Toward the Understanding of Complex Biochemical Systems: the
Significance of Global Protein Structure and
Thorough Parametric Analysis
Except where reference is made to the work of others, the work described in this
dissertation is my own or was done in collaboration with my advisory committee. This
dissertation does not include proprietary or classifled information.
Robert Moore
Certiflcate of Approval:
Holly Ellis
Associate Professor
Chemistry and Biochemistry
Douglas Goodwin, Chair
Associate Professor
Chemistry and Biochemistry
Evert Duin
Associate Professor
Chemistry and Biochemistry
German Mills
Associate Professor
Chemistry and Biochemistry
George Flowers
Acting Dean
Graduate School
Toward the Understanding of Complex Biochemical Systems: the
Significance of Global Protein Structure and
Thorough Parametric Analysis
Robert Moore
A Dissertation
Submitted to
the Graduate Faculty of
Auburn University
in Partial Fulflllment of the
Requirements for the
Degree of
Doctor of Philosophy
Auburn, Alabama
August 10, 2009
Toward the Understanding of Complex Biochemical Systems: the
Significance of Global Protein Structure and
Thorough Parametric Analysis
Robert Moore
Permission is granted to Auburn University to make copies of this dissertation at its
discretion, upon the request of individuals or institutions and at
their expense. The author reserves all publication rights.
Signature of Author
Date of Graduation
iii
Vita
Robert Lee Moore was born to Robert Lynn Moore and Margaret Dale Moore on
September 6, 1981. Robert graduated from the Alabama School of Mathematics and Science
in Mobile, AL in 1998. From there, Robert attended Huntingdon College in Montgomery,
AL where he earned Bachelor of Arts degrees in mathematics, chemistry, and cell biology
in 2002. While at Huntingdon College, Robert was inducted into the Kappa Mu Epsilon
mathematics honor society, Beta Beta Beta biology honor society, and Alpha Beta national
honor society. Robert married his wife Emilia Anna Lu?snia, whom he met at Huntingdon
College, on July 19, 2003. Shortly after marrying, he began employment in the Department
of Chemistry and Biochemistry at Auburn University as a lab technician for the general
chemistry labs. Robert joined the graduate program in 2004 and began work on his Ph.D.
in biochemistry.
iv
Dissertation Abstract
Toward the Understanding of Complex Biochemical Systems: the
Significance of Global Protein Structure and
Thorough Parametric Analysis
Robert Moore
Doctor of Philosophy, August 10, 2009
(B.A. Chemistry, Huntingdon College, 2002)
(B.A. Cell Biology, Huntingdon College, 2002)
(B.A. Mathematics, Huntingdon College, 2002)
178 Typed Pages
Directed by Douglas Goodwin
Enzymes are a highly diverse set of macromolecules, and, given their size and biocat-
alytic signiflcance, each is an extremely complex system to study. As such, many assump-
tions must be made to simplify these systems to a manageable level to study. Unfortunately,
the more complex the system, the more simpliflcations need to be introduced. Most simpli-
flcations entail two difierent types of assumptions: structural assumptions, and parametric
assumptions. Since catalysis is generally limited to a small region of the enzyme (the active
site), most non-active site structures are ignored if there is no evidence for substrate binding
or allosteric control. In regards to parametric analysis, the assumption is that as long as
one variable is held constant, that variable will have the same efiect on a system regard-
less of changes to other variables. Catalase-peroxidases provide an ideal system to analyze
these assumptions. Although having an active site identical to monofunctional peroxidases,
catalase-peroxidases have signiflcant catalase activity. Difierentiation between which cat-
alytic cycle is utilized appears to be linked to the pH of the environment. The commonly
held assumption was that by varying the pH at substrate-saturating conditions, the pH
v
optima of the two activities could be determined. Here, that assumption is dissected by
varying pH and substrate concentrations simultaneously. This revealed substrate-dependent
inhibition that had resulted in misidentiflcation of the pH optima and possible misinter-
pretations of structural data. Another assumption was that by only providing the enzyme
with the substrates required for one activity, the two difierent catalytic cycles could be
studied and understood separately. Here, by placing the system in an environment where
both activities could occur, it became clear that the two activities are synergistic and re-
sult in broadening the catalase pH range of the enzyme. Furthermore, the synergy of the
two catalytic cycles rather than competition emphasized that the classical representation of
catalase-peroxidase activity could not be true, leading to the proposal of a new mechanism.
Previous studies have shown that an entire domain absent in monofunctional peroxidases
is necessary for any catalysis in catalase-peroxidases. Here, by creating variants of residues
25 ?A (and further) away from the active site in two hydrogen bonding networks at the
interface of the two domains, the signiflcance of these non-active site networks is shown to
be as great as some of the active site structures.
vi
Acknowledgments
I could not begin to pretend that any of my achievements could have been possible
without being in uenced and surrounded by virtually innumerable, incredibly giving people.
Listed here are only a few of those.
My advisor, Dr. Douglas Goodwin, has established the ideal setting for growing in-
dependent researchers. He is clearly conscientious about his role as a mentor, not only to
aspiring scientists, but also to future mentors. Dr. Holly Ellis has taken time to preview
and comment every manuscript I have written, and has appreciated as much as anyone
my sarcasm and humor. Dr. Evert Duin has been of great assistance with EPR, and is a
model of genuine friendliness. It is obvious that his role as an educator is not limited to his
students, but encompasses who he is. I have shared many conversations with Dr. German
Mills about my research, science, academia, and beyond - and in every one his passion is
evident. I always walked away thinking how much fun it was to talk, and more inspired
and passionate about my work.
My laboratory predecessors and contemporaries (Drs. Li, Varnado, Baker-Hartfleld,
and Cook) were always very giving and patient in training me, and seeing how they ap-
proached research and how they matured as scientists gave me references to place standards
and expectations for myself. Dr. Carma Cook provided KatGC and KatGN data and the
undergraduates Luke Powell and Rachel Williams assisted me in the pH proflling of KatG
and preparation of the Y111A KatG mutant.
More personally, I absolutely must recognize my parents who opened every door they
could possibly open for me, and never shut a single one.
My amazing wife, Emilia: no one has been more supportive of me than you. Everything
I do, you are right beside me. I am blessed beyond measure. This work is dedicated to our
sons Isen Benjamin and Montana Drew.
vii
Style manual or journal used Biochimica et Biophysica Acta (together with the style
known as \aums").
Computer software used The document preparation package TEX (speciflcally LATEX)
together with the departmental style-flle aums.sty, Microsoft Excel for data analysis,
GraphPad Prism 4.0c for graph preparations, Swiss-PdbViewer 3.7 and MegaPOV 1.2.1
for protein structures, ChemDraw 10.0 and WinDrawChem 1.6.2 for reaction schemes and
molecularstructures, andMicrosoftPowerpoint2003andGnuImageManipulationProgram
2.4.4 for image conversions.
viii
Table of Contents
List of Tables xii
List of Figures xiii
1 Literature Review 1
1.1 Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Enzymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Structural Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Analysis Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.2.3 Therapeutics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1.3 Catalase-peroxidases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
1.3.1 Reactive Oxygen Species . . . . . . . . . . . . . . . . . . . . . . . . . 41
1.3.2 Monofunctional Catalases . . . . . . . . . . . . . . . . . . . . . . . . 43
1.3.3 Monofunctional Peroxidases . . . . . . . . . . . . . . . . . . . . . . . 45
1.3.4 Catalase-peroxidases . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2 Complexity of KatG Kinetics Revealed by pH Analysis 56
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.2.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.2.2 Expression and Puriflcation of wtKatG . . . . . . . . . . . . . . . . 57
2.2.3 Peroxidase Activity Assays . . . . . . . . . . . . . . . . . . . . . . . 61
2.2.4 Catalase Activity Assays . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.2.5 Circular Dichroism Spectroscopy . . . . . . . . . . . . . . . . . . . . 62
2.2.6 pKa Determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.3.1 Kinetic Parameters for Peroxidase Activity of KatG . . . . . . . . . 66
2.3.2 Kinetic Parameters for Catalase Activity of KatG . . . . . . . . . . 72
2.3.3 Kinetic Parameters and pKas . . . . . . . . . . . . . . . . . . . . . . 76
2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.4.1 Optimal Peroxidase Activity and ABTS-dependent Inhibition . . . . 78
2.4.2 Difierent pH Optima for Binding and Activity in Catalase Cycle . . 82
3 Presence of Reducing Substrates Broadens Catalase Activity pH
Range 84
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.2.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
ix
3.2.2 Expression, Puriflcation, and Reconstitution of EcKatG . . . . . . . 85
3.2.3 Activity Assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.2.4 End-point Assays and UV-visible Spectra . . . . . . . . . . . . . . . 86
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.3.1 Efiect of Reducing Substrates on Oxygen Production . . . . . . . . . 87
3.3.2 Evaluation of Role of pH on Activation Efiects . . . . . . . . . . . . 89
3.3.3 Efiect of Reducing Substrate Presence on Reacted Enzyme Spectra . 92
3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4 Generation of Mixed Spin-state Population via Y111A Substitution 101
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.2.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.2.2 Cloning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.2.3 Expression and Puriflcation . . . . . . . . . . . . . . . . . . . . . . . 104
4.2.4 Absorption Spectra and Activity Assays . . . . . . . . . . . . . . . . 104
4.2.5 Stopped- ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.2.6 Magnetic Circular Dichroism . . . . . . . . . . . . . . . . . . . . . . 105
4.2.7 Electron Paramagnetic Resonance . . . . . . . . . . . . . . . . . . . 106
4.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.3.1 UV-visible Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.3.2 Cyanide Binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.3.3 Magnetic Circular Dichroism . . . . . . . . . . . . . . . . . . . . . . 110
4.3.4 Electron Paramagnetic Resonance . . . . . . . . . . . . . . . . . . . 113
4.3.5 Steady-state Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5 Comprehensive Analysis of Interdomain Interface Single Variants 120
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.2.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.2.2 Cloning of R117A, R479A, D482A, and D597A . . . . . . . . . . . . 121
5.2.3 Expression and Puriflcation . . . . . . . . . . . . . . . . . . . . . . . 122
5.2.4 UV-visible Absorption Spectra and Activity Assays . . . . . . . . . . 122
5.2.5 Circular and Magnetic Circular Dichroism . . . . . . . . . . . . . . . 123
5.2.6 Electron Paramagnetic Resonance Spectroscopy and Spin Quantiflca-
tion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.3.1 Mutagenesis, Expression, and Puriflcation of KatG Interdomain Vari-
ants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.3.2 UV-visible Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . 124
5.3.3 Magnetic Circular Dichroism . . . . . . . . . . . . . . . . . . . . . . 130
5.3.4 Electron Paramagnetic Resonance . . . . . . . . . . . . . . . . . . . 130
5.3.5 Steady-state Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
x
6 Summary 139
6.1 Assumption: pH-proflling at Saturating Substrate Concentrations . . . . . . 140
6.2 Assumption: Catalysis Can Be Difierentiated by pH and Substrate Avail-
ability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.3 Assumption: Global Features Play Structural Roles, Active Site Features
Play Functional Roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
bibliography 144
xi
List of Tables
1.1 Examples of Non-heme Iron Cofactors. . . . . . . . . . . . . . . . . . . . . . 7
1.2 Enzyme Commission Classes. . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.3 Antibiotic Classes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.1 Lowest Concentration of ABTS at Each pH Where Inhibition Was Observed. 70
2.2 Observed Kinetic Parameters for E. coli KatG Reducing Substrates. . . . . 74
2.3 Approximated Rate Constants for E. coli KatG. . . . . . . . . . . . . . . . 79
4.1 Spectral Features of Ferric and Ferrous Wild Type and Y111A KatG. . . . 109
4.2 Apparent Catalase Kinetic Parameters of Wild Type and Y111A KatG. . . 116
4.3 Apparent Peroxidase Kinetic Parameters of Wild Type and Y111A KatG. . 116
5.1 Ratios of Various EPR Signals Observed in wtKatG and Variants. . . . . . 132
5.2 Apparent Catalase Kinetic Parameters of wtKatG and Interdomain Interface
Variants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.3 Apparent Peroxidase Kinetic Parameters of wtKatG and Interdomain Inter-
face Variants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5.4 Ratio of Catalase to Peroxidase Activity Relative to Wild Type. . . . . . . 138
xii
List of Figures
1.1 Heme. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Various Heme b Ligand Environments. . . . . . . . . . . . . . . . . . . . . . 11
1.3 Efiects of Ligands on FeIII d-electrons. . . . . . . . . . . . . . . . . . . . . . 20
1.4 Electronic Transitions Observed in Heme Peroxidases. . . . . . . . . . . . . 22
1.5 Electron Paramagnetic Absorption. . . . . . . . . . . . . . . . . . . . . . . . 23
1.6 EPR Absorption Derivatives. . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.7 Origin of MCD Signals: A-term. . . . . . . . . . . . . . . . . . . . . . . . . 27
1.8 Origin of MCD Signals: C-term. . . . . . . . . . . . . . . . . . . . . . . . . 29
1.9 Origin of MCD Signals: B-term. . . . . . . . . . . . . . . . . . . . . . . . . 30
1.10 Relative Concentrations of S, P, E, and ES During Reaction Progression. . 33
1.11 Comparison of Catalase-peroxidase Active Site to Monofunctional Catalase
and Peroxidases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
1.12 Identiflcation of Interhelical Insertions and C-terminal Domain in Catalase-
peroxidase. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
1.13 Classic Catalase-peroxidase Scheme. . . . . . . . . . . . . . . . . . . . . . . 55
2.1 Peroxidase Cycle Scheme with pH Dependence. . . . . . . . . . . . . . . . . 58
2.2 Catalase Cycle Scheme with pH Dependence. . . . . . . . . . . . . . . . . . 59
2.3 Initial Velocity of Peroxidase Activity Under Saturating Substrate Conditions. 67
2.4 Observed Peroxidase Kinetic Parameters Under Saturating ABTS Conditions. 69
2.5 Evidence of ABTS Inhibition. . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.6 Observed Peroxidase Kinetic Parameters Versus pH: Constant [ABTS]. . . . 71
xiii
2.7 Observed Peroxidase Kinetic Parameters Versus pH: Constant [H2O2]. . . . 73
2.8 Evidence of Protein Unfolding at Low pH. . . . . . . . . . . . . . . . . . . . 75
2.9 Observed Catalase Kinetic Parameters Versus pH. . . . . . . . . . . . . . . 77
3.1 Efiects of Reducing Substrate on Catalase Activity at pH 5.0. . . . . . . . . 88
3.2 Evidence of Peroxidatic Consumption of Reducing Substrates. . . . . . . . . 90
3.3 pH-dependence of Activation with ABTS Compared to Catalase and Perox-
idase Activities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.4 Efiect of ABTS on Apparent KM for H2O2 at pH 5.0. . . . . . . . . . . . . 93
3.5 Efiect of ABTS on Linearity of Catalase Initial Rates. . . . . . . . . . . . . 94
3.6 \Dead-end" Species Does Not Accumulate When ABTS Is Present. . . . . . 96
3.7 A Proposed Scheme Accounting for Difierentiation Between Catalase and
Peroxidase Prior to Substrate Interaction. . . . . . . . . . . . . . . . . . . . 98
4.1 Interactions Between the N-terminal BC Interhelical Loop and C-terminal
Domain of KatG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.2 UV-visible Absorption Spectra of Native and Reduced Y111A KatG. . . . . 108
4.3 Stopped- ow Cyanide Binding Shows Two Distinct Species in Y111A KatG. 111
4.4 Prediction of Y111A KatG Ferrous Heme MCD Spectrum. . . . . . . . . . . 112
4.5 EPR Spectrum of Y111A KatG Compared to wtKatG. . . . . . . . . . . . . 114
4.6 Y111A KatG Catalase Activity. . . . . . . . . . . . . . . . . . . . . . . . . . 117
4.7 Y111A KatG Peroxidase Activity. . . . . . . . . . . . . . . . . . . . . . . . . 118
5.1 Far-UV Circular Dichroism of wtKatG and Interdomain Interface Variants. 125
5.2 UV-vis Spectra of wtKatG, KatGN, and Interdomain Interface Variants: Fer-
ric Heme. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
5.3 UV-vis Spectra of wtKatG, KatGN, and Interdomain Interface Variants: Fer-
rous Heme. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.4 MCD Spectra of KatGN, wtKatG, and Interdomain Interface Variants: Fer-
rous Heme. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
xiv
5.5 MCD Spectrum of R479A KatG Ferric Heme Compared to wtKatG and
KatGN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.6 EPR Spectra of wtKatG, KatGN, and Interdomain Interface Variants. . . . 131
5.7 Simulation of Y111A KatG Spectrum. . . . . . . . . . . . . . . . . . . . . . 132
xv
Chapter 1
Literature Review
1.1 Proteins
DNA is colloquially referred to as the blueprint of the cell. Comparing two organisms?
DNA is a means of quantifying how closely related the two are, and has even been employed
in determining approximately how long ago two species diverged from a common ancestor.
Like a blueprint, DNA is only information. The true difierence between organisms arises in
their proteomes, or the protein content of the cell - a direct consequence of difierences in
DNA.
Proteins are a diverse group of biological molecules composed of a linear sequence of
amino acids (directed by the DNA) joined by peptide linkage. This is known as the pri-
mary structure. Stretches of amino acids can form ordered structures such as alpha-helices,
beta-sheets, and random (not to be confused with disordered) coils. These are known as
secondary structural features. The tertiary structure is the overall three-dimensional fold
of a protein that is achieved spontaneously after expression. Proteins that are composed of
multiple subunits not linked through the primary structure are considered to have quater-
nary structure. Variations on all four structural levels give rise to the diversity in protein
function.
The primary types of proteins based on their function are structural proteins, proteins
involved in cell signaling, transport proteins, and catalysts or enzymes. Keratin in hair and
flngernails, collagen in connective tissues, and tubulin in the cytoskeleton are examples of
structural proteins. The role of antibodies in immune response and insulin in glucose uptake
control are examples of cell signaling by proteins. Transport proteins include hemoglobin,
1
which is responsible for transporting oxygen in the blood, and ion channels, which are re-
sponsible for transporting ions across the cell membranes. Transport proteins are frequently
grouped with cell signaling proteins, as transport and cell signaling often refer to the same
thing, such as the ligand-gated ion channels involved in muscular action.
1.2 Enzymes
The catalytic group of proteins, or enzymes, is the working force of the cell, driving
all the chemical reactions without which life would not occur. Enzymes have signiflcant
advantages over more common small-molecule catalysts. They do not require extreme tem-
peratures, pH, or pressure to be fully functional, but instead are commonly adapted to
function optimally at the conditions of an organism?s ecosystem. In spite of functioning
under commonly experienced conditions, enzyme action is still tightly controlled within
the cell through expression regulation, post-transcriptional modiflcation, and through sub-
strate, inhibitor, or cofactor availability. Enzymes catalyze reactions with high speciflcity,
frequently distinguishing between stereoisomers and recognizing entire molecules rather
than just functional groups.
Enzymology is a major aspect of biochemistry. Understanding the mechanisms by
which enzymes gain their specialization and how they undertake their task is crucial in
our fundamental knowledge of the chemistry of life. Yet the diversity achieved in the
specialization is astounding considering that proteins are all made from the same 22 amino
acids and are all coded from DNA that has only a four-letter alphabet.
Inherent in any mechanistic elucidation is relating the function of the enzyme to its
structure (internal efiects) and environment (external efiects). Knowledge of the structure
can reveal electronic, chemical, and steric information regarding an enzyme?s active site (the
site where activity occurs) and how it translates into function. It is also not uncommon to
study what is known as second-sphere efiects, that is, the role of amino acids not inside the
active site but still in proximity to or in contact with active site amino acids. Attempting
to relate structural information from sites on the enzyme distant from the active site to
2
function is more di?cult and rarely observed in literature. Still, distant features are clearly
signiflcant. Small molecule active site mimics rarely demonstrate catalytic ability compara-
ble to a full enzyme, and early life organisms could ill-afiord tying up so much energy and
resources in developing large molecules if smaller molecules could have su?ced. Learning
to relate features seemingly distant from the relevant portions of an enzyme to its function
is a daunting, but crucial step in enzymology.
The scope of external efiects on function is even broader than that of the internal efiects.
The presence of substrates, inhibitors, and cofactors all have obvious roles and are typically
the flrst to be evaluated in a new characterization. Parameters not directly involved in the
overall reaction, however, frequently are just as important. For example, temperature can
afiect folding and stability of an enzyme, and pH can afiect the protonation of active site
acids or bases as well as possible pH-induced structural changes. It is common to study
most of these parameters one variable at a time, but this assumes that the efiects of the
variables are independent of each other (i.e., that varying A will not change the efiects of B
on function). This is indubitably an oversimpliflcation of external efiects that can lead to
many pitfalls, but due to the large number of environmental parameters, it is possibly an
inevitable simpliflcation. With current techniques, even varying two parameters such as pH
and substrate concentration is extremely time intensive to get su?cient data points. The
more complex the system gets (multiple substrates, multiple functions, more parameters to
vary), the time required to eliminate this assumption grows exponentially. This is not to say
that there has been no attempt to resolve this disparity between assumption and reality.
Much has been progressed in methods of data collection and analysis of multisubstrate
systems and pH efiects, such as the efiorts of W. W. Cleland [1]. Still, the application of
this progress could stand to be propagated in practice more widely than it has been, and
needs to be further developed to account for more variables.
In order to increase our ability to relate the function of an enzyme to its internal and
external afiecters, we will flrst lay the groundwork by looking in more depth at general
structural features and enzymological analysis techniques. From there, we will consider a
3
model of a complex biochemical system, the multi-functional catalase-peroxidase (KatG),
and demonstrate how limited parametric analysis has led to misidentiflcation of KatG func-
tional characteristics and how features in the global structure are as crucial to activity as
many active site features.
1.2.1 Structural Features
Active Sites
The region of the enzyme where activity occurs is referred to as the active site. The
minimal requirements of an active site are a substrate recognition and binding mechanism
and a mechanism for catalytic activity. Both of these are typically achieved by the arrange-
ment and identity of amino acids inside the active site.
The current theory of substrate binding is called the \induced-flt model". A  exible
activesitewillinteractwithasubstratetooptimizebindinguponinteraction. Thiscanlower
the activation energy needed for the reaction by binding the substrate in a conformation
closer to the transition state, by stabilizing the transition state through polar interactions,
decreasing entropy by properly orienting the substrate for reaction, or by providing an
alternate pathway that may not be available without the presence of a catalyst.
Once binding has occurred, the reaction can occur via acid-base catalysis, covalent
catalysis (where the substrate is temporarily covalently attached to the enzyme), by bringing
molecules together that would be less likely to flnd each other in solution, or by using
cofactors. Most of these reactions can occur in solution chemistry; but by using an active
site, the enzyme decreases the concentration needed and efiectivelyeliminates anyvariability
in collision orientation and kinetic energy.
Cofactors
Some enzymes require the presence of certain molecules called cofactors for activity,
and use them for the transferring of functional groups or electrons. The role of the cofactor
is dependent upon how it is bound or interacts with the enzyme and the structure of the
4
active site around. Cofactors that are loosely bound are called coenzymes and those that
are tightly bound are called prosthetic groups. Although the boundary between loosely
and tightly bound is fuzzy, a simplifled deflnition is tightly bound cofactors are those that
cannot leave the active site once incorporated. Some prosthetic groups are even covalently
bound to the enzyme. An enzyme that requires a cofactor that has yet to associate is an
apoenzyme. Conversely, an enzyme with an incorporated cofactor is termed holoenzyme.
Cofactors are divided into two categories, non-metallo or organic cofactors and metallo
cofactors. Many organic cofactors are vitamins (such as biotin) or use vitamins as precursurs
(such as FAD from B2 or NAD+ from B3). ATP and S-adenosyl methionine are common
organic cofactors that are not vitamin derivatives. Metallo cofactors are metal ions that can
be either directly ligated to the enzyme or have an organic component that also interacts
with the enzyme. This category of cofactors is highly variegated from mononuclear centers
such as the iron in Rieske dioxygenases [2], to binuclear centers (both homonuclear [3] and
heteronuclear [4]), to metal clusters (such as the iron-sulfur clusters found in numerous
enzymes [5] or the large FeMo-co clusters in nitrogenases [6]), and to the large class of
tetrapyrrole-based metallo cofactors [7] including the porphyrins (such as iron-containing
hemes), the nickel-containing F430, the cobalt-containing cobalamin, and the magnesium
containing chlorophylls.
The identity of the cofactor can frequently be used to determine part of the function
or mechanism of an enzyme, but even those with nearly exclusive roles cannot be assured
of that role without investigation. For example, ATP is commonly used as an energy
source during a reaction by transferring a phosphate group. In the biosynthesis of another
cofactor, S-adenosyl methionine, ATP transfers the adenosine moiety [8]. Similarly, S-
adenosyl methionine is most associated as the methyl donor cofactor for methyltransferases,
but it also has biosynethetic precursor roles and is a source of 5?-deoxyadenosyl radicals used
in the synthesis of L-methionine and 5?-deoxyadenosine [9].
Metallo cofactors can hardly be considered to have \exclusive" roles like many organic
cofactors. The nature of the d-orbitals in transition metals, particularly, allows for multiple
5
oxidation states and a broad range of reduction potentials - both of which are primarily
determined by the ligand environment. The major advantage transition metals (especially
iron, copper, and manganese) have is the ability to catalyze oxygen requiring reactions
that would be spin-forbidden with most organic compounds or cofactors. Molecular (or
di-) oxygen exists in a triplet ground state, that is, two unpaired electrons in antibonding
orbitals; but most organic compounds exist in the singlet ground state. This parity problem
is fortuitous in that it prevents oxygen from spontaneously oxidizing all organic compounds
it encounters, but is also the reason that nature requires transition metals that can reach
multiple oxidation states and reduction potentials to aid in the catalysis of many essential
biochemical reactions.
Iron is possibly the most versatile of these transition metals cofactors. A summary of
non-heme iron centers and some of the representatives and reactions catalyzed is shown in
Table 1.1 [2{6, 10{13]. Table 1.1 lists only resting oxidation states, but FeIV?O (ferryl
iron) is also seen in many intermediates. Reduction potentials for the various cofactors can
range from as low as -500 mV in 2Fe2S ferredoxins [13] to as high as 1.1 V in the diiron(III)
intermediate in toluene/o-xylene monooxygenase hydroxylase [3], illustrating how broad the
electronic states are that iron and iron derived cofactors can reach.
Heme iron is distinct from other iron cofactors in that it contains an organic component
in the form of a porphyrin ring. This supplies the iron with four nitrogen ligands and leaves
two iron binding sites available for ligation by the protein, solvent, or substrate. Oxidation
states are stabilized and reduction potentials are set by the identity of the heme, the nature
of the protein ligand and associated amino acids, and the interactions of the protein with
the porphyrin itself.
Although many other forms exist, the three most abundant forms of heme are heme
a, heme b, and heme c (Figure 1.1). Other types of heme vary in more positions than the
three shown in Figure 1.1. Heme b is the most widely used in nature, and is the cofactor
used by the model system catalase-peroxidase that we will be studying further on. The
difierences between heme a and b occur at the R1 and R3 position. In both cases, the heme
6
Cofactor Ligands Ox. St. Representative Reactions
Fe 2His/1CO2 II extradiol cleaving catechol
dioxygenases
oxidative ring cleav-
age
Reiske dioxygenases arene hydroxylation
fi-ketoglutarate dependent
enzymes
demethylation, hy-
droxylation, desat-
uration, ring reac-
tions, epimerization
pterin-dependent enzymes monohydroxylation
of aromatic amino
acids
II/III sulfur oxygenase reductase S ??! SO2{3 +HS{
Hpp epoxidase oxidative cyclization
3His II cysteine dioxygenase cysteine ! cysteine
sulflnic acid
3His/1CO2 III superoxide dismutase O?{2 ??! O2 +H2O2
Fe-Fe 2His/4CO2 II/II hydroxylase arene hydroxylation
Fe-Mn 2His/4CO2 II class 1c ribonucleotide re-
ductases
oxidation and elec-
tron transfer
2Fe2S 4Cys II/III ferredoxins electron transfer
2His/4Cys II/III Reiske proteins electron transfer
4Fe4S 4Cys II/III ferredoxins electron transfer
3Cys II/III quinolate synthase condensation
7FeMo9S 1His/1Cys II/III nitrogenase N2 ??! NH3
Table 1.1: Examples of Non-heme Iron Cofactors.
