

25. Dunlap, C. A., and Tsai, M. D. (2002) Biochemistry 41, 11226-11235.

26. Johnson, K. A. (1993) Annu. Rev. Biochem.62, 685-713.

27. Zhong, X. J., Patel, S. S., and Tsai, M. D. (1998) J. Am. Chem. Soc. 120, 235-236.

28. Shah, A. M., Conn, D. A., Li, S. X., Capaldi, A., Jager, J., and Sweasy, J. B. (2001) Biochemistry 40, 11372-11381.

29. Shah, A. M., Li, S. X., Anderson, K. S., and Sweasy, J. B. (2001) J. Biol. Chem. 276, 10824-10831.

30. Vande Berg, B. J., Beard, W. A., and Wilson, S. H. (2001) J. Biol. Chem. 276, 3408-3416.

41

31. Kim, S. J., Beard, W. A., Harvey, J., Shock, D. D., Knutson, J. R., and Wilson, S. H. (2003) J. Biol. Chem. 278, 5072-5081.

32. Yang, L. J., Beard, W. A., Wilson, S. H., Broyde, S., and Schlick, T. (2002) J. Mol. Biol. 317, 651-671.

33. Yang, L. J., Beard, W. A., Wilson, S. H., Roux, B., Broyde, S., and Schlick, T. (2002) J. Mol. Biol. 321, 459-478.

34. Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 946-950.

35. Merritt, E. A., and Murphy, M. E. P. (1994) Acta Crystallogr. D50, 869-873.

36. McRee, D. E. (1999) J. Struct. Biol. 125, 156-165.

42

CHAPTER 3

Crystal structure of a novel carboxypeptidase from the

hyperthermophilic archaeon Pyrococcus furiosus

3.1 Introduction

Carboxypeptidases hydrolyze peptide bonds from the C-terminus of peptides and proteins. Metallocarboxypeptidases, the largest class of carboxypeptidase, are ubiquitous in nature and typically have a single zinc ion bound to the active site. Different metallocarboxypeptidases can be classified by their substrate specificities. For instance, (CPA) is specific for neutral, preferably hydrophobic amino acids, (CPB) is for basic amino acids, and carboxypeptidase T (CPT) is for basic and hydrophobic amino acids (1, 2). Most of the carboxypeptidases studied to date are mesophilic enzymes, but several thermostable carboxypeptidases have been purified and characterized (3-8). In the biotechnological industry, there is growing interest in

developing applications for hyperthermophilic enzymes since they are often more stable

in the presence of severe environmental factors: elevated temperatures, denaturing agents,

and organic solvents. One such application is the use of a thermostable carboxypeptidase

43 for the C-terminal amino acid ladder sequencing of peptides and proteins using matrix-

assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS)

(6).

Early studies of the thermophilic Thermus aquaticus carboxypeptidase (TaqCP)

revealed distinct differences from any known carboxypeptidase. Biochemical

characterization of the TaqCP demonstrated that it had a novel size, sequence, and an

unexpected metal-binding motif (3). TaqCP was found to be missing the HXXE(X)123-

132H motif characteristic of classical metallocarboxypeptidases (CPA, CPB, and CPT), and instead contained the HEXXH motif (4, 5), common to other families of metalloproteases (thermolysin and deformylase) (9, 10), but not previously observed in a carboxypeptidase. These features were also later found in the hyperthermophilic P. furiosus carboxypeptidase that shares 37% sequence identity with TaqCP.

Pyrococcus furiosus carboxypeptidase (PfuCP) has one of the highest optimal

activity temperatures (90-100oC) studied to date. It was purified to homogeneity by

Cheng et al. and shown to be an unusual metalloprotease, in that the zinc-bound form is

inactive, while binding of other metals such as cobalt promotes its catalytic activity (6).

PfuCP has a broad substrate specificity that includes basic, aromatic, neutral, and polar

amino acids, but is unable to digest peptides with glycine, proline, or acid residues

(aspartate, glutamate) at the C-terminus. Based on SDS-PAGE, gel filtration, and

MALDI-TOF MS, PfuCP was found to be a homodimer of 59 kDa subunits (6).

A structural investigation of PfuCP was undertaken to classify this novel type of

carboxypeptidase and to elucidate the role of various structural elements in promoting its

44 catalytic activity. Here we present structures of three different forms of PfuCP that provide a clear picture of the overall aggregation-state and fold, and aid in identifying the metal-binding active site. A mechanism is proposed based on a computational model of substrate bound to the active site and from the locations of critical residues conserved in the overall family.

3.2 Materials and methods

3.2.1 Purification and Crystallization

PfuCP was purified from P. furiosus as previously described (6). For the crystallization, PfuCP was concentrated to 19 mg/mL in 50 mM Tris buffer (pH 8.0) with

10% glycerol. Several different crystal forms were obtained using the hanging drop method at 4o C with a single crystallization condition originally identified from the

Hampton Screen Kit I (11). The conditions were subsequently improved to the final solution consisting of 25-30% polyethylene glycol 4000, 100 mM Tris (pH 8.5), 20-40 mM MgCl2. Of the crystals, two forms were the easiest to reproduce and of sufficient quality for X-ray characterization. These crystals grew within 10 days (form I) and 6 weeks (form II). Form I crystals are triclinic and belong to the space group P1 with four molecules per asymmetric unit. Form II crystals are monoclinic and belong to the space group C2 with one molecule per asymmetric unit.

45 3.2.2 Data collection and cryoprotection

The structure was solved from MAD data collected on a single Yb derivatized

crystal (12). The crystal was prepared by soaking a triclinic crystal for 24 hours in an

artificial mother liquor containing 20mM YbCl3 in place of MgCl2. The crystal was

transferred to paratone-N (Hampton Research) and cryo-cooled in liquid nitrogen (13).

Data were collected at two wavelengths, the peak of the anomalous absorption and a

remote wavelength at 100 K. The wavelengths were chosen based on an X-ray

fluorescence spectrum of the crystal, close to the ytterbium LIII edge. The Pb derivatized

crystal was prepared by soaking a monoclinic crystal for 3 days in an artificial mother

liquor containing 1mM PbBr2. Both the apo-PfuCP and Pb-PfuCP crystals were transferred to solutions containing increasing concentrations of glycerol up to 25% prior liquor in flash freezing in liquid nitrogen. All diffraction data (Table 1) were collected in oscillation mode using synchrotron radiation with an ADSC-Q4 CCD detector.

Diffraction images were indexed, integrated, and scaled using the HKL program suite

(14).

3.2.3 Structure solution

Determination of the Yb binding-sites was made by inspecting Harker sections from a Bijvoet difference patterson map, calculated from the 1.386 Å wavelength data set using the program package XtalView (15). The initial MAD phases from 20 to 3.1 Å were obtained using the program PHASES (Table 3.1) (16). The electron map calculated from the initial phase was used to identify the locations of several long alpha helices

46 whose positions in turn could be utilized to identify the non-crystallographic symmetry

elements for the four different molecules in the cell (17). Following non-crystallography

symmetry averaging using the software package RAVE from the Uppsala Software

Factory (18), the quality of the maps was sufficient to build a polyalanine model

containing about 80% of the protein residues. After several cycles of phase combination

and rebuilding, the side chains could be assigned (16). The apo-PfuCP and Pb-PfuCP

structures were solved by molecular replacement using the program CNS (19).

3.2.4 Model building and refinement

Model building was performed using the program O from both the original and

averaged density maps (20). The entire polypeptide chain could be traced except for the

two N-terminal residues. The program package CNS was used to perform the structure

refinement originally using data from 20 to 3.1 Å resolution, and then extending to

include all the data from 20 to 2.3 Å resolution while applying an overall bulk solvent

correction (19). The refinement consisted of several rounds of least square minimization

and annealing (at 2500 K), followed by manual rebuilding with the program O (19, 20).

Ten percent of the data were excluded for calculating the Rfree (21). Individual B factors

were initially set to 15 Å2 for the protein atoms and were refined by applying an overall anisotropic B-factor scaling. Water molecules were excluded from this early refinement.

Final refinement included water molecules and individual anisotropic B-factor refinement. The final Yb derived structure contains 16652 nonhydrogen atoms and 610 water molecules in the asymmetric unit. The final R factor is 21.2 with an Rfree factor of

47 26.7. All of the residues lie within allowed regions of a Ramachandran plot. The

deviation from ideal of the bond lengths is 0.007 Å, and the deviation of bond angles is

1.237o. The geometry was checked by the program PROCHECK, and all parameters were in the acceptable range with the majority being better than acceptable (22). The secondary structure was assigned by PROCHECK and by visual inspection of model: α1,

residues 7-36; α2, residues 39-61; α3, residues 64-75; α4, residues 81-99; α5, residues

102-124; α6, residues 128-149; α7, residues 155-163; α8, residues 169-192; α9, residues

201-204; α10, residues 209-223; β1, residues 230-235; β2, residues 239-245; β3,

residues 248-253; α11, residues 261-278; α12, residues 282-285; α13, residues 295-306;

α14, residues 313-320; α15, residues 323-332; α16, residues 335-343; α17, residues

352-354; α18, residues 359-375; α19, residues 380-395; α20, residues 401-404; α21,

residues 411-414; α22, residues 419-441; α23, residues 443-450; α24, residues 454-466;

α25, residues 473-481; α26, residues 488-499.

3.2.5 Substrate modeling

The probable binding-site for the peptide substrate was modeled by comparing the

structure of PfuCP with those of transition-state analogs bound to astacin and

thermolysin, PDB codes 1QJI and 4TMN, respectively (10, 23). The modeling strategy

involved superimposing the active site HEXXH motifs of these proteins with that of

PfuCP using the program O (20) followed by the placement of a 10-mer polyalanine

substrate to the active site groove with the scissile carbonyl oxygen of the penultimate

alanine in proper distance and orientation to coordinate with the active site metal. The N-

48 terminus of the polyalanine model was extended in an anti-parallel orientation to strand

β2 of the beta sheet near the active site in a similar orientation to that of the reference structures.

3.3 Results and discussions

3.3.1 The overall protein fold

Three different PfuCP structures have been determined from two different crystal forms. In all, the overall protein fold is primarily helical, except for a 3-stranded β-sheet near each active site (α-helix: 68%, β-strand: 3.2%). In each case, as predicted from the biochemical studies (6), the structure appears as a dimer with an overall shape reminiscent of a butterfly (Figure 3.1). The dimensions of the dimer are 100 Å x 65 Å x

55 Å with an overall surface area of 34,800 Å2. A total of 4400 Å2 of surface area is buried between the dimer interface (24). Data collection, phasing, and refinement statistics are summarized in Table 3.1.

The dimer interface is primarily formed between the two N-terminal helices (α1 and α2) from each monomer that come together to form a stack of four helices. The residues that form this interface are primarily hydrophobic (A22, I23, A26, V29, L30,

P39, G42, V48, A49, L53, F286), although two pairs of salt bridges between R19--E59 and D33--R250 contribute to the dimer interaction. Some support for the importance of these residues in dimerization is provided by its sequence alignment with TaqCP (Figure

3.2). Unlike PfuCP, TaqCP exists as a monomer in solution (3). The alignment reveals

49 several sequence differences at the dimer interface that likely contribute to the differences in the aggregation state, such as the substitution of some of the hydrophobic residues with more hydrophilic counterparts and loss of one of the salt bridges in TaqCP.

Each of the two PfuCP subunits that forms the dimer is essentially identical (root mean square deviation of Cα atoms = 0.473 Å) having dimensions of 75 Å x 55 Å x 55

Å. Their distinguishing feature is a deep groove that transverses the length of the protein

(Figure 3.3). This groove is approximately 40 Å long and 30 Å deep at its lowest point.

The groove is widest at the open end measuring 15 Å (bottom of Figure 3.3). The other end is closed with a helical cap formed by helix α4 (top of Figure 3.3). Each subunit can be divided into two domains, a left domain (L) and a larger right domain (R) (Figure 3.3), based on its position relative to this substrate groove.

The first two N-terminal residues were not found and are likely disordered. The first secondary structure begins with a collection of three α−helices (α1-3, residues 7-75) that compose approximately one-half of the groove wall in the R domain. The following

α−helix (α4, residues 81-99) crosses over from the R domain to the L domain towards the top of the molecule acting as a cap to the top-end of the groove. The groove wall of the L domain is made up mostly of the next three helices (α5-7, residues 102-163). A long α−helix (α8, residues 169-192) forms the underside of the groove wall of the L domain before crossing back to the R domain. The rest of the right groove wall is composed of a helix followed by a three-stranded β-sheet (α10, β1-3, residues 209-253).

The next set of α−helices (α11-13) forms the active site. The first (α11, residues 261-

278) contains the HEXXH motif with its two histidines (H269 and H273) coordinated to

50 the active site metal and the last (α13, residues 295-306) contains the glutamate residue

(E299) that serves as the third protein ligand. Following these active site α−helices, the chain passes across the groove bottom in the form of three α−helices (α14-16, residues

313-343) and then backs again to the left wall of the groove, which is completed by five

α−helices (α17-21, residues 352-414). The C-terminal end of the polypeptide finishes with a long α−helix (α22, residues 419-441) that extends from the left wall to the bottom of the groove at which point the remaining α−helices (α23-26, residues 443-499) complete the groove floor below the active site. Based on the distance between the two active sites (~50 Å), the two subunits are believed to act independently.

3.3.2 The protein active site

The HEXXH motif is a common motif found in numerous metalloproteases including thermolysin, astacin, deformylase, and the M1 aminopeptidases (1, 9). The first identification of this motif in the PfuCP family of carboxypeptidases was in TaqCP from a HEMGH sequence located at positions 276-280 (4). This subsequently led to the placement of TaqCP in the newly formed M32 family of metallopeptidases (1). It was later demonstrated by mutagenesis studies that the two histidines of the HEXXH motif

(H276 and H280) are two of the ligands that bind the catalytic metal ion (5).

PfuCP has a HEFGH sequence at positions 269-273. The location of this motif places the active site at the bottom of the groove near the right wall just underneath the beta sheet of the loop connecting strands β1 and β2. Unexpectedly, this places the active

51 site midway along the groove rather than at its back, a feature more reminiscent of an than an . While current data appear convincing that PfuCP is a carboxypeptidase, a dual function for a specific polypeptide cannot be ruled out.

Consistent with the role of the HEXXH in metal ion binding, the structure of Pb-

PfuCP reveals a Pb ion bound to the two histidines of the HEXXH motif (His 269 and

His 273). The third protein ligand to the metal ion based on the structure is glutamate

E299. Based on the sequence alignment, this residue is conserved for the entire PfuCP family of carboxypeptidases placing them in the MA clan of HEXXH containing metalloproteases (Figure 3.2) (1). In TaqCP this would correspond to E306. Interestingly, a weaker electron density is found at the metal- in the structures of apo PfuCP and the Yb-PfuCP derivative. This is presumed to be from Mg2+, which is found in the crystallization condition. In each of the three structures we have determined, the protein ligands (H269, H273, E299), the catalytic water, and the “proton shuttle” E270 adopt conformations similar to those found in other peptidases containing the HEXXH motif

(25) (Figure 3.4).

While the PfuCP structures clearly determine the location of metal active site, the identity of the native metal itself remains elusive. While most metallopeptidases belonging to the HEXXH superfamily, including TaqCP, use a single Zn+2 ion as the active site metal, Cheng et al. found that PfuCP was activated by Co+2 ion and not zinc

(6). More recently, PfuCP has been reconstituted with Mn+2 with good activity (26).

Interestingly these features are reminiscent of another HEXXH metalloprotease, peptide deformylase, which has been shown to contain ferrous ion as its native metal. Here, the

52 zinc-bound form was shown to be 100 times less active than the native protein containing iron (27) while the cobalt and nickel forms were found to have similar activities as wild type. The origin of the metal-dependent differences in the activities of these enzymes remains unresolved.