7
a functional groups are more electron withdrawing than those on heme b (hydroxyfarnesyl
is a long chain hydrocarbon), increasing the reduction potential of a-type hemes. Heme
c difiers from heme b by becoming covalently attached to the protein via two thioether
linkages at the proximal carbons of the R1 and R2 vinyl groups. This covalent linkage is
believed to primarily serve as structural support to the protein; however, given the same
axial ligand set, heme c reduction potentials span a broader range than heme b [15, 16]. This
may be due to the rigidity with which the heme is held afiecting the interaction between
the iron and proximal (typically histidine) ligand. Forcing the heme into speciflc positions
or shapes would allow the protein to tune the reduction potential more speciflcally, which
would be advantageous in the primary role of electron transfer in heme c containing proteins
(cytochromes c).
Heme b is a prime example of how protein interactions afiect the versatility of the
cofactor, and also is the cofactor of the model enzyme KatG that this research is focused
around. Heme b-containing proteins perform sensing or transport of O2, NO, and CO,
electron transport, heme transport, and a variety of metabolic and redox reactions. This
is a direct result of the wide variety of protein folds (at least 20), heme anchoring residues,
heme face interacting residues, heme ligands, and extended ligand environments [17]. For
example, the propionate groups frequently form salt-bridges with arginines, but lysines,
histidines, tyrosines, and even serines, threonines, and backbone amines can all be utilized
to hold the heme propionate.
Even more in uential on setting the reduction potential of the iron are the heme face
interacting residues and the axial ligands. Figure 1.2 shows a sample of the various types of
heme b ligand environments. The more anionic in nature the axial ligand is, the lower the
reduction potential of the iron will be [16, 26, 27]. From most anionic to least would be the
tyrosinate ligand in catalases, followed by the thiolates observed in the P450 cytochromes,
then the histidine-aspartate couple in peroxidases, and flnally the neutral histidine in the
globin proteins [27]. The low reduction potential in catalases and peroxidases stabilizes
higher oxidation states, which are used in their reactions with hydrogen peroxide. On the
8
N N
NN
R 1
R 2
R 3
F e
O H
O
O H
O
heme a: R 1 =  hyd ro xyfarnesyl
R 2 =
R 3 =
C H C H 2
C H O
C H C H 2
C H 3
heme b : R 1 =
R 2 =
R 3 =
C H C H 2
C H 3
C H (C H 3) S Hheme c: R 1 =
R 2 =
R 3 =
C H (C H 3) S H
Figure 1.1: Heme. Structures were retrieved from the PubChem Public Chemical Database
using the following CID numbers in the chemical search: heme a - 5288529, heme b - 4973,
heme c - 25202875 [14].
9
other hand, the higher reduction potential in the globin proteins destabilizes the higher
oxidation states and results in FeII as being the native state for hemoglobin and myoglobin
and allows for the transport of dioxygen. Oxidation of the heme iron to FeIII even converts
hemoglobin to methemoglobin, increasing the dissociation constant for O2 to the point
that it is considered incapable of binding oxygen, thus interfering with its role of oxygen
transport [28].
In spite of being able to identify relationships between the identity and ligand environ-
ment of cofactors with their chemical and electronic properties (as exemplifled particularly
by the hemoproteins), knowledge of an active site and cofactor is rarely su?cient to predict
precisely the reaction catalyzed. Even the peroxidase family of proteins, which all undergo
the same type of reaction and follow the same general reaction scheme, metabolize a wide
variety of substrates that are often speciflc to individual members of the family. This is the
limitation of analyzing active sites alone, and again reemphasizes the need to determine the
roles of more distant enzyme features in catalysis.
Domains
Protein domains are discrete folding units of approximately 50 to 200 amino acids
that are spatially distinct in the three-dimensional structure [29{32]. Many proteins are
not large enough to have multiple domains, but others are so large that without multiple
domains the folding time would become excessive. By having stretches of amino acids that
fold spontaneously and independently of other sequences in the primary structure, a protein
can fold much more rapidly.
Domains are often the functional units of a protein, like organelles are the functional
units of the cell and organs are the functional units of a body [33]. Although being treated
as functional units, domains are categorized based on structural similarities. Interestingly
enough, not all similar domains share high sequence similarity, and not all those that share
high sequence similarity have the same function. In light of these, analysis of protein
domains is the source of many evolutionary studies [29, 31, 32].
10
Figure 1.2: Various Heme b Ligand Environments. All structures have propionate
groups oriented to the right. Amino acids are coded to the following color scheme: H -
pink, R - light blue, W - light green, D/E - red, F - dark green, Y - yellow, N - dark blue,
C - tan, T - brown, M - orange. Structures were taken from the following PDB accession
numbers: cytochrome c peroxidase - 2CYP [18], hemoglobin - 1A3N [19], catalase - 7CAT
[20], cystathionine fl-synthase - 1JBQ [21], cytochrome P450 monooxygenase - 2BVJ [22],
cytochrome b562 - 1QPU [23], hemopexin - 1QHU [24], cytochrome b5 - 1EUE [25].
11
Some domains are well-characterized. Most notably are those that appear with high
frequency such as the zinc flnger domains responsible for binding DNA [32, 34], or domains
observed in well-characterized metabolic pathways such as the TIM-barrel in triosephos-
phate isomerase in the glycolytic pathway.
The zinc flnger domain is the most prevalent domain in eukaryotes. The domain
sequence is only about 30 amino acids in length, and being so small, the role of nearly all of
the amino acids is known. It is an example of a domain where both sequence and structure
are highly conserved in spite of such wide species distribution. The structure contains two
anti-parallel fl-strands followed by an fi-helix, with a hydrophobic and an aromatic residue
working concomitantly with the zinc-binding to stabilize the fold. The zinc is coordinated
by four residues, most commonly two cysteines and two histidines, with the alternative
having the C-terminal histidine replaced by a cysteine. Zinc flnger proteins have multiple
repeats of the zinc flnger domain, and, when in complex with DNA, the fi-helices flt into the
major groove. The identity of the amino acids on the fi-helices determine what sequence of
DNA will be recognized [34]. The understanding of zinc flngers has progressed to the point
that it has been proposed that by choosing the order of zinc flngers in in a protein, one
can engineer proteins to recognize desired DNA sequences [34, 35]. The structure of zinc
flngers appear to be ideal for DNA interactions and it is unsurprising that these domains
are commonly associated with transcription regulators. More recently, however, efiorts are
being made to better characterize the protein-protein interactions of zinc flnger domains,
such as the zinc flnger protein FOG1 (Friend Of GATA1) that uses four of its nine flngers
to interact with the GATA1 transcription factor [36, 37].
The TIM-barrel domain is named for the enzyme in which it was flrst characterized
(triosephosphate isomerase). Consisting of eight fi-helices and eight parallel fl-strands, it
is the most common domain in all living organisms, found in approximately 10% of all
enzymes. The SCOP database (a structural database that will be addressed during the
discussion of bioinformatics below) recognizes 33 difierent superfamilies that share this
domain [38]. Eight of these are indicated to share a similar phosphate binding site, but
12
variations extend all the way to the Fe-S containing radical SAM enzymes, cobalamin
requiring enzymes, copper homeostasis CutC protein, luciferases, and many others. Our
current understanding of TIM-barrel proteins actually stems from the widespread utilization
of the fold. Although there is very little sequence similarity among difierent superfamilies,
the highest sequence conservation within a superfamily occurs in association with the fl-
strands. This is due to a clustering of the amino acid side chains stabilizing the core of the
protein. The clustering locations are conserved in all TIM-barrel proteins, even though the
sequence identity is not [39]. The loops connecting the ends of the fl-strands to the helices
typically carry conserved catalytic residues and form the active site [40{42].
The extent of our knowledge of the global features of zinc flngers and TIM-barrels is
directly linked to the small size of the zinc flngers and the prevalence in nature of both
domains. Zinc flngers are the smallest folding motifs that behave as actual domains. As
the size increases, so does the complexity. Even with the widespread distribution of TIM-
barrels, the fi-helices have gone largely untouched in the literature - yet they compose the
majority of the domain sequence. Most domains are large like the TIM-barrels, yet a scan of
the SCOP database would reveal that most domains are unique to a single protein. In these
cases, large-scale comparisons cannot be performed, such as those used to identify the clus-
tering patterns in TIM-barrel proteins. The large sizes and limited comparable examples
are what make studying the global features of domains so arduous and potentially un-
fruitful. Multidomain enzymes, however, can provide insight into global structure/function
relationships.
Multidomain enzymes are believed to arise from the fusion of two or more enzyme
precursors. These commonly appear to be the result of a gene duplication event, but this
is not always the case. There are many known beneflts to having multiple domains in
an enzyme. One example of multidomain proteins has already been considered in the zinc
flnger proteins. Another beneflt is that each domain could bind difierent substrates required
for activity, such as the PEP-binding TIM-barrel domain and nucleotide binding domain of
pyruvate kinase [43]. Pyruvate kinase also has a third domain that binds allosteric regulators
13
[44]. Multiple domains can also serve to protect reaction intermediates or bring catalytic
sites into close proximity to prevent the release of product prior to its next metabolic step,
such as seen in the enzyme TrifGART in purine synthesis in higher eukaryotes [45]. By
being covalently linked through the primary sequence, multiple domains impart structural
stability to each other and have the potential to alter each others structures from what they
would have been isolated in solution [46]. By asking whether a speciflc task carried out by
a multidomain enzyme could still be performed as e?ciently if the domains were separated,
and by analyzing the efiects that the interdomain interfaces have on structure and function,
one can make substantial strides in correlating global structure to function.
Post-translational Modiflcations
Post-translational modiflcations (PTMs) of proteins are not uncommon. Many modi-
flcations are seen regularly, such as the formation of disulflde bridges between cysteines or
the addition of a lipid to an amino acid. The main categories of PTMs are the addition of
a functional group (i.e.: methylation of a lysine or arginine, converting the carboxyl moiety
at the C-terminus to an amide, attachment of heme c), changing the nature of an amino
acid (i.e.: arginine to citrulline), and structural changes (i.e.: sulflde bridges, proteolysis)
[47].
The di?culty PTMs pose to enzymology is that the genetic sequence cannot be used
to predict PTMs, even though it is used frequently to predict the protein sequence. Since
PTMs can only be found empirically, there will always be the possibility of encountering
modiflcations that are not yet described; but setting out in search for a PTM would be
wasteful in terms of time and resources. Instead, PTMs are usually discovered during
regular protein analyses, however many approaches do have potential pitfalls requiring
conflrmation from an array of approaches. Some PTMs will afiect the molecular weight
of a protein through either the addition of a functional group or the removal of some
part of the protein. These can be detected in any sort of size analysis such as (in increasing
usefulness) somesize-based chromatographictechniques, some electrophoretic techniques, or
14
bio-mass spectroscopy. Similarly, covalent modiflcations could afiect the apparent molecular
weight and would produce unexpected mass fragments in hard (molecule fragmenting) mass
spectroscopy. Possibly the most straightfoward method of detecting a PTM is through
electron density mapping during X-ray crystallography. As this produces an actual image
of the protein, PTMs can be observed directly. The pitfall of hard mass spectroscopy
and X-ray crystallography is that they have the potential to induce covalent modiflcation,
reinforcing the need for corroboration from other techniques.
Once a PTM is observed, the next step is to deduce its role in the protein. Some PTMs
are obvious and well-described, particularly those known to act as cell signalers to direct
the protein to its location in the cell. Others may be more ambiguous, such as the role
of the disulflde bridge. It may serve only a structural role, or the highly oxidized nature
may play a role in catalysis or cofactor regeneration. Novel PTMs are even more di?cult
to assign a role as there is not always a good comparison in the existing literature. We will
observe one such novel PTM in our model enzyme.
1.2.2 Analysis Techniques
Bioinformatics
Bioinformatics is a useful "flrst step" in enzymology. Essentially, it is the process
of utilizing databases of information to compare a given enzyme under investigation to
other studied enzymes in order to identify common and unique features, establish trends,
make functional or structural predictions, or determine evolutionary relationships. We have
already considered an example of the use of bioinformatics with the discovery of the side
chain clustering in the fl-strands of TIM-barrels. The clustering commonality was identifled
by applying graph theory algorithms to the structures of 36 TIM-barrel proteins that shared
less than 10% sequence identity with any other protein in the data set [39]. Here we will
survey some of the most common databases and applications including enzyme classiflcation
and nomenclature, structural classiflcations and databases, and sequence alignments.
15
The standard nomenclature system for enzymes is derived from the reaction which is
catalyzed. Each enzyme is assigned a four-component Enzyme Commission (EC) number,
the flrst component describing the main class of reaction catalyzed (Table 1.2). The second
two components are the sub-class and the sub-sub-class, each specifying in more detail the
reaction catalyzed. For example, the sub-class of oxidoreductases in most cases specifles
the electron donor in the reaction and the sub-sub-class specifles the electron acceptor,
with the exceptions being sub-classes 1.11. (peroxide as acceptor) and 1.15. (superoxide
as acceptor).
While knowing the EC number of an enzyme is not all that valuable by itself, ExPASy
(http://www.expasy.ch/) is a proteomics database where the EC number can be used to
cross-reference many valuable tools. Using ENZYME database, for example, a user can
access the list of all of the enzymes that share the same sub-sub-class for any comparison
purpose. Basic information can be obtained about cofactors and substrates. Each enzyme
has links to the UniProt database where sequences of that enzyme from speciflc organisms
can be found, and links to the journal database PubMed to search the literature on the
enzyme.
With the IUPAC nomenclature of enzymes being completely based on function, the EC
number is the exclusive function-based classiflcation system. Structural-based classiflcations
are not as unifled, resulting in two primary classiflcation systems: Structural Classiflcation
of Proteins (SCOP) and Class Architecture Topology Homology (CATH). The main difier-
ences between the two classiflcation systems are the treatment of domains containing both
fi-helices and fl-sheets, and an extra layer of architectural classiflcation in the CATH system.
There are also other structural databases that can evaluate proteins and suggest structural
neighbors, such as Families of Structurally Similar Proteins (FSSP) and Vector Alignment
Search Tool (VAST). These rely on the protein databank (PDB) for their comparisons. PDB
is a database where researchers can upload solved protein structures and other researchers
can download the 3D structure for their own viewing and analysis. PDB entries include
16
EC number Class Reaction type
1. Oxidoreductase oxidation-reduction reactions
2. Transferase functional group transfers
3. Hydrolase hydrolysis reactions
4. Lyase double bond formations through functional
group elimination
5. Isomerase isomerization reactions
6. Ligase bond formation requiring ATP hydrolysis
Table 1.2: Enzyme Commission Classes.
17
the downloadable structure, the techniques used in the structure determination along with
associated publications, and both SCOP and CATH classiflcations.
The classic technique in bioinformatics in its application to enzymology is sequence
alignment. Since the flrst protein sequence in 1955 [48] and the flrst complete DNA sequence
of a bacteriophage in 1977 [49], sequencing of proteins and DNA is now commonplace, and
massive databases exist that can be utilized for alignments. Aligning sequences of proteins
or DNA can reveal conserved residues (and, if predicted from the sequence, conserved
secondary structural elements) and establish evolutionary proximity. Two of the most
common alignment tools are the Basic Local Alignment Search Tool (BLAST) and Clustal.
With BLAST, a user may enter a nucleotide or amino acid sequence and have it compared to
all existing sequences within the database of the users choosing. This provides the user with
closest matches and is particularly useful for identifying evolutionary relationships. Clustal
allows for the entry of multiple sequences to be uploaded and aligned. This is particularly
useful for comparing enzymes from difierent species or closely related enzymes to identify
their similarities and difierences.
Instrumentation
Clearly, much can be learned from bioinformatic approaches to enzymology, but in
order to flll the databases the data flrst must be accumulated. Also, the information gained
from bioinformatics must be verifled experimentally. For both reasons it is essential to have
a belt full of tools for studying enzymes directly. There is such a wide array of instruments
and techniques that can be used to study enzymes that an exhaustive review would be
excessive. Instead, we will survey those most applicable to the research presented here, in
particular the spectroscopic and kinetic approaches.
UV-visible spectroscopy has many applications in enzymology. A universal applica-
tion is estimating protein concentration based on the absorbance of tryptophan, tyrosine,
and cysteine near 280 nm [50]. In regards to heme-containing enzymes, such as catalase-
peroxidases, it is particularly useful. There are a handful of electronic transitions available
18
within the porphyrin or from the porphyrin to the iron, all of which are very sensitive to
the oxidation state of the iron and its ligand environment.
If we consider solitary FeIII, the 5 d orbital electrons are degenerate and each orbital is
occupied by a single electron. In the presence of ligands, however, the on-axis orbitals (those
that lie on the axis between the metal and ligand) will be destabilized relative to those that
are ofi-axis due to electron repulsion. Ligands that destabilize the on-axis orbitals to such a
degree that the electron pairing energy is less than the energy required to place an electron
in a destabilized orbital (or splitting energy) are called strong-fleld ligands. Ligands that do
not split the energy levels so severely are called weak-fleld ligands. In the presence of a weak
fleld ligand, FeIII has a 5/2 spin, or high spin. In the presence of strong fleld ligands, it will
be 1/2 spin, or low spin (Figure 1.3). In heme proteins, two sets of degenerate orbitals (on-
and ofi-axis orbitals) do not su?ciently describe the system, as the porphyrin nitrogens are
not equivalent to the axial (z-axis) ligands and are usually closer, destabilizing the dx2?y2
orbital further. Also, the conjugation of the porphyrin slightly destabilizes the d? (dxz and
dyz) orbitals relative to the dxy orbital.
In UV-visible absorption, an electron is excited by a photon from a ground energy
state (E0) to an excited energy state (E1). The most basic example would be an electron in
the ? (bonding) molecular orbital of ethene (H2C?CH2) excited to the ?? (anti-bonding)
molecular orbital. This transition requires energy equivalent to a photon with a wavelength
near 170 nm. This amount of energy is generally too high for practical purposes; but
the more conjugated a system is, the lower the energy requirement and absorption will
occur at higher wavelengths. The porphyrin of heme is highly conjugated, and the available
transitions in peroxidases can be seen in Figure 1.4. The Soret ( ) absorption band requires
the most energy (thus being observed at the lowest wavelengths) and is the most intense
signal in a heme spectrum. The Soret absorption band of free heme occurs at 380 nm.
When incorporated in a protein, the absorption will occur at higher wavelengths. The
charge transfer (CT) bands are observed when a porphyrin electron is excited into one of
the eg orbitals of the iron. These are not expected to be observed in low-spin systems
19
Figure 1.3: Efiects of Ligands on FeIII d-electrons.
20
and are usually too weak to see when the iron is in the ferrous state. The characteristic
absorption bands for KatG in its resting state occur at 408 nm (Soret), 502 nm (CT2), and
629 nm (CT1) [51].
The main advantage of UV-visible spectroscopy is how quickly data can be obtained.
Furthermore, the protein environment (such as bufier identity, pH, and temperature) can be
varied or set at the investigators choosing. In regards to heme proteins, general conclusions
about the oxidation and spin state of the heme iron can be arrived at quickly without
consuming a large amount of protein. UV-visible spectroscopy can be utilized for kinetic
measurements as well. If the substrate or product of catalysis is an absorbing species, it
can be monitored to measure the velocity of the reaction directly. Also, by monitoring
formation of heme intermediates, individual rate constants can be measured.
The primary disadvantage of UV-visible spectroscopy is the lack of detailed information
that can be extracted from a heme spectrum. If the heme iron is high-spin, is the sixth
coordination site open or is it occupied by a weak-fleld ligand? If it is low-spin, what is the
identity of the axial ligand? The charge transfer bands and fi and fl bands are weak and
can be easily obscured by light scattering. The heme absorption bands are broad enough
that the presence of multiple species in solution are di?cult to distinguish from each other,
if possible at all.
One way to address some of these shortcomings is by using a spectroscopic technique
called electron paramagnetic resonance (EPR). The basic premise behind EPR is that when
a paramagnet (such as an unpaired electron) is placed in a magnetic fleld, the magnetic
moment of the electron can be either aligned with the magnetic fleld (ms = +1/2) or against
the magnetic fleld (ms = -1/2). This results in an energy difierence between the two possible
states for that electron and can be deflned in terms of magnetic fleld strength:
?E = gflB (1.1)
21
Figure 1.4: Electronic Transitions Observed in Heme Peroxidases.
22
Figure 1.5: Electronic Paramagnetic Absorption.
23
where fl is the Bohr magneton, B is the magnetic fleld strength, and g is the splitting
factor. Unlike UV-visible absorption where the energy used to excite the electron is varied
by changing the wavelength of light, in EPR the resonance energy is held at a constant
microwave frequency and the magnetic fleld is swept (Figure 1.5). Absorption occurs when
the energy described in equation 1.1 is equivalent to that supplied by the microwave source.
The nature of absorption in EPR spectra make them more easily interpretable by
observing the absorption derivative. This is due to the fact that the g-factor is in uenced
by changes in the environment around the paramagnet. When the g-factor is identical in
all directions, the magnetic moment is not in uenced by its orientation in relation to the
magnetic fleld and absorption occurs at a single fleld strength. This is called an isotropic
signal. Anisotropic signals occur when the local magnetic fleld around the electron is not
uniform. Anisotropic signals can be either axial or rhombic (Figure 1.6). As the system can
be oriented in any direction during the spectrum accumulation, this leads to a continuum of
absorption between the three g-values. The g-values can be determined from the derivative
spectrum by flnding the zero crossing point of an isotropic signal, the peak (or trough) and
zero crossing point of an axial signal, or the peak, zero crossing point, and trough of a
rhombic signal.
A signiflcant portion of EPR spectra analysis is empirical, particularly in bio-EPR. A
metal center and certain properties can usually be rapidly identifled from a spectrum. In
the case of heme proteins, high-spin hexacoordinate (axial) and high-spin pentacoordinate
(rhombic) species are easily distinguishable from both each other and low-spin species.
Resolution of these species using UV-visible spectroscopy is nearly impossible. Furthermore,
by taking the double integral of the absorption derivatives of the individual components,
it is possible to quantify the relative amounts of each species. If a standard is available,
concentrations can even be determined.
Although EPR can give high resolution and quantitative information concerning the
metal center environment, it does have certain drawbacks. The most apparent is that the
species must be a paramagnet. This limits detectable mono-iron or mono-heme species
24
Figure 1.6: EPR Absorption Derivatives.
25
to FeIII. Another limitation is the temperature sensitivity. For the purposes of the work
here, spectra were recorded at 10K. Higher temperatures result in depopulation of the jms
= ?1/2i doublet which has the largest signal intensity [52]. Although samples are frozen
rapidly in liquid nitrogen, there is no guarantee that they are preserved in their native
state without precipitation or protein damage. Also, the helium required to operate at such
low temperatures is becoming more di?cult to acquire as demand and prices increase [53].
Plus, instrument preparation time (vacuum pumping and cooling) and sample storage (in
liquid nitrogen) add to the relative inconvenience when compared to other spectroscopic
techniques.
A compromise between the easy-to-use but low resolution UV-visible and the compli-
cated but high resolution EPR is magnetic circular dichroism (MCD). Circular dichroism
(CD) spectroscopy measures the absorption difierence between left and right circularly po-
larized light. This is an inherent property of chiral molecules. As such, CD itself has many
practical biological applications, including determination of DNA conformation, protein-
DNA complexes, and secondary structure content of proteins [54]. MCD is the measurement
of a CD spectrum in the presence of a magnetic fleld. MCD absorption is a property of the
magnetic fleld interacting with the electronic orbitals, not chirality, making MCD signals
independent of CD signals. As a result, a raw spectrum is composed of both MCD and CD
absorbance, and the CD component must be subtracted to obtain the MCD spectrum.
There are three types of MCD signals, each one based on difierent ways the magnetic
fleld might interact with the molecular orbitals. The most straightforward of the three is
called an A-term signal. In this case, the magnetic fleld splits the excited state S1 such that
excitation into one of the split orbitals requires the absorbance of left circularly polarized
(lcp) light and the other requires right circularly polarized (rcp) light. At the wavelength
where lcp light is absorbed, there will be a strong positive signal, and there will be a strong
negative signal at the wavelength where rcp light is absorbed. This results in a signal that
resembles the isotropic EPR absorbance derivative (Figure 1.7).
26
Figure 1.7: Origin of MCD Signals: A-term.
27
The C-term signal is very similar to the A-term in principal, but instead of the magnetic
fleld splitting the excited state, the magnetic fleld splits the ground state. Again, excitation
from one of the split states will absorb lcp light and the other rcp light, but C-term signals
can be distinguished from A-term signals via temperature variations. At low temperatures,
only the lower energy state will be populated and the resulting signal will appear like a
simple absorbance band. At high enough temperatures, only the higher energy state will be
populated and the signal will appear as an upside-down absorption band. At intermediate
temperatures, more derivative-like signals will occur (Figure 1.8).
The origin of B-term signals is less straightforward. In a magnetic fleld, two transitions
may experience a mixing of the electronic states such that one of the two transitions ab-
sorbs lcp light more strongly and the other absorbs rcp light equally more strongly. There
is no splitting of degenerate orbitals such as with the A- and C-terms and therefore no
temperature dependence is observed (Figure 1.9).
The easiest application of MCD to heme proteins is distinguishing between high- and
low-spin ferrous heme. The transitions observed in ferrous heme MCD are due to the B-
bands ( ) and Q-bands (fi and fl) (Figure 1.4). The low-spin ferrous heme Q-band produces
a strong A-term signal centered near 550 nm. High-spin ferrous heme is not featureless
in this region, but the features are complicated and far less intense. The difierence in
intensity between low- and high-spin ferrous heme Q-band signals is substantial. Even
a small portion of low-spin ferrous heme would be detectable in a predominantly high-
spin population. Considering that both B-band and Q-band transitions involve exciting an
electron into the same orbital, it stands to reason that the B-band signal is also A-term. This
is correct, but at room temperature the signal is highly susceptible to linewidth broadening
[55], and difierences between low- and high-spin ferrous heme of difierent proteins cannot
be established from the B-band signals without correlation with the Q-band region of the
spectrum.
It is not quite as easy to distinguish between low- and high-spin ferric heme species as
it is with ferrous, but it is still possible. The primary difierence is how the CT bands afiect
28
Figure 1.8: Origin of MCD Signals: C-term.
29
Figure 1.9: Origin of MCD Signals: B-term.
30
the spectra. Recall from Figure 1.4 that the CT bands arise from excitation of a heme
?-electron into a d? (eg) orbital of the iron. As shown in Figure 1.3, in the presence of a
strong fleld ligand (low-spin) the d? orbitals are at a lower energy level, and CT occurs at
very high wavelengths (> 900 nm). High-spin iron will have both higher energy d? orbitals
and more transitions available. This shifts the transitions into the B- and Q-band region,
complicating the spectrum. Empirically, the B-band transition for low-spin is also more
intense than that of high-spin ferric heme.
For MCD, sample preparation is as simple as with UV-visible spectroscopy, with the
exception of possibly needing higher sample concentration. It still does not require as much
sample as EPR, and adequate spectra can be acquired without freezing the sample. Also,
the instrument itself only needs a 30 minute warm-up time - lengthier than UV-visible
spectroscopy, but much shorter than the overnight required by EPR after liquid helium
is obtained. The drawback is that it is most useful when the heme is in the FeII state,
which requires reduction of the native state for many proteins such as peroxidases. Just
like with freezing, there is no guarantee that reduction does not damage the sample or alter
the structure of the active site or protein. The common reductant sodium dithionite, for
example, in too high of concentration can acidify the solution which may cause protein to
precipitate.
Overall, UV-visible spectroscopy serves as a good preliminary tool, providing basic
information quickly and can be used for rapid concentration determination. MCD is a good
intermediate tool. It does require more sample than UV-vis and data acquisition is on the
order of hours rather than minutes, but it can more easily distinguish between high- and
low-spin heme. EPR is an advanced technique that can not only easily distinguish between
high- and low-spin ferric heme, but can also give information about the coordination sphere
of the iron and relative concentrations of the difierent species in the population.
There are many other useful techniques that are not addressed here but are worth
mentioning. Separation methods including gel electrophoresis, capillary electrophoresis,
and liquid or gas chromatography have applications spanning from size determinations to
31
assay techniques. Many techniques directly probe the three-dimensional structure, such as
x-ray crystallography, nuclear magnetic resonance (mainly for small or membrane-bound
proteins), and x-ray absorption (for structural information just peripheral to the ligand
sphere of a metal center). Other spectroscopic methods include mass spectroscopy, which
was mentioned in relation to post-translation modiflcation detection, raman spectroscopy
to observe low-frequency or vibrational modes that give insight into the heme shape among
other things, and  uorescence spectroscopy.
Kinetics
In order to determine the role of distant structural features in an enzyme, the two
questions that need to be addressed are what is the impact on the active site structure,
and what is the impact on catalysis. To this point, the only application considered for the
instrumentation has been as spectroscopic probes of a metal center active site. The partner
application is their role in kinetic evaluations.
The simplest enzymatic mechanism describes the association of enzyme with substrate
followed by the release of product:
E+S k1??*) ?
k?1
ES k2?! E+P (1.2)
Named for the researchers who proposed this mechanism, the Michaelis-Menten equation
identifles the minimum of four reaction compounds: S - substrate, P - product, E - free
enzyme, and ES - the enzyme-substrate complex. Figure 1.10 is a basic representation of
the relative concentrations of these compounds during the reaction progression.
Shortly after the reaction begins, ES reaches a steady state concentration, that is:
k1[E][S] = (k2 +k?1)[ES] (1.3)
32
Figure 1.10: Relative Concentrations of S, P, E, and ES During Reaction Pro-
gression. Rate values are arbitrary.
33
Ifthereactionvelocityisdeterminedsoonaftersteadystateisreached, substitutingequation
1.3 into v = k2[ES] gives
v = [E]tot[S]k2[S]+(k
2 +k?1)=k1
(1.4)
where [E]tot is the total enzyme concentration. This rearranges to the classic relationship,
v
[E]tot =
[S]kcat
KM +[S] (1.5)
where kcat is the maximum number of cycles that the active site can perform per unit time
(the turnover number), and KM is the concentration required to reach half the maximum
reaction velocity (the Michaelis constant). The velocity at saturating substrate concentra-
tions is kcat. At low substrate concentration, the slope of v=[E]tot vs. [S] approaches the
apparent second-order rate constant kcat=KM, a measure of e?ciency of the enzyme with
a given substrate. Generally, kcat re ects the reactivity of the ES complex, and kcat=KM
re ects the reactivity of free enzyme and substrate [1].
Determining these values experimentally requires measuring the initial velocity at sev-
eral difierent substrate concentrations above and below KM. If either the substrate or
product is a chromophore, this can be measured continuously using UV-visible absorption
by monitoring its change in absorbance over time. Other continuous detection methods
that might be useful include  uorescence spectroscopy or potentiometric detection. If it
is not possible to monitor the evolution of the reaction directly (which is common with
isomerization reactions, for example), an end-point assay can be used where the reaction
proceeds for a flxed amount of time and then is quenched and analyzed.
Even in the simplest case, steady state kinetics can only be used to determine k2 in the
form of kcat and the dissociation constant of the ES complex if k?1 is much greater than k2.
Furthermore, since steady state kinetics rely on monitoring either the substrate consump-
tion or product formation, multiple enzyme-based intermediates would go undetected. In
order to get individual rate constants or detect multiple intermediates, pre-steady state or
transient state kinetics must be used.
34
The main di?culty in analyzing pre-steady state kinetics is the time scale on which
they occur. Mixing enzyme and substrate can require anywhere from a few seconds using
inversion by hand to a few tenths of a second with a stir bar. The time it takes for the entire
solution to be mixed is called the \dead time". If any intermediates are formed within this
dead time, rate constants cannot be determined without the use of rapid mixing techniques.
To achieve this, instruments have been developed that use automated mixing devices that
force solutions together using syringes. The mixed solution is forced through a tube past a
detector, and the dead time is determined by the  ow rate and the distance of the detector
from the mixing chamber. The most common instrument that uses this principle is the
stopped- ow. Enzyme and substrate are mixed together and forced past a detector when
 ow stops and the solution comes to rest. The dead time is determined by the time it takes
 ow to stop and can be less than one millisecond.
Due to the fact that the automated mixing apparatus requires space, detection meth-
ods must be compatible. Absorption and  uorescence detectors are example of compatible
approaches. Electrodes or magnetic approaches are unlikely to be compatible. If the in-
termediates are not observable with compatible detectors, or the investigator wants to use
other approaches to probe the intermediates, rapid quenching may be used. The same prin-
ciple is used, but after mixing the enzyme and substrate, a set amount of aging time passes
and the mixture is then combined with a quencher. A variation from chemical quenching
was proposed in 1961 by using freeze quenching [56]. Its immediate and still most com-
mon application has been in Rapid Freeze Quench (RFQ) EPR. This is especially useful
when reaction intermediates involve radical reactions that are di?cult to observe with other
techniques and too rapid to be detected with steady state approaches.
Parametric Efiects
In the simplest mechanism (equation 1.5), the enzyme reacts with only one substrate
and no external afiectors are considered. Unsurprisingly, this is rarely the case. A glance
at the EC classes (Table 1.2) should make it obvious that a number of enzymes must
35
require multiple substrates. Ligases require ATP plus a substrate, transferases need both a
functional group donor and acceptor, and oxidoreductases very rarely use the same substrate
for both oxidation and reduction. Ionizable groups on the enzyme can afiect binding or
catalysis, requiring consideration of pH and bufiering. Temperature and ionic strength can
also afiect catalytic rates and folding properties. Even inhibitors introduced intentionally or
not (such as inhibition by substrate, product, or bufier) can prove signiflcant in developing
a kinetic understanding of the enzyme. Here we will consider a few of these parameters not
accounted for in the simplest Michaelis-Menten mechanism, particularly multiple substrates
and pH efiects.
There are only a few mechanisms by which multiple substrates bind. In the ping-
pong mechanism, a substrate is bound and product is released resulting in a modifled,
stable active site to which another substrate can bind. Sequential mechanisms require the
binding of all substrates before catalysis can occur. Some sequential mechanisms require
binding to be in a speciflc order, in others it can be random. These mechanisms can be
distinguished kinetically by varying all substrate concentrations simultaneously; but once
the mechanism is determined, it is common to discontinue with such extensive protocols.
Instead, it is common practice to transform the reactions into pseudo-flrst order kinetics by
holding one substrate constant and treating the reaction as the simple Michaelis-Menten
type mechanism [1]. This can be seen in the rate equations of the two types of mechanisms.
The ping-pong mechanism rate equations is
v
[E]tot =
[A][B]kcat
KA[B]+KB[A]+[A][B] (1.6)
and the sequential mechanism rate equation is
v
[E]tot =
[A][B]kcat
K0AKB +KA[B]+KB[A]+[A][B] (1.7)
36
where A and B are the substrates and KA and KB are their respective Michaelis con-
stants. By keeping one substrate at a constant concentration in great excess of its Michaelis
constant, it is clear that these equations simplify to the form seen in equation 1.5.
Many enzymes have what is called a \pH proflle", indicating that at some pH a transi-
tion occurs that afiects the rate of catalysis. This transition is often ascribed to an ionizable
group that may play a part in acid-base catalysis. Other transitions could be macroscopic
(the composition of multiple related changes or equilibria) or the result of protein unfold-
ing or reorganization. Fortunately, mechanisms accounting for pH transitions can also be
simplifled to Michaelis-Menten type rate equations of the form
v
[E]tot =
[S]kobscat
KobsM +[S] (1.8)
where kobscat includes consideration of protonation of the ES complex and KobsM includes
consideration of protonation of both the free enzyme and ES complex. Thorough derivations
of two complex examples can be found in chapter 2 where rate equations are described that
involve both pH dependence and multi-substrate reactions.
As seen in chapter 2, these examples of parametric kinetic analyses become more com-
plicated when combined with each other or other considerations such as substrate or product
inhibition. Only the simplest complications to the rate equations can be found in textbooks,
leaving it up to the researcher to derive the appropriate rate equation in most cases. The
model enzyme that we will be considering has the additional novel complication of being
able to catalyze two reactions within the same active site. The implications of this for
kinetics and parametric analysis will be discussed further on in this chapter and in chapter
3.
1.2.3 Therapeutics
Before proceeding on to consider the model enzyme that is the centerpiece of this
research, itisnotable to mentionhowcatalase-peroxidasescame intothe spotlight. Studying
37
enzymes has many purposes beyond the sheer curiosity of understanding living organisms
and how they operate. One of the most desired results is therapeutic applications.
Early advances in medicine came directly from biological studies, with the biochemical
understanding coming later. Two of the most notable include insulin, a protein secreted
by the pancreas involved in the regulation of blood glucose, and penicillin, a fl-lactam
antibiotic that acts by inhibiting the flnal step in peptidoglycan layer synthesis in gram-
positive bacteria.
Insulin has been a hallmark of many biochemical advances, generating multiple Nobel
prizes along the way. Not only was a Nobel Prize in Medicine awarded for its treatment
of diabetes (1923), it was the flrst protein to have its sequence determined (Nobel Prize in
Chemistry, 1958) [48], it was key in the development of x-ray crystallography technique of
isomorphous crystals used in protein crystal structure determination, and was the motiva-
tion behind the development of the radioimmunoassay for detecting hormone levels (Nobel
Prize in Medicine, 1977) [57].
Penicillin has also generated its share of Nobel Prizes and was involved in the iso-
morphous crystal technique, but the understanding of the biochemistry behind the action
of penicillin is still in constant use today in designing new antibiotics. Penicillin-binding
proteins (PBPs) catalyze the cross-linking of peptidoglycan in gram-positive bacterial cell
walls. Penicillin is an analogue of the D-alanyl-D-alanine that is removed during the reac-
tion. When PBPs act on penicillin, however, the penicillin irreversibly inactivates the PBP
by forming a covalent bond to an active site serine. With this knowledge, a long list of
penicillin derivatives have been developed with the goal of optimizing the binding of the
derivative to the PBP by altering the fl-lactam functional groups. Included in the list of
these derivatives is ampicillin, which is widely used in molecular biology laboratories to
select target strains that have intentionally had ampicillin resistance introduced.
This process of using biochemical knowledge as a foundation for drug discovery is
now common practice. There are four primary cellular processes that antibiotics disrupt:
DNA replication or RNA transcription, protein synthesis, metabolic processes, and cell
38
wall synthesis (Table 1.3) [58{60]. Each class of antibiotics targets a speciflc enzymatic
process by interacting with either the enzyme or substrate, and as with the fl-lactam class,
the antibiotics are designed to optimize the interaction or prevent resistance to it. More
recently, bactericidal antibiotics (those whose action kill the pathogen rather than halt
its growth) have been shown to have the commonality of creating reactive oxygen species
during their primary action, aiding in their bactericidal properties [60].
With drug resistance constantly arising, the search for new antibiotics is vigorous.
Established targets (those that have been studied and are fairly well understood) require
understanding of the mechanism of resistance so that new drugs can work around them.
New targets (those whose disruption could be potentially harmful to the pathogen) need to
be characterized for antibiotic development strategies. This is where catalase-peroxidases
come in.
Mycobacterium tuberculosis is a pathogen responsible for the disease tuberculosis (TB).
Tuberculosis is known to have afiected humans since at least around 1500 BC (as verifled
by DNA analysis), and possibly much earlier [61]. According to WHO, around one-third
of the world?s population is infected with TB [62]. In the 1950s, an organic compound
called isoniazid (INH) was found to be anti-tubercular and was immediately implemented
in sanitariums. Since then it has been one of two front-line drugs for TB treatment alongside
rifampicin. Credited to improper treatment regiments, drug resistance has reached the point
where TB strains resistant to both front-line drugs [63] and more than half of the second
line drugs have emerged. It was not until 1992 that the antitubercular mechanisms began to
be determined when isoniazid resistance was traced to mutations in the katG gene [64, 65].
Isoniazid is activated by the multi-functional enzyme KatG [64, 66, 67] by creating
an isonicotinoyl radical that can form adducts with NAD+ and NADP+. These adducts
inhibit mycolic acid synthesis, a key ingredient in mycobacterial cell walls [68]. One speciflc
adduct has also been shown to inhibit dihydrofolate reductase, a key enzyme in nucleotide
production [68, 69]. Oxidation by KatG also produces NO? radicals that can interfere with
respiratory enzymes [68].
39
Cellular process Target Antibiotic Class
DNA replication or DNA gyrase or topoisomerase IV quinolones
RNA transcription
Protein synthesis Peptidyl tRNA transferase aminoglycosides
macrolides
amphenicols
Transpeptidase lincosamides
Association of tRNA with tetracyclines
mRNA-ribosome complex
Metabolic processes Tetrahydrofolate synthesis sulfonamides
pathway (dihydropteroate
synthase)
Cell wall synthesis PBP fl-lactams
Peptidoglycan incorporation glycopeptides
Table 1.3: Antibiotic Classes.
40
Mutant TB strains in most cases contain a modiflcation to the katG gene that reduces
INH binding in KatG. S315T KatG is the most prevalent example [65, 70, 71]. By preventing
binding, INH cannot be activated and loses all antitubercular properties. As KatG is the
only catalase (or hydroperoxidase) active enzyme in Mtb [72] and is crucial in the ability
of Mtb to withstand the hosts oxidative burst [73], it is unfortunate that mutations such
as the S315T only prevent INH binding but do not disrupt the ability of the enzyme to
break down hydrogen peroxide [70, 71]. On the other hand, it is fascinating that non-active
site mutations can selectively eliminate one activity from this multifunctional enzyme. As
a result of the relation between isoniazid and KatG, a strong push has been made to better
understand the enzyme itself in hopes of being able to design drugs that can still be activated
by variants that cannot activate INH. Interest has continued to deepen as other periplasmic
catalase peroxidases have been linked to the virulence of pathogens such as Escherichia coli
O157:H7 (food poisoning) [74], Legionella pneumophila (Legionnaires disease) [75, 76], and
Yersinia pestis (bubonic plague) [77, 78].
1.3 Catalase-peroxidases
KatG is commonly called catalase-peroxidase after its two most studied activities. It
was mentioned in the previous section that KatG is the only catalase in Mtb and that
makes it indispensable during the hosts oxidative burst defense mechanism. To expand on
that, and to introduce the multi-functionality of KatG, we will flrst look at the products of
the oxidative burst (reactive oxygen species) which serve as a substrate for KatG, and then
consider the monofunctional catalases and monofunctional peroxidases to introduce possible
mechanisms and roles for KatG and provide a basis for structural and kinetic comparisons.
1.3.1 Reactive Oxygen Species
The reduction potential of O2 to H2O is around 0.8 V, indicating that molecular oxygen
is a strong oxidant. As mentioned before, however, in its ground state it is not very reactive.
Activating molecular oxygen allows access to its abilityto oxidize and enables cells to harvest
41
the energy from its conversion to water, such as in the generation of ATP during respiration.
Another application of activating oxygen is the generation of reactive oxygen species (ROS).
ROS refers primarily to three compounds: superoxide (O?{2 ), hydrogen peroxide (H2O2),
and the hydroxyl radical (HO?). Superoxide and hydrogen peroxide are less reactive than
the hydroxyl radical, which can react detrimentally with proteins, lipids, and DNA in the
cell. Phagocytes attempt to overwhelm a pathogen?s ability to neutralize ROS with what
is called a respiratory or oxidative burst [79].
In mammals, the initial step in oxidative burst is the generation of extreme concentra-
tions of superoxide by the multi-component enzyme NADPH oxidase:
NADPH+2O2 NADPHoxidase??????????! NADP+ +2O??2 +H+ (1.9)
The superoxide is disproportionated into hydrogen peroxide and molecular oxygen by the
enzyme superoxide dismutase (SOD):
2O??2 +2H+ SOD???! H2O2 +O2 (1.10)
Hydrogen peroxide, being a neutral molecule, is more able to difiuse across the lipid mem-
brane into the invading cell. Reaction of hydrogen peroxide with FeII (the Fenton reaction)
within the cell generates the destructive hydroxyl radical:
FeII +H2O2 ??! FeIII +OH? +OH? (1.11)
The goal of the phagocyte is to generate as much H2O2 as possible to difiuse into the
invading cell. The peroxide will damage metal-containing proteins, initiating a cascade of
reactions such as heme degradation that results in the release of iron which can then initiate
the Fenton reaction. Cells capable of breaking down H2O2 before or as it is able to difiuse
into the cell have a signiflcant advantage over our immune system. Disabling a cells ability
42
to deal with H2O2 is lethal to the cell, hence the signiflcance of the KatG mutations that
lead to INH resistance but not KatG inactivation.
ROS are generated on a much smaller scale during normal cell processes, as well,
and cells are equipped to prevent the accumulation of these using catalase or peroxidase
enzymes. Catalases disproportionate hydrogen peroxide into water and oxygen. Peroxidases
use hydrogen peroxide to oxidize a wide variety of reactions. Although these enzymes both
break down hydrogen peroxide, they are very difierent enzymes that serve difierent purposes.
1.3.2 Monofunctional Catalases
Function
The presence of catalase in living organisms was identifled over a century ago by its
ability to remove hydrogen peroxide from cells at very high rates. More recently other
possible roles have been identifled. Catalases were found to protect cells from UVB light
that can directly damage DNA and other cellular molecules by absorbing the UVB energy
to generate ROS that can be broken down by other anti-oxidant enzymes [80]. Catalase
deflciency has also been linked to graying hair as hydrogen peroxide is able to accumulate
and irreversibly inactivate the enzyme tyrosinase, leading to a decrease in the melanogenesis
(pigmenting) pathway [81].
Active site
Out of the over 300 known catalases, more than 90% contain heme (mostly heme b),
with the exceptions containing a dimanganese cluster [82]. The heme is coordinated to
the enzyme via a tyrosinate ligand (Figure 1.2), which signiflcantly lowers the reduction
potential and stabilizes high oxidation states. The key active site residues are a histidine
and an asparagine. The histidine is parallel to the heme surface, and is oriented above
either pyrrole ring III (as in Figure 1.2) or pyrrole ring IV.
43
Global structure
Catalases crystalize as homotetramers. Each monomer has either four or flve domains:
N-terminal, fl-barrel core (where the heme resides), fi-helical, wrapping, and (in large sub-
unit catalases) a C-terminal domain. The N-terminal extension of one domain is interweaved
with the wrapping domain of an adjacent subunit, causing the tetramer to consist of two
sets of tightly interlocked subunit pairs. The N-terminal domains in large subunit catalases
are larger than the N-terminal domains in the small subunit catalases [83].
Activity and kinetics
Catalase breaks down peroxide via a ping-pong mechanism. The flrst step (equation
1.12) is a two-electron oxidation of the enzyme with one molecule of hydrogen peroxide
to create a ferryl porphyrin cation radical intermediate known as Compound I (CpdI) and
releasing water. CpdI is then reduced back to the resting state (equation 1.13) with another
molecule of H2O2 releasing another molecule of water and a molecule of oxygen.
E?Por?FeIII +H2O2 ??! E?Por?+?FeIV?O+H2O (1.12)
E?Por?+?FeIV?O+H2O2 ??! E?Por?FeIII +H2O+O2 (1.13)
The second-order rate constant for catalases has been found to be as high as 6.6 ? 107 M {1
s {1 [84], approaching the difiusion limit for enzyme-substrate association (109 M {1s {1)
[1]. Unbufiered studies also showed catalase to maintain maximal activity from pH 4.75
to 10.5 [84]. Kinetic determinations for catalases are complicated by peroxide dependent
inactivation at high peroxide concentrations [82]. Large subunit catalases are more resistant
to this inactivation [84].
44
1.3.3 Monofunctional Peroxidases
A peroxidase is any enzyme capable of catalyzing the following general reaction:
R1OOR2 +2e? +2H+ ??! R1OH+R2OH (1.14)
The electrons in equation 1.14 could be donated from a wide variety of donors, including
organic compounds such as phenols, anilines, and azines or large molecules such as the pro-
tein cytochrome c. Some peroxidases have high substrate speciflcity such as the cytochrome
c peroxidases [85{87], while others have very broad speciflcity such as chloroperoxidase [88].
Functions
In catalases, the function is obvious from the mechanism: eliminate hydrogen peroxide.
With peroxidases, however, the function is not always as obvious. Still, the role of perox-
idases is often ascribed as protection of the cell against oxidative damage [86, 87, 89, 90].
Other roles are not always as easy to detect, but have been identifled.
First, variations of the peroxide substrate can give insight into alternative roles. For
example, the selenium-containing glutathione peroxidases (Gpx) act on organic or lipid
peroxides, but not hydrogen peroxide. Gpx-4 is membrane-associated and acts speciflcally
on phospholipid peroxides, which also serve as activators of the lipoxygenase pathway. 5-
lipoxygenase metabolizes arachidonic acid to leukotrienes. Improper Gpx-4 levels adversely
afiect the leukotriene balance, indicating that it has a role in 5-lipoxygenase regulation [90].
The other perspective is to consider the consequence of the oxidation of the electron
donor. In fact, many peroxidases are named based on the identity of the reducing sub-
strate. NADH peroxidase could easily be considered as having a role in the redox cycling of
NADH and NAD+. Haloperoxidases (such as chloroperoxidase, myeloperoxidase, lactoper-
oxidase, and thyroperoxidase) are named such based on their ability to oxidize halogens
into molecules that can halogenate organic compounds. These products have wide-ranging
efiects. Halogenated nucleic acids, for example, have anti-cancer and anti-viral activity as
45
they inhibit DNA synthesis [91]. Oxidation of thiocyanide and halogens by lactoperoxidase
(found in milk) is bacteriostatic to gram-positive bacteria and potentially bacteriocidal to
gram-negative bacteria, a feature crucial to infants during their immune system develop-
ment [92]. Thyroperoxidase iodinates tyrosine residues in thyroglobulin (Tg) to synthesize
the thyroid hormones T3 and T4 [93]. Manganese-dependent and lignin peroxidases depoly-
merize lignin in the biodegradation of wood by white rot fungus [94{97].
Active site
Like catalases, monofunctional peroxidase are found in nearly all organisms. Structures
of monofunctional peroxidases, however, are not nearly as uniform as catalases [85, 86]. Sub-
strates and mechanisms are equally diverse. As evident from equation 1.15, peroxidases are
oxidoreductases and have nearly the entire EC subclass 1.11.x.x devoted to them. Being
oxidoreductases, they need a redox active catalytic site. A few of the more uncharacteristic
catalysic centers can be seen in glutathione peroxidases, NADH peroxidase, chloroperoxi-
dase, class 1c ribonucleotide reductases, and some of the cytochrome c peroxidases. Five of
the six glutathione peroxidases (enzymes that use glutathione as the electron donor) rely
on a selenocysteine residue as the redox center [90, 98, 99]. The NADH peroxidase active
site contains oxidized cysteine (cysteine-sulfenic acid) and the cofactor  avin adenine din-
ucleotide (FAD) [85, 100]. The chloroperoxidases are diverse in themselves, some having a
vanadium metal center, some being metal-free and using the serine-histidine-aspartate cat-
alytic triad, and some containing heme with glutamic acid as the peroxide cleaving catalytic
base rather than the histidine seen in most heme-peroxidases [91]. Although not classifled
as peroxidases, class 1c ribonucleotide reductases demonstrate peroxidase activity at the Fe-
Mn cofactor as a part of activation and maintenance of the metal center [4]. Cytochrome
c peroxidases from Pseudomonas aeruginosa, Pseudomonas nautica, and Nitrosomonas eu-
ropaea have all been shown to have di-heme centers with both hemes being c-type hemes
with very difierent reduction potentials [86, 87, 101].
46
Yeast cytochrome c peroxidase (Ccp), on the other hand, only has one b-type heme,
as do most other peroxidases. It is the heme b containing peroxidases that we will be
focusing on, in particular the plant structural superfamily. The Ccp shown in Figure 1.2
is stereotypical for this family. All share the catalytic histidine and arginine on the distal
side of the heme, as well as the triad aspartate-histidine-iron on the proximal side. The
tryptophan shown is characteristic of the class that Ccp belongs to, but is a phenylalanine
in the other classes.
Global structure
In 1992, Karen Welinder identifled three classes within the plant peroxidase superfamily
[86]. Class I is described as of prokaryotic origin and includes yeast Ccp and ascorbate
peroxidase. Class II is the fungal peroxidases such as the manganese-dependent peroxidases,
lignin peroxidases, and versatile peroxidases. Class III is the secretory plant peroxidases,
encompassing the classical peroxidases such as horseradish peroxidase. Sequence similarity
within a class can be as low as 32%, but all classes show consistent overall folding patterns.
All have 10 helices, which are labeled A through J beginning at the N-terminus. The
proximal histidine ligand is invariable and located on the F-helix. Catalytic residues are
found on the B-helix.
Structural difierences are primarily limited to the interhelical loop regions, but one
major difierence is the relationship each of the classes has with calcium ions. Class I
peroxidases do not require any structural calcium. Crystal structures of class II and III
peroxidases do indicate the presence of calcium ions [86, 97], but calcium depletion com-
pletely inactivates class II peroxidases while only decreasing class III peroxidase activity
by around 50% [86, 94{97, 102]. The distal calcium ion is coordinated by residues and
carbonyls on the loop connected to the active site histidine [97], and in lignin peroxidases
the loss of calcium is concomitant with bis-histidine coordination of the heme as evidenced
by low-spin iron features in the UV-vis, MCD, and EPR spectra [94, 96]. Calcium depleted
47
versatile peroxidase does still contain high-spin heme, but has a more  exible heme cavity
that can become bis-histidine coordinated at certain pHs [102].
Activity and kinetics
The peroxidase catalytic cycle begins the same way as the catalase cycle with the
peroxide oxidizing the heme center to create Compound I (equation 1.15). Instead of a
two-electron reduction, peroxidases typically undergo two single-electron reduction steps.
The flrst electron creates a second ferryl intermediate known as Compound II (equation
1.16), and the second electron reduction restores the enzyme to its native state and releases
water (equation 1.17).
E?Por?FeIII +H2O2 ??! E?Por?+?FeIV?O+H2O (1.15)
E?Por?+?FeIV?O+RH ??! E?Por?FeIV?O+R? +H+ (1.16)
E?Por?FeIV?O+RH+H+ ??! E?Por?FeIII +R? +H2O (1.17)
Some of the earliest work on the kinetics of horseradish peroxidase used the flrst version of
the stopped- ow apparatus [103]. CpdI formation occurs with k = 1 ? 107 M {1 s {1, which
is not much slower than the apparent second-order rate constant for catalase activity (6.6
? 107 M {1 s {1). Oxidation of the substrates took place with a low turnover in the range
of 5 s?1, but with a second-order rate constant near 2 ? 107 M {1 s {1 [104]. These same
experiments found that although peroxidases followed a ping-pong like mechanism, it still
conformed to and conflrmed the Michaelis-Menten and Briggs-Haldane theories [103, 104].
1.3.4 Catalase-peroxidases
In the 1970s, an enzyme from Escherichia coli was found to have both catalase and
o-dianisidine peroxidase activity [105]. This was unique considering that known catalases
48
up to that point had been shown to be very poor peroxidases [85, 105]. Likewise, per-
oxidases were known to only have minimal catalase activity, a feature shared by many
heme-containing enzymes partly due to the ability of heme to disproportionate peroxide
non-enzymatically [85]. This enzyme, later to be named a catalase-peroxidase, has since
been isolated from a number of other bacteria and fungi [51, 106{114].