3.3.3 Structural similarity with neurolysin

Structure homology searches were performed with TOP, VAST, FUGUE, and

DALI (28-31). One hit was found with DALI and FUGUE (30, 31), revealing that PfuCP has a similar fold to Rattus norvegicus neurolysin (PDB code: 1I1I) (32). This zinc- dependent endopeptidase cleaves neurotensin, a 13-residue neuropeptide, between residues 10 and 11 (32). PfuCP can be superimposed on neurolysin using DALI with a

RMSD of 2.1 Å for the 251 Cα atoms used in the superposition. This structure-based sequence alignment is provided in Figure 3.2. Neurolysin (681 residues) is larger than

PfuCP (499 residues) and has a number of extra structural elements. These include an additional alpha helix at its N-terminus (α1) and a cork-like sub-domain (α7-8 and β1-2) that has the overall effect of narrowing the bottom-end of the substrate groove, as well as elongating the left groove wall (Figure 3.5A). The last major difference between the two structures is that the beta sheet near the active site is composed of five strands in neurolysin while only three for PfuCP. In light of the clear structural similarity of PfuCP and neurolysin, it seems likely that PfuCP and neurolysin evolved from a common ancestor by divergent evolution, much like that predicted for leucine aminopeptidase and

CPA (33).

53 Despite this high structural similarity, however, BLAST showed no sequence similarity between these two proteins (34). Using the structure alignment based on the amino acids used in the structure superposition, we find that the amino acid identity between PfuCP and neurolysin is low (<16%). Even with the low degree of similarity eleven of the conserved residues of the PfuCP family of carboxypeptidases are also conserved in neurolysin (PfuCP/neurolysin: H238/425, H269/474, E270/475, G272/477,

H273/478, E299/503, E306/510, H411/601, G418/608, Y423/613, and G468/650) (Figure

3.5A). Six of these residues are proposed to play an important role in catalysis (Figures

3.5b and 3.6).

3.3.4 Sequence homologs of PfuCP

A BLAST search (34) using the PfuCP sequence identified 14 other members that span all three kingdoms of life - archaea, bacteria, and eukarya (included in the supplemental material). The homology between PfuCP and each of the proteins is fairly high (31-87% identity) and extends over the entire protein (Figure 3.2). This suggests that these sequence homologs share a common fold for which PfuCP provides the structural prototype. The paucity of these enzymes as compared to the much larger family of CPA/CPB homologs is notable. This may suggest a very specific role for these enzymes in biology. In support of this, in many of the organisms that contain a PfuCP homolog, a gene appearing to belong to the much larger family of carboxypeptidases was also identified (organism, PIR code: P. horikoshii, G71097 and C71119; D. radiodurans,

A75364 and C75531; B. halodurans, E83851; P. abyssi, A75042; and B. subtilis,

54 B70076, H69817, and E69640) (8). These genes are homologs of the carboxypeptidase from S. solfataricus (CPS), which is itself a homolog of CPB. For those organisms where no homolog was found, the genome was found to be incomplete.

The sequence alignment of PfuCP and its homologs (Figure 3.2), reveals five conserved amino acid sequences in addition to the characteristic metalloprotease HEXXH motif. Using the PfuCP sequence as a reference, they are: (1) an HPF sequence (residues

238-240) mostly on strand β2; (2) a DXRXT sequence (residues 248-252) on strand β3,

(3) a HESQ sequence (residues 298-301) on helix α13, (4) an IRXXAD sequence

(residues 350-355) on helix α17, and (5) a GXXQDXHW sequence (residues 405-412) located on the loop between helices α20 and α21. Roles for these motifs could be inferred from their location in the PfuCP structure by modeling a 10-mer polyalanine substrate to the active site based the crystal structures of astacin and thermolysin in complex with transition-state analogs (10, 23). The putative functions of all these sequences in regard to the catalysis will be discussed in our proposed mechanism with the exception of the DXRXT sequence, which is far removed from the active site and whose catalytic role, if any, is unclear.

3.3.5 Proposed catalytic mechanism

The mechanism of catalysis of this family of carboxypeptidase is unclear at this time. The fundamental issue is the direction of substrate approach. One possibility is based on the shape of the groove. The groove has opened and closed ends (opened: bottom of Figure 3.3; closed: top of Figure 3.3). One would predict, therefore, that based

55 on the greater ease in accessing the active site that C-terminal end of peptide substrates should approach from the open end. Unfortunately, this possibility disagrees with the direction suggested from homology modeling studies with other HEXXH metalloproteases. In all structures of HEXXH metalloproteases bound to peptides, the peptide binds to the metal site in the opposite orientation. Based on the orientation suggested by this homology model, the C-terminus would feed into the short gap between the closed end of the groove (helical cap α4) and the metal active site. We currently favor the latter model and present this below, but clearly additional studies will be required to confirm this.

All HEXXH proteins reported to date have a similar binding pocket around the active site with the substrate binding in a similar orientation relative to the HEXXH motif. Indeed, the protein ligands, H269 and H273 of the HEXXH motif and E299 of the conserved HESQ signature sequence, adopt conformations similar to those found in other

HEXXH-containing . Based on this homology, a 10-mer polyalanine oligopeptide was modeled to the PfuCP active site using the locations of inhibitors bound in structures of astacin and thermolysin transition-state inhibitor complexes as a guide

(10, 23). The oligopeptide was modeled as an extended β-strand that interacts with the

β2 strand of the PfuCP protein in an anti-parallel orientation forming a ladder of four hydrogen bonds. This has the effect of expanding the β-sheet near the active site by an additional strand, as has been observed in astacin and thermolysin (10, 23).

Using this orientation, a mechanism can be proposed for PfuCP that is similar to other HEXXH proteins (Figure 3.6) (10, 23). Here, the metal serves to stabilize a bound

56 water/hydroxide and/or activate the scissile carbonyl of the substrate by serving as a

Lewis acid. The active site glutamate, E270 of the HEXXH motif, assists in the

nucleophilic attack of the activated water/hydroxide on the carbonyl to form the

tetrahedral intermediate, by acting as a general acid/base that can shuttle the hydrogen

atom from the activated water to the scissile amide nitrogen of the substrate. Protonation

of this amide nitrogen makes it a better leaving group, thereby facilitating cleavage of the

amide bond.

The H238 and P239 of the HPF signature sequence may play a role in orienting

the amide nitrogen to accept this hydrogen and in stabilizing the proper transition-state.

The carbonyl oxygen of P239 on strand β2 could hydrogen-bond the scissile N-H of the

substrate to position it for protonation by the proton shuttle E270, as does its counterparts

in other HEXXH metalloproteases (9, 10, 23). Upon protonation, H238 could serve a role in transition-state stabilization. Modeling studies reveal that H238 can be positioned to serve as hydrogen bond acceptor that would stabilize the tetrahedral transition-state.

H238 is conserved for all members of the PfuCP family of carboxypeptidases as well as neurolysin (H425) (Figure 3.5). Consistent with such a role, in thermolysin, an asparagine (N112) is suggested to serve a similar role and is located at nearly the same position (10). It should be noted that modeling studies also reveal that H238 can be positioned to hydrogen bond the C-terminal carboxylate. Its presence in the endopeptidase neurolysin, however, argues against the importance of this interaction.

Y423 is another residue that could play a role in both positioning the C-terminal carboxylate and stabilizing the tetrahedral intermediate. This residue is conserved for the

57 entire family of PfuCP carboxypeptidases as well as neurolysin (Y613) (Figure 3.5), and

is located at a similar position in the active site as residues Y149 and H231 of astacin and

thermolysin, respectively. These residues have been suggested to stabilize the negative

charge of the oxyanion intermediate by hydrogen bonding one of the coordinating

oxygens (10, 23). In PfuCP, this can occur by reorientation of χ1 torsional angle of Y423

(Figure 3.6). The OH hydrogen of the phenolic side chain can also be positioned to

hydrogen bond the C-terminal carboxylate of the modeled substrate.

The IRXXAD and GXXQDXHW signature sequences are proposed work

together in concert to bind the substrate C-terminal carboxylate and promote release of

the cleaved amino acid product. In the absence of substrate, R351 and D409 from the

IRXXAD and GXXQDXHW motifs, respectively, form an intra-subunit salt bridge.

Modeling studies indicate, however, that R351 can be repositioned to form a salt bridge

with the C-terminal carboxylate of the bound substrate by a minor adjustment of its χ1 and χ2 torsional angles. This brings R351 within 3.1 Å of the C-terminal carboxylate.

We propose that R351 serves a role in stabilizing substrate binding, but that following

hydrolysis, it returns to reform its salt bridge with D409 releasing the C-terminal amino

acid (Figure 3.6).

3.4 Conclusion

Carboxypeptidases catalyze the C-terminal hydrolysis of polypeptides. While the

most ubiquitous are those with homology to the metalloprotease, carboxypeptidase A

(CPA), a new class of metallocarboxypeptidase has been identified that exhibits no

58 sequence similarity to CPA. This class contains a HEXXH motif characteristic of aminopeptidases and endopeptidases. Several members of this new family of carboxypeptidases originate from hyperthermophilic organisms. These carboxypeptidases, including the Pyrococcus furiosus carboxypeptidase (PfuCP), are of particular interest due to their thermostability and their potential for use in C-terminal

MALDI/MS ladder sequencing.

The structure of PfuCP has been determined to 2.2 Å resolution using MAD methods. The overall structure is comprised of a homodimer. Each subunit is mostly helical with its most pronounced feature being a deep substrate-binding groove. The active site lies at the bottom of this groove and contains an HEXXH motif that coordinates the metal ion required for catalysis. Surprisingly, the structure is similar to the recently reported rat neurolysin. Comparison of these structures as well as sequence analyses with other homologous proteins reveal several conserved residues. Based on their locations, roles for these conserved residues in the catalytic mechanism are inferred.

The structure of PfuCP represents the first structure of this new family of carboxypeptidases. It allows for the characterization of its metal-binding site, permits its structural comparison to neurolysin, and facilitates the identification of the conserved residues involved in catalysis. Based on the structural features of its active site, we suggest a mechanism for PfuCP in C-terminal peptide hydrolysis.

59

Table 3.1. Crystallographic data for PfuCP.

60 61

Figure 3.1. Ribbons representation of PfuCP homodimer in stereo (β-strands red, α- helices cyan, and loops green) with the active site metals in pink. This figure was prepared using the programs MOLSCRIPT and Raster-3D (35, 36).

62

Figure 3.2. Sequence alignment of the M32 family of metallocarboxypeptidases. Sequences of PfuCP and 8 other carboxypeptidases of the M32 family were aligned with the program ClustalW (37). This is followed by the structure-based alignment of rat neurolysin using the residues used in the superposition (29). The numbering and secondary structure above the alignment corresponds to that of PfuCP. Residues that are conserved between all the homologs are in bold and highlighted in yellow. Motifs are indicated with colored lettering as follows: HPF, red; DXRXT, green; HEXXH, blue; HESQ, brown; IRXXAD, violet; and GXXQDXHW, orange. Accession numbers with percent identity to PfuCP: B. subtilus, p50848, 36%; T. aquaticus, p42663, 37%; Halobacterium sp., q9hmb7, 34%; D. radiodurans, q9rrr3, 33%; R. rickettsii, cac33677, 30%; L. major, cac37108, 33%; V. cholerae, q9ks46, 31%; and R. loti, bab50714, 32%.

63 64 65

Figure 3.3. PfuCP monomer structrures. (A) Ribbons diagram of PfuCP monomer directed towards substrate groove in stereo with the active site metal in pink; (B) surface diagram of PfuCP in stereo (negatively charged residues are in red, positively charged residues are in blue) modeled with 10mer polyalanine substrate; and (C) a rainbow stereo th plot of the Cα trace of PfuCP, every 20 residue is labeled with the N-terminus in blue, the C-terminus in red, and active site metal in pink. Figures were prepared using the programs MOLSCRIPT, Raster-3D, and GRASP (24, 35, 36).

66

Figure 3.3. PfuCP monomer structrures.

67

Figure 3.4. Electron density of active site. A 2.3Å 2Fo-Fc electron density map (1.0σ, blue and 4σ, magenta) superimposed on the final active site model of PfuCP (Yb phased crystal). Carbon atoms are colored yellow; nitrogen, cyan; oxygen, red; and active site metal, green (modeled as Mg+2). Figure was prepared with XtalView and Raster-3D (15, 35).

68

Figure 3.5. Structural comparison with neurolysin. (A) Backbone trace of the overall structures of PfuCP in blue superimposed on neurolysin in red with conserved residues and active site metals (PDB code: 1I1I). (B) Active site comparison of PfuCP and neurolysin with proposed catalytic residues. Figures were prepared using the programs MOLSCRIPT and Raster-3D (35, 36).

69

Figure 3.6. Proposed catalytic mechanism. Schematic diagram of a possible catalytic mechanism for peptide binding and hydrolysis.

70

Chapter 3 References

1. Rawlings, N. D., and Barrett, A. J. (1995) Methods Enzymol. 248, 183-228.

2. Stepanov, V. M. (1995) Methods Enzymol. 248, 675-683.

3. Lee, S. H., Minagawa, E., Taguchi, H., Matsuzawa, H., Ohta, T., Kaminogawa, S., and Yamauchi, K. (1992) Biosci. Biotechnol. Biochem. 56, 1839-1844.

4. Lee, S. H., Taguchi, H., Yoshimura, E., Minagawa, E., Kaminogawa, S., Ohta, T., and Matsuzawa, H. (1994) Biosci. Biotechnol. Biochem. 58, 1490-1495.

5. Lee, S. H., Taguchi, H., Yoshimura, E., Minagawa, E., Kaminogawa, S., Ohta, T., and Matsuzawa, H. (1996) Protein Eng. 9, 467-469.

6. Cheng, T. C., Ramakrishnan, V., and Chan, S. I. (1999) Protein Sci. 8, 2474- 2486.

7. Ishikawa, K., Ishida, H., Matsui, I., Kawarabayasi, Y., and Kikuchi, H. (2001) Appl. Environ. Microbiol. 67, 673-679.

8. Colombo, S., Dauria, S., Fusi, P., Zecca, L., Raia, C. A., and Tortora, P. (1992) Euro. J. Biochem. 206, 349-357.

9. Chan, M. K., Gong, W. M., Rajagopalan, P. T. R., Hao, B., Tsai, C. M., and Pei, D. H. (1997) Biochemistry 36, 13904-13909.

10. Holden, H. M., Tronrud, D. E., Monzingo, A. F., Weaver, L. H., and Matthews, B. W. (1987) Biochemistry 26, 8542-8553.

11. Jancarik, J., and Kim, S. H. (1991) J. Appl. Crystallogr. 24, 409-411.

12. McDonald, N. Q., Panayotatos, N., and Hendrickson, W. A. (1995) EMBO J. 14, 2689-2699.

71 13. Teng, T. Y. (1990) J. Appl. Crystallogr. 23, 387-391.

14. Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, 307-326.

15. McRee, D. E. (1999) J. Struct. Biol. 125, 156-165.

16. Furey, W., and Swaminathan, S. (1997) Methods Enzymol. 277, 590-620590-&.

17. Kleywegt, G. J., and Jones, T. A. (1996) Acta Crystallogr. D52, 826-828.

18. Kleywegt, G. J., and Jones, T. A. (1999) Acta Crystallogr. D55, 941-944.

19. 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., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Acta Crystallogr. D54, 905-921.

20. Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M. (1991) Acta Crystallogr. A 47, 110-119.

21. Brunger, A. T. (1993) Acta Crystallogr. D49, 24-36.

22. Laskowski, R. A., Macarthur, M. W., Moss, D. S., and Thornton, J. M. (1993) J. Appl. Crystallogr. 26, 283-291.

23. Grams, F., Dive, V., Yiotakis, A., Yiallouros, I., Vassiliou, S., Zwilling, R., Bode, W., and Stocker, W. (1996) Nat. Struct. Biol. 3, 671-675.

24. Nicholls, A., Bharadwaj, R., and Honig, B. (1993) Biophys. J. 64, A166-A166.

25. Monzingo, A. F., and Matthews, B. W. (1984) Biochemistry 23, 5724-5729.

26. Cheng, T. C. (2003) Dissertation.

27. Rajagopalan, P. T. R., Yu, X. C., and Pei, D. H. (1997) J. Am. Chem. Soc. 119, 12418-12419.

28. Lu, G. G. (2000) J. Appl. Crystallogr. 33, 176-183.

29. Gibrat, J. F., Madej, T., and Bryant, S. H. (1996) Curr. Opin. Struct. Biol. 6, 377- 385.

30. Holm, L., and Sander, C. (1998) Nucleic Acids Res. 26, 316-319.

72 31. Shi, J. Y., Blundell, T. L., and Mizuguchi, K. (2001) J. Mol. Biol. 310, 243-257.

32. Brown, C. K., Madauss, K., Lian, W., Beck, M. R., Tolbert, W. D., and Rodgers, D. W. (2001) Proc. Natl. Acad. Sci. U. S. A. 98, 3127-3132.