Structural comparison of bifunctional catalase-peroxidase to monofunctional catalases and
peroxidases
Sequence alignments placed catalase-peroxidases clearly within the plant peroxidase
superfamily, speciflcally in class I [86, 115]. The prediction that catalase-peroxidase would
contain the 10 helices in approximately the same positions as monofunctional peroxidases
was conflrmed by the crystal structure of catalase-peroxidase from Haloarcula marismortui
[116]. Since this flrst crystal structure, others have been solved for Burkholderia pseu-
domallei [117], Mycobacterium tuberculosis [118], and Synnechoccus PCC 7942 [119], all
signifying that the classiflcation of catalase-peroxidase as a class I peroxidase within the
plant superfamily was accurate. The active sites of catalase-peroxidases are identical to
monofunctional class I peroxidases with all signiflcant residues strictly conserved [115] (Fig-
ure 1.11). The bifunctional catalase-peroxidase shows no sequence or structural similarity
to monofunctional catalases [122].
The superimposability of the active sites and structural similarities between catalase-
peroxidases and monofunctional peroxidases makes the dual activity of catalase-peroxidases
even more unexpected as non-active site features must somehow tune the active site for cata-
lase activity without changing its structure. This has led to a search for and investigation
of conserved structures of KatGs that are absent in monofunctional peroxidases [123{128].
The most obvious three structures were revealed through sequence alignment: two interhe-
lical insertions and a C-terminal domain that is proposed to have originated from a gene
duplication and fusion event [86, 115, 122, 129] (Figure 1.12).
49
Figure 1.11: Comparison of Catalase-peroxidase Active Site to Monofunctional
Catalase and Peroxidases. All structures have propionate groups oriented to the right.
Amino acids are coded to the following color scheme: H - pink, R - light blue, N - dark
blue, F - dark green, Y - yellow, D/E - red, W - light green. Structures were taken from the
following PDB accession numbers: catalase-peroxidase - 1SJ2 [118], bovine liver catalase
- 7CAT [20], cytochrome c peroxidase - 2CYP [18], lignin peroxidase [120], horseradish
peroxidase [121].
50
Figure 1.12: Identiflcation of Interhelical Insertions and C-terminal Domain in
Catalase-peroxidase. N-terminal domain shown in blue, C-terminal domain shown in
green, DE insertion shown in yellow, FG insertion and heme shown in dark red. Structure
is from Mycobacterium tuberculosis (accession number 1SJ2)[118].
51
If the C-terminal domain is a result of a gene duplication and fusion event, it has
since lost the ability to bind heme and has no apparent catalytic capability. There is also
no evidence of any substrate binding, allosteric control, or signaling purpose. Appearing
to have no role, the C-terminal domain came to be considered \inactive", a rare trait for
protein domains considering their treatment as the functional units of proteins. Deletion
mutagenesis, however, has elucidated possible roles for this \inactive" domain.
Deletion of the C-terminal domain results in enzyme preparation that is highly dis-
turbed. The separately expressed N-terminal domain (KatGN) expresses in inclusion bodies
and has to be refolded. Once refolded, it still has no activity as the active site is collapsed
and the catalytic distal histidine (106 by Escherichia coli numbering) coordinates the heme
iron directly, reminiscent of calcium-depleted manganese-dependent and lignin peroxidases
[130]. Introducing the separately expressed and purifled C-terminal domain (KatGC) reacti-
vates the enzyme concomitant with removing the active site histidine from the coordination
sphere of the heme iron [46]. The C-terminal domain thus has a signiflcant structural role
by supporting the integrity of the active site, and may also play a role in proper folding
during expression. Chapters four and flve will deal speciflcally with investigations of the
residues at the interdomain interface.
The FG intehelical insertion in KatG begins near residue 270 and ends around residue
310 [115]. It has an ascending and descending arm with the base of each arm positioned
between the active site and the solvent, and the apex protruding into a hydrophobic pocket
in the C-terminal domain of the other subunit in the dimer. The insertion runs parallel to
a catalase-essential hydrogen bonding network involving the catalytic His106 with Asp152
and Asn153 [123, 131, 132]. Deletion of the FG insertion results in enzyme that retains
almost no catalase activity. The peroxidase turnover is half of that of the wild-type, but the
apparent second order rate constant in respect to the reducing substrate ABTS is increased
by an order of magnitude. It has been proposed that the increased apparent second order
rate constant results from greater accessability of the heme edge to reducing substrates
[123].
52
The DE interhelical insertion begins near residue 195 and ends near residue 235 [115],
and also constricts access to the active site [133]. This loop connects the proximal and
distal catalytic halves of the N-terminal domain and has a stretch of highly conserved
residues near the end of the insertion. The backbone nitrogen of Ile225 is within hydrogen
bonding distance of the catalase-essential Asp152 [134]. Deletion of the DE insertion also
eliminated catalase activity, but enhanced peroxidase activity [135]. The DE insertion
deletion variant is susceptible to hydrogen peroxide inactivation, and in the absence of a
reducing substrate readily forms an inactive intermediate known as Compound III. This is a
heme dioxygen (FeII?O2 ?! FeIII?O?{2 ) intermediate that is observed in monofunctional
peroxidases when Compound II reacts with hydrogen peroxidase [85]. It has also been
observed in catalase-peroxidases, but only at high concentrations of hydrogen peroxide [66].
Another feature of the DE insertion revealed in the crystal structures is the presence of a
covalent PTM involving the active site tryptophan (Trp105), a conserved tyrosine (Tyr226)
on the DE insertion, and a conserved methionine (Met252) [116{119]. Mass spectroscopy
and SDS-PAGE showed that the cross-link existed in solution and was required for catalase
activity [136{138]. Disruption of the covalent adduct by creating variants of the participants
also conflrmed that the adduct is essential for catalase activity, but not peroxidase activity
[124, 126, 139{141]. The exact nature of the role of this adduct is still contested with
some proponents arguing that it may donate an electron to the heme initially forming
a tryptophanyl radical that quickly transfers to the tyrosine [142, 143] whose radical is
stabilized by an arginine residue [127]. Others argue that the tryptophanyl radical observed
could not have been on the Trp106 as that radical is only stable during crosslink formation
[144, 145]. Further possible incorporation of this adduct into the KatG mechanism will be
dealt with in the discussion of chapter three.
Activity and kinetics
Considering that monofunctional catalases and monofunctional peroxidases share the
same flrst step in catalysis (equations 1.12 and 1.15), directly combining the two results
53
in the classical representation of the KatG catalytic cycles schematically and textually and
is the most common representation in literature (Figure 1.13) [51, 109{111, 123, 126, 127,
130, 145{147].
NADH can act as a reducing substrate in the peroxidase reaction, but KatG also has
NADH oxidase activity. This activity is quite distinct from the peroxidase reaction in that
it has a pH optimum of 8.5 and consumes oxygen while producing ROS [146]. KatG also has
hydrazinolysis activity, which is part of the INH activation. These two activities appear to
be synergistic based on studies involving an indicator radical that appears faster when both
substrates are present than it does cumulatively with only one substrate present at a time.
Manganese ion also accelerates radical production. KatG from Mycobacterium tuberculosis
also is an e?cient isonicotinoyl-NAD synthase, IN-NAD being the compound that interferes
with mycolic acid synthesis. Binding sites for these substrates have yet to be determined,
but the S315T variant in MtbKatG inhibits the hydrazinolysis reaction by reducing INH
binding. Only preliminary determinations of catalytic residues have been performed on
residues known to be involved in the catalase and peroxidase activities.
Of the peroxide-dependent activities, catalase and peroxidase, the two most in uential
factors governing which activity is observed are pH and the availability of a reducing sub-
strate. To date, the evaluation of pH-dependency has involved observing initial rates using
single concentrations of substrates, typically at or near saturating conditions [51, 111, 146].
This method consistently identifles pH optima of 6.5-7.0 for catalase activity and 4.5-5.0 for
peroxidase activity. Like many monofunctional peroxidases, KatGs are able to utilize a wide
variety of electron donors to complete the peroxidase catalytic cycle [106, 107, 111, 126].
Chapter two covers a much more extensive analysis of the pH-dependency of these activities,
and chapter three investigates the activity of the enzyme when both pathways are available.
54
Figure 1.13: Classic Catalase-peroxidase Scheme.
55
Chapter 2
Complexity of KatG Kinetics Revealed by pH Analysis
2.1 Introduction
Many of the searches for catalase speciflc KatG features are based on the reasonable
assumption that some structures will have a difierent conflguration based on pH, considering
that the dominant reaction catalyzed by KatG changes with pH. To be able to accurately
relate the structure of the enzyme to function, a complete understanding of its kinetic
behavior is necessary. This step has been primarily omitted, particularly in relation to pH.
Steady-state parameters have been consistently determined only at the pH determined to
be \optimal" based on velocities at expected saturating conditions.
In efiorts to provide a basis for more thoroughly correlating structure and function and
its dependence on pH, this chapter covers a full-scale kinetic evaluation to determine the
kinetic parameters at pH intervals of 0.25 across the ranges of observed catalytic activity.
For peroxidase kinetics measurements, one of the most common reducing substrates used
in vitro 2,2?-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (ABTS) was selected for the
wealth of data to compare it to as well as the ease of measurements. Through this we
discovered the flrst evidence of ABTS-dependent inhibition of peroxidase activity, most
evident at pH values below 4.5. This subsequently revealed that while ABTS acts as the
reducing substrate, the optimal pH for peroxidase activity is masked by the inhibition and
is not, in fact, between 4.5 and 5.0. Also, we present results demonstrating that the lower
pH limit for activity is due to an unfolding event.
These results have led to establishing the pH-dependent schemes for peroxidase cycle
(Figure 2.1) and catalase cycle (Figure 2.2). Calculation of the catalase cycle pKas for
the various stages of catalysis for both the interactions between HCpdI and H2O2 and the
56
HCpdI-H2O2 complex (Figure 2.2) reveal that the catalase pH optimum of 6.5 is actually
the intersection of the optimum for binding (the former) and the optimum for catalysis (the
latter).
2.2 Materials and Methods
2.2.1 Materials
Hydrogen Peroxide (30%), imidazole, hemin, Sephacryl 300 HR, ampicillin, chloram-
phenicol, citric acid, phenylmethylsulfonyl  uoride (PMSF), 2,2?-azino-bis(3-ethylbenzthia-
zoline-6-sulfonicacid)(ABTS),pyrogallol, and3,3?-dimethoxybenzidine(o-dianisidine)were
purchasedfromSigma(St. Louis, MO).Isopropyl-fl-D-thiogalactopyranoside(IPTG),mono-
and di-basic sodium phosphate, acetic acid, hydrochloric acid, and sodium acetate were
obtained from Fisher (Pittsburgh, PA). Sodium tartrate was purchased from J.T. Baker
Chemical Company (Phillipsburg, NJ). Bugbuster and benzonase were purchased from No-
vagen (Madison, WI). The E. coli strain BL-21 [DE3] pLysS was obtained from Stratagene
(La Jolla, CA). Nickel-nitrilotriacetic acid (Ni-NTA) resin was purchased from Qiagen (Va-
lencia, CA). All bufiers and media were prepared using water purifled through Millipore
Q-PakII system (18.2 M?/cm resistivity).
2.2.2 Expression and Puriflcation of wtKatG
Expression of wtKatG was done using plasmid that had been modifled (pKatG1) to
incorporate a six-histidine tag on the C-terminus of the protein. The pKatG1 plasmid
carries the gene for ampicillin resistance. Expression was carried out in E. coli (BL-21
[DE3] pLysS) cells that are resistant to chloramphenicol. Cells were grown in 0.5 L of
Luria-Bertani broth that had been supplemented with ampicillin and chloramphenicol at
37 ?C with constant shaking. Expression was induced with IPTG once cells had grown to
mid-log phase (OD600 = 0.5) and monitored hourly to ensure growth. Cells were harvested
by centrifugation 4 hours after induction and stored at -80 ?C until puriflcation.
57
Figure 2.1: Peroxidase Cycle Scheme with pH Dependence.
58
Figure 2.2: Catalase Cycle Scheme with pH Dependence.
59
Expression analysis was performed using the TCA precipitation technique. A quantity
of cells su?cient to yield a 0.05 OD600 reading (when diluted to 1 ml) was treated with
an equal volume of 10% trichloroacetic acid (4 ?C) followed by centrifugation. The pellet
was washed with 1 ml acetone, then dried and resuspended in SDS-PAGE loading bufier,
adjusting the pH as appropriate with trizma base. Samples were then separated by SDS-
PAGE using a 7.6% acrylamide resolving gel.
Puriflcation began by resuspending the cell pellets in Bugbuster reagent in the pres-
ence of PMSF (0.1 mM) and benzonase (250U). The cell lysate was centrifuged, and the
supernatant loaded onto a Ni-NTA column by recirculating the solution with a  ow rate of
1 mL/min through the column overnight. Once loaded, the column was washed with Tris
bufier pH 8.0, followed by 50 mM phosphate bufier / 200 mM NaCl pH 7.0, then a repeat
of the phosphate / NaCl bufier that also contained 2 mM imidazole. Protein was eluted
using a gradient of the phosphate / NaCl bufier from 2 to 200 mM imidazole. Fractions
were analyzed by SDS-PAGE, and those containing the purest target were combined and
concentrated with an Amicon ultraflltration concentrator. Excess imidazole was removed
by gel flltration using a column packed with Sephacryl 300 HR. The elute was collected in
6 mL fractions and analyzed by SDS-PAGE. Protein-containing fractions were combined,
concentrated, aliquotted, and stored at -80 ?C.
Concentration of purifled enzyme was estimated according to the method of Gill and
von Hippel [50] ("280 = 1.44 ? 105 M?1 cm?1). Even though no heme precursors had been
added to the expression media, a small portion of the purifled enzyme already had heme
incorporated while inside the cells. Adding 0.9 to 0.95 heme equivalents directly to the
enzyme solution was su?cient to introduce heme to the rest of the enzyme. Reconstituted
enzyme solution sat for 72 hours to allow unincorporated heme to settle out of solution. The
solution was then spun and removed from the precipitated heme. This was to ensure that
free heme did not interfere with the accumulation of spectral or kinetic data. The concen-
tration of the reconstituted enzyme was then determined using the Soret heme absorption
band ("408 = 120.7 mM?1 cm?1) [123].
60
2.2.3 Peroxidase Activity Assays
Peroxidase activity was evaluated by monitoring the production of ABTS radical ("417
= 34.7 mM?1 cm?1), pyrogallol oxidation ("430 = 2.47 mM?1 cm?1), or o-dianisidine ox-
idation ("460 = 11.3 mM?1 cm?1) over time in the presence of 20 nM wtKatG; initial
ABTS concentrations were also determined spectrophotometrically ("340 = 3.66 ? 104 M?1
cm?1) [148]. All assays were carried out at room temperature on a Shimadzu UV-1601
spectrophotometer (Columbia, MD). Initial velocities were determined across a range of
ABTS concentrations while keeping peroxide concentration constant, as well as a range of
peroxide concentrations while keeping ABTS concentration constant. Enzyme inhibition
(as evidenced by suppressed activity levels) was observed at hydrogen peroxide concen-
trations greater than 1.0 mM. It was therefore necessary to select a concentration below
that of inhibition for performing assays for ABTS-dependent parameter determination; the
peroxide concentration, however, was still slightly higher than twice the observed Michaelis
constant for peroxide. For the substrates pyrogallol and o-dianisidine, initial velocities were
determined across a range of reducing substrate concentrations at 0.4 mM H2O2. The ini-
tial velocities (excluding those determined under inhibiting substrate conditions) were flt
to the Michaelis equation using a non-linear regression analysis to determine apparent ki-
netic parameters. Peroxidase assays with ABTS were initially carried out in 50 mM acetate
bufier, pH 3.50 to 6.00 in 0.25 pH increments. Recognizing that acetate has little bufiering
capacity at the extremes of this range, assays were also carried out in 50 mM tartate bufier,
pH 3.50 and 3.75 and 100 mM phosphate bufier, pH 6.00. The data obtained using these
bufiers were the same as the data from assays performed in the acetate bufier. Peroxidase
assays with pyrogallol and o-dianisidine were carried out in 50 mM acetate bufier, pH 4.00
to 6.00 in 0.5 pH increments.
2.2.4 Catalase Activity Assays
Catalase activity was evaluated by monitoring the decrease in H2O2 concentration
over time at 240 nm ("240 = 39.4 M?1 cm?1) [149] in the presence of 20 nM wtKatG. All
61
assays were carried out at room temperature on a Shimadzu UV-1601 spectrophotometer
(Columbia, MD). Initial velocities were flt to the Michaelis equation using non-linear regres-
sion analysis to determine kinetic parameters. Accurate determination of catalase activity
became increasingly di?cult at pH > 7 due to the small Michaelis constant for hydrogen
peroxide. Measurements carried out below the calculated KM from preliminary data had
initial absorbance below 0.1, and complete consumption of peroxide occurred within 30
seconds of initiation. To slow down the reaction (to obtain better initial velocities), en-
zyme concentration was reduced to 5 nM. Catalase assays were carried out in the following
bufiers: 50 mM acetate bufier, pH 5.0 to 6.0 in 0.25 pH increments; 100 mM phosphate
bufier pH 6.0 to 8.0 in 0.25 pH increments.
2.2.5 Circular Dichroism Spectroscopy
All spectra were obtained using 5 ?M enzyme in 5 mM citrate (pH 3.10, 3.30, 3.50, 3.60,
3.85) or acetate (pH 3.75, 4.05, 4.20, 4.60, and 4.90) bufier to minimize bufier interference
below 200 nm. Acetate had little bufiering capacity below pH 3.75, necessitating the overlap
of the two bufiers. Spectra were recorded at 23 ?C in a quartz cell (0.5 mm path length)
from 250 - 195 nm on a Jasco J-810 spectrophotometer (Tokyo, Japan). Baselining and
analysis were done using Jasco J-720 software.
2.2.6 pKa Determination
The scheme in Figure 2.1 produced the equations (derived here) that gave the best flt
to the peroxidase kinetic parameters across the pH ranges observed. The mechanism is a
modiflcation of the conventional mechanism [150] to account for pH efiects and gives rise
to the following equation:
v0
[E]tot =
fi[S][H2O2]
fl1[S]+fl2[H2O2] (2.1)
62
where fi, fl1 and fl2 are as follows:
fi = 2k4k5k6 + k2k4k6KH2 +k3k4k5KH3[H+] (2.2)
fl1 = k5k6
 
1+ KH1[H+]
?
(2.3)
fl2 = k4k6
 
1+ KH2[H+]
?
+k4k5
 
1+ KH3[H+]
?
(2.4)
Equation 2.1 can be rewritten as
v0
[E]tot =
kcat[S][H2O2]
KH2O2M [S]+KSM[H2O2]
(2.5)
Empirically, k2 and k3 (rates of product formation by the unprotonated intermediates)
can be assumed to be very small (Figures 2.6 and 2.7), simplifying equation 2.2 to the
following:
fi = 2k4k5k6 (2.6)
If we also assume that k5  k4 and k6, then we can relate equation 2.5 to the classical
Michaelis-Menten type kinetics by holding one substrate constant:
v0
[E]tot =
kobscat[S]
[S]+KobsM (2.7)
where
kobscat = kcat[H2O2]
KH2O2M
= fifl
1
[H2O2] = 2k4K
H1
[H+] +1
[H2O2] (2.8)
KobsM = K
S
M[H2O2]
KH2O2M
(2.9)
 k
cat
KM
?
obs
= kcatKS
M
= fifl
2
= 2k6K
H3
[H+] +1
(2.10)
63
and similarly:
v0
[E]tot =
kobscat[H2O2]
[H2O2]+KobsM (2.11)
where
kobscat = kcat[S]KS
M
= fifl
2
[S] = 2k6K
H3
[H+] +1
[S] (2.12)
KobsM = K
H2O2
M [S]
KSM (2.13)
 k
cat
KM
?
obs
= kcat
KH2O2M
= fifl
1
= 2k4K
H1
[H+] +1
(2.14)
The relation between equations 2.8 and 2.14 and the relation between equations 2.10
and 2.12 are apparent and lead to the following equations that will be used to flt the
peroxidase kinetic parameters:
kobscat
[H2O2] =
?
kcat
KH2O2M
!obs
= 2k410?pK
H1
10?pH +1
(2.15)
kobscat
[S] =
 k
cat
KSM
?obs
= 2k610?pK
H3
10?pH +1
(2.16)
These allow for the determination of pKH1 and pKH3 as well as k4 and k6. Furthermore,
given that kcat (that is, fi) is a constant, relative values for the two Michaelis constants can
be assigned.
The kinetic evaluation of the catalase pathway can be done by treating it as a ping-pong
bi-bi reaction. The general solution for such pathways is:
v0
[E]tot =
fi[A][B]
fl1[A]+fl2[B]+fl3[A][B] (2.17)
64
Since both substrates ?A? and ?B? are hydrogen peroxide, this simplifles further to:
v0
[E]tot =
fi[H2O2]
fl1 +fl2 +fl3[H2O2] (2.18)
or
v0
[E]tot =
kcat[H2O2]
KM +[H2O2] (2.19)
where according to the scheme in Figure 2.2, under the assumption that k6  k?5, k8  
k?7, k8  k6 [150{152], and that the pKas for Compound I formation in peroxidases lie
well outside the pH range used in current experiments [150],
fi = k5k6k7k8 (2.20)
fl1 = k6k7k8 (2.21)
fl2 = k5k6k8qF (2.22)
fl3 = k5k7(k6qFO +k8) (2.23)
kcat = fifl
3
= k6k8k
6qFO +k8
(2.24)
KM = fl1 +fl2fl
3
(2.25)
kcat
KM =
fi
fl1 +fl2 =
k5k7
k5qF +k7 (2.26)
and
qF = KF2[H+] +1+ [H
+]
KF1 (2.27)
qFO = KFO2[H+] +1+ [H
+]
KFO1 (2.28)
As the pH optimum is approached for maximum turnover and the apparent second-
order rate constant, kcat approaches k6 and (kcat=KM) approaches k5. As pH varies from
the optimum, the two values approach (k8/qFO) and (k7/qF) respectively. This can be used
65
to make approximations of k5 and k6, however estimations of k7 and k8 are more inaccurate
considering that q values vary logarithmically with pH. The pKa values can be estimated
by fltting the data to a simple 2-pKa model:
kobscat = kcat10?pK
FO2
10?pH +1+
10?pH
10?pKFO1
= k6k8k
6 +k8
 1
qFO
?
(2.29)
 k
cat
KM
?obs
=
kcat
KM
10?pKF2
10?pH +1+
10?pH
10?pKF1
= k5k7k
5 +k7
 1
qF
?
(2.30)
2.3 Results
2.3.1 Kinetic Parameters for Peroxidase Activity of KatG
Dramatic and difierential efiects of pH on the two most prominent activities of catalase-
peroxidases have been widely observed [51, 127, 128, 146]. The standard approach for
determining these pH proflles has been to vary pH, measuring activity at a single unvaried
concentration of substrate(s) - often one approaching saturation. Using this approach,
peroxidase activity has typically shown a maximum near pH 4.5. Indeed, in our hands
KatG from E. coli shows precisely this behavior (Figure 2.3). Likewise, E. coli KatG and
other catalase-peroxidases show maximal catalase activity at a substantially higher pH
(?6.5).
Interestingly, a thorough kinetic evaluation of this phenomenon has not been under-
taken, opening the possibility that mechanistic conclusions have been drawn based on an
incomplete picture of pH efiects on catalysis. In order to flll in some of the blanks, we
measured the efiect of pH on activity across a range of substrate concentrations in order
to obtain apparent kinetic parameters for peroxidase activity with respect to H2O2 and
reducing substrate concentration as well as for catalase activity. At flrst glance, efiect of
H2O2 concentration on peroxidase activity showed few surprises over pH ranging from 3.5
to 6.0. High concentrations of H2O2 (> 1.0 mM) resulted in inhibition of the enzyme
66
Figure 2.3: Initial Velocity of Peroxidase Activity Under Saturating Substrate
Conditions. [ABTS] = 1.2 mM, [H2O2] = 0.51 mM.
67
which in peroxidase kinetics has been attributed to irreversible inactivation by formation
of Compound III, a heme dioxygen (FeII-O2) complex [26]. This complex results from the
reaction of hydrogen peroxide with Compound II. It is unlikely that the inhibition is a
result of competition by catalase activity since it was observed at all pH values. As a result
of this inactivation, kinetic studies were carried out at concentrations below the inhibition
threshold of hydrogen peroxide. The maximum catalytic output (as measured by kcat) with
respect to H2O2 concentration was observed at pH 4.5 (Figure 2.4). Likewise, optimum
peroxidase e?ciency (as estimated by apparent kcat=KM) was observed around pH 4.5.
Comparable experiments to evaluate the efiect of pH on ABTS-dependent kinetic pa-
rameters showed that ABTS behaved not only as a substrate, but also as an inhibitor
(Figure 2.5). Its abilities to act as a substrate and inhibitor were both dependent upon
pH. A progressive decrease in pH led to more substantial inhibition by ABTS. Table 2.1
shows the lowest concentration of ABTS observed to result in inhibition under various pH
conditions. No inhibitory efiect was observed above pH 4.75, but at pH 3.75 as little as 0.1
mM ABTS resulted in appreciable inhibition of peroxidase activity.
In light of this phenomenon, the H2O2-dependent kinetic parameters for peroxidase
activity were reevaluated at the greatest concentration of ABTS that was never observed
to inhibit peroxidase activity (0.05 mM). Under these conditions, the apparent kcat and
apparent kcat=KM for peroxidase activity with respect to H2O2 continued to increase as
pH was lowered below 4.5. In terms of turnover number and e?ciency, a maximum for
both was now observed at pH 3.75, 175 s?1 and 1.2 ? 106 M?1 s?1 respectively (Figure
2.6). Below pH 3.75, an abrupt shift in kinetic parameters was observed consistent with a
substantial loss of enzymatic activity.
Similar behavior in peroxidase activity was observed when the kinetic parameters with
respect to reducing substrate (ABTS) were evaluated (Figure 2.7). A continuous increase
in apparent kcat was observed from pH 6.0 down to pH 3.75, reaching a peak value near 200
s?1. As with H2O2-dependent parameters, there was an abrupt decrease below pH 3.75.
The increase in apparent kcat=KM between pH 6.0 and 3.75 was more dramatic, reaching
68
Figure 2.4: Observed Peroxidase Kinetic Parameters Under Saturating ABTS
Conditions. (A) pH dependence of KatG peroxidase kinetic parameters kcat (?) and KM
(N) as determined by varying H2O2 concentration at 1 mM ABTS. (B) pH dependence of
KatG peroxidase kinetic parameters kcat=KM as determined by varying H2O2 concentration
at 1 mM ABTS.
69
Figure 2.5: Evidence of ABTS Inhibition. Initial rates as a function of ABTS concen-
tration at pH 4.75 (?) and 4.25 (N). Dashed line demonstrates observed activity under
inhibiting conditions.
pH [ABTS]
4.5 1.2 mM
4.25 0.6 mM
4.0 0.3 mM
3.75 0.1 mM
Table 2.1: Lowest Concentration of ABTS at Each pH Where Inhibition Was
Observed. As observed at 0.5 mM hydrogen peroxide in 50 mM acetate bufier.
70
Figure 2.6: Observed Peroxidase Kinetic Parameters Versus pH: Constant
[ABTS]. (A) Observed kcat flt to equation 2.16 (?) and observed KH2O2M (N) of perox-
idase when [ABTS] = 0.05 mM. (B) Observed kcat=KH2O2M ([ABTS] = 0.05 mM) flt to
equation 2.15. Open symbols denote data not used in fltting.