33. Artymiuk, P. J., Grindley, H. M., Park, J. E., Rice, D. W., and Willett, P. (1992) FEBS Lett. 303, 48-52.

34. Madden, T. L., Tatusov, R. L., and Zhang, J. H. (1996) Methods Enzymol. 266, 131-141.

35. Merritt, E. A., and Murphy, M. E. P. (1994) Acta Crystallogr. D50, 869-873.

36. Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 946-950.

37. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994) Nucleic Acids Res. 22, 4673-4680.

73

CHAPTER 4

Isolation, characterization, and crystallization of the acetyl coenzyme-A

decarbonylase/synthase (ACDS) complex and its comprising subunits

4.1 Introduction

Methanogens are organisms that release methane as a byproduct of their

anaerobic respiration. While all methanogens can grow on CO2, some methanogens such as Methanosarcina barkeri can grow exclusively on acetate as the sole energy source (1).

These organisms have a significant impact on the environment, since the approximately

109 tons of methane produced annually are thought to result from this methanogenic growth on acetate (2).

The protein complex that initiates the conversion of acetate into methane and CO2 is the acetyl coenzyme-A (acetyl-CoA) decarbonylase/synthase (ACDS) (3). This multi- subunit complex catalyzes the reversible degradation of acetyl-CoA into CO2 and CoA as

indicated in equation [1], where CH3-H4SPt is methyl-tetrahydrosarcinapterin and H4SPt is tetrahydrosarcinapterin (4, 5).

+ - [1] CH3(CO)-CoA + H4SPt + H2O ↔ CO2 + 2H + 2e + CH3-H4SPt + CoA

74 The M. barkeri ACDS complex is comprised of the five different subunits with

subunit molecular weights (in kDa) of 88, 53, 47, 51, and 19, respectively. The mass of

the complex (2.1 MDa) suggests an (αβγδε)8 oligomer (5-7). This multimeric complex

can be dissociated into three components by partial proteolysis, a β component, a α2ε2 complex, and a γδ complex (5). Each of these components exhibits one of the fundamental activities of the whole complex (5, 8, 9).

The first reaction involving the cleavage of the acetate C-C bond in to a methyl group and CO has been localized to Ni-Fe-containing cluster, known as the A-center, located in the β subunit (5, 6, 10). This reversible acetyl transfer and cleavage activity is in the opposite direction of the acetyl-CoA synthase (ACS) of acetogens and is commonly referred to in the reverse direction. The A-center also catalyzes the transfer of the methyl group to a corrinoid protein yielding the reaction shown in equation [2].

[2] CH3(CO)-CoA + corrinoid (I) protein ↔ CH3-corrinoid (III) protein + CoA + CO

(A-center)

The second reaction is the reversible oxidation of CO to CO2 that is known as

carbon monoxide dehydrogenase (CODH) activity as indicated in equation [3]. The

CODH activity has been localized at α and ε subunits and a stable heterotetramer α2ε2 component can be isolated and purified (11). This reaction occurs at a second Ni-Fe- containing cluster, known as the C-center, located in the α subunit.

+ - [3] CO + H2O ↔ CO2 + 2H + 2e (C-center)

The reversible methylcobalamide/tetrahydropterin methyltransferase activity

(equation 4) is localized in the δ and γ subunits, which consist of a corrinoid protein and

75 an iron-sulfur containing protein (FeSP) (5, 8). The corrinoid cofactor is believed to be associated with the δ subunit while a Fe4S4 is proposed to be with the γ subunit (12).

+ [4] CH3-corrinoid (III) protein + H4SPt ↔ corrinoid (I) protein + CH3-H4SPt + H

Interest in the ACDS complex stems in part from the fact that its reaction involves organometallic intermediates that are formed at these metal clusters (8, 12). Particular interest is focused on its A- and C-clusters that possess unusual Ni-Fe-S centers whose structures were until recently unknown.

While structures of the methanogenic ACDS complex are still not available some progress has been made with the CODH/ACS enzymes from acetogenic bacteria. In these organisms, however, CODH/ACS is used to convert CO2 to acetyl-CoA, the reaction in this direction is known as the Wood-Ljungdahl carbon fixation pathway (13).

Regardless of the source, these nickel enzymes are evolutionarily related, although they often vary in respect of their metabolic role, subunit content, and catalytic activity (14).

Based on the activities, these enzymes have been divided into two major classes: the monofunctional variety that only possess CODH activity, and the bifunctional class that contain both ACS and CODH activity (8). ACS/CODH-like enzymes have been the focus of much study since they are found only in primitive anaerobic organisms (8, 12-

14). Thus they have been implicated as participating in the origins of life since the prehistoric atmosphere is believed to have been rich in CO and CO2 and deficient in oxygen (15). Indeed they play a critical role in the global carbon cycle and the degradation of environmental pollutants (16, 17).

76 Two structural prototypes are available of the monofunctional CODHs from the

phototropic Rhodospirillum rubrum and the hydrogenogenic Carboxydothermus

hydrogenoformans (18, 19). The structures of these two CODHs are very similar in

terms of the their overall fold and the arrangement of their five metal centers, called B,

B’, C, C’, and D (Figure 4.1A). The homodimer structures are mostly helical with a

butterfly-like shape with the D-cluster found at the subunit interface. Each subunit is

composed of three domains, the N-terminal domain, and the middle and C-terminal and

Rossman-like domains. Notably, the two structure differ in respect to their active site C-

clusters, which are modeled as the Drennan’s [Ni-Fe4S4] and the Dobbeks’s [Ni-Fe4S5] clusters (Figure 4.2A, B).

Two bifunctional CODH/ACS structures have also recently been determined, both of which are from the acetogenic bacterium Moorella thermoacetica (formerly called

Clostridium thermoaceticum) (20, 21). The structures are virtually identical and reveal a

αββα heterotetramer with an overall shape reminiscent of a bow tie with the CODH

activity residing on the core β subunits and the ACS activity on the flanking α subunits

(Figure 4.1B). The CODH components are similar to the monofunctional CODHs

described above. The ACS component is comprised of three domains, as indicated in

Figure 4.1C (21). The N-terminal domain I (residues 1-314) contains a Rossman-like

fold and a hydrophobic channel that leads to the A-cluster found in domain III (residues

495-729), while the function of domain II (residues 315-494) is not known. Differences

were once again found at the active site metal clusters not only at the C-cluster of the

77 CODH component, but also at the A-cluster of the ACS component. The Darnault

structure reveals a third model for the C-cluster and a [Ni-Ni-Fe4S4] A-cluster, while the

Doukov structure shows a [Ni-Cu-Fe4S4] A-cluster (Figure 4.2C,E,D, respectively).

Structural studies were initiated on the M. barkeri ACDS complex in an effort to

understand the mechanisms of acetyl-CoA degradation and CO/CO2 oxidation in

methanogens, and to reveal the atomic structures of the A- and C-clusters. Three parallel

approaches were chosen in this effort. First was crystallization of the entire complex,

second was isolation and crystallization of the individual components of the complex,

and third was crystallization of cloned components from related complexes from different

organisms. Since there are no methanogen structural prototypes yet available these

studies are of particular interest. To date, four different crystal forms of M. barkeri α2ε2 have been identified, however, due to inherent problems with cryo-cooling and mosaicity, its structure determination is still in progress. In addition, for the first time the M. barkeri components of β, γ, and δ have each been individually isolated. Crystallization screening is still in progress, though the first crystals of the β component have been found.

4.2 Materials and methods

4.2.1 ACDS complex from M. barkeri

4.2.1.1 Preparation of cells and cell extracts

M. barkeri strain MS (DSM 800) cells were prepared and donated by the laboratory of Dr. Krzycki (11). Cell culturing was performed under anaerobic conditions in 40 L flasks in a phosphate-buffered medium supplemented with 80 mM sodium

78 acetate. The cells were harvested anaerobically after 14 days inoculation by

centrifugation, and washed three times with 50 mM MOPS buffer, pH 7.0, and 1 mM

sodium dithionite, and stored at -80 ˚C.

Prior to preparation of the cell extracts, the clumps of frozen cells (~350 g each

prep) were first made into a fine powder by suspending the frozen cells with liquid

nitrogen and then chopping in a blender for approximately 30 seconds. The fine slurry of

cells were transferred to a 2 L sidearm flask and made anaerobic by repeated cycles of

evacuation and flushing with N2 for 120 minutes or until completely thawed.

Cell extracts were prepared by suspending the degassed cells into 50 mM Bis-Tris

buffer pH 7.0, 1 mM dithiolthreitol (DTT), and 1 mM sodium dithionite (1.8 grams of

frozen cell per 1 mL of buffer). Cells were lysed with a French pressure cell (Thermo

Spectronic, 40K cell) at an operating pressure of 25,000 psi. The cell lysate was

collected anaerobically in a stoppered bottle under nitrogen. The lysate was clarified by

ultracentrifugation in a 60-ti Beckman rotor at (45K rpm, 160,000 g) for 2 hours at 4 oC.

The same procedure was used for the preparation of the cell extracts for the isolation of the individual subunits of the ACDS complex.

4.2.1.2 Purification of the M. barkeri ACDS complex

Purification of the M. barkeri ACDS complex required strict anaerobic conditions. All steps were performed in an anaerobic glove bag containing 98% N2 and

2% H2 (Coy Laboratories Products Inc., Grass Lake, MI) at room temperature. All

chromatography columns and solutions were made anaerobic by repeated cycles of

79 evacuation and flushing with purified N2 (passed through a column of copper-containing

BASF catalyst R3-11 to remove trace O2). Sodium dithionite (1 mM) and DTT (1 mM) were added to the buffers prior to use. These strict precautions were also followed during the anaerobic purification of the individual components of the ACDS complex.

The ACDS complex was isolated from M. barkeri cells following a previously published procedure with modifications (22). It involved the chromatography columns of

Superose 6B and Mono Q. Since limitations to this first chromatographic step are the large volume and protein content in the cell extract, a crucial ultrafiltration step was implemented to reduce the volume and to remove the low molecular weight proteins prior to running the column. In the procedure, cell extract was concentrated 10X using an

Amicon concentrator equipped with a 500 kDa polysulfone membrane. The concentrated cell extract was loaded onto a 500 mL bed of Superose 6B prep grade column (2.6 x 95 cm) pre-equilibrated in 50 mM Bis-Tris buffer, pH 7.0, containing 100 mM NaCl. Care was taken so that the volume of the extract loaded did not exceed 3% of the bed volume.

Once loaded, the protein was eluted with 600 mL of equilibration buffer at a flow rate of

1 mL/min. Protein fractions were analyzed by SDS-PAGE. Fractions containing the

ACDS complex were combined and loaded onto a Mono-Q HR 10/10 column pre- equilibrated with 100 mM NaCl in 50 mM Bis-Tris buffer pH 6.5. The sample loaded on the column should not exceed 50 mg protein. The protein was eluted using a 200 mL linear gradient of 100-500 mM NaCl at a flow rate of 1 mL/min. The resulting protein was concentrated (90-100 mg/mL) and stored at -80 ˚C. All protein concentrations were determined by the Lowry method using bovine serum albumin as a standard (23).

80 The chromatographic resolution after each column (Figures 4.3, 4.4) was checked

by SDS-PAGE (data not shown). The ACDS complex isolation was performed in triplicate with total of 673 mg of pure ACDS obtained from 690 grams of cells. The final protein samples were checked by SDS-PAGE and dynamic light scattering (DLS), as shown in

Figure 4.5. The SDS-PAGE gel shows a complex consisting of five subunits. The DLS results suggested that the protein was monodispersive based on the low baseline (1.004),

SOS error (0.739), and the narrow peak of polydispersity (3.0%), as indicated by the percent mean radius (M.R.). The average molecular mass, 1.93 MDa, is consistent with previous studies that have suggested that the ACDS complex is an (αβγδε)8 oligomer.

4.2.1.3 Crystallization of the ACDS complex

Unless otherwise noted, all crystallization attempts of the ACDS complex and its

subunits were performed in a Coy anaerobic chamber at room temperature using

precipitating agents that were made anaerobic by repeated cycles of evacuation and

flushing with purified N2, followed by the addition of enough sodium dithionite to yield a

final concentration of 1 mM. Three different crystallization methods were used with the

ACDS complex: batch capillary, and vapor diffusion by hanging drop and sitting drop

methods. For the batch capillary method, a protein sample (90-100 mg/mL, 50 mM Bis-

Tris, pH 6.5, 450 mM NaCl, 7-10 µL) was first injected into the capillary (KIMAX, 1.5-

1.8 x 90 mm) followed by 15 µL of precipitating agent that was carefully injected on top of the protein solution.

81 The hanging drop method was performed using a 24 well tray (VDX Plate,

Hampton). The protein sample (1 µL) was placed on a siliconized cover slip (22 mm square, Hampton) followed by 1 µL of from the reservoir solution (0.5–1.0 mL). Each well was sealed and then allowed to equilibrate over time (24).

The sitting drop method was performed on a 24 well plate (Cryschem Plate,

Hampton) filled with different reservoir solutions (0.5–1.0 mL). The protein sample (2

µL) was placed on the sitting drop post followed by 2-3 µL of reservoir solution. The tray was sealed with sealing tape (Crystal Clear, Manco Inc.) and allowed to equilibrate over time.

Crystals of M. barkeri ACDS complex have yet to be identified. Initial crystal screens suggest that the most promising conditions are those using polyethylene glycol

(PEG) as a precipitant and a pH range of 6.5-8.5. The fact that no crystals have been obtained could be due to the large molecular weight of the complex and its possible dissociation under the crystallization conditions.

4.2.2 The CODH α2ε2 component of the ACDS complex from M. barkeri

4.2.2.1 Purification of α2ε2 CODH component

The α2ε2 CODH subunit was purified following previously published procedures with only minor modification (11, 25). The α2ε2 was isolated directly from M. barkeri cells by anaerobic purification involving DEAE-Sepharose, HAP, HiLoad Q, and Mono

Q columns. The cell extract (section 4.2.1.1) was loaded on to a DEAE-FF Sepharose column (5.0 x 35 cm) pre-equilibrated with 100 mM NaCl in 50 mM Bis-Tris buffer, pH

82 7.0. The column was eluted with a 2800 mL linear gradient of 100-500 mM NaCl at a

flow rate of 3 mL/min. Fractions containing the most of the CODH activity were pooled,

while fractions containing crude ACDS complex, as identified by SDS-PAGE and CODH

activity, were combined in a separate pool. The crude ACDS pool was used for later

processing of the individual subunits (section 4.2.3). The CODH pool was concentrated

by Amicon (30K, regenerated cellulose membrane) 3X and desalted with a HiPrep

Sephadex G-25 desalting column (2.6 x 10 cm) pre-equilibrated with 1 mM KH2PO4 buffer, pH 6.8. The desalted CODH pool was immediately applied to a hydroxyapatite

(HAP) (macro-prep type I, Bio-Rad) column (2.6 x 18 cm) that had been pre-equilibrated with 1 mM KH2PO4 buffer, pH 6.8. The sample loaded on the column should not exceed

3 mg protein/mL of HAP matrix. The sample on the HiLoad Q column was eluted with a

360 mL linear gradient of 1-500 mM KH2PO4 at a flow rate of 2 mL/min. The active

fractions were combined and immediately loaded to a HiLoad Q column (2.6 x 13 cm)

that had been pre-equilibrated with 50 mM Bis-Tris buffer, pH 7.0. The column was

eluted with a 520 mL linear gradient of 0-500 mM NaCl at a flow rate of 2 mL/min. The

active fractions were combined, diluted with an equal volume of the column equilibration

buffer, and applied to a Mono-Q HR 10/10 column pre-equilibrated with 50 mM Bis-Tris

buffer, pH 7.0. Here, the sample load should not exceed 50 mg protein. Elution was

performed using a 200 mL linear gradient of 100-500 mM NaCl at a flow rate of 1

mL/min. The protein was concentrated (100 mg/mL) and stored at -80 ˚C. Protein purity

was checked by SDS-PAGE and DLS.

83 The chromatographic resolution after each chromatographic column (Figures 4.6-

4.9) was checked by SDS-PAGE (Figure 4.10). The isolation was performed in triplicate

yielding a total of 514 mg of pure active CODH protein from 850 grams cells. The final

α2ε2 protein samples were checked by SDS-PAGE and DLS (Figure 4.10). The SDS-

PAGE gel shows a complex consisting of both α and ε subunits. It is believed that α

subunit contains the C-cluster and can catalyze oxidation of CO to CO2, however, the

function of ε component remains unknown. The DLS results indicate that the isolated

protein is monodispersive based on its low baseline (1.000), M.R. (10.0%), and SOS

error (0.283) with an average molecular mass around 163 kDa. This approximately

agrees with the theoretical molecular weight of the α2ε2 dimer.