71
a maximum at pH 3.75 of 9 ? 106 M?1 s?1. Although the exact values of the Michaelis
constanttowardsABTSandhydrogenperoxidecannotbedeterminedfromthedataathand,
under the assumptions explained in the derivations, changes in kcat=KM should primarily
be due to changes in KM. This indicates that decreasing the pH is accompanied by a sharp
decrease in KM towards ABTS. This is consistent with increasing propensity of ABTS
to act as an inhibitor of peroxidase activity at lower pH. Similar trends in observed KM
were seen with other reducing substrates (Table 2.2). The H2O2- and ABTS-dependent
kinetic parameters for the peroxidase activity of KatG indicated an abrupt shift in catalytic
behaviorbelowpH3.75, indicatingasubstantialdisruptionofproteinconformation. Far-UV
CD spectra showed a dramatic loss of secondary structure with a mid-point for unfolding
occurring at pH 3.6 (Figure 2.8). Spectra recorded between pH 5.0 and pH 3.75 were
indistinguishable. Below 3.75, however, a dramatic change in the spectrum was detected,
and below pH 3.5, the CD spectrum of KatG was relatively featureless. Furthermore, this
is clearly not a bufier efiect considering that both the folded and disrupted protein were
observed in citrate bufier. A control assay at pH 3.5 containing heme and substrates but
no enzyme had no signiflcant activity. Therefore, any residual activity could be explained
by a fraction of the protein remaining folded or partially folded.
2.3.2 Kinetic Parameters for Catalase Activity of KatG
The observed KM for hydrogen peroxide determined at pH 7.0 was 3.8 mM, which is
consistent with other KatGs (BpKatG - 5.9 mM [70], SynKatG - 4.2 mM [151], MtbKatG
- 2.5 mM [141]). The KMs determined below pH 6.0, however, elevated to the point that
saturating hydrogen peroxide concentrations could not be used due to inaccuracies inherent
inabsorbancevaluesgreaterthan1. Inthosecases, theKM shouldbetreatedasanestimate.
The initial catalase velocity data points versus pH at hydrogen peroxide concentrations
above the observed KMs, (22 mM), below the KMs (2.3 mM), and at an intermediate
concentration (11 mM) did, in fact, follow typical activity trends (Figure 2.9A), that is,
they indicate an enzymatic pH optimum at 6.50. This is true for all concentrations used
72
Figure 2.7: Observed Peroxidase Kinetic Parameters Versus pH: Constant
[H2O2]. (A) Observed kcat flt to equation 2.15 (?) and observed KABTSM (N) of perox-
idase when [H2O2] = 0.5 mM. (B) Observed kcat=KABTSM ([H2O2] = 0.5 mM) flt to equation
2.16. Open symbols denote data not used in fltting.
73
pH 6.00 5.50 5.00 4.50 4.00
ABTS
kcat (s?1) 11 ? 1 20 ? 1 55 ? 1 92 ? 3 151 ? 6
KM (mM) 0.06 ? 0.02 0.09 ? 0.03 0.09 ? 0.01 0.053 ? 0.006 0.032 ? 0.003
pyrogallol
kcat (s?1) 69 ? 2 64 ? 2 56 ? 1 44 ? 2 27 ? 1
KM (mM) 1.6 ? 0.2 1.5 ? 0.2 1.3 ? 0.1 1.1 ? 0.2 1.1 ? 0.2
o-dianisidine
kcat (s?1) 71 ? 4 69 ? 6 58 ? 5 37 ? 5 24 ? 2
KM (mM) 0.12 ? 0.02 0.11 ? 0.03 0.09 ? 0.02 0.05 ? 0.03 0.04 ? 0.01
Table 2.2: Observed Kinetic Parameters for E. coli KatG Reducing Substrates.
Assays with ABTS contained 0.5 mM hydrogen peroxide. Assays with pyrogallol and o-
dianisidine contained 0.4 mM hydrogen peroxide.
74
Figure 2.8: Evidence of Protein Unfolding at Low pH. (A) CD spectra obtained in
5 mM acetate bufier at pH 3.75, 4.05, 4.20, 4.60, and 4.90. CD spectra obtained in 5 mM
citrate bufier at pH 3.10, 3.30, 3.50, 3.60, and 3.85. (B) CD signal at 207 nm as a function
of pH.
75
in data collection, not just the three shown here. Curves generated from the catalase
kinetic parameters, if they were to simulate the velocity curves, would have both exhibited
optima at pH 6.5. Instead, the highest kcat came at pH 5.75, while the greatest kcat=KM was
observed at pH 7.00 (Figure 2.9B-C), indicating that understanding of a pH optimum is more
complex than simply monitoring the velocity while varying the pH at saturating substrate
conditions. A better understanding of the kinetic parameters is vital. Although the fltting
of the kinetic parameter curves had surprisingly low R2 values, the values obtained from
the fltting were able to be used to generate the initial velocity curves with high accuracy in
Figure 2.9A. This indicates that the pH optimum of 6.5 is an artifact of the balance between
the optimum pH for reaction of the CpdI-H2O2 complex (5.75) and the optimum pH for
the interaction of CpdI with H2O2 (7.00) (Figure 2.2).
2.3.3 Kinetic Parameters and pKas
Not all kinetic constants could be determined from the steady-state approach taken
here, however, the data do provide reasonable estimates of some of the slower rate constants.
In the peroxidase cycle (Figure 2.1), HCpdI formation (k4) and HCpdII reduction (k6) can
be determined from fltting the observed (kcat=KM) data for constant peroxide and constant
ABTS to equations 1 and 2 respectively (Table 2.3). To eliminate protein-unfolding efiects,
data obtained at pH 3.5 and below were omitted in the analysis (Figures 2.6 and 2.7). The
flt of the curves in Figures 2.6B and 2.7A (which were used for the determination of k4)
can both be seen to be less than ideal due to shoulders around pH 5.0. Alternative models
(such as two pKas with highest activity at lowest pH) were used to flt the data, but they
did not improve the goodness of flt provided with a single pKa model. As a result of loss of
secondary structure, kinetic data could not be obtained below pH 3.75. This led to a degree
of uncertainty in the peroxidase pKas. It is clear from the data, however, that 4.5 can be
established as an upper limit of the pKas considering that the highest observed activity and
e?ciency were all at least twice the values of observed activity and e?ciency at pH 4.5.
76
Figure 2.9: Observed Catalase Kinetic Parameters Versus pH. (A) Initial velocity
as a function of pH at 22 mM H2O2 (currency1), 11 mM H2O2 (?) and 2.3 mM H2O2 (H). Fit lines
are simulations of initial velocities given the above concentrations based on parameters
determined by equations 2.29 and 2.30. (B) Observed kcat flt to equation 2.29 (?) and
observed KM (N) for catalase activity. (C) Observed kcat=KM for catalase activity flt to
equation 2.30.
77
For catalase activity (Figure 2.2), the forward binding rate of peroxide to form E-H2O2
(k5) and its subsequent reaction to form HCpdI (k6) can be very nearly approximated from
the fltting of the observed (kcat) and observed (kcat=KM) data to equations 3 and 4 (Table
2.3). The data showed four pKas for catalase activity - the HCpdI-H2O2 complex had a
lower pKa of 4.8 and upper pKa of 7.2, whereas the HCpdI + H2O2 interactions had a
lower pKa of 5.7 and upper pKa of 7.9. None of the catalase pKas are close enough to
possibly be attributed to the same residue, further emphasizing the separate pH optima for
the HCpdI-H2O2 complex and the interactions between HCpdI and hydrogen peroxide. No
efiort is made here to assign pKa values to any speciflc residue. The rate constant data
reported in Table 2.3, along with the pKas reported here were used to calculate hypothetical
initial velocities at the concentrations used in Figure 2.9A. The curves generated from these
calculations are overlaid on the actual data, demonstrating a flt that validates the model
and reported pH-independent constants.
2.4 Discussion
2.4.1 Optimal Peroxidase Activity and ABTS-dependent Inhibition
The kinetic behavior of KatG is not simple enough for a cursory evaluation. The
multiple functions catalyzed by the same active site, the use of the same substrate to
initiate two difierent catalytic cycles, inactivation of the enzyme by the presence of an
excess of the oxidizing substrate peroxide, and an apparent control by pH as to which
activity is dominant all lend to the kinetic complexity of KatG. The recognition that the
catalytic pathway seemed to be directed by pH made it evident that optimal pH ought to be
determined for each activity. To simplify the method of gaining this kinetic information, pH
optima have been determined only by measuring initial velocities under highly saturating
conditions across a wide range of pH values. This has consistently generated data that
report catalase activity of KatG to be optimal at pH 6.5 and peroxidase activity of KatG
to be optimal between pH 4.5 and 5.0.
78
Peroxidase
k4 0.67 ? 106 M?1 s?1
k6 38.3 ? 106 M?1 s?1
Catalase
k5 3.63 ? 106 M?1 s?1
k6 2.1 ? 104 s?1
Table 2.3: Approximated Rate Constants for E. coli KatG.
79
A factor that has a profound in uence on the activity of KatG that has not been
considered is ABTS-dependent inhibition. Decreasing the pH below 4.5 correlates with a
decreasing KM for ABTS. This also correlates with an increasing efiect of ABTS-dependent
inhibition. In E. coli KatG lacking the large loop insertion between the F and G helices
(or large loop 2), a low KM for ABTS comparable to the KM seen at low pH in the wild
type KatG is observed for ABTS at pH 5.0, and with it strong evidence of ABTS-dependent
inhibition[123]. TheappearanceofasmoothactivitycurvecenteredatpH4.5forperoxidase
activity (as seen in Figures 2.3 & 2.4) can be explained by an apparent decrease in KI for
ABTS with decreasing pH. As long as ABTS concentration was held constant, the activity
would be further inhibited. It should be noted that there have been instances where the
optimal peroxidase activity of KatG has been reported below pH 4.5, but in those instances
the optimal pH corresponds to the lowest pH at which the concentration of ABTS used
does not inhibit (Table 2.1) [146].
By evaluating the kinetic parameters below the inhibition threshold, we saw that the
pKas for the observable peroxidatic steps were below pH 4.5 with greater activity in the
lower pH region, and that there was greater activity and e?ciency at pH 3.75 than at any
other pH or combination of substrate concentrations. Similarly, reduction rates for mono-
functional peroxidases also increase as pH decreases for all substrates with the exception
of substrates that are responsible for donation of both the proton and the electron for the
formation of water. For these substrates, at su?ciently low pH, donation of a proton from
solvent becomes competitive with substrate proton donation. ABTS is solely an electron
donor, and proton transfer is required from the solvent [150]. This would suggest that the
second-order rate constant and kcat for reaction with ABTS should continue to increase as
the pH decreases, contrary to a pH optimum centered at pH 4.5. In contrast, both pyrogal-
lol and o-dianisidine are expected to transfer a proton [150, 153] leading to the anticipated
decrease in activity as pH decreases (Table 2.2). As we have shown, maximum observed
activity and catalytic e?ciency do in fact increase and even accelerate as pH decreases to
pH 3.75. Below pH 3.75, however, a sharp drop in activity was observed. Similar efiects in
80
lignin peroxidase were ascribed to protein instability [154]. The loss of activity between pH
3.75 and 3.5 is concomitant with a signiflcant loss of secondary structural content of KatG
as shown by circular dichroism.
With the assumption that optimal peroxidase activity of KatG was in the range of pH
4.5 to 5.0, work has been done indicating a partial structural rearrangement of R426 in the
BpKatG sequence [128]. They report that at pH 4.5 R426(411 E. coli) is predominately
(90%) oriented towards R492(479), R497(484), and D495(482) (R conflguration); at pH 8.5
R426 is predominately (95%) oriented towards Y238(226) (Y conflguration); and at pH 6.5
the orientation is approximately 50:50 R:Y. Ultimately, it is suggested that the optimal
activities of peroxidase (4.5), catalase (6.5), and NADH-oxidase (8.75) may result from how
the position of R426 afiects the movement of electrons around the active site. In light of
the data presented here, the structure = mechanism conclusions based on pH behavior may
need to be explored further. Due to the structural disruption and loss of activity, the true
optimum of peroxidase activity could not be observed, and we can only place an upper
limit on the apparent pKas (4.5). Although this does not negate the explanation given
by Carpena et al [128], it does indicate that the orientation of the R426 may be only one
of multiple pH-dependent features that optimize peroxidase activity. We will revisit this
arginine movement in chapter 5 within the context of the roles of interdomain interface
residues R479 and D482.
Our data show that pH control over which mechanistic pathway is utilized by KatG
may be just as related to changes in substrate binding as it is to turnover mechanisms.
Signiflcant in the capacity of the enzyme to catalyze the peroxidase reaction is the decrease
in KM for ABTS simultaneous with the increase in apparent KM for H2O2 in the catalase
cycle as the pH is lowered. This would tend to favor a shift from catalase to peroxidase
catalysis. Considering that peroxidase and catalase share the same initial oxidation step,
this implies that binding of hydrogen peroxide to the ferryl-oxo form of the enzyme becomes
less facile, disfavoring compound I reduction by H2O2. As pH continues to drop, excess
ABTS even interferes with activity itself. Although the binding site for ABTS is unknown,
81
it can be postulated that decreasing the pH below 4.5 results in a change in the binding site
that allows for alternative inhibitory binding modes by ABTS. This would be facilitated
with a more open and  exible binding site in this pH range - a  exibility that perhaps
even foreshadows the secondary structural loss below pH 3.75. According to the CD data,
however, any change in the ABTS binding site does not involve any observable change in
secondary structural content of the protein. It is possible that increased a?nity for ABTS
is due to the negatively charged sulfonic acid groups being attracted to the more protonated
enzyme at lower pHs, however, this would not explain the similar decrease in KM of the
neutral substrates pyrogallol and o-dianisidine (Table 2.2).
2.4.2 Difierent pH Optima for Binding and Activity in Catalase Cycle
With regards to catalase activity, it was expected that our data would simply be an
addition to what was already known about catalase kinetics. It was surprising when our pH
optima did not coincide with previously published results. This, however, resulted from the
more exhaustive approach used to gather the data. Once it was realized that the data we
had at any constant hydrogen peroxide concentration led to a pH proflle comparable to those
obtained by others, we were able to more fully accept the parameters as being as complex
as they appeared. Difierent pH optima for binding and activity is certainly not uncommon.
Recent spectral data lends support to this observation. Work done with SynKatG, BpKatG,
and MtbKatG show that catalase intermediates generated at pH 8.5 (and 7.0 for SynKatG)
have UV-visible spectral maxima at 418 and 520 nm, but intermediates generated at pH 5.6
have maxima at 415, 545, and 580 nm [151]. Our work with E. coli KatG indicates maximal
catalase e?ciency at pH 7.0, but maximal turnover number at pH 5.75. Furthermore,
considering that saturating hydrogen peroxide concentrations were not feasible at the lower
end of the pH range used, the kcat values may be underestimated and the optimum pH
for turnover would then be even lower than 5.75. Attempting to understand the structure-
function relationship of catalase-peroxidases with the assumption of a pH optimum for
catalase is 6.5 will yield unsatisfactory results if not coupled with the understanding that
82
this optimum results as the overlap of binding and activity pH curves. The multiple pKas
of catalase activity give it a more complex appearance than peroxidase, which may have
only one. The pH optimum for binding occurring where catalase turnover is not optimal
(and vice-versa) provides a kinetic basis for what has primarily been a structural-based
argument (concerning features such as the covalent cross-link and duplicated domain) that
catalase activity was obtained by ancestral monofunctional peroxidases (possibly related to
lignin peroxidases) and is not an activity that is indigenous to the enzyme.
83
Chapter 3
Presence of Reducing Substrates Broadens
Catalase Activity pH Range
3.1 Introduction
Peroxidase kinetic characterizations have largely neglected the fact that some fraction
of the enzyme population is going through the catalase cycle according to the classical
scheme (Figure 1.13). This results in artiflcially lowering the peroxidase activity ascribed
to KatG. In chapter 2, we saw that the pH optima of Escherichia coli KatG (EcKatG) for
catalase and peroxidase were separated by nearly 3 pH units (3.75 for peroxidase, 6.5 for
catalase), and from pH 4.5 to 5.5 the enzyme exhibited only a fraction (<50%) of either
activity. To date, KatG kinetics have been performed by taking advantage of this large
difierence by performing them so that only the necessary substrates were present and at a
pH appropriate to minimize the interference by the alternative catalytic cycle; but with the
majority of published peroxidase data being acquired between pH 4.5 and 5.0, interference
by catalase activity should not be considered negligible. Quantiflcation of this interference
would be beneflcial in interpreting published kinetic data. Furthermore, understanding the
kinetics of KatG while all catalytic pathways are available is much more biologically relevant
than the classical assay.
We monitored the oxygen production attributed to catalase activity with and without a
wide number of known or potential reducing substrates. We found that reducing substrates
not only fail to inhibit catalase turnover around pH 5.0, but they actually enhance oxygen
production rates. Evaluating the pH range over which this occurs suggests that this efiect
serves to broaden the pH range of optimal catalase activity. Finally, we suggest that the
presence of the reducing substrate activates oxygen production by protecting the enzyme
84
from inactivation, corroborating recently published data and further supporting the concept
that the classic portrayal of the KatG catalytic cycles is largely insu?cient.
3.2 Materials and Methods
3.2.1 Materials
Hydrogen peroxide (30%), imidazole, hemin, Sephacryl 300 HR, ampicillin, chloram-
phenicol, sodium dithionite, phenylmethylsulfonyl  uoride (PMSF) 2,2?-azino-bis(3-ethyl-
benzthiazoline-6-sulfonic acid) (ABTS), L-tyrosine, phenazine, chlorpromazine (CPZ), 3,3?-
dimethoxybenzidine hydrochloride (o-dianisidine), guaiacol, ferulic acid, L-ascorbic acid,
and pyrogallol were purchased from Sigma (St. Louis, MO). Isopropyl-fl-D-thiogalactopyra-
noside (IPTG), mono- and di-basic sodium phosphate, acetic acid, sodium acetate, ethanol,
and sodium hydroxide were obtained from Fisher (Pittsburgh, PA). L-tryptophan was pur-
chased from J.T. Baker Chemical Company (Phillipsburg, NJ). Bugbuster and benzonase
were purchased from Novagen (Madison, WI). The E. coli strain BL-21 [DE3] pLysS was
obtained from Stratagene (La Jolla, CA). Nickel-nitrilotriacetic acid (Ni-NTA) resin was
purchased from Qiagen (Valencia, CA). All bufiers and media were prepared using water
purifled through Millipore Q-PakII system (18.2 M?/cm resistivity).
3.2.2 Expression, Puriflcation, and Reconstitution of EcKatG
Expression and puriflcation of EcKatG was carried out as described in chapter 2.
Heme reconstitution and holoenzyme concentration determination was also carried out as
described in chapter 2.
3.2.3 Activity Assays
Data for the pH-dependent peroxidase and catalase assays come from chapter 2. To
monitor the efiects of peroxidatic substrate on catalase activity, assays were performed by
85
monitoring oxygen production with a Hansatech DW1 oxygen electrode chamber (Pent-
ney, Norfolk, England) using accompanying Oxygraph Plus software. Reducing substrates
ABTS, ascorbate, CPZ, ferulic acid, guaiacol, o-dianisidine, phenazine, pyrogallol, tryp-
tophan, and tyrosine were each added to the assays to determine their efiects on oxygen
production. Ferulic acid and tyrosine stock solutions were prepared as 10 mM in 0.1 M
NaOH. Guaiacol stock solution was prepared as 10 mM in 5% ethanol. Phenazine stock
solution was prepared as 10 mM in 50% ethanol. All other stock solutions were prepared in
water. Unless otherwise noted, assays were performed using 2.5 nM enzyme and appropriate
volume of solvent (baseline) or stock solution in 50 mM acetate bufier and initiated with
10 ?L of the appropriate concentration of hydrogen peroxide (1 mL flnal reaction volume)
at 21 ?C. Efiect of reducing substrate was represented using the following equation:
efiect = activity with substrateactivity without substrate ?1 (3.1)
A positive efiect would indicate oxygen production enhancement, a negative number would
indicate inhibition, and zero would indicate that the substrate had no efiect on oxygen
production. Due to the high concentration of ethanol required to dissolve phenazine, assays
involving 1 mM phenazine could not be performed due to enzyme precipitation.
3.2.4 End-point Assays and UV-visible Spectra
End-point assays were used to determine which of the ten substrates listed above were
serving as peroxidase substrates. Two samples were prepared for each substrate containing
2.5 nM enzyme and 0.1 mM substrate at pH 5.0. To one of the samples, 1.0 mM hydro-
gen peroxide was added. These reacted for 2 hours, and then their absorbance spectra
were taken at room temperature using a Shimadzu UV-1601 spectrophotometer (Columbia,
MD) with a cell pathlength of 1.0 cm. The difierence between the two spectra was then
86
divided by the maximum absorbance to normalize the difierence spectra of the various sub-
strates to fraction-changed difierence spectra. All other spectra were also recorded at room
temperature on the Shimadzu UV-1601 spectrophotometer.
3.3 Results
3.3.1 Efiect of Reducing Substrates on Oxygen Production
To evaluate what portion of the enzyme population was still contributing to catalase
activity during peroxidase assays at pH 5.0, we measured catalase activity by oxygen pro-
duction with and without ten difierent possible peroxidatic reducing substrates (Figure 3.1).
Activity efiects were determined at 0.1 and 1.0 mM reducing substrate and hydrogen per-
oxide. Of the ten, only the amino acids tyrosine and tryptophan did not have an efiect on
catalase activity. Of the remaining eight, only 1 mM o-dianisidine was actually observed to
inhibit catalase activity, and it surprisingly served to enhance oxygen production at 0.1 mM,
as did the remaining seven in all cases. Phenazine enhanced oxygen production by about
25%, and ferulic acid and guaiacol were only marginally more potent. ABTS, ascorbate,
and pyrogallol, however, enhanced oxygen production by over 150% at certain concentra-
tions of substrate and 1.0 mM hydrogen peroxide. The strongest activation was observed
with the substrate CPZ where oxygen production was more than quadrupled at 1.0 mM
hydrogen peroxide. In the absence of enzyme, reducing substrate and hydrogen peroxide
did not generate molecular oxygen (data not shown).
A survey of the characterization of eight catalase-peroxidases from a variety of sources
showed that o-dianisidine and pyrogallol have been shown to be a peroxidase substrate
in nearly all cases [105{112]. ABTS has been reported as a substrate in many of those
and is commonly used due to its high solubility in water and large absorption coe?cient
[106, 107, 111, 112]. No instances were found in which CPZ, ferulic acid, phenazine, or
the amino acid tryptophan were tested, although all have been used as substrates for
some monofunctional peroxidases [155{158]. Guaiacol was found to be a substrate for the
87
Figure 3.1: Efiects of Reducing Substrate on Catalase Activity at pH 5.0. Change
in % activity based on Equation 3.1
88
catalase-peroxidase isolated from E. coli [105], Synechocystis PCC 6803 [109], and Penicil-
liumsimplicissimum[107], butnotfromRhodopseudomonascapsulatus[108]orBurkholderia
cenocepacia [106]. Ascorbate was not found to be a peroxidase substrate for Rhodopseu-
domonas capsulatus [108] or Synechocystis PCC 6803 [109], but has been used as a perox-
idatic electron donor along with tyrosine in the Y249F variant of Synechocystis PCC6803
KatG [126]. Using end-point assays to observe substrate consumption or oxidized product
formation, we observed that ABTS, ascorbate, guaiacol, pyrogallol, and o-dianisidine all
showed evidence of being peroxidase substrates with the predicted increase in absorbance
at the wavelength corresponding to oxidized product: ABTS - 417 nm, guaiacol - 450 nm,
o-dianisidine - 460 nm, pyrogallol - 430 nm (Figure 3.2). Oxidized ascorbate decomposes
rapidly and could not be visualized, but a clear decrease in absorbance at 270 nm was
observed. Small changes in the spectral features of CPZ and ferulic acid were also ob-
served, but were too slight to be considered signiflcant. Tyrosine and tryptophan showed
no evidence of oxidation or consumption.
3.3.2 Evaluation of Role of pH on Activation Efiects
It was important to establish whether this activation was isolated to the pH region
where both peroxidase and catalase activities were diminished when measured indepen-
dently, or if it also occurred close to their pH optima. ABTS was used to determine the
efiect of pH on the enhancement of oxygen production for the reasons listed above and the
wealth of data already assembled for EcKatG peroxidase activity with ABTS. At pH 4.0,
ABTS does act as an inhibitor of catalase activity, but the most signiflcant evidence of oxy-
gen production enhancement occurs from pH 4.5 to 5.5 (Figure 3.3). This falls directly in
the range where catalase and peroxidase activity fall below 50% of their individual optima.
It has been established that the increase in Michaelis constant (KM) for hydrogen per-
oxide contributes signiflcantly to the decrease in observed catalase activity as pH decreases
[151, 159]. Determination of the KM for hydrogen peroxide at pH 5.0 both with and with-
out the presence of ABTS did reveal that having the electron donor present decreased the
89
Figure 3.2: Evidence of Peroxidatic Consumption of Reducing Substrates. Each
line is the difierence spectrum of 0.1 mM substrate incubated with 2.5 nM EcKatG at pH
5.0 for 2 hrs with and without 1.0 mM H2O2 added normalized to the maximum absorbance
for the respective species in that wavelength range. Those with most signiflcant changes are
noted. The ten substrates shown are ABTS (red), ascorbate (orange), CPZ (yellow), ferulic
acid (green), guaiacol (cyan), o-dianisidine (blue), phenazine (purple), pyrogallol (pink),
tryptophan (gray), and tyrosine (black).
90
Figure 3.3: pH-dependence of Activation with ABTS Compared to Catalase and
Peroxidase Activities. Peroxidase activity (?) determined with 0.06 mM ABTS, 1.0 mM
H2O2. Maximum peroxidase activity observed was 145 s?1 at pH 3.75. Catalase activity
(N) determined with 1.0 mM H2O2. Maximum catalase activity observed was 1200 s?1 at
pH 7.5. Oxygen activation (?) determined with 0.1 mM ABTS, 1.0 mM H2O2. Maximum
activity (917 s?1) observed with ABTS present was 2.6 times activity without ABTS (353
s?1) at pH 5.0.
91
KM (Figure 3.4). Although this would account for the activation at low hydrogen per-
oxide concentrations, it could not explain how all of the assays that evidenced activation
had more linear traces than the slower baseline assays that did not contain the electron
donor. Velocities are much more sensitive to changes in substrate concentration when it is
present at sub KM concentrations, and even peroxide consumption rates by KatG become
more hyperbolic as the KM increases relative to a given initial peroxide concentration [151].
Figure 3.5 shows an example of two assays with very similar rates, yet the more linear
of the two includes the substrate ABTS and peroxide at sub KM concentration, and the
more hyperbolic trace has no reducing substrate and contains peroxide in excess of the KM
(ten times the concentration of peroxide in the ABTS-containing assay). Bearing in mind
that the total peroxide consumption in these two assays is approximately 100 ?M, reactant
consumption should have a negligible efiect on the oxygen production rate of the 10 mM
peroxide assay and the assay would be expected to be more linear. Furthermore, with the
products of catalase activity being water and molecular oxygen, it is equally unlikely that
product inhibition could account for the hyperbolic nature of the trace. This leaves enzyme
inactivation as the most likely suspect for the rate deceleration.
3.3.3 Efiect of Reducing Substrate Presence on Reacted Enzyme Spectra
Assuming an inactive form of the enzyme would be produced in the presence of hydro-
gen peroxide, but would be prevented or rescued by the presence of reducing substrates, we
took a series of absorption spectra to attempt to visualize any potential inactive intermedi-
ate (Figure 3.6). Spectrum 1 contained only EcKatG at pH 5.0. Spectrum 2 was a control
showing that the presence of ABTS and ascorbate did not afiect the spectrum after a 60
second incubation. Ascorbate was included to prevent the accumulation of ABTS radical,
a strong absorber in the heme Soret region. Spectrum 3 was taken following 60 seconds
of reaction of EcKatG with 1.0 mM hydrogen peroxide. A red-shift in the Soret band to
407.8 was observed, along with a loss of features at 516 and 543 and a blue-shift of the
CT band at 633 to below 620. Also, an increase in light-scattering was observed which can
92
Figure 3.4: Efiect of ABTS on Apparent KM for H2O2 at pH 5.0. Activity assays
with (N) and without (?) ABTS flt to standard Michaelis-Menten curve.