4.2.2.2 Activity assay of α2ε2 CODH component

The CODH activity was measured using a CO oxidation assay that monitors the

colorimetric reduction of methyl viologen at 578 nM over time. All reactions were

performed under N2 as described previously (11, 26). Briefly, disposable cuvettes were

sealed with a rubber stopper in the anaerobic chamber and then flushed and evacuated

with N2. The reaction mixture contained 50 mM NaH2PO4 buffer (pH 7.0) and 5 mM

methyl viologen (Sigma). The reaction mixture (1 mL) and 0.4 mL of CO were then

added to the cuvette sequentially and allowed to equilibrate at room temperature for

about 15 minutes. The reaction mixture was then titrated with sodium dithionite solution

until the methyl viologen turned light blue confirming the absence of oxygen. The

84 reaction was initiated by adding the enzyme. The methyl viologen reduction rate

(increase of blue color) was monitored at 578 nM using a Beckman DU-70

spectrophotometer for 30 seconds.

4.2.2.3 Crystallization and data collection of α2ε2 CODH component

The crystallization of the α2ε2 was performed using three different crystallization

techniques: batch capillary, and hanging drop and sitting drop vapor diffusion methods,

as described above. Likewise, the crystallization was also performed at 4o C in an

inflatable glove chamber (I2R Glove Bag, Instruments for Research and Industry) under a

nitrogen atmosphere by the batch capillary method and by hanging drop vapor diffusion.

For the batch capillary method, 7-10 µL of the protein sample (50-100 mg/mL, 50 mM

Bis-Tris, pH 7.0, 150-325 mM NaCl) was injected into the capillary, followed by 15 µL of precipitating agent. A variation of this method was performed in which a 2-3 µL silica plug (Hampton) was added between the protein solution and the precipitating agent. For crystallization by the hanging drop method, 1-2 µL protein sample (20-100 mg/mL) was mixed with 2-3 µL of precipitating agent on a cover slip prior to sealing. For the sitting drop method, 2-5 µL of protein sample (50-100 mg/mL) was mixed with 2-3 µL of precipitating agent.

Prior to data collection, the crystals were flash cooled in liquid nitrogen by one of four methods: [1] direct freezing from mother liquors without cryo-protectant; [2] transferring the crystal to Paratone-N (Hampton); [3] stepwise transferring the crystal to mother liquors with increasing concentrations of glycerol (final 25%); and [4] transfer of

85 the crystal to mother liquors with saturated Li2SO4. The diffraction data for the different crystal forms was collected at 100 K at the beamline 9-2 at the Stanford Synchrotron

Radiation Laboratory (SSRL) and at the beamline 17-ID at the Industrial Macromolecular

Crystallography Association (IMCA) at the Advanced Photon Source (APS) at Argonne

National Laboratory. In addition, crystals were screened at our home source, consisting of a Rigaku/MSC rotating anode generator equipped with an R-AXIS IV++ area detector.

4.2.3 The acetyl-CoA synthase β component of the ACDS complex from M. barkeri

4.2.3.1 Purification acetyl-CoA synthase β component

The β component of the ACDS complex was isolated from M. barkeri cells under anaerobic conditions by multistep chromatography involving DEAE-Sepharose, HiLoad

Q, HAP, and Mono Q columns. First, the crude ACDS-DEAE pool from the α2ε2 purification (section 4.2.2.1) was diluted with 50 mM Bis-Tris buffer, pH 7.0, and loaded onto a HiLoad Q column (2.6 x 13 cm) that had been pre-equilibrated with 50 mM Bis-

Tris buffer, pH 7.0. The column was eluted with a 520 mL linear gradient of 0-500 mM

NaCl at a flow rate of 2 mL/min. Fractions were analyzed by SDS-PAGE and those fractions that were enriched in M. barkeri β, δ, and γ subunits were pooled accordingly.

The δ and γ subunit pools were further processed to obtain their individual components

(sections 4.2.4 and 4.2.5, respectively). The M. barkeri β pool was desalted with a

HiPrep Sephadex G-25 desalting column (2.6 x 10 cm) pre-equilibrated with 1 mM

KH2PO4 buffer, pH 6.8. The desalted M. barkeri β pool was immediately applied to a

HAP column (1 x 6.4 cm) that had been pre-equilibrated in 1 mM KH2PO4 buffer, pH

86 6.8. The loaded protein was eluted with a 50 mL linear gradient of 1-500 mM KH2PO4 at a flow rate of 0.5 mL/min. Fractions were analyzed by SDS-PAGE and those enriched in

M. barkeri β were combined. The HAP β pool was applied to a Mono-Q HR 10/10 column pre-equilibrated with 50 mM Bis-Tris buffer, pH 7.0. Elution was performed using a 200 mL linear gradient of 100-500 mM NaCl at a flow rate of 1 mL/min. The final purified protein was concentrated (50 mg/mL) and stored at -80 ˚C.

The chromatographic resolution after each chromatographic column was checked by SDS-PAGE (Figures 4.11-4.13). Two separate isolations were performed and total of

17 mg of pure M. barkeri β protein was obtained from 850 grams of cells. The purity and quality of the final protein was checked by SDS-PAGE and DLS. The SDS-PAGE gel shows that the isolated M. barkeri β consists of a single band of equivalent size to the

β in the M. barkeri ACDS complex (Figure 4.14). The DLS results indicated that the protein was monodispersive based on the low baseline (1.002), M.R. (23.9%), and SOS error (0.905). The DLS predicted molecular mass of 88 kDa is greater than the theoretical molecular weight (53 kDa) indicating that the oligomer state is either a monomer or dimer.

The M. barkeri β was also analyzed by mass spectrum analysis, N-terminal sequencing, and metal analysis. Mass assignment was made by an electrospray ionization mass spectrometer (MS-Q2 TOF ES) at the Mass Spectrometry and

Proteomics Facility of Campus Chemical Instrument Center at The Ohio State University.

The results indicate a molecular weight of 50.5 kDa (Figure 4.15). This is smaller than the theoretical weight (52.8 kDa), as predicted by the amino acid sequence of M. barkeri

87 β from the fusaro strain. The sequence of the MS strain is not available, but is expected to be nearly identical to that of fusaro. N-terminal sequencing of the M. barkeri β protein confirmed that the six amino acids at the N-terminus were identical relative to that of fusaro (AEFPFE). The N-terminal sequencing was done by the Edman method and performed by the Molecular Structure Facility at University of California, Davis (27).

The metal analysis was performed with a Perkin Elmer Elan 6000 inductively couple plasma (ICP) spectrometer equipped with a mass spectrometer (MS) detector system at

Laboratory of Environmental Analysis at the University of Georgia, Athens.

4.2.3.2 Activity assay of acetyl-CoA synthase β component

M. barkeri β acetyltransferase activity was measured by the redox dependent acetyl exchange assay in which the acetyl group from acetyl-CoA is transferred to

3’dephospho-CoA to form CoA and acetyl-3’dephospho-CoA (equation 5). This method

(which does not require the use of radioactive coenzyme A) was developed by Grahame and is based on an earlier method from Ragsdale (5, 28).

[5] acetyl-CoA + 3’dephospho-CoA ↔acetyl-3’dephospho-CoA + CoA

The analysis of acetyltransferase activity was performed in a Coy anaerobic chamber with modifications to a previously described protocol (5, 6, 29). Briefly, the standard reaction mixture (1.2 mL final) contained aliquots of M. barkeri β (0.1-1 µg), 1 mM Ti3+-citrate, 100 mM MOPS buffer (pH 7.2), 100-200 µM of acetyl-CoA, 50 µM aquacobalamin, and 100-200 µM of 3’dephospho-CoA. The acetyltransferase reactions were initiated with 3’dephospho-CoA and allowed to proceed at 23o C for up to 60

88 minutes. Aliquots (60 µL) of the reaction mixture were quenched as a function of time by the addition of 60 µL of a stop solution containing 2 mM TiCl3 and 0.5 M sodium

citrate, pH 4.0. The quenched reaction mixtures were frozen and stored at –80 oC prior to

HPLC analysis.

Quantitative determination of the substrates (acetyl-CoA and 3’dephospho-CoA)

and the products (CoA and acetyl-3’dephospho-CoA) was carried out by reverse phase

HPLC analysis, as described previously (29). In the modified procedure, the column

(C18 Discovery, Supelco) was pre-equilibrated in 10 mM potassium phosphate buffer

(pH 6.5) containing 50 µM 2-mercaptoethanesulfonate. A linear gradient of 0-10.9 % acetonitrile in the same buffer was applied over a period of 19 minutes at a flow rate of 1 mL/min. Reactant and product peak areas, obtained by integration at 254 nm, were used for calculation of the reaction rates.

The protein was demonstrated to be active by the acetyl exchange reaction and had a turnover rate for acetyl transfer of 4460 min-1 (Figure 4.16) (2nd prep 4040 min-1).

This agrees closely to that of the M. barkeri β fusaro strain (3100 min-1) obtained by

proteolytic digestion of the ACDS complex by Grahame (29) and the recombinant M.

thermophila β after nickel reconstitution (4500 min-1) (6).

89 4.2.3.3 Metal dependence on acetyltransferase activity

In an effort to test the effect of Cu(I) on the acetyltransferase activity in M.

barkeri β, two experiments were initiated: copper addition to M. barkeri β protein

containing substoichiometric levels of copper by reconstitution methods; and copper

subtraction from M. barkeri β protein with a copper selective chelator.

4.2.3.3.1 Copper reconstitution

Copper reconstitution of M. barkeri β was performed in the presence of a Cu(I)-

glutathione (GSH) complex (30, 31). Briefly, the Cu(I)-GSH complex was prepared by adding an anaerobic solution of GSH in 0.1 M phosphate buffer, pH 7.0, to an appropriate amount of CuCl powder under nitrogen and hydrogen (98% and 2%, respectively) in a Coy anaerobic chamber. The final GSH-Cu(I) molar ration was 3:1.

Prior to reconstitution, the M. barkeri β protein solution was exchanged into an anaerobic buffer of 100 mM phosphate, pH 7.0. Copper reconstitution was performed by incubating 8.1 µM M. barkeri β protein solution in the presence of an equal molar

concentration of GSH-Cu(I) for 3.5 hours. After reconstitution, the protein was desalted

with a G-25 desalting column. The acetyltransferase activity assay was performed, as

discussed previously, on the M. barkeri β protein with and without copper reconstitution.

90 4.2.3.3.2 Copper chelation

Copper removal and quantification from M. barkeri β was performed with the

Cu(I) specific chelator of bathocuproine disulfonate (BCS). BCS allows for monitoring of the extent of copper chelation due to the linear quenching of the BCS emission at 770 nm with Cu(I) (32). Briefly, anaerobic solutions of BCS (Aldrich) were prepared in the presence of 50 mM phosphate buffer, pH 7.5, with 10 µM ascorbate. Fluorometric determinations were performed using a submicro quartz cell (Starna Cells Inc., 16.100-Q-

10). Emission spectra were recorded at 25 oC on a FluoroMax-3 spectrofluorometer. The amount of copper in the protein samples was calculated from their emission at 770 nm

(λex 580 nm) in comparison with a standard curve of BCS fluorescence versus copper concentration. The adequate BCS concentration for a given protein amount was determined by recording the emission spectra of different BCS concentrations (from 0.1-

45 µM) in the absence and presence of 18.4 µM M. barkeri β. A plot of the emission decrease caused by the protein sample at each BCS concentration shows a plateau above

15 µM BCS (Figure 4.17B), indicating that a chelator concentration of 15 µM was suitable for complete chelation of the Cu(I) in the protein sample. Thus, the amount of

Cu(I) in the protein was calculated form the emission decrease of 15 µM BCS in the presence of 18.4 µM M. barkeri β, in comparison with a plot of the of 15 µM BCS emission versus standard Cu(I) concentrations. The acetyltransferase activity assays were then performed, as discussed previously, on the M. barkeri β protein presence and absence of 15 µM BCS.

91 In addition, metal reconstitution with nickel and zinc was performed on the BCS chelated protein. Briefly, the chelated protein was incubated in the presence of either

NiCl2 or ZnCl2 for 2 hours under anaerobic conditions (6). The final molar ratio of the metal to protein was 2:1. The acetyltransferase activity assays were also performed on the reconstituted protein samples, as discussed previously.

4.2.3.4 Crystallization and data collection of acetyl-CoA synthase β component

The crystallization of the M. barkeri β was performed by the hanging drop and sitting drop vapor diffusion methods. For the hanging drop method, 1 µL of protein sample (20-50 mg/mL, 50 mM Bis-Tris, pH 7.0, 400 mM NaCl) was mixed with 1-2 µL of precipitating agent. For the sitting drop method, 2-5 µL of protein sample (50 mg/mL) was mixed with 2-3 µL of precipitating agent.

Two crystals were prepared for data collection by cryo-cooling with liquid nitrogen either by freezing of the crystal directly from mother liquor without added cro- protectant or by transferring the crystal to Paratone–N (Hampton) prior to cryo-cooling.

Diffraction data was collected at the beamline 17-ID at IMCA at APS. In addition, crystals were screened at our home source.

4.2.4 The cobalamin methyltransferase δ component of the ACDS complex

4.2.4.1 Purification cobalamin methyltransferase δ component

The δ component of the ACDS complex was isolated from M. barkeri cells under anaerobic conditions by multistep chromatography involving DEAE-FF Sepharose,

92 HiLoad Q, HAP, and Mono Q columns. Continuing the purification of δ obtained from

HiLoad Q column (section 4.2.3.1), the δ pool was desalted with a HiPrep Sephadex G-

25 desalting column (2.6 x 10 cm) pre-equilibrated with 1 mM KH2PO4 buffer, pH 6.8.

The δ pool was pink and had the characteristic UV/Vis absorption of a corrinoid-

containing protein. Since there is no available activity assay for M. barkeri δ, the identity

of the protein was based on color and size. The desalted δ pool was immediately applied

to a HAP column (1 x 6.4 cm) that had been pre-equilibrated in 1 mM KH2PO4 buffer, pH 6.8. The column was eluted with a 50 mL linear gradient of 1-500 mM KH2PO4 at a

flow rate of 0.5 mL/min. Fractions were analyzed by SDS-PAGE and those enriched in δ

were combined. The HAP δ pool was applied to a Mono-Q HR 5/5 column pre-

equilibrated with 50 mM Bis-Tris buffer, pH 7.0. Elution was performed using a 20 mL

linear gradient of 0-500 mM NaCl at a flow rate of 0.5 mL/min. The purified protein was

concentrated and stored at -80 ˚C. The purity of the protein was checked by SDS-PAGE

and DLS.

The chromatographic resolution after each column (Figures 4.19, 4.20) was

checked by SDS-PAGE (data not shown). The final protein eluted from the Mono Q

column in two elution peaks (Figure 4.20), which were identical by SDS-PAGE analysis

(Figure 4.21A). DLS analysis indicated that the two forms maybe due to different

aggregation states of the enzyme (Figure 4.21B,C). The more abundant form had

molecular mass of 55 kDa, while other was 112 kDa. This agrees with the theoretical

molecular weight of the δ monomer and δ2 homodimer, respectively. Both protein solutions were monodispersive based on the following DLS results: the monomer had a

93 baseline (1.003), M.R. (24.6%), and SOS error (1.861); and dimer had a baseline (1.001),

M.R. (23.8%), and SOS error (2.435). A total of 7 mg of pure M. barkeri δ protein was obtained from 400 grams of cells, 4 mg of the monomer and 3 mg of the dimer.

4.2.4.2 Crystallization cobalamin methyltransferase δ component

The crystallization of the M. barkeri δ was performed by the hanging drop vapor diffusion method. It consisted of 1 µL protein sample (20-50 mg/mL, 50 mM Bis-Tris, pH 7.0, 250 mM NaCl) mixed with 1 µL of precipitating agent.

Crystals of M. barkeri δ have yet to be identified. Initial crystal screens suggest that the most promising conditions are those using PEG as a precipitant within a pH range of 6.5-8.5.

4.2.5 The iron-sulfur methyltransferase γ component of the ACDS complex from

M. barkeri

4.2.5.1 Purification iron-sulfur methyltransferase γ component

The γ component of the ACDS complex was isolated from M. barkeri cells under anaerobic conditions by multistep chromatography involving DEAE-Sepharose, HAP,

HiLoad Q, and Mono Q columns. Continuing the purification of γ obtained from HiLoad

Q column (section 4.2.3.1), the γ pool was desalted with a HiPrep Sephadex G-25 desalting column (2.6 x 10 cm) pre-equilibrated with 1 mM KH2PO4 buffer, pH 6.8. The

γ pool was brown and had the characteristic UV/Vis absorption of a Fe-S containing protein. Since there is no available activity assay for M. barkeri γ, the identity of the 94 protein is based on color and size. The desalted γ pool was immediately applied to a HAP

column (1 x 6.4 cm) that had been pre-equilibrated with 1 mM KH2PO4 buffer, pH 6.8.

The column was eluted with a 50 mL linear gradient of 1-500 mM KH2PO4 at a flow rate

of 0.5 mL/min. Fractions were analyzed by SDS-PAGE and those enriched in γ were

combined. The HAP γ pool was applied to a Mono-Q HR 10/10 column pre-equilibrated

with 50 mM Bis-Tris buffer, pH 7.0. Elution was performed using a 200 mL linear

gradient of 0-500 mM NaCl at a flow rate of 1 mL/min. The final protein was

concentrated (40 mg/mL) and stored at -80 ˚C. Protein purity was confirmed by SDS-

PAGE and DLS.

The chromatographic resolution after each column (Figures 4.22, 4.23) was

checked by SDS-PAGE (data not shown). A total of 20 mg of pure γ protein was obtained

from 400 grams of cells. The final protein sample was checked by SDS-PAGE and DLS

(Figure 4.24). The DLS results indicated that the protein was monodispersive based on

the low baseline (1.000), M.R. (14.3%), and SOS error (0.558). The predicted molecular

mass, 51 kDa, closely agrees with the theoretical molecular weight of the enzyme.

4.2.5.2 Crystallization iron-sulfur methyltransferase γ component

Attempts to crystallize the M. barkeri γ were performed using the hanging drop

and sitting drop vapor diffusion methods. For the hanging drop method, 1 µL protein

sample (40 mg/mL, 50 mM Bis-Tris, pH 7.0, 425 mM NaCl) was mixed with 1-2 µL of

precipitating agent. For the sitting drop method, 1-2 µL protein sample was mixed with 2

µL of precipitating agent.

95 Crystals of M. barkeri γ have yet to be identified. Initial crystal screens suggest

that the most promising conditions are those that contain PEG within a pH range of 6.5-

8.5.

4.2.6 Recombinant ACS β from M. thermoautotrophicum

4.2.6.1 Cell culturing and expression

Methanobacterium thermoautotrophicum ACS β was cloned into E. coli by Dr.

Xuejun Zhong (unpublished results) and stock cells were provided. Cell culturing and

protein expression was performed under aerobic conditions. Briefly, ten 1 L cultures of

LB containing 1% glucose, 30 µg/mL kanamycin, and 17 µg/mL chloramphenicol were

each inoculated with 20 mL of overnight culture. The cells were grown at room

temperature, supplemented with 0.5 mM NiSO4 and 25 mg FeSO4 at induction, and

induced with 0.2 mM isopropyl-β-D-thiogalactopyranoside (IPTG) when OD600 reached

0.6. Cells were grown for an additional 8 hours, and then harvested by centrifugation yielding 17 g of cells.

4.2.6.2 Purification of recombinant M. thermoautotrophicum ACS β

Prior to preparation of the cell extracts, cells were made anaerobic by washing with 50 mM Tris buffer pH 8.0, 1 mM DTT, and 20 mM sodium dithionite (1 grams of frozen cell per 5 mL of buffer) before centrifugation. The anaerobic cells were then suspended in lysis buffer (50 mM Tris buffer pH 8.0, 1 mM DTT, and 2 mM sodium dithionite) and lysed with a French pressure cell (Thermo Spectronic, 40K cell) at an

96 operating pressure of 25,000 psi. The cell lysate was collected anaerobically in a

stoppered bottle under nitrogen. The cell debris was separated by ultracentrifugation in a

60-ti Beckman rotor at (35K rpm, 160,000 g) for 80 minutes at 4 oC, and decanted to give

a soluble extract.

The recombinant M. thermoautotrophicum ACS β was isolated under anaerobic conditions by multistep chromatography involving HiLoad Q, Mono Q, and Superose 6B columns. In the first step, cell extract was immediately loaded onto a HiLoad Q column

(2.6 x 13 cm) that had been pre-equilibrated with buffer consisting of 50 mM Tris, pH

8.0, 1 mM DTT, and 2 mM sodium dithionite. The proteins were separated using a 1000 mL linear gradient of 0-700 mM NaCl at a flow rate of 2-4 mL/min. Fractions were analyzed by SDS-PAGE and fractions enriched in β were combined. The HiLoad Q β pool was diluted with an equal volume of the HiLoad Q buffer then loaded onto a Mono-

Q HR 10/10 column that had been pre-equilibrated with 50 mM Bis-Tris buffer pH 6.2, 1 mM DTT, and 1 mM sodium dithionite, and washed with 100 mL. The elution was performed using a 200 mL linear gradient of 0-700 mM NaCl at a flow rate of 2 mL/min.

Fractions were analyzed by SDS-PAGE and fractions enriched in β were combined. The

Mono Q β pool was concentrated by Amicon (30K, regenerated cellulose membrane) and loaded to a 500 mL bed of Superose 6B prep grade column (2.6 x 95 cm) pre-equilibrated in 50 mM Tris buffer, pH 8.0, containing 100 mM NaCl and 1 mM DTT. The protein was eluted with 600 mL of the equilibration buffer at a flow rate of 1 mL/min and then

97 concentrated and stored at -80 ˚C. Protein concentrations were determined by the Lowry method using bovine serum albumin as standard. Protein purity was determined by SDS-

PAGE and DLS.

A total of 24 mg of purified M. thermoautotrophicum ACS β protein was obtained from 17 grams of cells. The chromatographic resolution of the Superose 6B size exclusion column indicated several aggregation states of the enzyme (Figure 4.25). The final protein sample was checked by SDS-PAGE and DLS (Figure 4.26). The SDS-

PAGE gel shows that the isolated M. thermoautotrophicum ACS β consists of a single band with an estimated molecular weight of 70 kDa, which is higher than the predicted molecular weight of 51.7 kDa. The DLS results were ambiguous since on one hand they indicated that the protein was monodispersive based on the low baseline (1.002) and

M.R. (20.5%). However, the SOS error (5.436) was relatively high suggesting slight polydispersity, while the large estimated molecular mass, 669 kDa, indicated multimerization of the protein.

4.2.6.3 Characterization of recombinant M. thermoautotrophicum ACS β

The activity of M. thermoautotrophicum ACS β acetyltransferase activity was measured by the redox dependent acetyl exchange assay, as performed with M. barkeri β

(section 4.2.3.2). The protein was demonstrated to be active, but at a rate that was approximately 10,000 times slower than the rate observed for M. barkeri ACS β. The low activity may be due to an incomplete A-cluster assembly, as suggested by its light

98 color. Compared to the isolated M. barkeri β protein the color of the M.

thermoautotrophicum β was much less brown. Reconstitution attempts with nickel and

iron, nevertheless, failed to increase acetyl exchange activity (data not shown).

To investigate the aggregation state of the enzyme chemical crosslinking studies were initiated. Five different crosslinkers were evaluated. The progress of the crosslinking was checked by SDS-PAGE. The crosslinking agents are listed below with their classification.

1. Glutaraldehyde (Sigma): homofunctional –amine/amine reactive

2. 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC, Pierce):

heterofunctional with zero length spacer– amine/carboxylate reactive

3. dimethyl pimelimidate hydrochloride (DMP, Pierce): homofunctional with 9.2 Å

spacer – amine/amine reactive

4. N-succinimidyl iodoacetate (SIA, Pierce): heterofunctional with 1.5 Å spacer–

amine/thiol reactive)

5. 1,8-bis-maleimidotriethyleneglycol (BM(PEO)3, Pierce): homofunctional with 14.7 Å

spacer – thiol/thiol reactive

99 4.2.6.4 Crystallization

The crystallization of the M thermoautotrophicum β was attempted using the

batch capillary and hanging drop vapor diffusion methods. For the batch capillary

method each condition consisted of the protein sample (50 mg/mL, 50 mM Bis-Tris, pH

6.5, 300 mM NaCl, 7-10 µL) followed by 15 µL of precipitating agent. The hanging

drop method consisted of the protein sample (1 µL) followed by the precipitating agent (1

µL).

Crystals of M. thermoautotrophicum β have yet to be identified. Initial crystal

screens do not suggest anything promising. The fact that no crystals have been obtained

could be due to polydispersity of the protein as suggested by DLS and gel filtration.

4.2.7 Recombinant cobalamin methyltransferase δ from Archeoglobus fuligidus

4.2.7.1 Transformation and cell culturing

Archeoglobus fuligidus δ was cloned into E. coli by Bill Mannet (unpublished results) and the plasmid was provided. E. coli cells (BL-21 DE3) were transformed with the A. fuligidus δ plasmid by electroporation. Cell culturing and protein expression was performed under aerobic conditions. Briefly, three 1 L cultures of LB containing 1% glucose, 30 µg/mL kanamycin, and 17 µg/mL chloramphenicol were each inoculated with 20 mL of overnight culture. The cells were grown at 37 oC and induced with 1 mM

IPTG when OD600 reached 0.6.

100 4.2.7.2 Isolation and protein refolding of recombinant A. fuligidus δ

Though overexpression was achieved, the protein was entirely concentrated in inclusion bodies. Two strategies were undertaken to reduce the formation of the inclusion bodies, reduction in the concentration of IPTG from 1 to 0.1 and 0.025 µM, and reduction of the induction temperature from 37 to 21 oC. Since these strategies failed to reduce inclusion formation, the recombinant A. fuligidus δ was isolated directly from the inclusion bodies. This strategy takes advantage of the large amount of protein present in the inclusions while benefiting from the efficient enrichment from the soluble E. coli proteins. Briefly, the inclusion bodies were isolated and washed three times with 20 mM

Tris, pH 8.0, 0.5 mM EDTA, and 2% Triton 100, and then solubilized with 6 M guanidine HCl, 100 mM Tris, pH 8.0, 1 mM EDTA, and 100 mM β-mercaptoethanol

(33). Three methods were pursued for refolding of the solubilized inclusion bodies (34-

37).

1. Removal of the denaturant by dialysis

2. Removal of the denaturant by rapid dilution in the presence of stabilizing agents:

arginine, detergents, and non-detergent sulfo-betaines (NDSBs)

3. Rapid buffer exchange with simultaneous dilution by gel filtration

101 The recombinant A. fuligidus δ was isolated from E. coli cells as inclusion bodies.

The protein was refolded in the absence of cobalamin by a variety of methods, but none resulted in formation of a stable aggregation state as determined by size exclusion chromatography and DLS (data not shown). Crystallization of A. fuligidus δ was not explored for this reason.

4.3 Results and discussions

4.3.1 Isolation of the components of ACDS

4.3.1.1 M. barkeri α2ε2 CODH component

The M. barkeri α2ε2 CODH component was previously purified in our laboratory by Bing Hao, albeit by a different method (25). Her purification was particularly time- consuming since it involved DE-52, DEAE-Sepharose, Sepharose CL-6B, and multiple

Mono-Q chromatographic columns. The final purification step proved to be most challenging since the most prominent impurity, the three subunit complex of methylreductase, co-eluted with the CODH on the Mono-Q column resulting in the need for the multiple runs to better resolve the two proteins. In an attempt to streamline the purification the following changes to the previous method were implemented that improved the CODH purity and isolation efficiency. First, the DE-52 anion exchange and the Sepharose CL-6B size exclusion columns were removed. The columns were eliminated from the purification scheme because the back-to-back weak anion exchange columns (DE-52, DEAE-Sepharose) were unnecessary, while the size exclusion column did not improve the protein quality (data not shown). Secondly, a subsequent HAP column

102 was implemented into the purification allowing the removal of most of the problematic methylreductase (Figure 4.7). In fact, a HAP column was used in the original purification protocol (11). Thirdly, the HiLoad Q column was added as a quick method for exchanging the phosphate buffer, which caused the protein to precipitate upon freezing or to a lesser extent upon concentration. These changes in the purification protocol resulted in CODH protein that could be purified more efficiently and crystallized more readily.

4.3.1.2 The recombinant Archeoglobus fuligidus δ cobalamin-dependent methyltransferase

The recombinant Archeoglobus fuligidus δ cobalamin-dependent methyltransferase was isolated directly from inclusion bodies. Subsequent refolding studies failed to yield a stable oligomer state. Upon further literature review, it is suggested that future work with this project might be directed at refolding in the presence of the cobalamin cofactor. Furlow and Ferry demonstrated the recombinant δ from M. thermophila was also overexpressed in E. coli as inclusion bodies (38). They were able to reconstitute the enzyme by refolding in the presence of hydroxycobalamin. Despite their success, the authors were never able to reproduce the characteristic UV spectra of the base-off configuration of the native enzyme purified from M. thermophila suggesting nonspecific or incorrect orientation of the bound corrinoid.

103 4.3.2 Oligomer structures of ACDS and its components

4.3.2.1 M. barkeri ACDS complex

The isolated M. barkeri ACDS complex was found to have an average molecular

mass of 1.93 MDa, as predicted by DLS measurements. As previously mentioned, this is

in close agreement to the (αβγδε)8 oligomer suggested by Grahame and coworkers (5-7).

Though the DLS results do indicate a highly stable complex, close examination of the

final chromatographic step indicates otherwise. Figure 4.4 shows that the Mono Q

column resolves additional α2ε2 CODH components from the ACDS complex. This is

evident to an even greater extent when the pH of the running buffer is raised above 7.0

(data not shown). It is believed that this CODH has dissociated from the ACDS complex and is not attributed to the large excess of CODH present in the cell extracts since the smaller α2ε2 is well resolved from the much larger ACDS complex on the Superose 6B

column. These results may suggest that the CODH component is on the outer surface of

the ACDS complex.

Further insights into the ACDS oligomer structure can be gleaned from the

characterization of M. barkeri ACS β. Considering the N-terminal sequencing results

indicate an intact N-terminus and the MS molecular weight was 2.3 kDa smaller than the

anticipated theoretical weight (Figure 4.15), the truncation that was observed in M.

barkeri β is therefore expected to be at the C-terminus. It should be noted that Grahame

observed that the ACDS complex of M. barkeri (fusaro strain) treated with chymotrypsin

resulted in partial dissociation of the complex into three components (5). This resulted in

a β protein that was truncated by approximately 45 amino acids from the C-terminus (6).

104 Since dissociation of the ACDS complex by proteolytic digestion does have precedence,

it is believed that the truncation of our M. barkeri β is caused by intrinsic proteases that co-purify with the crude ACDS complex early in the purification. This may indicate that

β serves as the core of the ACDS complex in methanogens. In fact since the overall direction of the net reaction is reversed in methanogens as compared to acetogens, the relative positions of the CODH/ACS components may also prove to be switched yielding a methanogen β core with flanking α2ε2 subunits. This hypothesis is consistent with the

observed dissociation of the α2ε2 component from the isolated ACDS complex mentioned previously, however additional studies are obviously required to confirm this.

4.3.2.2 Recombinant M. thermoautotrophicum ACS β

The aggregation state of M. thermoautotrophicum β was probed by the use of chemical crosslinking agents. The oligomeric state of DNA helicase from the hepatitis C virus was examined by such techniques (39). The chemical crosslinking agent of

BM(PEO)3 was identified as successfully crosslinking M. thermoautotrophicum β. This belongs to the homofunctional class of crosslinking agent, which covalently attach free cysteines of the protein via a 14.7 Å spacer. Two oligomeric protein species were observed with a size consistent of a tetramer and octamer (Figure 4.26). These free cysteines of the protein may also be contributing to the multiple aggregation states observed by DLS and gel filtration by disulfide bond formation. This may explain the low acetyl exchange activity if the disulfides are formed with the ligands of the A-cluster.