93
Figure 3.5: Efiect of ABTS on Linearity of Catalase Initial Rates. Solid line is assay
performed with 1.0 mM H2O2 and 0.1 mM ABTS. Dashed line is assay with 10 mM H2O2
without ABTS.
94
be characteristic of oxygen bubble production accompanying catalase turnover. With the
high concentration of enzyme used, this was not unexpected. Spectrum 4 was also taken
following 60 seconds of reaction with 1.0 mM hydrogen peroxide, but included ABTS and
ascorbate in the reaction. The light scattering was even more evident, but the spectral
features of the spectra 1 and 2 were retained, indicating a prevention of the accumulation
of a non-resting state species. Assignment of the new spectral features to a species is not
necessary for the purposes here and would be di?cult due to low accumulation. However,
further investigation into the identity of the accumulated species would provide more insight
into the possible mechanism.
3.4 Discussion
The mechanism of difierentiation between catalase and peroxidase activity in KatG
is a process which is undergoing much discussion. Catalase-peroxidase isolated from Ar-
chaeoglobus fulgidus undergoes rapid inactivation in the presence of H2O2, but it has also
been noted that ABTS has a \stabilizing efiect" on the enzyme, increasing the half-life
three-fold at pH 4.5 [111]. It was not suggested what this inactivation might be. Of the
eight substrates expected to be good peroxidatic electron donors (this excludes tyrosine
and tryptophan), all enhanced oxygen production rates with only one exception that will
be discussed below. There is nothing in our data to suggest that this is anything other than
a result of an increase in catalase turnover. The lack of oxygen production in the absence of
enzyme eliminates the possibility of some non-enzymatic reaction as an explanation. Fur-
thermore, considering that accumulation of oxidized substrate is not required to observe
enhanced oxygen production rates, we can conclude that it is the reducing substrate itself,
and not the oxidized counterpart, that activates the oxygen production (Figure 3.5).
The only substrate to demonstrate signiflcant inhibition of oxygen production was o-
dianisidine. This was also observed in Escherichia coli catalase peroxidase by Claiborne
et al in the flrst characterization of a catalase-peroxidase [105]. They used the catalatic
inhibition by o-dianisidine to monitor peroxidase activity without the interfering efiects of
95
Figure 3.6: \Dead-end" Species Does Not Accumulate When ABTS Is Present.
(1) 1.4 ?M EcKatG. (2) 1.4 ?M EcKatG incubated with 0.1 mM ABTS and 0.5 mM ascorbic
acid for 60 seconds. (3) 1.4 ?M EcKatG reacted with 1.0 mM H2O2 for 60 seconds. (4) 1.4
?M EcKatG reacted with 1.0 mM H2O2 for 60 seconds in the presence of 0.1 mM ABTS and
0.5 mM ascorbic acid. Italicized numbers indicate wavelengths of features more prominent
in spectrum 3; standard numbers indicate wavelengths of features diminished in spectrum
3.
96
hydrogen peroxide consumption by catalase activity. Enzyme inactivation by 0.33 mM o-
dianisidine was observed by monitoring peroxidase activity, but inactivation by numerous
other electron donors (including guaiacol and pyrogallol) was not observed. Their results
indicated that o-dianisidine at su?ciently high concentrations can irreversibly inactivate the
enzyme. As such, inhibition by o-dianisidine can be treated as unique and occurs through
a mechanism not yet described and unavailable to other electron donors.
At this point, two very surprising and signiflcant facts become clear regarding both
the enzyme mechanism and the ability of KatG to consume peroxide. First, difierentiation
between the catalase and peroxidase cycles must occur prior to interaction with the second
substrate, a phenomenon not previously suggested for catalase-peroxidases and obviously
unnecessary to suggest for monofunctional peroxidases. Secondly, nature has flne-tuned
KatG to maintain high levels of catalatic hydrogen peroxide consumption across its entire
range of pH, more similar to monofunctional catalases, in spite of its remarkable structural
similarity to monofunctional peroxidases. We will discuss each of these individually.
If peroxidatic reducing substrates interacted with an intermediate involved in cata-
lase turnover (such as the ferryl heme porphyrin cation radical), they also must act as
competitive inhibitors of oxygen production. This is evidently not the case. If, however,
difierentiation between catalase and peroxidase turnover occurs following reaction with the
flrst molecule of peroxide but prior to reaction of the second substrate, in the absence of a
reducing substrate a fraction of the enzyme population would dedicate itself to a \dead-end"
peroxidase pathway, progressively inactivating the population (Figure 3.7). The presence of
an electron donor would, however, allow for the completion of the peroxidase cycle, opening
a pathway back to the resting state, preventing inactivation and resulting in higher observed
catalase rates. This is consistent with our results. The accumulation of a \dead-end" would
explain the hyperbolicity seen in Figure 3.5 and spectral changes in Figure 3.6.
Coupledwithresearchdoneoncatalaseintermediatesfoundintheliterature, ourresults
have led us to propose the scheme found in Figure 3.7 as a possible mechanistic pathway.
The inclusion of protein-based radicals in catalase-peroxidase mechanistic schemes has been
97
Figure 3.7: A Proposed Scheme Accounting for Difierentiation Between Cata-
lase and Peroxidase Prior to Substrate Interaction, and including possible dead-end
pathway available when reducing substrate is not present.
98
discussed at length [124, 139{142, 144, 145, 160], particularly following the identiflcation
of the covalent adduct involving a methionine, tyrosine, and tryptophan. Heme-dioxygen
intermediates in the catalase cycle have also been discussed more recently, and are the most
recently proposed intermediates in the mechanism based on rapid freeze quench electron
paramagnetic resonance studies of catalase intermediates [143]. Building on these recently
published mechanisms, we propose that peroxidase activity also requires a rapid radical
transfer prior to turnover completion. In variants incapable of forming the covalent adduct,
rapid accumulation of a species with a Compound III-like spectrum in the presence of excess
hydrogen peroxide has been observed in our lab (data not shown) and others [126, 139].
This supports the idea that rapid radical transfer occurs following reaction with the initial
hydrogen peroxide. If the radical is transferred to the nearby covalent adduct, the heme-
dioxygen intermediate formed upon further reaction with hydrogen peroxide can be rescued
by the electron hole on the adduct. If the radical is transferred to some other (possibly
external) site, reaction with peroxide would form an inactive heme-dioxygen species. Fur-
thermore, an external site would explain the ability of large molecules like ABTS and CPZ
to be such e?cient electron donors in spite of the narrow access channel to the heme active
site.
Prior to the observations made here, with catalase activity decreasing as pH dropped
and peroxidase activity not being substantial until below the pKa 4.5, there was a large
apparent pH gap in activity, and thus a pH gap in peroxide consumption ability. A gap
such as this would be detrimental to an organism?s ability to deal with oxidative stress
and would be expected to be unfavored by nature. At pH 5.0, we observed that EcKatG
in the presence of 0.1 mM ABTS generates molecular oxygen at 250% of the rate of the
enzyme with hydrogen peroxide alone (Figure 3.1). This also happens to be where catalase
is around 30 - 40% of optimal activity (Figure 3.3), restoring oxygen production to 75 -
100% of optimal. ABTS only inhibited below the pKa for peroxidase activity. Essentially,
the presence of ABTS expanded the efiective pH range of catalase activity all the way to the
point where peroxidase turnover is most active, eliminating the gap in e?cient hydrogen
99
peroxide consumption. This activity broadening mechanism would be particularly crucial
for an organism such as M. tuberculosis, which does not have any other catalase active
protein to deal with oxidative stress. Overall, it is striking to observe how nature has
maximized this enzymes ability to decompose hydrogen peroxide using two mechanisms,
utilized by the same active site, that work synergistically rather than competitively.
100
Chapter 4
Generation of Mixed Spin-state Population
via Y111A Substitution
4.1 Introduction
To this point, the focus has been on the parametrics afiecting the kinetic properties
of KatG. In these next two chapters, our attention will turn to the role of global structure
in catalytic ability, particularly the interactions between the N- and C-terminal domains of
KatG.
Just as the calcium in class II peroxidases interact with residues on the catalytic histi-
dine bearing B helix and BC interhelical loop [94{97, 102], the C-terminal domain in KatG
interacts with the N-terminal domain in the same region. Investigation of the interdomain
interface reveals that the residues along the interface are highly conserved and create two
hydrogen bonding networks between 25 and 30 ?A away. The closer network involves in-
teractions of Tyr111 and backbone carbonyls on the BC interhelical loop with Asp482 and
Arg479 on the B?C? interhelical loop respectively. The other contact exists between Arg117
on the BC interhelical loop with Asp597 on the E? helix (Figure 4.1).
Elucidating the roles of the individual interdomain interactions is anticipated to help us
in our understanding of how the C-terminal domain restructures the N-terminal active site
and imparts activity to the enzyme. The interaction nearest the active site is the Y111-D482
pair. We chose to disrupt this interaction flrst by mutating only Y111 to an alanine (Y111A)
since D482 is within hydrogen-bonding distance of R479 and may afiect its positioning for
interaction with the backbone carbonyls of the BC interhelical loop. We then observed the
spectral and kinetic properties of Y111A. The spectral results demonstrated that prepara-
tions of this variant yielded a mixture of high- and low-spin heme states, thus creating the
101
Figure 4.1: Interactions Between the N-terminal BC Interhelical Loop and C-
terminal Domain of KatG. All C-terminal domain structures are labeled with primed
letters (e.g., E?-helix), and all structures from the N-terminal domain are shown with unac-
cented letters (e.g., B-helix). Coordinates from the structure of Burkholderia pseudomallei
KatG were used (PDB accession number 1MWV) [117]. Numbering re ects Escherichia
coli KatG sequence.
102
appearance of a transition between wild type (high-spin) and C-terminal domain-lacking
(low-spin) KatG (KatGN). Concurrently there was a loss of activity, suggesting that the
Y111-D482 does play a role in preventing His106 coordination. Furthermore, we showed
that peroxidase activity becomes proportionally more favorable than catalase activity when
compared to the wild type, insinuating that the Y111-D482 interaction is also involved in
properly conflguring the enzyme for bifunctionality.
4.2 Materials and Methods
4.2.1 Materials
Hydrogen peroxide (30%), imidazole, hemin, ampicillin, chloramphenicol, sodium dithi-
onite, phenylmethylsulfonyl  uoride (PMSF), and 2,2?-azino-bis(3-ethylbenzthiazoline-6-
sulfonic acid) (ABTS) were purchased from Sigma (St. Louis, MO). Isopropyl-fl-D-thioga-
lactopyranoside (IPTG), urea, mono- and di-basic sodium phosphate, acetic acid, and
sodium acetate were obtained from Fisher (Pittsburgh, PA). Bugbuster and benzonase
were purchased from Novagen (Madison, WI). All restriction enzymes were purchased from
New England Biolabs (Beverly, MA). All oligonucleotide primers were purchased from In-
vitrogen (Carlsbad, CA). All E. coli strains (BL-21 [DE3] pLysS and XL-1 Blue) and Pfu
polymerase were obtained from Stratagene (La Jolla, CA). Nickel-nitrilotriacetic acid (Ni-
NTA) resin was purchased from Qiagen (Valencia, CA). Desalting chromatography columns
were purchased from Bio-Rad. All bufiers and media were prepared using water purifled
through Millipore Q-PakII system (18.2 M?/cm resistivity).
4.2.2 Cloning
The plasmid pKatG(Y111A) was prepared using mutagenic primers [5?-GCG GGG
ACT GCA CGT TCA ATC GAT GG-3? (coding) and 3?-CGC CCC TGA CGT GCA AGT
TAG CTA CC-5? (non-coding)] for pKatG according to the QuikChange procedure (Strata-
gene, La Jolla, CA). Ampliflcation products were used to transform E. coli XL-1 Blue by
103
electroporation (BIO-RAD MicroPulser, Hercules, CA). Plasmids from candidate colonies
were evaluated by diagnostic restriction digest and DNA sequence analysis. Correctly mu-
tated plasmid was used to transform E. coli (BL-21 [DE3] pLysS).
4.2.3 Expression and Puriflcation
Expression and puriflcation of Y111A KatG was carried out similar to wtKatG as
described in chapter 2 with few exceptions. Prior to loading the cell lysate onto the Ni-
NTA column, urea was added to a concentration of 2M to relax the protein for better
column binding. Due to higher a?nity of the protein for the Ni-NTA column than wild
type, the protein did not require a gradient elution. Instead, the column was washed with
2, 20, and 200 mM imidazole in 50 mM phosphate bufier / 200 mM NaCl pH 7.0. The 20
and 200 mL washes were demonstrated to contain the target protein by SDS-PAGE and
were combined. Also, due to such low levels of non-target protein contamination, the eluted
protein only needed to be separated from the imidazole and urea using a desalting column.
Protein fractions were combined and concentrated with Amicon ultraflltration concentrator.
Concentration of purifled enzyme was estimated according to the method of Gill and von
Hippel [50] ("280 = 1.43 ? 105 M?1 cm?1).
4.2.4 Absorption Spectra and Activity Assays
Enzyme was reconstituted with 0.9 equivalents of hemin. Reconstituted enzyme solu-
tion incubated at 4 ?C for 72 hours to allow unincorporated heme to settle out of solution.
The solution was then centrifuged and the precipitated heme and other insoluble mate-
rial discarded. Concentration of reconstituted enzyme was determined using the pyridine
hemichrome assay [161]. Protein containing heme in the ferrous state was prepared by
adding a small amount (<10 mg) of sodium dithionite to the native enzyme. All spectra
were obtained at room temperature on a Shimadzu UV-1601 spectrophotometer (Columbia,
MD) with a cell pathlength of 1.0 cm.
104
Catalase and peroxidase activity assays were performed as described in chapter 2.
Initial velocities were flt to Michaelis-Menten equation by non-linear regression analysis to
determine apparent kinetic parameters. If inhibition was evident, the fltting equation was
modifled to a general excess substrate-dependent inhibition model:
v?
[E]tot =
kcat[S]
KM +[S]+ [S]
2
KN
(4.1)
where KN is a macroscopic constant that re ects the ability of the substrate to act as an
inhibitor.
4.2.5 Stopped- ow
Binding of CN? by Y111A KatG was monitored using an SX.18 MV Stopped- ow
Rapid Reaction Analyzer (Applied Photophysics, Surrey, UK) in single mixing mode. One
syringe contained 5 ?M Y111A KatG and the second contained KCN (20?400 ?M). Spectra
were recorded by diode array. Kinetic constants were determined from triplicate measure-
ments of single wavelength data recorded at 414 nm. All reactions were carried out at 25
?C in 100 mM phosphate bufier, pH 8.0.
4.2.6 Magnetic Circular Dichroism
Y111A KatG, wtKatG, and KatGN spectra were obtained using 15 ?M enzyme in 50
mM phosphate 50 mM NaCl bufier, pH 7.0. Spectra were recorded at 23 ?C in a quartz cell
(5.0 mm path length) with 1.4 Tesla magnetic cell holder from 700?350 nm on a Jasco J-810
spectropolarimeter (Tokyo, Japan). Baselining and analysis were done using Jasco J-720
software. Y111A KatG containing ferrous heme was prepared by adding a small amount
(<10 mg) of sodium dithionite to the native enzyme.
105
4.2.7 Electron Paramagnetic Resonance
Spectra were recorded using a Bruker EMX instrument equipped with an Oxford ESR
900 cryostat and ITC temperature controller. Additional sample concentration was per-
formed using Amicon Ultra-4 centrifugal devices. The settings for the spectrometer were
as follows: temperature, 10 K; microwave frequency, 9.38 GHz; microwave power, 0.1 mW;
modulation amplitude, 10 G; modulation frequency, 100 kHz; time constant, 655.36 ms;
conversion time, 655.36 ms; and receiver gain, 1.0 ? 105.
4.3 Results and Discussion
4.3.1 UV-visible Absorption
The heme absorption spectra for native (ferric) and reduced (ferrous) Y111A KatG
were taken and compared with the spectra of the wild type KatG (Figure 4.2 and Table
4.1). The wild type ferric heme spectrum is consistent with a mixture dominated by penta-
and hexa-coordinate high-spin heme iron [162]. The shift of the Y111A Soret band to 411
nm and the red shift of the charge transfer two (CT2) band into the beta region suggest a
greater contribution from a low-spin species in the Y111A ferric heme spectrum. In fact, the
band at 525 nm most likely has contributions from both CT2 and fl-band absorption. These
spectral features were not in uenced by the identity of the bufier, the ionic strength of the
solution, or by formation of the crosslink through reaction with 100 equivalents of hydrogen
peroxide. Changes in pH also had no efiect on the spectrum from 8.0 to 6.0. Between pH 6.0
and 4.5, however, the spectra indicated signs of increased protein aggregation as evidenced
by light scattering and precipitation, but no change in the position of spectral features after
centrifugation. At pH 4.0 removal of the precipitate through centrifugation resulted in a
spectrum similar to wtKatG, indicating that the population of Y111A contributing low-spin
heme was completely insoluble at this pH.
The ferrous Y111A Soret band appeared at 427 nm, consistent with a low-spin heme
system, but also demonstrated a substantial shoulder near 440 nm where high spin ferrous
106
heme absorbs. In the alpha and beta region of the ferrous spectrum (520 nm - 600 nm),
high-spin heme exhibits an absorption band near 560 nm and a shoulder between 580 and
590 nm, such as is observed with the wild type KatG. Low-spin ferrous heme exhibits
absorption bands near 530 and 560 nm. Y111A absorbed at 560 nm with shoulders near
530 and 585 nm, also indicative of a mixture of both high and low-spin ferrous heme.
4.3.2 Cyanide Binding
Changes in the rates or mechanism by which a KatG variant binds cyanide compared
to the wild type can provide insight into changes in accessibility of the active site. Y111A
KatG, unlike the wild type [51, 130], demonstrated a two-phase binding process including a
fast phase and a slow phase. Previous cyanide binding experiments with a novel periplasmic
catalase-peroxidase, KatP, also showed a two-phase process. One phase was dependent on
cyanide concentration, but the other was not [51]. In Y111A, however, both phases were
linearly dependent on cyanide concentration (Figures 4.3A and B). The fast phase had an
apparent rate constant (or on-rate - kon) of 3.5 ? 105 M?1 s?1, which is comparable to
the apparent rate constant of wtKatG - 5 ? 105 M?1 s?1 [51]. The slow phase had an
apparent rate constant of only 4% of the fast phase - 1.7 ? 104 M?1 s?1. The amplitudes of
the two phases were independent of cyanide concentration (Figure 4.3C), with 64.1% of the
amplitude coming from the faster phase and the remaining 35.9% coming from the slower
phase (determined from the average amplitude across all concentrations). When considered
alongside the UV-Vis spectral data, it is reasonable to suggest that the two phases in the
Y111A stopped- ow experiment are in fact two distinct species - each undergoing a single-
phase binding step. The fast phase would correspond to a high-spin species, and the slow
phase a low-spin species.
The dissociation rate of cyanide (koff) is determined by the y-intercept of the kobs
vs. [CN?] plots (Figures 4.3 A and B), allowing for calculation of the dissociation constant
(KD) for cyanide from the koff=kon. The KD corresponding to the flrst exponential was
determined to be 11.1 ?M. Within error, koff for the slow phase was zero, suggesting that
107
Figure 4.2: UV-visible Absorption Spectra of Native and Reduced Y111A KatG.
Native Y111A (?) contains heme in the ferric state. Reduced Y111A (??) contains heme
in the ferrous state. (A) Soret region of heme absorption spectrum. (B) fi, fl, and charge
transfer region of heme absorption spectrum.
108
Heme protein Absorption band maxima [nm (mM?1 cm?1)]
state Soret( ) fl fi CT2 CT1
ferric wtKatG 408 (120.7) - - 502 (17.5) 629 (9.7)
Y111A 411 (92.7) 525 (9.98) - - 634.5 (2.91)
ferrous wtKatG 439 (79.8) 565 (15.1) 581 (10.7) - -
Y111A 427 (71.5) 535 (10.3)/ 586.5 (8.07) - -
559 (13.8)
Table 4.1: Spectral Features of Ferric and Ferrous Wild Type and Y111A KatG.
109
this CN? binding event is irreversible. Performing an equilibrium titration with cyanide
revealed only one KD around 10 ?M, consistent with the stopped- ow data.
4.3.3 Magnetic Circular Dichroism
In order to more efiectively difierentiate the high and low-spin contributions to Y111A
KatG spectra, we employed magnetic circular dichroism. As noted in chapter 1, this tech-
nique is preferable to UV-vis absorption in that the difierences between ferrous high-spin
and ferrous low-spin are signiflcantly more pronounced. Empirically, low-spin ferrous heme
is easily distinguished by a very strong A-term corresponding to the Q-band (i.e., the fi-
band). This signal is not present in high-spin ferrous heme. The B-bands (from Soret
absorption) are also typically distinct for the high- and low-spin states. Examples of low-
and high-spin ferrous heme spectra can be found in [163{166] as well as in reviews [55, 167].
The spectrum of the ferrous Y111A KatG (Figure 4.4) contained obvious low-spin
features, but not exclusively. The A-term centered at 557 nm was clearly present, but it
was accompanied by other features in the 575 to 625 nm range consistent with high-spin
ferrous heme. Furthermore, the B-band more closely resembled that of high-spin ferrous
heme than low-spin, albeit with less intensity. Similar to the UV-vis spectra, these features
were unchanged after formation of the crosslink through reaction with 100 equivalents of
hydrogen peroxide.
We were able to correlate this spectrum with the cyanide-binding data from our
stopped- ow investigation by generating a simulated Y111A KatG MCD spectrum using
ferrous wtKatG and ferrous KatGN as endpoints on a continuum of spin-state populations
observed in KatG, the wtKatG being dominated by high-spin species and KatGN repre-
senting the exclusively hexacoordinate low-spin complex (Figure 4.4A). By calculating the
weighted average of the two using the amplitude fractions from the stopped- ow data, a
simulated spectrum was generated:
calculated = 0:64?(wtKatG spectrum)+0:36?(KatGN spectrum) (4.2)
110
Figure 4.3: Stopped- ow Cyanide Binding Shows Two Distinct Species in Y111A
KatG. (A) kon = 3.5 ? 105 M?1 s?1. (B) kon = 1.4 ? 104 M?1 s?1. (C) The fractions of
amplitude of fast (?) and slow (N) phases are independent of cyanide concentration. Fast
phase: 64.1 ? 0.7%. Slow phase: 35.9 ? 0.7%.
111
Figure 4.4: Prediction of Y111A KatG Ferrous Heme MCD Spectrum. (A) The
calculated Y111A KatG spectrum ( ) overlaid with the reduced wild type spectrum ( )
and the reduced KatGN spectrum (???). (B) Actual Y111A KatG spectrum ( ) overlaid
with the calculated Y111A KatG Spectrum (??).
112
The actual MCD spectrum of ferrous Y111A KatG was virtually indistinguishable from the
calculated (Figure 4.4B). In the absence of the C-terminal domain, His106 coordinates the
heme iron [130]. Likewise, in Y111A the sixth ligand would most likely be that of histidine.
Further, the ferrous low-spin A-term features of the Y111A and KatGN variants are most
similar to those of known bis-histidine coordinated heme systems such cytochrome b5 [163]
and an analogous ascorbate peroxidase variant [164].
4.3.4 Electron Paramagnetic Resonance
Although MCD verifled the ratio of high-spin to low-spin species predicted by the
stopped- ow, in order to obtain spectral evidence of both species in the native FeIII state
we employed EPR spectroscopy (Figure 4.5). The most prominent feature is the high-spin
axial signal (g? = 5.97, gk = 1.99) with rhombic contributions (gx;z = 6.65, 1.95; gy is
obscured by the axial signal at 5.97). These features are nearly identical to wtKatG. In
contrast to wtKatG, low-spin signals are also easily detectable (gz;y;x = 2.92, 2.27, 1.53).
This latter set of signals is indistinguishable from KatGN [46].
4.3.5 Steady-state Kinetics
Substitution of Y111 afiects catalase and peroxidase activity difierently. We observed
a flve-fold reduction in the apparent kcat for catalase activity and an eight-fold decrease in
the apparent second-order rate constant (Figure 4.6 and Table 4.2). For peroxidase activity,
the H2O2- and ABTS-dependent peroxidatic apparent kcat were reduced by 66% and 40%
percent respectively, however, this did not translate into any signiflcant change in apparent
second-order rate constant due to a reduction in apparent KM for both substrates (Figure
4.7 and Table 4.3). The oxidation rate of ABTS is least likely to be afiected by subtle
changes in the active site, and therefore a 40% loss of activity is most likely a result of 36%
of the Y111A protein containing low-spin heme. It is possible that the decrease in catalase
activity is due to a decrease in the rate of compound I formation or a decrease in ability to
form the Met-Tyr-Trp covalent cross-link necessary for catalase activity. A decrease in the
113
Figure 4.5: EPR Spectrum of Y111A KatG Compared to wtKatG. Relative inten-
sity magnifled by 4 at fleld strength greater than 1800 Gauss. g-values report Y111A KatG
signals.
114
rate of compound I formation may or may not afiect the peroxidase kinetics, but a lesser
ability to form the covalent crosslink would have no efiect as KatG is still peroxidase active
even without the ability to form the covalent cross-link [126].
In chapter 2, we observed that ABTS can act as an inhibitor to the peroxidase reaction
in wtKatG. ABTS a?nity was assessed based on the apparent Michaelis constant and the
lowest concentration of ABTS at which inhibition was evident. ABTS a?nity was shown
to increase with decreasing pH. ABTS inhibition was also observed in the Y111A variant
(Figure 4.7B). The a?nity of Y111A KatG for ABTS at pH 5.0, however, was most similar
to that of wtKatG at pH 4.25 (KM = 0.06, [ABTS] required for evidence of inhibition =
0.6 mM). It was also shown that wtKatG is much more active at pH 4.25 (and below) than
it is at pH 5.0 - suggesting a shift in Y111A to a more peroxidase-like enzyme.
4.4 Discussion
Catalase-peroxidase lacking its C-terminal domain (KatGN) contains a collapsed ac-
tive site due to the energetically favorable coordination of histidine 106 to the heme iron
[46, 130]. Upon introduction of the C-terminal domain as a separately expressed and iso-
lated protein, however, the active site reorganizes into its functional form [46]. It is thus
proposed that despite its distance from the active-site, the C-terminal domain serves to
support an architectural framework necessary for proper active-site conflguration by pro-
viding a series of interactions on the BC-loop that together are more favorable than the
Fe-His coordination. Because of its strict conservation and its position near the interdo-
main interface, Y111 may have an important role in C-terminal domain-dependent support
of the active site. Another consideration is that the C-terminal domain is not found in the
monofunctional peroxidases and may have a role in imparting bifunctionality to KatG. The
behavior of our Y111 variant supports both hypotheses.
Substitution of Y111 with an alanine resulted in enzyme preparations that contained a
mixture of two heme spin states. All three spectroscopic techniques used (UV-visible absorp-
tion, MCD, and EPR spectroscopy) contained features seen in both wtKatG and KatGN.
115
Protein Catalase cycle parameters
kcat (s?1) KM (mM H2O2) kcat=KM (M?1 s?1)
wtKatG 11000 ? 200 3.5 ? 0.2 3.2 ? 106
Y111A 2140 ? 50 5.2 ? 0.3 4.1 ? 105
Table 4.2: Apparent Catalase Kinetic Parameters of Wild Type and Y111A
KatG. Assays included 20 nM wtKatG or 70 nM Y111A KatG, 100 mM phosphate bufier,
pH 7.0, 23 ?C
Protein Peroxide-dependent parameters
kcat (s?1) KM (mM H2O2) kcat=KM (M?1 s?1) KN (mM H2O2)
wtKatG 76 ? 9 0.22 ? 0.04 4.5 ? 105 2.9 ? 1.1
Y111A 24.1 ? 0.8 0.081 ? 0.007 3.0 ? 105 8.3 ? 2.2
ABTS-dependent parameters
kcat (s?1) KM (mM ABTS) kcat=KM (M?1 s?1) KN (mM ABTS)
wtKatG 55.2 ? 1.3 0.087 ? 0.008 6.34 ? 105 not detected
Y111A 33.3 ? 0.9 0.061 ? 0.004 5.46 ? 105 1.6 ? 0.1
Table 4.3: Apparent Peroxidase Kinetic Parameters of Wild Type and Y111A
KatG. Assays included 20 nM wtKatG or 70 nM Y111A KatG, 50 mM acetate bufier, pH
5.0, 23 ?C
116
Figure 4.6: Y111A KatG Catalase Activity. Initial rates of catalase activity of wild
type (H) and Y111A (?) KatG as a function of H2O2 concentration.