105 Due to the multiple aggregation states and the low acetyl exchange activity of M.

thermoautotrophicum ACS β, it is suggested that the future work should be directed

toward anaerobic culturing followed by nickel reconstitution. Since completing my work

on M. thermoautotrophicum β, Gencic and Grahame have cloned and expressed the β

component of the ACDS complex from Methanosarcina thermophila in E. coli (6). The

recombinant M. thermophila β obtained from anaerobic culturing contained a [Fe4S4] center, however, the enzyme was inactive by the acetyl exchange activity because it lacked nickel. They were able to reconstitute the recombinant M. thermophila β with

nickel yielding enzyme with acetyltransferase turnover rates that slightly exceeded that of

the native enzyme. Such an approach with M. thermoautotrophicum β could minimize

the putative disulfide-mediated aggregation and provide a more active enzyme.

4.3.3 Progress of crystallization of the ACDS components

4.3.3.1 Crystallization of M. barkeri α2ε2

Four different crystal forms of M. barkeri α2ε2 were identified from the conditions summarized in table 4.1. Briefly, form 1 crystals (Figure 4.27) were grown by both sitting drop and batch capillary methods, which contained ammonium fluoride and

PEG 3350 (Hampton). Form 2 crystals (Figure 4.28) were also grown by the sitting drop and batch capillary methods. They contained isopropyl alcohol (IPA), NaCl, PEG 4000

(Fluka), glycerol, and HEPES buffer. Form 3 crystals were grown by the sitting

106 drop, hanging drop, and batch capillary methods. They contained NaSCN and PEG

3350. Form 4 crystals (Figure 4.29) were grown by the sitting and hanging drop methods

contained Mg(NO3)2, glycerol, and PEG 3350.

Crystal Salt pH additive Precipitant form 1 0.2 M NH4F pH 6.2 --- 20% PEG 3350

2 2M NaCl 0.1 M HEPES, pH 5-10% IPA, 15-20% PEG 4000 7.5 0-10% glycerol 3 0.2-0.8M pH 6.9 --- 14-16% PEG 3350 NaSCN 4 0.4M Mg(NO3)2 pH 5.8 0-10% glycerol 9-12% PEG 3350

Table 4.1: CODH crystallization conditions

Briefly, efforts of crystal optimization were directed at pH, salt type, precipitant

choice, and additives. As shown in the above table, crystals were obtained from a pH

range of 5.8-7.5. CODH crystals could not be obtained at any condition with a pH below

5, though some could be obtained for the form 2 crystals as high as pH 8.5. Nonetheless,

in general pH optimization did little to improve the crystal quality. Salts were a common

requirement in all the crystal forms. Though the salt concentrations were critical for

obtaining CODH crystals, the salt type could often be substituted. For instance, the NaCl

in form 2 was successfully replaced with both NaI and CsCl, suggesting that neither the

sodium cation nor chloride anion are essential. In addition, KSCN was successfully

substituted for the NaSCN in the form 3 crystals, and MgCl2 and MgAc2 were both

substituted form the Mg(NO3)2 in the form 4 crystals. Regardless, the salts listed in the above table had superior quality than their substituted counterparts. The precipitant choice was strict and limited to PEGs in the range of 3350-4000. The supplementation of 107 the additive glycerol did not affect the crystal quality, but did aid in its cryo-cooling.

Though many additives have been screened without success (transition metals, alcohols,

NDSBs, and salts), it is suggested that commercially available additive screens

(Hampton) be explored in the future. Recently, a related CODH from Methanosarcina thermophila has been successfully crystallized. Addition of the additive spermidine identified from Hampton Additive Screen 2 improved the quality of the CODH crystals, albeit not enough to solve the crystal structure (40).

Form 1 crystals had a plate-like morphology and grew relatively slowly, typically within 1-6 months. The best crystals diffracted to 2 Å resolution were frozen with

Paratone-N, although the crystals even under these conditions were highly mosaic while slight ice rings were observed (Figure 4.30). The high mosaicity prevented accurate determination of the space group and cell dimensions, however, from the spacing of the diffraction spots the cell dimensions were estimated to be a = 60 Å, b = 68 Å, c = 250 Å.

Form 2 crystals had two different morphologies, one being a plate-like and the other was hexagonal. The plate-like crystals grew within 2-6 weeks, while the hexagonal crystals typically grew within 1 month. Of the form 2 crystals the hexagonal shaped crystals diffracted to the highest resolution (2 Å), but they too were highly mosaic (Figure 4.32).

The best crystal (Figure 4.28) was frozen with Paratone-N, however prior to transferring the crystal to the Paratone-N the crystal was washed with mother liquors, which caused it to dissolve and fracture. As with the form 1 crystals, the form 2 crystals had high mosaicity that prevented accurate determination of the space group and cell dimensions.

Nevertheless, the pattern of the diffraction spots suggested a six-fold axis. The form 2

108 platelet crystals grew more quickly (1-6 weeks) and were easier to reproduce. They had a similar diffraction pattern, but diffracted to a lower resolution (5 Å) and were even more mosaic than the hexagonal shaped one (Figure 4.31). Form 3 crystals had also had plate- like morphology and grew within 1-6 weeks. The crystal diffracted poorly only to 17 Å resolution (data not shown). Thus an accurate cell assignment was not possible. The form

4 crystals had three different crystal morphologies: irregular, leaf-like, and plates (Figure

4.29). The irregular and leaf-like crystals were the most prominent and grew quickly, typically 2-7 days, while the plates generally took longer, approximately 1 month. The plates, cryo-cooled in mother liquors containing 25% glycerol, diffracted more strongly than that of the irregular and leaf-like crystals (7 Å versus 20 Å), but they too were very mosaic (Figure 4.33). As in the other crystal forms, the high mosaicity prevented accurate cell determination, however from the spacing of the diffraction spots the cell dimensions were estimated to be a = 58 Å, b = 66 Å, c = 250 Å. The estimated cell dimensions is very similar to that of crystal form I and is believed to be of the same space group. In summary, the high mosaicity observed in all of the crystal forms is believed to be an inherent problem in all of the fast growing crystals. Furthermore, cryo-cooling of the plate-type crystals often exasperated the mosaicity problem. Balbo recently reported similar difficulty in the cryo-cooling of CODH crystals from M. thermophila (40). In general, it is believed that crystals that grow slower and are not thin plates, such as the hexagonal-type form 2 crystals, are the most conducive for the structure elucidation.

109 These are, however, the most difficult to reproduce. It is recommended that if additional

crystals of this form are obtained that they be transferred directly to Paratone-N without

dilution in the well solution due to their highly fragile nature.

4.3.3.2 Crystallization of M. barkeri ACS β

Two crystal forms of the M. barkeri β have been identified. The first was grown from condition 13 of the Hampton Screen Kit II. It contained 0.2 M ammonium sulfate,

0.1 M sodium acetate (pH 4.6), and 30% PEG 2000 monomethyl ether. The two crystals grown from this condition had a plate-like morphology and were observed only after 5 months. The best crystal, cryo-cooled from Paratone-N, diffracted to 3 Å resolution

(Figure 4.34). Direct freezing from the mother liquors without cry-protection resulted in lower diffraction. Both crystals were partially dried due to evaporation of the well solution prior to crystal mounting, which may explain the high mosaicity of the crystals.

This prevented accurate determination of the space group and cell dimensions, nonetheless, the spacing of the diffraction spots suggested that the cell dimensions were a

= 105 Å, b = 105 Å, c = 81 Å.

The second crystal form was grown from condition 36 of the Hampton Screen Kit

I. It contained 0.1 M Tris (pH 8.5) and 30% PEG 8000. The crystals grown from this condition were small plate-like clusters, which were observed only after 9 months. The crystals diffracted weakly to resolution of 9 Å, due to their small size, preventing their cell assignment.

110 4.3.4 Models of the subunits in lieu of the available structures

4.3.4.1 Modeling and comparison with other CODHs

While working on this project, four CODH crystal structures have been

determined: two of the monofunctional class, and two from a bifunctional CODH/ACS

(18-21). No structures, however, are available from a methanogen. Comparison of the

sequence of the M. barkeri CODH with these proteins reveals a low overall similarity

despite conserving most of the protein ligands of the metal clusters (14, 25). In addition,

there is no protein equivalent for the M. barkeri ε subunit found in these other CODHs.

Due to the low sequence homology between the different CODHs, it was not possible to

model the 3D structure of M. barkeri α subunit greater than 66 residues with SWISS-

MODEL or Geno3D (model not shown) (41, 42). Nevertheless, secondary structural information was obtained by performing a sequence alignment with MultiAlign in comparison with M. thermoacetica β, which again is equivalent to M. barkeri α component (43). This allowed for the superimposition of the secondary structure of the

M. thermoacetica β subunit (PDB # 1OAO) that was added with ESPript (Figure 4.35)

(44). Examination of the alignment reveals several regions that have additional amino acids in the M. barkeri α. In general, these regions are mostly confined to the helical domain I of M. thermoacetica β, as defined by residues 4-237 (19). For example, helix

α8 is 9 residues longer in M. barkeri. In M. thermoacetica β this helix is located at the

inter-subunit interface where it contacts domain I of the ACS subunit. Though this would

cause a steric clash between the CODH and ACS subunits of M. thermoacetica, since

extending this helix would result in it crashing into domain I of ACS. This domain is,

111 however, not present in the M. barkeri ACS (Figure 4.37, discussed in the section of

modeling and comparison with other ACSs), so a longer helix could supposedly be

accommodated. Other regions that have addition amino acids are near helix α10 and β-

strand β4. These structural elements are buried deep in the CODH molecule and are far

remove from the cluster sites. The last and most prominent difference between the two

sequences is that the C-terminus of M. barkeri α extends 82 additional residues. By

examining the structure of M. thermoacetica β, the C-terminus is also at the αβ interface.

An additional sequence alignment performed with Clustal-W, reveals limited sequence homology between the C-terminus of M. barkeri α and some of the N-terminus comprising part of domain I of M. thermoacetica α (Figure 4.36), which is equivalent to

M. barkeri β (45). The longer C-terminus may contribute to the hydrophobic channel connecting the C-cluster of CODH to the A-cluster of ACS, as does domain I in M. thermoacetica α (21). This may have arisen from domain shuffling at the gene level. In support of this hypothesis, Doukov et al. noticed a high structural similarity between domain I of the ACS α subunit and the Rossman domains of the CODH β subunit (21).

They proposed that this could be due to gene duplication.

4.3.4.2 Modeling and comparison with Moorella thermoacetica ACSs

As mentioned previously, two CODH/ACS structures are available from M.

thermoacetica (20, 21). The overall sequence homology between the M. barkeri β and

the M. thermoacetica α is high, 41 % identical and 61 % similar as performed by Clustal-

W (Figure 4.37) (45). Furthermore, the protein ligands for the A-clusters are completely 112 conserved. One noteworthy difference is that M. barkeri β completely lacks the amino acids comprising domain I. The high sequence homology between the two enzymes permits the structural modeling of M. barkeri β. The structure was modeled with the programs SWISS-MODEL and Geno-3D using the atomic coordinates from the Darnault structure (PDB # 1OAO) (41, 42). As shown in Figure 4.38, the proposed model is composed of only two domains, II and III, with the A-cluster being largely unprotected as compared to M. thermoacetica α. Since methanogen CODHs lack this domain I yet they contain the additional ε subunit, it is tempting to speculate that ε may contribute in directing the toxic CO gas through the hydrophobic channel as does domain I of M. thermoacetica ACS. However, such speculation is largely unfounded since no sequence homology is found between the M. barkeri ε and domain I of M. thermoacetica α.

Clearly additional studies will be required to confirm this hypothesis.

4.3.5 Identity of labile metal in the A-cluster of the ACS β component

While pursuing my research with M. barkeri β, Doukov reported the crystal structure of CODH/ACS enzyme from acetogenic bacteria Moorella thermoacetica to 2.2

Å resolution (21). It has long been believed that the α ACS subunit contains a novel Ni-

Fe-S A-cluster required for acetyl-CoA synthesis, however to the surprise of the community that studies CODH/ACS enzymes their structure revealed an A-cluster consisting of a Ni-Cu-[Fe4S4] center (Figure 4.2D). The Cu ion is proposed to be the site of binding of the CO and acetyl ligands. In support of their hypothesis, Seravalli et al. recently reported XANES and EXAFS evidence that the Cu ion at the A-cluster can

113 coordinate with the CoA analog of Se-CoA via a Se-Cu bond (46). Contrary to these

findings, another crystal structure of M. thermoacetica CODH/ACS was solved by

Darnault et al. to 1.9 Å resolution (20). This structure revealed Ni-Ni-[Fe4S4] and Ni-Zn-

[Fe4S4] centers at the A-cluster. The Ni-Ni form is believed to be the active form of the

enzyme (Figure 4.2E). Support for the Ni-Ni-[Fe4S4] A-cluster was provided with the recent the cloning and metal reconstitution of β from M. thermophila (6). In this study an inactive apo enzyme containing only a [Fe4S4] cluster was reconstituted with two mole equivalents of nickel yielding a fully active enzyme. In an effort to probe this nickel- copper controversy, copper reconstitution and chelation experiments of β from M. barkeri were initiated.

Considering copper was routinely found in the CODH/ACS complex from

Moorella thermoacetica (21, 46), it was not completely unexpected when copper was also found in both of the preparations of M. barkeri β (Figure 4.39). Since the copper was at substoichiometric levels, experiments were initiated at increasing the copper levels in the protein by reconstitution with GSH-Cu(I) complex. The redox-dependent acetyltransferase activity assay was performed on the β protein with and without copper reconstitution. Comparison of the relative rates of activity indicate that the β reconstituted with GSH-Cu(I) complex had almost a 50% decrease in the acetyltransferase activity (Figure 4.40).

To probe the other extreme, removal of copper, chelation experiments were initiated with the copper selective BCS. The copper chelation results first indicate that the copper in M. barkeri β is labile and can be quantified. As shown in Figure 4.17A, the

114 BCS emission at 770 nm is quenched by M. barkeri β providing evidence that an

interaction between the chelator and the Cu(I) from the enzyme. The quenching effect of

18 µM M. barkeri β was measured at the BCS emission in the absence and the presence of the protein with increasing concentration of BCS (Figure 4.18). A plot of the difference in the BCS emission between the two shows an increase as the BCS emission is raised to 15 µM and remains constant from 15 to 44 µM BCS (Figure 4.17B). Thus, a

BCS concentration of 15 µM was used for quantifying the Cu(I) from the 18 µM M. barkeri β, since this amount of chelator is sufficient for the copper present in the protein sample. Data from two experiments such as that in Figure 4.17B indicate that the 18 µM

M. barkeri β causes a decrease in the BCS emission at 770 nm, which is equal to that provided by a 6.4 µM standard Cu(I) under identical conditions. Thus it appears that 18

µM M. barkeri β contains 6.4 µM Cu(I). This agrees closely with the value obtained by

MS-ICP of 5.5 µM. It should be noted that other metal ions such as Fe, Zn, and Ni do not interfere in the BCS fluorescence nor in the quenching by Cu(I) (32).

In the first chelation experiment performed on the first protein preparation (Figure

4.39), quenching of the BCS fluorescence was nearly immediate, however repeating the experiment with the other preparations required longer hold times (12 hours) to achieve the same quenching effect. We believe the low iron content in the first protein preparation resulted in a more labile copper site.

In combination with the acetyltransferase activity assay, the copper chelation results also indicate that the copper in M. barkeri β has an inhibitory effect on the acetyl exchange reaction and that it can be replaced with a more functional nickel ion. The 115 acetyltransferase activity assays were performed on 18 µM M. barkeri β in the presence and absence of 15 µM BCS. In addition, metal reconstitution with nickel and zinc was performed on the BCS chelated protein with NiCl2 and ZnCl2. As shown in Figure 4.40, the chelated M. barkeri β has a 9% higher relative rate of acetyl exchange activity than untreated protein and that reconstituted with nickel has a 21% increase. The zinc reconstituted M. barkeri β only had a moderate increase in activity. These results provide additional evidence that nickel is the functional ion in the acetyltransferase reaction in M. barkeri β. Furthermore, the Cu that was observed in M. barkeri β is shown to have an inhibitory effect on this reaction. Taken together, results consistently support the

Darnault Ni-Ni-[Fe4S4] A-cluster (Figure 4.2E).