117
Figure 4.7: Y111A KatG Peroxidase Activity. (A) Initial rates of peroxidase activity
of wild type (?) and Y111A (N) KatG as a function of H2O2 concentration ([ABTS] = 0.05
mM). (B) Initial rates of peroxidase activity of wild type (?) and Y111A (N) KatG as a
function of ABTS concentration ([H2O2] = 0.5 mM).
118
We established that the mixture was approximately 2:1 high-spin to low-spin using stopped-
 ow cyanide binding, and reduction (as demonstrated by MCD) did nothing to change that
proportion. Furthermore, the accurate prediction of the MCD spectrum emphasizes that
the high- and low-spin species present in the Y111A variant contain active-sites resembling
wtKatG and KatGN, respectively. The transition of one-third of the enzyme to a low-spin
state indicates that Y111 is crucial in the interdomain interactions, and loss of this residue
results in an increased tendency toward Fe-His coordination (KatGN-type active site). Such
a structural modiflcation predicts diminished catalase and peroxidase activity, an efiect we
also observed. Indeed, the 40% decrease in ABTS-dependent peroxidase turnover re ects
closely the mixture of the spin states. Catalase activity, however, was more substantially
decreased than peroxidase activity, and the behavior of the variant towards the reducing
substrate, ABTS, in regards to binding and inhibition at pH 5.0 was more akin to wtKatG
at its more active pH 4.25 than wild-type at pH 5.0. Both suggest that Y111 and perhaps
the interaction of the C-terminal domain with Y111 play at least an indirect role (such
as through crosslink formation ability or some other mechanism) in tuning the enzyme for
bifunctionality.
119
Chapter 5
Comprehensive Analysis of Interdomain Interface Single Variants
5.1 Introduction
The results from chapter 4 provide a good framework for continuing on with variants of
the other interdomain interface residues. The Y111A variant existed in a mixed population.
The majority (64%) retained wild-type active site conflguration, whereas the remaining
resembled enzyme lacking the C-terminal domain (KatGN). Also, catalase activity was
afiected to a degree that could not be explained simply by accounting for a population with
only 64% active species. Although these results were substantial considering the distance
of the variation from the active site, this variation did not entirely replicate the efiects
observed by deleting the C-terminal domain.
To follow up on this, we have additionally created alanine variants of the other four
residues on the interdomain interface (R117A, R479A, D482A, and D597A) and evaluated
all variants spectroscopically and kinetically for comparison to wild type KatG (wtKatG)
and KatGN. We will refer to sets of variants based on two criteria: the distance of the
hydrogen bonding network from the active site (Y111A, R479A, and D482A are the near-
network variants; R117A and D597A are the distant-network variants), or the domain on
which the residue exists (Y111A and R117A are the N-terminal variants; R479A, D482A and
D597A are the C-terminal variants). All single variants expressed as soluble proteins and all
had catalase and peroxidase activity, distinguishing them immediately from KatGN. The
complexity of the hydrogen bonding networks is re ected in the observation that disrupting
either side of the amino acid pairs (Y111-D482 and R117-D597) produces difierent results.
While all variants exhibited some spectroscopic and kinetic disruption compared to wtKatG,
120
the efiects of the near-network variants were more substantial and even comparable to
certain active site variants.
5.2 Materials and Methods
5.2.1 Materials
Hydrogen peroxide (30%), imidazole, hemin, ampicillin, chloramphenicol, sodium di-
thionite, phenylmethylsulfonyl  uoride (PMSF) and 2,2?-azino-bis(3-ethylbenzthiazoline-6-
sulfonic acid (ABTS) were purchased from Sigma (St. Louis, MO). Isopropyl-fl-D-thiogalac-
topyranoside (IPTG), urea, mono- and di-basic sodium phosphate, acetic acid, and sodium
acetate were obtained from Fisher (Pittsburgh, PA). Bugbuster and benzonase were pur-
chased from Novagen (Madison, WI). All restriction enzymes were purchased from New
England Biolabs (Beverly, MA). All oligonucleotide primers were purchased from Invitro-
gen (Carlsbad, CA). AllE. coli strains (BL-21 [DE3] pLysS and XL-1 Blue), Pfu polymerase,
and T4 DNA ligase were obtained from Stratagene (La Jolla, CA). Nickel-nitrilotriacetic
acid (Ni-NTA) resin was purchased from Qiagen (Valencia, CA). Desalting chromatography
columns were purchased from Bio-Rad (Hercules, CA). All bufiers and media were prepared
using water purifled through Millipore QPakII system (18.2 M?/cm, resistivity).
5.2.2 Cloning of R117A, R479A, D482A, and D597A
The preparation of the plasmid pKatG(Y111A) was described in chapter 4. The plas-
mids pKatG(R117A) and pKatG(D597A) were prepared using mutagenic primers [5?-ACT
GCT GAT CGC GAA AGC ACA GC-3? (R117A coding), 5?-GCT GTG CTT TCG CGA
TCA GCA GT-3? (R117A non-coding), 5?-GTT CAA TCG ATG GCG CCG GTG GCG
CGG G-3? (D597A coding), and 5?-CCC GCG CCA CCG GCG CCA TCG ATT GAA
C-5? (D597A non-coding)] for pKatG according to the QuickChange procedure (Strata-
gene, La Jolla, CA). Ampliflcation products were used to transform E. coli XL-1 Blue by
electroporation (Bio-Rad Micropulser, Hercules, CA). The plasmids pKatG(R479A) and
121
pKatG(D482A) were prepared using mutagenic primers [5?-Phos-GCG GCG ACA AAC
GCG GTG G-3? (R479A coding), 5?-Phos-CGG CGA AGG TAG AAG CAG ATG CCC-
3? (R479A non-coding), 5?-Phos-CGC GGT GGT GCC AAC G-3? (D482A coding), and
5?-Phos-TTT GGC GCC ACC ACG GAA GGT AG (D482A non-coding)] for pKatG ac-
cording to the ?Round-the-Horn procedure. ?Round-the-Horn difiers from the QuickChange
procedure in that the mutagenic primers are not complementary. Instead, if the 5? end of
the forward primer is at position \x" in the sequence, the 5? end of the reverse primer is
at position \x-1". As a result, the ampliflcation products must undergo blunt-end ligation
with T4 DNA ligase, after which they were used to transform E. coli XL-1 Blue by heat
shock. Plasmids from candidate colonies were evaluated by diagnostic restriction digest and
DNA sequence analysis (Davis Sequencing, Davis, CA). Correctly mutated plasmids were
used to transform E. coli (BL-21 [DE3] pLysS) expression hosts.
5.2.3 Expression and Puriflcation
Expression and puriflcation of all variants was carried out identically to Y111A as
described in chapter 4. Concentration of purifled enzyme was estimated according to the
method of Gill and von Hippel [50] ["280(Y111A) = 1.43 ? 105 M?1 cm?1, "280(R117A,
R479A, D482A, and D597A) = 1.44 ? 105 M?1 cm?1].
5.2.4 UV-visible Absorption Spectra and Activity Assays
Enzyme was reconstituted with 0.9 equivalents of hemin. Reconstituted enzyme so-
lution incubated at 4 ?C for 72 h to allow unincorporated heme to settle out of solution.
The solution was then centrifuged and the precipitated heme and other insoluble mate-
rial discarded. Concentration of reconstituted enzyme was determined using the pyridine
hemichrome assay [161]. Protein containing heme in the ferrous state was prepared by
adding a small amount (<10 mg) of sodium dithionite to the native enzyme. All spectra
were obtained at room temperature on a Shimadzu UV-1601 spectrophotometer (Columbia,
MD) with a cell pathlength of 1.0 cm.
122
Catalase and peroxidase assays were performed as described in chapter 2. Initial ve-
locity fltting was performed as described in chapter 4.
5.2.5 Circular and Magnetic Circular Dichroism
Circular dichroism spectra were obtained in 5 mM phosphate bufier, pH 7.0. Spectra
were recorded at 23 ?C in a quartz cell (0.5 mm path length) from 300-180 nm on a Jasco
J-810 spectropolarimeter (Tokyo, Japan). Baselining and analysis were done using Jasco
J-720 software.
Magnetic circular dichroism spectra were obtained in 50 mM phosphate 50 mM NaCl
bufier, pH 7.0. Spectra were recorded at 23 ?C in a quartz cell (5.0 mm path length) with 1.4
Tesla magnetic cell holder from 700-350 nm on a Jasco J-810 spectrapolarimeter. Baselining
and analysis were done using Jasco J-720 software. Enzyme containing ferrous heme was
prepared by adding a small amount (<10 mg) of sodium dithionite to native enzyme.
5.2.6 Electron Paramagnetic Resonance Spectroscopy and Spin Quantiflcation
Spectra were recorded using a Bruker EMX instrument equipped with an Oxford ESR
900 cryostat and ITC temperature controller. Additional sample concentration was per-
formed using Amicon Ultra-4 centrifugal devices. The settings for the spectrometer were
as follows: temperature, 10 K; microwave frequency, 9.38 GHz; microwave power, 0.1 mW;
modulation amplitude, 10 G; modulation frequency, 100 kHz; time constant, 655.36 ms;
conversion time, 655.36 ms; and receiver gain, 1.0 ? 105. Spin quantiflcation was carried
out using the Biomolecular EPR Spectroscopy Software package available online [168].
5.3 Results
5.3.1 Mutagenesis, Expression, and Puriflcation of KatG Interdomain Variants
We successfully cloned single alanine variants of all flve interdomain interface residues
Y111, R117, R479, D482, and D597 as conflrmed by sequencing of the plasmid isolates. Each
123
of the variants over-expressed in a soluble form and had detectable catalase and peroxidase
activity. There was no observable change to secondary structural content based on the
circular dichroism of the variants when compared to wild type (Figure 5.1).
5.3.2 UV-visible Spectroscopy
Spectra of the variants in the resting ferric state show only slight difierences from the
wild type enzyme (Figure 5.2). The red-shifting of the Soret and charge transfer 2 bands
in the near-network variants suggest the presence of low-spin heme. R117A exhibited the
largest shoulder at 380 nm consistent with a more penta-coordinate (5-c) heme environment.
The signal at 380 nm is most concave in the D482A and D597A variants, suggesting a more
hexa-coordinate (6-c) system in these variants.
Difierences among the variants were more pronounced in the ferrous form (Figure 5.3).
In wtKatG, the primary absorption bands are located at 439 nm (Soret), 559 nm (fl), and
589 nm (fi), characteristic of high-spin ferrous heme. KatGN absorption bands occur at 427,
530 and near 560 nm, characteristic of low-spin ferrous heme. The near-network variants
contained all features found in both wtKatG and KatGN. The near-network variants all
had blue-shifted Soret bands (realtive to wtKatG) with shoulders at 439, indicative of a
mixture of high spin (439 nm) and low spin (? 425 nm) species. Evidence of shoulders at
both 530 and 589 also conflrmed the presence of a mixed spin-state population in all near-
network variants. Additionally, the R479A and D482A variants were less stable and prone
to precipitation in the ferrous state, resulting in more light-scattering evident in the spectra
and broadening of the signals. This can be particularly well seen in the less pronounced
absorption bands in the R479A variant in relation to Y111A, for example. Centrifugation
resulted in samples that produced purely high-spin spectra (data not shown). The distant-
network variants more closely resembled the wild type, although the D597A Soret band was
slightly blue-shifted and the spectrum was less concave around 535 nm.
124
Figure 5.1: Far-UV Circular Dichroism of wtKatG and Interdomain Interface
Variants.
125
Figure 5.2: UV-vis Spectra of wtKatG, KatGN, and Interdomain Interface Vari-
ants: Ferric Heme.
126
Figure 5.3: UV-vis Spectra of wtKatG, KatGN, and Interdomain Interface Vari-
ants: Ferrous Heme.
127
Figure 5.4: MCD Spectra of KatGN, wtKatG, and Interdomain Interface Vari-
ants: Ferrous Heme.
128
Figure 5.5: MCD Spectrum of R479A KatG Ferric Heme Compared to wtKatG
and KatGN. R479A (??), wtKatG ( ), and KatGN ( ).
129
5.3.3 Magnetic Circular Dichroism
Distinction between high- and low-spin heme species in magnetic circular dichroism
(MCD) is more facile in the ferrous heme state than ferric. Magnetic circular dichroism
on the ferrous forms of the enzyme corroborated the evidence of low-spin heme observed
in the ferrous UV-vis in Y111A and D482A, but R479A did not have any more low-spin
contribution than the wild type or the other high-spin variants (Figure 5.4). The distant-
network variants also appeared to be purely high-spin, like the wild type. Although R479A
appeared high-spin after reduction with dithionite, the ferric MCD spectrum revealed a clear
low-spin presence similar to KatGN and unlike the wild type (Figure 5.5). The discrepancy
between the ferric and ferrous MCD spectra is consistent with the precipitation of low-spin
species upon addition of dithionite observed during the UV-vis studies.
5.3.4 Electron Paramagnetic Resonance
EPR spectra conflrm the presence of low-spin heme iron in the ferric state of the near-
network variants and the absence of low-spin in the distant-network variants (Figure 5.6).
To quantify the relative proportion of each species, we simulated the spectra and integrated
the individual signals. The best simulations required three high-spin signals and one low-
spin. The high spin signals were a rhombic signal (RHS) with g-values approximately
6.64, 4.95, and 1.955, an axial signal (AHS1) with g-values approximately 5.95 and 1.995,
and a second axial signal (AHS2) with g-values approximately 5.65 and 1.995. The two
axial signals could also be treated as one rhombic signal, but not without less satisfactory
simulations. The low-spin signal was rhombic (RLS) with approximate g-values of 2.93, 2.3,
and 1.53. The amount of low-spin signal present in the distant-network variants was small
enough that simulations could not be improved by adding in the RLS signal. The RLS in
the near-network variants accounted for approximately twenty percent of the signal, two to
three times that of wild type (Table 5.1). Y111A, as an example of a spectrum that includes
substantial contribution of all four signals, is shown with its simulated spectrum overlain
in Figure 5.7. Although all of the interface variants had more rhombic contribution to the
130
Figure 5.6: EPR Spectra of wtKatG, KatGN, and Interdomain Interface Variants.
All spectra, except KatGN, magnifled by four above 2000 Gauss. R117A and D597A spectra
have had copper impurity signal subtracted.
131
Figure 5.7: Simulation of Y111A KatG Spectrum. Y111A KatG( ) and simulated
( ) spectra magnifled by four above 2000 Gauss.
Species HS : LS RHS : AHS AHS1 : AHS2 AHS1:AHS2:RHS:RLS
wt 91 : 9 60 : 40 51 : 49 28 : 27 : 36 : 9
KatGN 2 : 98 74 : 26 100 : 0 1 : 0 : 1 : 98
Y111A 80 : 20 50 : 50 55 : 45 22 : 18 : 40 : 20
R117A 100 : 0 73 : 27 52 : 48 14 : 13 : 73 : 0
R479A 78 : 22 75 : 25 49 : 51 10 : 10 : 58 : 22
D482A 84 : 16 53 : 47 54 : 46 21 : 19 : 44 : 16
D597A 100 : 0 40 : 60 55 : 45 33 : 27 : 40 : 0
Table 5.1: Ratios of Various EPR Signals Observed in wtKatG and Variants.
Expressed as percentages.
132
overall signal than the wild type, only R117A and R479A had a greater RHS to AHS ratio
than wild type. This corroborates the more prominent shoulder at 380 nm in UV-vis ferric
spectra of these two variants.
5.3.5 Steady-state Kinetics
Substituting any of the residues along the interdomain interface resulted in a decrease in
the apparent kcat for catalase activity with the exception of D597, the residue most distant
from the active site (Table 5.2). The apparent Michaelis constant was largely unafiected in
the C-terminal residue variants (R479A, D482A, D597A), but was slightly elevated in the
N-terminal residue variants (Y111A, R117A). Apparent peroxidase parameters are shown
in Table 5.3. With the exception of the R117A ABTS-dependent kcat, all variants showed
a decrease in kcat and kcat/KM to varying degrees.
5.4 Discussion
The non-heme binding C-terminal domain of KatG is crucial for proper active site
conflguration of the enzyme. Without it, the active site is collapsed due to the energetically
favorable coordination of histidine 106 to the heme iron, but can be restored with the
introduction of the separately expressed and isolated C-terminal domain [46, 130]. This
suggests that the C-terminal domain is providing an architectural framework necessary for
proper active site conflguration. The strict conservation of the interdomain interacting
residues suggests that the interactions are speciflc and crucial to the ability of the C-
terminal domain to act as this framework. Therefore, disruption of these interdomain
contacts ofiers an opportunity to understand the nature of how the C-terminal domain
provides this support from such a great distance from the active site.
KatGN is expressed in inclusion bodies and must be refolded before becoming soluble,
signifying that the C-terminal domain is essential for structural stability. The single alanine
variants of all of the interdomain interface residues were all soluble as expressed. This
indicates that no single interaction is completely responsible for the structural stability
133
Protein Catalase cycle parameters
kcat (s?1) KM (mM H2O2) kcat=KM (106 M?1 s?1)
wtKatG 11000 ? 200 3.5 ? 0.2 3.2
Y111A 2140 ? 50 5.2 ? 0.3 0.41
R117A 3820 ? 230 4.2 ? 0.7 0.91
R479A 4370 ? 110 3.4 ? 0.3 1.3
D482A 5840 ? 120 3.7 ? 0.2 1.6
D597A 10000 ? 400 3.2 ? 0.4 3.1
Table 5.2: Apparent Catalase Kinetic Parameters of wtKatG and Interdomain
Interface Variants. Assays included 20 nM wtKatG, R117A, D597A, 50 nM R479A,
D482A, or 70 nM Y111A KatG, 100 mM phosphate bufier, pH 7.0, 23 ?C
134
Protein Peroxide-dependent peroxidase cycle parameters
kcat (s?1) KM (mM H2O2) kcat=KM (105 M?1 s?1) KN (mm H2O2)
wtKatG 76 ? 9 0.22 ? 0.04 4.5 2.9 ? 1.1
Y111A 24.1 ? 0.8 0.081 ? 0.007 3.0 8.3 ? 2.2
R117A 30 ? 5 0.14 ? 0.05 2.1 5.8 ? 5.0
R479A 10.9 ? 0.4 0.19 ? 0.01 0.57 3.1 ? 0.3
D482A 28.4 ? 4 0.57 ? 0.13 0.49 2.1 ? 0.6
D597A 32 ? 5 0.8 ? 0.2 0.40 not detected
ABTS-dependent peroxidase cycle parameters
kcat (s?1) KM (mM ABTS) kcat=KM (105 M?1 s?1) KN (mm ABTS)
wtKatG 55.2 ? 1.3 0.087 ? 0.008 6.3 not detected
Y111A 33.3 ? 0.9 0.061 ? 0.004 5.5 1.6 ? 0.1
R117A 76 ? 4 0.13 ? 0.01 5.8 6.8 ? 3.0
R479A 8.7 ? 0.2 0.025 ? 0.001 3.5 0.92 ? 0.05
D482A 20.7 ? 1.0 0.046 ? 0.004 4.5 0.39 ? 0.04
D597A 33 ? 3 0.11 ? 0.04 3.0 not detected
Table 5.3: Apparent Peroxidase Kinetic Parameters of wtKatG and Interdomain
Interface Variants. Assays included 20 nM wtKatG, R117A, D597A, 50 nM R479A,
D482A, or 70 nM Y111A KatG, 50 mM acetate bufier, pH 5.0, 23 ?C
135
that the C-terminal domain imparts to the enzyme. The tendency to precipitate upon
addition of dithionite, however, suggests that some of the stability has been lost in the
R479A and D482A variants.
The fact that no two variants had identical properties, despite four of the residues
acting as single contact bridges (Y111-D482 and R117-D597), was surprising at flrst. The
extensiveness of the hydrogen bonding networks, however, provides some explanation. In
the available structures, Y111 is only within hydrogen bonding distance of D482. Similarly,
D597 is only within hydrogen bonding distance of R117. However, D482 also is within
hydrogen bonding distance with R479 and R484, which are also within hydrogen bonding
distance of each other, and R117 is within hydrogen bonding distance of D115. This does
not even consider the intradomain hydrogen bonds between these mentioned residues and
backbone carbonyls. As such, it is impossible to make a single variation without possibly
afiecting multiple interdomain contacts simultaneously, and each residue must primarily
be considered based on what it individually contributes to the whole network. In fact, it
could be considered more surprising that the kinetic and spectroscopic characteristics of
the variants can be grouped based on the proximity to the active site or on the domain on
which they reside.
The distant-network variants remained spectroscopically similar to the wild type with
the exception of rhombic contribution to the high spin signal observed in EPR (R117A
appearing more rhombic, D597A less rhombic). The near-network variants all demonstrated
a mixture of high- and low-spin heme in all spectroscopic evaluations, whereas the distant-
network variants exhibited purely high-spin heme. This suggests that the nearer hydrogen
bonding network is more crucial to the prevention of active site collapse, while the more
distant hydrogen bonding network plays more of a flne-tuning role.
Concomitant with the shift to a greater relative population of low spin species in
the near-network variants is a loss in overall activity due to inactive species. This makes
direct kinetic comparison to wild-type di?cult. To determine if the residue has a role in
more than just active site conflguration, we looked at the ratio of catalase to peroxidase
136
maximum activity (Table 5.4). Interestingly, none of the variants had the same ratio as the
wildtype. The N-terminal residue variantsreducedcatalase activitymore substantiallythan
peroxidase activity. Most KatG variants show the same behavior [117, 125, 128, 169]. This is
expected, considering that the overall structure of KatG is more similar to a monofunctional
peroxidase, and if a change was made that disrupted one activity more than another, the
novel activity should be more afiected than the original. Surprisingly, the substitution of
the C-terminal residues reduced peroxidase activity more than catalase activity. The only
other instance of a KatG variant selectively and substantially reducing peroxidase more
than catalase activity was W321F in M. tuberculosis KatG, however speciflc activity of the
W321F was measured for catalase and peroxidase at pH 7.2, well outside of the reasonable
range for peroxidase activity measurements [170].
The transition of the population to mixed spin states and the loss of between 50 and
75% activity seen in most of these variants is quite substantial given their distance from the
active site. In fact, these changes are comparable to variations made to the many active site
residues in Synechocystis PCC 6803 [147, 171]. The Soret and CT2 band shifts in the ferric
heme UV-vis spectra of H123E and R119A (H106 and R102 in E. coli numbering) in the
distal cavity and D402N and D402E in the proximal triad (Trp341 - D402 - His290, SynKatG
numbering) were similar to what is seen here in the near-network interface residues variants.
The ferrous UV-vis spectra of R119N, R119A, H123E, and H123Q were also very similar
to the ferrous spectra of the near-network interface variants having Soret peaks around 425
nm with a shoulder around 439 nm and three bands evident between 525 and 590 nm.
With the exception of the proximal histidine, variants of the proximal triad retained at
least 25% peroxidase activity, while R479A here lost 85% activity [147]. These comparisons
demonstrate that the roles of the speciflc interactions along the interdomain interface are as
substantial in proper active site conflguration and functionality as many of the active site
residues themselves, giving the C-terminal domain the same signiflcance to KatG function
as the proximal triad and distal catalytic residues.
137
Protein ratio relative to wtKatG
wtKatG 167.7 1
Y111A 74.6 0.445
R117A 71.7 0.428
R479A 445.9 2.659
D482A 239.8 1.43
D597A 307.7 1.835
Table 5.4: Ratio of Catalase to Peroxidase Activity Relative to Wild Type. Ratio
is catalase kcat divided by the average of peroxide- and ABTS-dependent kcat.
138
Chapter 6
Summary
The complicating features of the catalase-peroxidase system are numerous. With a
single active site, the enzyme is able to catalyze at least two difierent reaction cycles. This
active site is indistinguishable from the active site in monofunctional peroxidases which can
catalyze only one reaction. Furthermore, the KatG active site bears little resemblance to
catalase-active enzymes. Therefore, catalase catalytic capability is imparted to the enzyme
from some non-active site structure or collection of structures, requiring an expansion from
the typical active site-limited scrutiny.
The next complicating factor is that the catalytic mechanisms are not strictly Michaelis-
Menten type reactions as they require multiple substrates. Each cycle on its own can easily
be handled, as the kinetics have been worked out for the monofunctional equivalents in the
past. Putting the two activities inside the same enzyme greatly complicates the kinetics,
particularly with the two activities sharing the same flrst step. KatG seems to ofier some
relief in this in that pH and the availability of a peroxidatic reducing substrate can be
used to difierentiate between the two reaction cycles. The appropriateness of this approach,
however is debatable. Is it safe to assume that the catalase cycle behaves the same when
no reducing substrate is present? Are the two activities competitive? Have the pH efiects
been su?ciently characterized to know which pH is best to study each activity and why?
These were questions that were overlooked during the attempts to simplify this system to
a manageable level of complexity.
139
6.1 Assumption: pH-proflling at Saturating Substrate Concentrations
Instead, pH-proflling was carried out disregarding the kinetic complexity of the enzyme.
Using saturating substrate concentrations, KatG was found to have maximal peroxidase ac-
tivity between pH 4.5 and 5.0, maximal catalase activity near pH 6.5, and maximal NADH
oxidase activity at pH 8.5. This information was then used to interpret structural infor-
mation, such as an arginine that is oriented towards the hydrogen bonding near-network
residues on the C-terminal domain at pH 4.5, oriented towards the tyrosine involved in the
covalent crosslink at pH 8.5, and is split between the two orientations at pH 6.5. The con-
clusion was that this residue is responsible for \switching on" peroxidase activity. Knowing
that a full-scale kinetic solution including protonation efiects had not been carried out for
KatG, we took this task on and found that these conclusions could not be the whole story
since peroxidase is still not completely \turned on" at pH 4.5.
What we found was that the approach used to generate the pH-proflles overlooked a
signiflcant feature: ABTS-dependent inhibition. This led to highly skewed pH-proflles for
peroxidase activity, with what was claimed to be maximum activity actually being inhibited
catalysis. We found that by using sub-saturating conditions at pH 3.75, velocities could
be achieved in great excess of those found using any substrate concentration at pH 4.5 or
higher. In fact, it appeared that activity would be even much higher at lower pHs if not
for the loss of structural stability and unfolding of the enzyme at pH 3.6. Although precise
pKas could not be assigned for the peroxidase cycle, 4.5 could easily be established as
an upper-limit. This would suggest that the arginine-switch either is less signiflcant in the
difierentiation between activities, or is only one of multiple pH-dependent features (possibly
independent or in cascade) involved in difierentiation. The trends in a?nity towards the
reducing substrate and H2O2 suggest that difierentiation may be closely linked to changes
in binding.
Similarly, by using the mechanistic solution for pH-dependent catalase kinetics, we
found it to be more complicated than just a simple 6.5 pH optimum. It revealed that even
140
though all single concentration pH-proflles had an optimum of 6.5, the actual optimum
for maximum turnover was pH 5.75 or lower, and the optimum for e?ciency was pH 7.0.
Again, this information will be valuable in evaluating pH-based structural and spectroscopic
changes.
6.2 Assumption: Catalysis Can Be Difierentiated by pH and Substrate Avail-
ability
Simply from the classical representation of the catalase-peroxidase mechanistic scheme
(Figure 1.13), it is obvious that the two catalytic cycles were assumed to be competitive.