4.4 Conclusions

In this project, we report the isolation and crystallization screening of M. barkeri

ACDS complex and its comprising subunits, as well as two cloned subunits from related methanogens. The M. barkeri α2ε2 CODH component has yielded four crystal forms while two crystal forms were found for the M. barkeri β ACS component, however their structure determination is ongoing due to difficulties with crystal disorder and cro- cooling. Biochemical evidence is provided that supports the [Ni-Ni-Fe4S4] A-cluster of

M. barkeri β subunit and furthermore refutes the controversial [Ni-Cu-Fe4S4] A-cluster.

Support for this contention is provided by the observation of increased acetyltransferase activity of M. barkeri β after the Cu removal and a further increase after nickel reconstitution. Sequence comparison of M. barkeri β and α2ε2 components with those of 116 related organisms with available structures, as well as a structural model of M. barkeri β, are presented and possible implications of the observed similarities and differences are discussed.

117

Figure 4.1. Available CODH/ACS structures: (A) Superposition of two known monofunctional CODH structures, PDB ID: 1JJY and 1JQK (pink and blue). (B) Bifunctional CODH/ACS structure (1OAO), ACS, pink and CODH, cyan. (C) ACS component (1OAO) and its domains (I, II, and III, colored blue green and red).

118 A.

B.

C.

Figure 4.1. Available CODH/ACS structures: 119

Figure 4.2. ACS/CODH active site metal clusters: (A) Drennan’s C-cluster from R. rubrum, (B) Dobbek’s C-cluster from C. hydrogenoformans, (C) Darnault’s C-cluster from M. thermoacetica, (D) Doukov’s A-cluster from M. thermoacetica with an acetate ligand, Ac, and (E) Darnault’s A-cluster from M. thermoacetica with an unidentified ligand, L.

120

Figure 4.3. Chromatographic resolution of the M. barkeri ACDS complex on Superose 6B size exclusion column as a function of elution volume, and absorbance at 280 nm (blue). The peak corresponding to the ACDS complex is indicated.

121

Figure 4.4. Chromatographic resolution of the M. barkeri ACDS complex on Mono Q column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The larger peak corresponding to the ACDS complex while the peak at ~70 mL corresponding to CODH.

122 A. B.

Figure 4.5. The quality of the M. barkeri ACDS complex. (A) SDS-PAGE gel of the purified ACDS complex: lane 1, ACDS complex; lane 2, MW marker. (B) DLS measurement of the ACDS complex (M.R. 3%, baseline 1.004, and SOS error 1.783).

123

Figure 4.6. Chromatographic resolution of the M. barkeri α2ε2 CODH and crude ACDS complex on DEAE-FF column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The peaks corresponding to CODH and ACDS complex are indicated.

124

B.

Figure 4.7. Chromatographic resolution of M. barkeri α2ε2 CODH on HAP column. (A) Elution profile as a function of elution time, absorbance at 280 nm (blue), and conductance (brown). The peak corresponding to the CODH is indicated. (B) SDS- PAGE gel of the HAP column fractions shows the separation between the methylreductase impurities from the CODH.

125

Figure 4.8. Chromatographic resolution of M. barkeri α2ε2 CODH on HiLoad Q column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The peak corresponding to the CODH is indicated.

126

Figure 4.9. Chromatographic resolution of M. barkeri α2ε2 CODH on Mono Q column as a function of elution time, absorbance at 280 nm (blue), and conductance (brown). The large peak corresponding to CODH.

127

A. B.

Figure 4.10. The quality of the M. barkeri α2ε2 CODH component of the ACDS complex. (A) SDS-PAGE gel: lane 1, cell extract; lane 2, pool from DEAA-FF; lane 3, pool from HAP; lane 4, pool from HiLoad Q, lane 5, final protein from Mono Q. (B) DLS measurement (M.R. 10%, baseline 1.000, and SOS error 0.283).

128

Figure 4.11. Chromatographic resolution of crude M. barkeri ACDS complex into its comprising subunits on HiLoad Q column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The peak corresponding to the individual subunits are indicated.

129

Figure 4.12. Chromatographic resolution of M. barkeri β on HAP column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The peaks corresponding to β are indicated.

130

Figure 4.13. Chromatographic resolution of M. barkeri β on Mono Q column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The larger peak corresponding to the β.

131

A. B.

Figure 4.14. The quality of M. barkeri β component of the ACDS complex. (A) SDS- PAGE gel of purified β: lane 1, the ACDS complex; lane 2, the purified β; lane 3, molecular weight marker (B) DLS measurement of β (M.R. 23.9%, baseline 1.003, and SOS error 0.589).

132

Figure 4.15. Mass spectrum of purified M. barkeri β.

133

acetyl transfer reaction -- M. barkeri β

350

300 Co-A 250 Acetyl Co-A

200 3' dephosphoCoA

150 Acetyl 3' dephosphoCoA 100

50

0 0246810121416 time (min)

Figure 4.16. Acetyl exchange activity of M. barkeri β. The exchange rate is 88.3 µmol acetyl/min/mg enzyme. The turnover number is 4460 min-1.

134 A.

B. M. barkeri β titration with BCS

700000

600000

500000

400000

300000

200000 difference emission intensity 100000

0 0 5 10 15 20 25 30 35 40 45 50 [BCS] µM

Figure 4.17. Quenching of BCS fluorescence of M. barkeri β (A) Emission spectra (λex 580nm) of 15 µM BCS in the absence (red) and presence (blue) of 18 µM β. (B) A plot of the difference emission of BCS at 770 nm caused by 18 µM β versus a blank buffer solution at different BCS concentrations.

135

A.

B.

Figure 4.18. Compilation of fluorescence emission spectra of BCS titration. (A) control buffer solution, (B) 18 µM M. barkeri β. Both the control and protein solutions were titrated with 0.4-44 µM BCS (λex 580nm).

136

Figure 4.19. Chromatographic resolution of the M. barkeri δ on HAP column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The prominent peak corresponding to the δ.

137

Figure 4.20. Chromatographic resolution of the M. barkeri δ on Mono Q column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The both peaks are δ, the first is the monomer and the second is dimer.

138 A.

B. C.

Figure 4.21. The quality of the M. barkeri δ component of the ACDS complex. (A) SDS- PAGE gel of the purified δ: lane 1, the purified δ; lane 2, molecular weight marker. Dynamic light scattering of measurements: (B) the δ monomer (M.R. 24.6%, baseline 1.003, and SOS error 2.434) and (C) the δ dimer (M.R. 23.8%, baseline 1.001, and SOS error 2.435).

139

Figure 4.22. Chromatographic resolution of the M. barkeri γ on HAP column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The peak corresponding to the γ is indicated.

140

Figure 4.23. Chromatographic resolution of the M. barkeri γ on Mono Q column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The peak corresponding to the γ is indicated.

141 A.

B.

Figure 4.24. The quality of the M. barkeri γ component of the ACDS complex. (A) SDS- PAGE gel of purified γ: lane 1, the purified γ; lane 2, molecular weight marker. (B) Dynamic light scattering measurement of the purified γ (M.R. 14.3%, baseline 1.000, and SOS error 0.558). 142

Figure 4.25. Chromatographic resolution of the recombinant M. thermoautotrophicum β on Superose 6B size exclusion column as a function of elution volume, absorbance at 280 nm (blue), and conductance (brown). The two larger peaks correspond to different aggregation states of β.

143

A. B.

Figure 4.26. The quality of recombinant M. thermoautotrophicum β. (A) SDS-PAGE gel of the purified and crosslinked β: lane 1, molecular weight marker; lane 2, the purified β: lane 3, the crosslinked β. (B) Dynamic light scattering measurement of the β (M.R. 20.5%, baseline 1.007, and SOS error 2.212).

144

Figure 4.27. M. barkeri α2ε2 CODH crystals form 1.

145 A.

B.

Figure 4.28. M. barkeri α2ε2 CODH crystals form 2: (A) hexagonal like crystal after partially dissolving, (B) plate-like crystal.

146 A. B.

C.

Figure 4.29. M. barkeri α2ε2 CODH crystal form 4: (A) irregular shaped crystal, (B) leaf-like shape crystal, (C) plate-like crystal.

147

Figure 4.30. Diffraction images of M. barkeri α2ε2 CODH crystal form 1.

148 A.

B.

Figure 4.31. Diffraction images of plate-like M. barkeri α2ε2 CODH crystal form 2 (B) is 90o relative to (A). 149 A.

B.

Figure 4.32. Diffraction images of hexagonal-like M. barkeri α2ε2 CODH crystal form 2 (B) is 90o relative to (A). 150

Figure 4.33. Diffraction image of plate-like M. barkeri α2ε2 CODH crystal form 4.

151

Figure 4.34. Diffraction image of the M. barkeri β crystal.

152

Figure 4.35. Structure-based sequence alignment of M. barkeri α CODH component and M. thermoacetica β CODH. The secondary structural elements of M. thermoacetica β (PDB # 1OAO) are placed on top of the sequence alignment (43, 44). The conserved residues are boxed in black and the similar residues are boxed in white. Turns are indicated by TT and every 10th residue is marked by a period. Sequence identity is 25% and similarity, 35%.

153

Figure 4.35 continued. 154

Figure 4.35 continued.

155

Figure 4.36. Sequence alignment of the C-terminus of M. barkeri α CODH component and a portion of the N-terminus of M. thermoacetica α ACS component (45). The conserved residues are boxed in yellow and the similar residues are colons and periods. Sequence identity is 21% and similarity 48%.

156 157

Figure 4.37. Sequence alignment of M. barkeri β vs. M. thermoacetica α (45). The conserved residues are indicated according to their amino acid type, while the similar residues are with a plus sign. The protein ligands to the A-cluster are in bold. The overall identity is 41% (61% similarity).

158

Figure 4.38. Model of the M. barkeri β ACS component (yellow) superimposed on M. thermoacetica α (blue), PDB 1OAO. The Fe, Ni, and S atoms of the A-cluster are colored black, green, and yellow, respectively. The structure was modeled with the programs Swiss-Model and Geno-3D (41, 42).

159

ICP Metal analysis metal atoms per β subunit Fe Ni Cu Zn expected 4 1 or 2 0 or 1 0 prep 1 1.47 0.88 0.30 1.86 prep 2 3.31 1.94 0.25 0.48

Figure 4.39. ICP metal analysis of M. barkeri ACS β. The expected values are based on the available models of the A-cluster from M. thermoacetica ACS α.

160

sample Relative acetyl exchange activity (as compared to control) β as isolated (control) 1 BCS chelated 1.09 BCS chelated + 2X Zn 1.04 BCS chelated + 2X Ni 1.21 1X Cu(I)-GSH 0.52 + 10X Cu 0.11 + 10X Zn 0.93 + 10X Ni 1.02

Figure 4.40. Relative activities of the M. barkeri ACS β after chelation with BCS and metal reconstitution.

161

Chapter 4 References

1. Krzycki, J. A., Lehman, L. J., and Zeikus, J. G. (1985) J. Bacteriol. 163, 1000- 1006.

2. Smith, P. H., and Mah, R. A. (1966) Appl. Microbio. 14, 368-71.

3. Ferry, J. G. (1992) J. Bacteriol. 174, 5489-5495.

4. Grahame, D. A. (1987) Federation Proceedings 46, 2179-2179.

5. Grahame, D. A., and DeMoll, E. (1996) J. Biol. Chem. 271, 8352-8358.

6. Gencic S, G. D. (2003) J Biol Chem 278, 6101-10.

7. Grahame, D. A., and Stadtman, T. C. (1987) J. Biol. Chem. 262, 3706-3712.

8. Ragsdale, S. W., and Kumar, M. (1996) Chem. Rev.96, 2515-2539.

9. Murakami, E., and Ragsdale, S. W. (2000) J. Biol. Chem. 275, 4699-4707.

10. Grahame, D. A., Dai, Y. R., and DeMoll, E. (1997) FASEB J. 11, 2613.

11. Krzycki, J. A., and Zeikus, J. G. (1984) J. Bacteriol. 158, 231-237.

12. Lindahl, P. A. (2002) Biochemistry 41, 2097-2105.

13. Ferry, J. G. (1995) Annu. Rev. Microbiol. 49, 305-333.

14. Lindahl, P. A., and Chang, B. (2001) Orig. Life Evol. Biosph. 31, 403-434.

15. Huber, C., and Wächtershäuser, G. (1997) Science 276, 245-247.

16. Zinder, S. H. (1993) in Methanogenesis: ecology, physiology, biochemistry & genetics (Ferry, J. G., Ed.), Chapman & Hall, New York.

162

17. Wolfe, R. S. (1993) in Methanogenesis: ecology, physiology, biochemistry & genetics (Ferry, J. G., Ed.) pp 1-32, Chapman & Hall, New York.

18. Dobbek, H., Svetlitchnyi, V., Gremer, L., Huber, R., and Meyer, O. (2001) Science 293, 1281-1285.

19. Drennan, C. L., Heo, J. Y., Sintchak, M. D., Schreiter, E., and Ludden, P. W. (2001) Proc. Natl. Acad. Sci. U.S.A. 98, 11973-11978.

20. Darnault, C., Volbeda, A., Kim, E. J., Legrand, P., Vernede, X., Lindahl, P. A., and Fontecilla-Camps, J. C. (2003) Nat. Struct. Biol. 10, 271-279.

21. Doukov, T. I., Iverson, T. M., Seravalli, J., Ragsdale, S. W., and Drennan, C. L. (2002) Science 298, 567-572.

22. Grahame, D. A. (1991) J. Biol. Chem. 266, 22227-22233.

23. Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chem. 193, 265-75.

24. Jancarik, J., and Kim, S. H. (1991) J. Appl. Crystallogr. 24, 409-411.

25. Hao, B. (2002) Dissertation.

26. Terlesky, K. C., Nelson, M. J. K., and Ferry, J. G. (1986) J. Bacteriol. 168, 1053- 1058.

27. Edman, P. (1950) Acta Chem. Scand., 4283.

28. Ragsdale, S. W. (1991) Crit. Rev. Biochem. Mol. Biol. 26, 261-300.

29. Bhaskar, B., DeMoll, E., and Grahame, D. A. (1998) Biochemistry 37, 14491- 14499.

30. Ciriolo, M. R., Desideri, A., Paci, M., and Rotilio, G. (1990) J. Biol. Chem. 265, 11030-11034.

31. Musci, G., DiMarco, S., Bellenchi, G. C., and Calabrese, L. (1996) J. Biol. Chem. 271, 1972-1978.

32. Rapisarda, V. A., Volentini, S. I., Farias, R. N., and Massa, E. M. (2002) Anal. Biochem. 307, 105-109.

163 33. Georgiou, G., and Valax, P. (1999) in Amyloid, Prions, and Other Protein Aggregates pp 48-58.

34. Middelberg, A. R. (2002) Trends Biotechnol. 20, 437-443.

35. Muller, C., and Rinas, U. (1999) J. Chromatogr. A 855, 203-213.

36. Clark, E. D. (2001) Curr. Opin. Biotechnol. 12, 202-207.

37. Wei, G., and Moreira, A. R. (1998) Abstr. Pap. Am. Chem. Soc. 216, 100-BTEC.

38. MaupinFurlow, J., and Ferry, J. G. (1996) J. Bacteriol. 178, 340-346.

39. Levin, M. K., and Patel, S. S. (1999) J. Biol. Chem. 274, 31839-31846.

40. Balbo P, O. M. (2003) Acta Crystallogr D59, 721-3.

41. Schwede, T., Diemand, A., Guex, N., and Peitsch, M. C. (2000) Res. Microbiol. 151, 107-112.

42. Combet, C., Jambon, M., Deleage, G., and Geourjon, C. (2002) Bioinformatics 18, 213-214.

43. Corpet, F. (1988) Nucleic Acids Res. 16, 10881-10890.

44. Gouet, P., Courcelle, E., Stuart, D. I., and Metoz, F. (1999) Bioinformatics 15, 305-308.

45. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994) Nucleic Acids Res. 22, 4673-4680.

46. Seravalli, J., Gu, W. W., Tam, A., Strauss, E., Begley, T. P., Cramer, S. P., and Ragsdale, S. W. (2003) Proc. Natl. Acad. Sci. U. S. A. 100, 3689-3694.

164

Bibliography

1. Lippard, S. J., and Berg, J. M. (1994) Principles of Bioinorganic Chemistry.

2. Cowan, J. A. (1997) Inorganic Biochemistry: An Introduction, 2 ed.

3. Messerschmidt, A., Huber, R., Poulos, T., and Wieghardt, K. (2001) in Handbook of Metalloproteins.

4. Matte, A., and Delbaere, L. T. J. (2001) in Handbook on Metalloproteins pp 59-91.

5. Sun, H. Z., Chen, R., and Che, C. M. (2002) Prog. Chem. 14, 257-262.

6. Huang, Z. X. (2002) Prog. Chem. 14, 318-322.

7. Frazzon, J., and Dean, D. R. (2003) Curr. Opin. Chem. Biol. 7, 166-173.

8. Dudev, T., and Lim, C. (2003) Chem. Rev.103, 773-787.

9. Cowan, J. A. (1995) Biological Chemistry of Magnesium.

10. Kobayashi, M., and Shimizu, S. (1999) Euro. J. Biochem. 261, 1-9.

11. Pratt, J. M. (2001) in Handbook on Metalloproteins pp 603-668.

12. Cheng, T. C., Ramakrishnan, V., and Chan, S. I. (1999) Protein Sci. 8, 2474-2486.

13. Natesh, R., Schwager, S. L. U., Sturrock, E. D., and Acharya, K. R. (2003) Nature 421, 551-554.

14. Sumner, J. B. (1926) J. Biol. Chem. 69, 435-441.

15. Dixon, N. E., Gazzola, C., Blakeley, R. L., and Zerner, J. B. (1975) J. Am. Chem. Soc. 97, 4131-4133.

165 16. Watt, R. K., and Ludden, P. W. (1999) Cell. Mol. Life Sci. 56, 604-625.

17. Walsh, C. T., and Ormejohnson, W. H. (1987) Biochemistry 26, 4901-4906.

18. Ciurli, S., and Mangani, S. (2001) in Handbook on Metalloproteins pp 669-707.

19. Bentrop, D., Capozzi, F., and Luchinat, C. (2001) in Handbook on Metalloproteins pp 357-459.

20. Rees, D. C. (2002) Annu. Rev. Biochem. 71, 221-246.

21. Rees, D. C., and Howard, J. B. (2003) Science 300, 929-931.

22. Cody, G. D., Boctor, N. Z., Filley, T. R., Hazen, R. M., Scott, J. H., Sharma, A., and Yoder, H. S. (2000) Science 289, 1337-1340.

23. Drennan, C. L., and Peters, J. W. (2003) Curr. Opin. Struct. Biol. 13, 220-226.

24. Hendrickson, W. A. (1997) FASEB J. 11, 1565.

25. Hendrickson, W. A. (1999) J. Synchrot. Radiat. 6, 845-851.

26. Hendrickson, W. A., and Ogata, C. M. (1997) Methods Enzymol. 276 494-523.

27. Hendrickson, W. A. (2000) Trends Biochem.Sci. 25, 637-643.

28. McDonald, N. Q., Panayotatos, N., and Hendrickson, W. A. (1995) EMBO J. 14, 2689-2699.

29. Hubscher, U., Maga, G., and Spadari, S. (2002) Annu. Rev. Biochem. 71, 133-163.

30. Beese, L. S., Derbyshire, V., and Steitz, T. A. (1993) Science 260, 352-355.

31. Showalter, A. K., Byeon, I. J. L., Su, M. I., and Tsai, M. D. (2001) Nat. Struct. Biol. 8, 942-946.

32. Wilson, S. H. (1998) Mutat. Res.-DNA Repair 407, 203-215.

33. Kunkel, T. A., and Bebenek, R. (2000) Annu. Rev. Biochem. 69, 497-529.

34. Pelletier, H., Sawaya, M. R., Kumar, A., Wilson, S. H., and Kraut, J. (1994) Science 264, 1891-1903.

166 35. Pelletier, H., Sawaya, M. R., Wolfle, W., Wilson, S. H., and Kraut, J. (1996) Biochemistry 35, 12742-12761.

36. Sawaya, M. R., Prasad, R., Wilson, S. H., Kraut, J., and Pelletier, H. (1997) Biochemistry 36, 11205-11215.

37. Doublie, S., Tabor, S., Long, A. M., Richardson, C. C., and Ellenberger, T. (1998) Nature 391, 251-258.

38. Kiefer, J. R., Mao, C., Braman, J. C., and Beese, L. S. (1998) Nature 391, 304-307.

39. Steitz, T. A. (1999) J. Biol. Chem. 274, 17395-17398.

40. Zhong, X. J., Patel, S. S., Werneburg, B. G., and Tsai, M. D. (1997) Biochemistry 36, 11891-11900.

41. Arndt, J. W., Gong, W. M., Zhong, X. J., Showalter, A. K., Liu, J., Dunlap, C. A., Lin, Z., Paxson, C., Tsai, M. D., and Chan, M. K. (2001) Biochemistry 40, 5368- 5375.

42. Wong, I., Patel, S. S., and Johnson, K. A. (1991) Biochemistry 30, 526-537.

43. Dunaway-Mariano, D., and Cleland, W. W. (1980) Biochemistry 19, 1506-1515.

44. Cleland, W. W. (1982) Methods Enzymol. 87, 159-179.

45. Teng, T. Y. (1990) J. Appl. Crystallogr. 23, 387-391.

46. Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, 307-326.

47. Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M. (1991) Acta Crystallogr. A 47, 110-119.

48. Brünger, A. T. (1993) X-PLOR Version 3.1 manual, Yale University, New Haven, CT.

49. Brünger, A. T. (1993) Acta Crystallogr. D49, 24-36.

50. Laskowski, R. A., Macarthur, M. W., Moss, D. S., and Thornton, J. M. (1993) J. Appl. Crystallogr. 26, 283-291.

51. Werneburg, B. G., Ahn, J., Zhong, X. J., Hondal, R. J., Kraynov, V. S., and Tsai, M. D. (1996) Biochemistry 35, 7041-7050.

167 52. Liu, J., and Tsai, M. D. (2001) Biochemistry 40, 9014-9022.

53. Dunlap, C. A., and Tsai, M. D. (2002) Biochemistry 41, 11226-11235.

54. Johnson, K. A. (1993) Annu. Rev. Biochem. 62, 685-713.

55. Zhong, X. J., Patel, S. S., and Tsai, M. D. (1998) J. Am. Chem. Soc. 120, 235-236.

56. Shah, A. M., Conn, D. A., Li, S. X., Capaldi, A., Jager, J., and Sweasy, J. B. (2001) Biochemistry 40, 11372-11381.

57. Shah, A. M., Li, S. X., Anderson, K. S., and Sweasy, J. B. (2001) J. Biol. Chem. 276, 10824-10831.

58. Vande Berg, B. J., Beard, W. A., and Wilson, S. H. (2001) J. Biol. Chem. 276, 3408- 3416.

59. Kim, S. J., Beard, W. A., Harvey, J., Shock, D. D., Knutson, J. R., and Wilson, S. H. (2003) J. Biol. Chem. 278, 5072-5081.

60. Yang, L. J., Beard, W. A., Wilson, S. H., Broyde, S., and Schlick, T. (2002) J. Mol. Biol. 317, 651-671.

61. Yang, L. J., Beard, W. A., Wilson, S. H., Roux, B., Broyde, S., and Schlick, T. (2002) J. Mol. Biol. 321, 459-478.

62. Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 946-950.

63. Merritt, E. A., and Murphy, M. E. P. (1994) Acta Crystallogr. D50, 869-873.

64. McRee, D. E. (1999) J. Struct. Biol. 125, 156-165.

65. Rawlings, N. D., and Barrett, A. J. (1995) Methods Enzymol. 248 183-228.

66. Stepanov, V. M. (1995) Methods Enzymol. 248, 675-683.

67. Lee, S. H., Minagawa, E., Taguchi, H., Matsuzawa, H., Ohta, T., Kaminogawa, S., and Yamauchi, K. (1992) Biosci. Biotechnol. Biochem. 56, 1839-1844.

68. Lee, S. H., Taguchi, H., Yoshimura, E., Minagawa, E., Kaminogawa, S., Ohta, T., and Matsuzawa, H. (1994) Biosci. Biotechnol. Biochem. 58, 1490-1495.

69. Lee, S. H., Taguchi, H., Yoshimura, E., Minagawa, E., Kaminogawa, S., Ohta, T., and Matsuzawa, H. (1996) Protein Eng. 9, 467-469.

168

70. Ishikawa, K., Ishida, H., Matsui, I., Kawarabayasi, Y., and Kikuchi, H. (2001) Appl. Environ. Microbiol. 67, 673-679.

71. Colombo, S., Dauria, S., Fusi, P., Zecca, L., Raia, C. A., and Tortora, P. (1992) Euro. J. Biochem. 206, 349-357.

72. Chan, M. K., Gong, W. M., Rajagopalan, P. T. R., Hao, B., Tsai, C. M., and Pei, D. H. (1997) Biochemistry 36, 13904-13909.

73. Holden, H. M., Tronrud, D. E., Monzingo, A. F., Weaver, L. H., and Matthews, B. W. (1987) Biochemistry 26, 8542-8553.

74. Jancarik, J., and Kim, S. H. (1991) J. Appl. Crystallogr. 24, 409-411.

75. McDonald, N. Q., Panayotatos, N., and Hendrickson, W. A. (1995) EMBO J. 14, 2689-2699.

76. Furey, W., and Swaminathan, S. (1997) Methods Enzymol. 277, 590-620.

77. Kleywegt, G. J., and Jones, T. A. (1996) Acta Crystallogr. D52, 826-828.

78. Kleywegt, G. J., and Jones, T. A. (1999) Acta Crystallogr. D55, 941-944.

79. 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., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Acta Crystallogr. D54, 905-921.

80. Grams, F., Dive, V., Yiotakis, A., Yiallouros, I., Vassiliou, S., Zwilling, R., Bode, W., and Stocker, W. (1996) Nat. Struct. Biol. 3, 671-675.

81. Nicholls, A., Bharadwaj, R., and Honig, B. (1993) Biophys. J. 64, A166-A166.

82. Monzingo, A. F., and Matthews, B. W. (1984) Biochemistry 23, 5724-5729.

83. Cheng, T. C. (2003) Dissertation.

84. Rajagopalan, P. T. R., Yu, X. C., and Pei, D. H. (1997) J. Am. Chem. Soc. 119, 12418-12419.

85. Lu, G. G. (2000) J. Appl. Crystallogr. 33, 176-183.

86. Gibrat, J. F., Madej, T., and Bryant, S. H. (1996) Curr. Opin. Struct. Biol. 6, 377-385.

169 87. Holm, L., and Sander, C. (1998) Nucleic Acids Res. 26, 316-319.

88. Shi, J. Y., Blundell, T. L., and Mizuguchi, K. (2001) J. Mol. Biol. 310, 243-257.

89. Brown, C. K., Madauss, K., Lian, W., Beck, M. R., Tolbert, W. D., and Rodgers, D. W. (2001) Proc. Natl. Acad. Sci. U. S. A. 98, 3127-3132.

90. Artymiuk, P. J., Grindley, H. M., Park, J. E., Rice, D. W., and Willett, P. (1992) FEBS Lett. 303, 48-52.

91. Madden, T. L., Tatusov, R. L., and Zhang, J. H. (1996) Methods Enzymol. 266, 131- 141.

92. Merritt, E. A., and Murphy, M. E. P. (1994) Acta Crystallogr. D50, 869-873.

93. Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 946-950.

94. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994) Nucleic Acids Res. 22, 4673-4680.

95. Krzycki, J. A., Lehman, L. J., and Zeikus, J. G. (1985) J. Bacteriol. 163, 1000-1006.

96. Smith, P. H., and Mah, R. A. (1966) Appl. Microbio. 14, 368-71.

97. Ferry, J. G. (1992) J. Bacteriol. 174, 5489-5495.

98. Grahame, D. A. (1987) Federation Proceedings 46, 2179-2179.

99. Grahame, D. A., and DeMoll, E. (1996) J. Biol. Chem. 271, 8352-8358.

100. Gencic S and Grahame DA. (2003) J Biol Chem 278, 6101-10.

101. Grahame, D. A., and Stadtman, T. C. (1987) J. Biol. Chem. 262, 3706-3712.

102. Ragsdale, S. W., and Kumar, M. (1996) Chem. Rev. 96, 2515-2539.

103. Murakami, E., and Ragsdale, S. W. (2000) J. Biol. Chem. 275, 4699-4707.

104. Grahame, D. A., Dai, Y. R., and DeMoll, E. (1997) FASEB J. 11, 2613.

105. Krzycki, J. A., and Zeikus, J. G. (1984) J. Bacteriol. 158, 231-237.

106. Lindahl, P. A. (2002) Biochemistry 41, 2097-2105.

170 107. Ferry, J. G. (1995) Annu. Rev. Microbiol. 49, 305-333.

108. Lindahl, P. A., and Chang, B. (2001) Orig. Life Evol. Biosph. 31, 403-434.

109. Huber, C., and Wächtershäuser, G. (1997) Science 276, 245-247.

110. Zinder, S. H. (1993) in Methanogenesis: ecology, physiology, biochemistry & genetics (Ferry, J. G., Ed.), Chapman & Hall, New York.

111. Wolfe, R. S. (1993) in Methanogenesis: ecology, physiology, biochemistry & genetics (Ferry, J. G., Ed.) pp 1-32, Chapman & Hall, New York.

112. Dobbek, H., Svetlitchnyi, V., Gremer, L., Huber, R., and Meyer, O. (2001) Science 293, 1281-1285.

113. Drennan, C. L., Heo, J. Y., Sintchak, M. D., Schreiter, E., and Ludden, P. W. (2001) Proc. Natl. Acad. Sci. U.S.A. 98, 11973-11978.

114. Darnault, C., Volbeda, A., Kim, E. J., Legrand, P., Vernede, X., Lindahl, P. A., and Fontecilla-Camps, J. C. (2003) Nat. Struct. Biol.10, 271-279.

115. Doukov, T. I., Iverson, T. M., Seravalli, J., Ragsdale, S. W., and Drennan, C. L. (2002) Science 298, 567-572.

116. Grahame, D. A. (1991) J. Biol. Chem. 266, 22227-22233.

117. Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chem. 193, 265-75.

118. Hao, B. (2002) Dissertation.

119. Terlesky, K. C., Nelson, M. J. K., and Ferry, J. G. (1986) J. Bacteriol. 168, 1053- 1058.

120. Edman, P. (1950) Acta Chem. Scand., 4283.

121. Ragsdale, S. W. (1991) Crit. Rev. Biochem. Mol. Biol. 26, 261-300.

122. Bhaskar, B., DeMoll, E., and Grahame, D. A. (1998) Biochemistry 37, 14491- 14499.

123. Ciriolo, M. R., Desideri, A., Paci, M., and Rotilio, G. (1990) J. Biol. Chem. 265, 11030-11034.

171 124. Musci, G., DiMarco, S., Bellenchi, G. C., and Calabrese, L. (1996) J. Biol. Chem. 271, 1972-1978.

125. Rapisarda, V. A., Volentini, S. I., Farias, R. N., and Massa, E. M. (2002) Anal. Biochem. 307, 105-109.

126. Georgiou, G., and Valax, P. (1999) in Amyloid, Prions, and Other Protein Aggregates pp 48-58.

127. Middelberg, A. R. (2002) Trends Biotechnol. 20, 437-443.

128. Muller, C., and Rinas, U. (1999) J. Chromatogr. A 855, 203-213.

129. Clark, E. D. (2001) Curr. Opin. Biotechnol. 12, 202-207.

130. Wei, G., and Moreira, A. R. (1998) Abstr. Pap. Am. Chem. Soc. 216, 100-BTEC.

131. MaupinFurlow, J., and Ferry, J. G. (1996) J. Bacteriol. 178, 340-346.

132. Levin, M. K., and Patel, S. S. (1999) J. Biol. Chem. 274, 31839-31846.

133. Balbo P, O. M. (2003) Acta Crystallogr D59, 721-3.

134. Schwede, T., Diemand, A., Guex, N., and Peitsch, M. C. (2000) Res. Microbiol. 151, 107-112.

135. Combet, C., Jambon, M., Deleage, G., and Geourjon, C. (2002) Bioinformatics 18, 213-214.

136. Corpet, F. (1988) Nucleic Acids Res. 16, 10881-10890.

137. Gouet, P., Courcelle, E., Stuart, D. I., and Metoz, F. (1999) Bioinformatics 15, 305-308.

138. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994) Nucleic Acids Res. 22, 4673-4680.

139. Seravalli, J., Gu, W. W., Tam, A., Strauss, E., Begley, T. P., Cramer, S. P., and Ragsdale, S. W. (2003) Proc. Natl. Acad. Sci. U. S. A. 100, 3689-3694.

172