Compound I could react with either H2O2 or a reducing substrate. The simple way of
studying the two cycles separately would be to only provide H2O2 to study catalase, and
do so at a pH where catalase is optimal. Similarly, to study peroxidase shift to a pH
where catalase is no longer optimal and provide the substrates necessary for peroxidase.
As a result of the pH-optima being misidentifled, peroxidase activity has been regularly
evaluated between pH 4.5 and 5.0. Catalase activity, however, is still easily measurable
down to pH 5.0 and peroxidase (as just discussed) is not very active until below pH 4.5.
Furthermore, based on this more exhaustive analysis of pH-dependence, pH 4.5 to 5.5 seems
to be a catalytic pH gap for KatG.
If at pH 5.0 both activities are present, it seemed beneflcial to determine what fraction
of the enzyme was utilizing the catalase cycle and what fraction was utilizing the peroxidase
cycle. This would provide a framework for interpreting past (and future) work as well
as being more biologically relevant in that it would require studying the enzyme while
both cycles are available. What we found was that the presence of peroxidase reducing
substrate did not act competitively with catalase activity, but enhanced it. The result of
this synergistic efiect is that KatG becomes a highly e?cient catalase over an extremely
broad pH range (4.5 to near 8.0), which can be used to suggest that its primary role is
perhaps H2O2 detoxiflcation.
141
The only way this synergy could occur would be if the classical scheme was a misrep-
resentation or oversimpliflcation of the KatG catalytic cycles. After establishing that the
presence of reducing substrate prevented the accumulation of an unreactive species during
catalase turnover, we proposed a mechanism that has difierentiation between catalase and
peroxidase based on electron transfer occurring prior to binding of the second substrate.
Direct evidence for this electron transfer and for part of our proposed mechanism came out
in some of the most recent published literature [143].
6.3 Assumption: Global Features Play Structural Roles, Active Site Features
Play Functional Roles
Overall, structure and function are intimately linked. Function arises from structure.
However, much like the equipment inside a factory determines what it produces and not the
shape of the factory itself, structure-function relationship studies in enzymes are usually
limited to the active site while global features are treated as the building framework. Lim-
iting consideration of functional players to active site features has been well-established to
be a fallacy by KatG studies. Of the KatG activities, non-active site mutations can selec-
tively knock out certain activities: the S315T substitution (and others) in Mycobacterium
tuberculosis knocks out the isonicotinic acid hydrazide activation, and substitutions block-
ing formation of the covalent adduct knock out catalase activity. Absence of the C-terminal
domain completely knocks out activity due to structural instability and active site collapse.
Here, even the C-terminal domain?s role is shown to be possibly functional based on the
interdomain interface variants.
Maintaining the proper fold and active site architecture (established roles of the C-
terminal domain) can still be considered primarily framework in nature. At flrst, this role
was conflrmed by the Y111A variant. This enzyme existed in a mixed population, with 64%
resembling wild-type KatG and 36% resembling KatGN. This ratio was consistent enough
to allow prediction of the MCD ferrous spectra from stopped- ow kinetic data. The steady-
state kinetics, however, revealed that catalase activity was decreased more than what could
142
be accounted for by only 36% having a collapsed active site. This foreshadowed some of
the results found by evaluating all of the other alanine variants of the interdomain interface
residues.
The near-network variants showed a population mixture similar to the Y111A variant.
This was conflrmed with UV-Vis, MCD, and EPR spectroscopy. The ratios of catalase to
peroxidase activity, however, showed that function was afiected in a way that could not
correlate to the population mixtures. Instead, the C-terminal variants demonstrated an
increase in the catalase:peroxidase ratio relative to wild-type, while the N-terminal variants
demonstrated a decrease in that ratio. This actually serves to conflrm that the arginine
switch mentioned earlier in relation to pH-induced structural changes is involved in catal-
ysis difierentiation to some degree. It also serves to indicate that speciflc residues on the
C-terminal domain (R479, D482) are involved in catalysis difierentiation, as when those
residues are modifled peroxidase activity is decreased more substantially than catalase ac-
tivity.
6.4 Conclusion
KatG has proved to be a superb model for evaluating some of the common assumptions
made in enzymology. From attempts to transform enzyme kinetics into pseudo-flrst order
through substrate concentration manipulation to how global features are treated in the
scope of structure-function relationships, we have been able to evaluate the necessity of
thorough parametric analysis and the signiflcance of global protein structure. Even in
these studies, however, not all assumptions could be eliminated. In the pH studies, pH
and one substrate concentration were varied while the other substrate was held constant.
The shear volume of data involved in this collection alone proved to be both tedious and
often elusive due to the time required to collect one data set. To have expanded this to
include another variable would have created a data set that required more points than could
be reasonably or even accurately collected in one sitting with current technology. In the
interdomain interface variants, peroxidase data was still obtained at pH 5.0 in spite of the
143
results demonstrated by the pH studies. With no structural studies and very little literature
involving kinetics carried out below pH 4.5, there would have been no reference point for
the work. Furthermore, performing the same pH-based evaluation of all the variants would
have been nearly as massive an undertaking as adding another variable. In the end, the
complexity of enzymes necessitates simplifying assumptions that can often be detrimental,
or at least misleading, in data interpretation. This emphasizes the need for technological
developments in the realm of data acquisition as it pertains to speed of collection, volume of
reagents, quantity of data, and number of variables. As these advancements are in progress,
it is the responsibility of the researcher (as it has always been) to be aware of the pitfalls
associated with making simpliflcations, eliminate them whenever possible and reasonable,
and recognize how those simpliflcations might afiect the data at hand and the interpretation
of that data.
144
Bibliography
[1] A Fersht, Structure and Mechanism in Protein Science. W. H. Freeman and Company,
New York, 1999, pp. 119-122, 164-215.
[2] PCA Bruijnincx, G van Koten, RJMK Gebbink, Mononuclear non-heme iron enzymes
with the 2-His-1-carboxylate facial triad: recent developments in enzymology and mod-
eling studies. Chem. Soc. Rev. 37 (2008) 2716-2744.
[3] LJ Murray, R Garca-Serres, MS McCormick, R Davydov, SG Naik, SH Kim, BM
Hofiman, BH Huynh, SJ Lippard, Dioxygen activation at non-heme diiron centers:
oxidation of a proximal residue in the I100W variant of toluene/o-xylen monooxygenase
hydroxylase. Biochemistry 46 (2007) 14795-14809.
[4] JM Bollinger, W Jiang, MT Green, C Krebbs, The manganese(IV)/iron(III) cofac-
tor of Chlamydia trachomatis ribonucleotide reductase: structure, assembly, radical
initiation, and evolution. Curr. Opin. Struct. Biol. 18 (2008) 650-657.
[5] AH Saunders, AE Gri?ths, KH Lee, RM Cicchillo, L Tu, JA Stromberg, C Krebs, SJ
Booker, Characterization of quinolate synthases from Escherichia coli, Mycobacterium
tuberculosis, and Pyrococcus horikoshii indicates that [4Fe-4S] clusters are common
cofactors throughout this class of enzymes. Biochemistry 47 (2008) 10999-11012.
[6] IDance, The consequences of an interstitial N atom in the FeMo cofactor of nitrogenase.
Chem. Commun. 3 (2003) 324-325.
[7] A Battersby, Tetrapyrroles: the pigments of life. Nat. Prod. Rep. 17 (2000) 507-526.
[8] GL Cantoni, The nature of the active methyl donor formed enzymatically from L-
methionine and adenosinetriphosphate. J. Am. Chem. Soc 74 (1952) 2942-2943.
145
[9] S Roje, S-Adenosyl-L-methionine: Beyond the universal methyl group donor. Phyto-
chemistry 67 (2006) 1686-1698.
[10] T Urich, TM Bandeiras, SS Leal, R Rachel, T Albrecht, P Zimmerman, C Scholz,
M Teixeira, CM Gomes, A Kletzin, The sulphur oxygenase reductase from Acidianus
ambivalens is a multimeric protein containing a low-potential mononuclear non-haem
iron centre. Biochem. J. 381 (2004) 137-146.
[11] SP de Visser, Elucidating enzyme mechanism and intrinsic chemical properties of
short-lived intermediates in the catalytic cycles of cysteine dioxygenase and taurine/fi-
ketoglutarate dioxygenase. Coord. Chem. Rev. 253 (2009) 754-768.
[12] LE Grove, J Xie, E Yikilmaz, A Karapetyan, AF Miller, TC Brunold, Spectro-
scopic and computation insights into second-sphere amino-acid tuning of substrate
analogue/active-site interactions in iron(III) superoxide dismutase. Inorg. Chem. 47
(2008) 3993-4004.
[13] EN Brown, R Friemann, A Karlsson, JV Parales, MMJ Couture, LD Eltis, S Ra-
maswamy, Determining Rieske cluster reduction potentials. J. Biol. Inorg. Chem. 13
(2008) 1301-1313.
[14] The PubChem Project. http://pubchem.ncbi.nlm.nih.gov/. Accessed June 2009.
[15] SEJ Bowman, KL Bren, The chemistry and biochemistry of heme c: functional bases
for covalent attachment. Nat. Prod. Rep. 25 (2008) 1118-1130.
[16] CM DiCarlo, LB Vitello, JE Erman, Efiect of active site and surface mutations on the
reduction potential of yeast cytochrome c peroxidase and spectroscopic properties of
the oxidized and reduced enzyme. J. Inorg. Biochem. 101 (2007) 603-613.
[17] S Schneider, J Marles-Wright, KH Sharp, M Paoli, Diversity and conservation of inter-
actions for binding heme in b-type heme proteins. Nat. Prod. Rep. 24 (2007) 621-630.
146
[18] BC Finzel, TL Poulos, J Kraut, Crystal structure of cytochrome c peroxidase reflned
at 1.7?A resolution. J. Biol. Chem. 259 (1984) 13027-13036.
[19] JRH Tame, B Vallone, The structures of deoxy human hemoglobin and the mutant Hb
Tyrfi42His at 120 K. Acta Crystallog., Section D: Biol. Crystallog. D56 (2000) 805-811.
[20] I Fita, MG Rossmann, The NADPH binding site on beef liver catalase. Proc. Nat.
Acad. Sci. 82 (1985) 1604-1608.
[21] M Meier, M Janosik, V Kery, JP Kraus, P Burkhard, Structure of human cystathionine
fl-synthase: a unique pyridoxal 5?-phosphate-dependent heme protein. EMBO J. 20
(2001) 3910-3916.
[22] DH Sherman, S Li, LV Yermaliskaya, Y Kim, JA Smith, MR Waterman, LM Po-
dust, The structural basis for substrate anchoring, active site selectivity, and product
formation by P450 PikC from Streptomyces venezuelae. J. Biol. Chem. 281 (2006)
26289-26297.
[23] F Arnesano, L Banci, I Bertini, J Faraone-Mennella, A Rosato, PD Barker, AR Fersht,
The solution structure of oxidized Escherichia coli cytochrome b562. Biochemistry 38
(1999) 8657-8670.
[24] M Paoli, BF Anderson, HM Baker, WT Morgan, A Smith, EN Baker, Crystal structure
of hemopexin reveals a novel high-a?nity heme site formed between two beta-propellor
domains. Nat. Struct. Biol. 6 (1999) 926-931.
[25] M Wirtz, V Oganesyan, X Zhang, J Studer, M Rivera, Modulation of redox potentials
in electron transfer proteins: Efiects of complex formation on the active site microen-
vironment of cytochrome b5. Faraday Discussions 116 (2000) 221-234.
[26] DB Goodin, DE McRee, The Asp-His-iron triad of cytochrome c peroxidase controls
the reduction potential electronic structure, and coupling of the tryptophan free radical
to the heme. Biochemistry 32 (1993) 3313-3324.
147
[27] S Adachi, S Nagano, K Ishimori, Y Watanabe, I Morishima, T Egawa, T Kitagawa, R
Makino, Roles of proximal ligand in heme proteins: replacement of proximal histidine
of human myoglobin with cysteine and tyrosine by site-directed mutagenesis as models
for P-450, chloroperoxidase, and catalase. Biochemistry 32 (1993) 241-252.
[28] J Umbreit, Methemoglobin{it?s not just blue: a concise review. Amer. J. Hematol. 82
(2007) 134-144.
[29] CP Ponting, RR Russell, The natural history of protein domains. Annu. Rev. Biophys.
Biomol. Struct. 31 (2002) 45-71.
[30] RB Russell, Classiflcation of protein folds. Mol. Biotech. 20 (2002) 17-28.
[31] AD Moore, AK Bj?orklund, D Ekman, E Bornberg-Bauer, A Elofsson, Arrangements
in the modular evolution of proteins. Trends Biochem. Sci. 33 (2008) 444-451.
[32] C Chothia, J Gough, Genomic and structural aspects of protein evolution. Biochem.
J. 419 (2009) 15-28.
[33] BC-H Lam, E Blumwald, Domains as functional building blocks of plant proteins.
Trends Plant Sci. 7 (2002) 544-549.
[34] SA Wolfe, L Nekludova, CO Pabo, DNA recognition by Cys2His2 zinc flnger proteins.
Annu. Rev. Biophys. Biomol. Struct. 3 (1999) 183-212.
[35] J Nardelli, TJ Gibson, C Vesque, P Charnay, Base sequence discrimination by zinc-
flnger DNA-binding domains. Nature 349 (1991) 175-178.
[36] KJ Brayer, DJ Segal, Keep your flngers ofi my DNA: protein-protein interactions me-
diated by C2H2 zinc flnger domains. Cell Biochem. Biopyss. 50 (2008) 111-131.
[37] AH Fox, C Liew, M Holmes, K Kowalski, J Mackay, M Crossley, Transcriptional cofac-
tors of the FOG family interact with GATA proteins by means of multiple zinc flngers.
EMBO J. 18 (1999) 2812-2822.
148
[38] SE Brenner, SCOP: Fold: TIM beta/alpha-barrel. http://scop.mrc-lmb.cam.ac.uk/
scop/data/scop.b.d.b.html. Last updated November 2007. Accessed April 2009.
[39] N Kannan, S Selvaraj, MM Gromiha, S Vishveshwara, Clusters in fi/fl barrel proteins:
implications for protein structure, function, and folding: a graph theoretical approach.
PROTEINS: Struct. Funct. Genet. 43 (2001) 103-112.
[40] C Chu, Y Lai, H Huang, Y Sun, Kinetic and structural properties of triosephosphate
isomerase from Heliobacter pylori. PROTEINS: Struct. Funct. Bioinf. 71 (2008) 396-
406.
[41] I Kursula, S Partanen, A Lambeir, RK Wierenga, The importance of the conserved
Arg191{Asp227 saltbridgeoftriosephosphate isomeraseforfolding, stability, andcatal-
ysis. FEBS Lett. 518 (2002) 39-42.
[42] B Schneider, T Knchel, B Darimont, M Hennig, S Deitrich, K Babinger, K Kirschner, R
Sterner, Role of the N-terminal extension of the (flfi)8-barrel enzyme indole-3-glycerol
phosphate synthase for its fold, stability, and catalytic activity. Biochemistry 44 (2005)
16405-16412.
[43] K Suzuki, S Ito, A Shimizu-Ibuka, H Sakai, Crystal structure of pyruvate kinase from
Geobacillus stearothermophilus. J. Biochem. 144 (2008) 305-312.
[44] MS Jurica, A Mesecar, PJ Heath, W Shi, T Nowak, BL Stoddard, The allosteric
regulation of pyruvate kinase by fructose-1,6-bisphosphate. Structure 6 (1998) 195-210.
[45] S An, R Kumar, ED Sheets, SJ Benkovic, Reversible compartmentalization of de novo
purine biosynthetic complexes in living cells. Science 320 (2008) 103-106.
[46] RD Baker, CO Cook, DC Goodwin, Catalase-peroxidase active site restructuring by a
distant and \inactive" domain. Biochemistry 45 (2006) 7113-7121.
[47] CT Walsh, Posttranslational Modiflcations of Proteins. Roberts & Company Publish-
ers, Greenwood Village, 2007.
149
[48] AOW Stretton, The flrst sequence: Fred Sanger and insulin. Genetics 162 (2002) 527-
532.
[49] F Sanger, GM Air, BG Barrell, NL Brown, AR Coulson, CA Fiddes, CA Hutchison,
PM Slocombe, M Smith, Nucleotide sequence of bacteriophage ` X174 DNA. Nature
265 (1977) 687-95.
[50] SC Gill, PH von Hippel, Calculation of protein extinction coe?cients from amino acid
sequence data. Anal. Biochem. 182 (1989) 319-326.
[51] CL Varnado, KM Hertwig, R Thomas, K Roberts, DC Goodwin, Properties of a novel
periplasmic catalase-peroxidase from Escherichia coli O157:H7. Arch. Biochem. Bio-
phys. 421 (2004) 166-174.
[52] WR Hagen, EPR spectroscopy as a probe of metal centres in biological systems. Dalton
Trans. (2006) 4415-4434.
[53] S Basu, Whatever happened to...? Helium shortage ? Cytokine Storm ? Martian dust
threat ? Novel HIV drug. Sci. Amer. 297(4) (2007) 18.
[54] WA Baase, WC Johnson, Circular dichroism and DNA secondary structure. Nucleic
Acids Res. 6 (1979) 797-814.
[55] MR Cheesman, C Greenwood, AJ Thomson, Magnetic Circular Dichroism of Hemo-
proteins. Adv. Inorg. Chem, 36 (1991) 203-255.
[56] RC Bray, Sudden freezing as a technique for the study of rapid reactions. Biochem. J.
81 (1961) 189-193.
[57] R Yalow, Development and proliferation of radioimmunoassay technology. J. Chem.
Ed. 76 (1999) 767-768.
[58] Scheering-Plough Animal Health Corp., Antibiotics, the basics.
http://www.nu or.com/ images/new/Antibiotic Classes.pdf. Last updated June
2008. Accessed March 2009.
150
[59] JD Buynak, Medicinal Outline. http://faculty.smu.edu/jbuynak/Medicinal Outline
11 4 04.pdf. Last updated November 2004. Accessed March 2009.
[60] MA Kohanski, DJ Dwyer, B Hayete, CA Lawrence, JJ Collins, A common mechanism
of cellular death induced by bactericidal antibiotics. Cell 130 (2007) 797-810.
[61] AR Zink, C Sola, U Reischl, W Grabner, N Rastogi, H Wolf, AG Nerlich, Charac-
terization of Mycobacterium tuberculosis complex DNAs from Egyptian mummies by
spoligotyping. J. Clin. Microbiol. 41 (2003) 359-367.
[62] World Health Organization, Tuberculosis Fact Sheet. http://www.who.int/mediacen-
tre/factsheets/fs104/en/index.html. Last updated March 2007. Accessed April 2009.
[63] A Rattan, A Kalia, N Ahmad, Multidrug-resistant Mycobacterium tuberculosis: molec-
ular perspectives. Emerging Infectious Diseases 4 (1998) 195-209.
[64] Y Zhang, B Heym, B Allen, D Young, S Cole, The catalase-peroxidase gene and
isoniazid resistance of Mycobacterium tuberculosis. Nature 358 (1992) 591-593.
[65] B Heym, PM Alzari, N Honor?e, ST Cole, Missense mutations in the catalase-peroxidase
gene, katG, are associated with isoniazid resistance in Mycobacterium tuberculosis. Mol.
Microbiol. 15 (1995) 235-245.
[66] S Chouchane, I Lippai, R Magliozzo, Catalase peroxidase (Mycobacterium tuberculosis)
catalysis and isoniazid activation. Biochemistry 39 (2000) 9975-9983.
[67] B Lei, CJ Wei, SC Tu, Action mechanism of antitubercular isoniazid: Activation by
Mycobacterium tuberculosis KatG, isolation and characterization of INH inhibitor. J.
Biol. Chem. 275 (2000) 2520-2526.
[68] GS Timmins, V Deretic, Mechanisms of action of isoniazid. Mol. Microbiol. 62 (2006)
1220-1227.
151
[69] A Argyrou, MW Vetting, B Aladegbami, JS Blanchard, Mycobacterium tuberculosis
dihydrofolate reductase is a target for isoniazid. Nat. Struct. Mol. Biol. 13 (2006) 408-
413.
[70] T Deemargan, X Carpena, R Singh, B Wiseman, I Fita, PC Loewen, Structural charac-
terization of the Ser324Thr variant of the catalase-peroxidase (KatG) from Burkholde-
ria pseudomallei, J. Mol. Biol. 345 (2005) 21-28.
[71] S Yu, S Girotto, C Lee, RS Magliozzo, Reduced a?nity for isoniazid in the S315T
mutant of Mycobacterium tuberculosis KatG is a key factor in antibiotic resistance. J.
Biol. Chem. 278 (2003) 14769-14775.
[72] B Heym, Y Zhang, S Poulet, D young, S Cole, Characterization of the katG gene en-
coding a catalase-peroxidase required for the isoniazid susceptibility of Mycobacterium
tuberculosis. J. Bacteriol. 175 (1993) 4255-4259.
[73] VH Ng, JS Cox, AO Sousa, JD MacMicking, JD McKinney, Role of KatG catalase-
peroxidase in mycobacterial pathogenesis: countering the phagocyte oxidative burst.
Mol. Microbiol. 52 (2004) 1291-1302.
[74] W Brunder, H Schmidt, H Karch, A novel catalase-peroxidase encoded by the large
plasmid of enterohaemorrhagic Escherichia coli O157:H7. Microbiology 142 (1996)
3305-3315.
[75] AM Junko, MA Sakurako, K Fumiaki, T Tomoko, W Haruo, Identiflcation of a novel
periplasmic catalase-peroxidase KatA of Legionella pneumophila. FEMS Microbiol.
Lett. 176 (1999) 339-344.
[76] P Bandyopadhvay, HM Steinman, Catalase-peroxidases of Legionella pneumophila:
cloning of the katA gene and studies of KatA function. J. Bacteriol. 182 (2000) 6679-
6686.
152
[77] E Garcia, YA Nedialkov, J Elliott, VL Motin, RR Brubaker, Molecular characterization
of KatY (antigen 5), a thermoregulated chromosomally encoded catalase-peroxidase of
Yersinia pestis. J. Bacteriol. 181 (1999) 3114-3122.
[78] BA Chromy, MW Choi, GA Murphy, AD Gonzales, CH Corzeett, BC Chang, JP Fitch,
SL McCutchen-Maloney, Proteomic characterization of Yersinia pestis virulence. J.
Bacteriol. 44 (2005) 15093-15105.
[79] P Wojtaszek, Oxidative burst : an early plant response to pathogen infection. Biochem.
J. 322 (1997) 681-692.
[80] DE Heck, AM Vetrano, TM Mariano, JD Laskin, UVB light stimulates production
of reactive oxygen species: unexpected role for catalase. J. Biol. Chem. 278 (2003)
22432-22436.
[81] JM Wood, H Decker, H Hartmann, B Chavan, H Rokos, JD Spencer, S Hasse, MJ
Thornton, MShalbaf, RPaus, KUSchallreuter, Senilehairgraying: H2O2-mediatedox-
idative stress afiects human hair color by blunting methionine sulfoxide repair. FASEB
J (2009) In Press.
[82] P Chelikani, I Fita, PC Loewen, Diversity of structures and properties among catalases.
Cell. Mol. Life Sci. 61 (2004) 192-208.
[83] P Nicholls, I Fita, PC Loewen, Enzymology and structure of catalases. Adv. Inorg.
Chem. 51 (2001) 51106.
[84] P Jones, A Suggett, The catalase-hydrogen peroxide system: kinetics of catalatic action
at high substrate concentrations. Biochem. J. 110 (1968) 617-620.
[85] HB Dunford, JS Stillman, On the function and mechanism of action of peroxidases.
Coord. Chem. Rev. 19 (1976) 187-251.
[86] KG Welinder, Superfamily of plant, fungal and bacterial peroxidases. Curr. Opin.
Struct. Biol. 2 (1992) 388-393.
153
[87] V F?ul?op, CJ Ridout, C Greenwood, J Hajdu, Crystal structure of di-haem cytochrome
c peroxidase from Pseudomonas aeruginosa. Structure 3 (1995) 1225-1233.
[88] LP Hager, PF Hollenberg, T Rand-Meir, R Chiang, D Doubek, Chemistry of peroxidase
intermediates. Ann New York Acad. Sci. 244 (1975) 80-93.
[89] GC Mills, Hemoglobin catabolism I. glutathione peroxidase, an erythrocyte enzyme
which protects hemoglobin from oxidative breakdown. J. Biol. Chem. 229 (1957) 189-
197.
[90] JR Arthur, The glutathione peroxidases. Cell. Mol. Life Sci. 57 (2000) 1825-1835.
[91] J Littlechild, Haloperoxidases and their role in biotransformation reactions. Curr. Opin.
Chem. Biol. 3 (1999) 28-34.
[92] R Floris, I Recio, B Berkhout, S Visser, Antibacterial and antiviral efiects of milk
proteins and derivatives thereof. Curr. Pharm. Des. 9 (2003) 1257-1275.
[93] J Ruf, P Carayon, Structural and functional aspects of thyroid peroxidase. Arch.
Biochem. Biophys. 445 (2006) 269-277.
[94] G Nie, SD Aust, Efiect of calcium on the reversible thermal inactivation of lignin
peroxidase. Arch. Biochem. Biophys. 337 (1997) 225-231.
[95] GRJ Sutherland, SD Aust, The efiects of calcium on the thermal stability and activity
of manganese peroxidase. Arch. Biochem. Biophys. 332 (1996) 128-134.
[96] SJ George, M Kvaratskhelia, MJ Dilworth, RNF Thorneley, Reversible alkaline in-
activation of lignin peroxidase involves the release of both the distal and proximal
site calcium ions and bishistidine co-ordination of the haem. J. Biochem. 344 (1999)
237-244.
[97] M Sundarmoothy, K Kishi, MH Gold, TL Poulos, The crystal structure of manganese
peroxidase from Phanerochaete chrysosporium at 2.06-?A resolution. J. Biol. Chem. 269
(1994) 32759-32767.
154
[98] R Margis, C Dunand, FK Texeira, M Margis-Pinheiro, Glutathione peroxidase family
- an evolutionary overview. FEBS J. 275 (2008) 3959-3970.
[99] J Lu, A Holmgren, Selenoproteins. J. Biol. Chem. 284 (2009) 723-727.
[100] A Claiborne, JI Yeh, TC Mallett, J Luba, EJ Crane, V Charrier, D Parsonage,
Protein-sulfenic acids: diverse roles for an unlikely player in enzyme catalysis and
redox regulation. Biochemistry 38 (1999) 15407-15416.
[101] SE Brenner, SCOP: Family: Di-heme cytochrome c peroxidase. http://scop.mrc-
lmb.cam.ac.uk/scop/data/scop.b.b.h.b.h.html. Last updated November 2007. Accessed
April 2009.
[102] J Verd??n, R Pogni, A Baeza, MC Baratto, R Basosi, R V?azquez-Duhalt, Mechanism
of versatile peroxidase inactivation by Ca2+ depletion . Biophys. Chem. 121 (2006)
163-170.
[103] N Kresge, RD Simoni, RL Hill, Classics: Britton Chance: olympian and developer of
stop- ow methods. J. Biol. Chem. 279 (2004) 109-111.
[104] B Chance, The kinetics of the enzyme-substrate compound of peroxidase. J. Biol.
Chem. 151 (1943) 553-577.
[105] A Claiborne, I Fridovich, Puriflcation of the o-dianisidine peroxidase from Escherichia
coli B. Physicochemical characterization and analysis of its dual catalatic and peroxi-
datic activities. J. Biol. Chem. 254 (1979) 4245-4252.
[106] P Charalabous, J Risk, R Jenkins, A Birss, C Hart, J Smalley, Characterization of
a bifunctional catalase peroxidase of Burkholderia cenocepacia. FEMS Immunol. Med.
Microbibol. 50 (2007) 37-44.
[107] M Fraaije, H Roubroeks, W Hagen, W van Berkel, Puriflcation and characterization of
an intracellular catalase-peroxidase from Penicillium simplicissimum. Eur. J. Biochem.
235 (1996) 192-198.
155
[108] A Hochman, A Shemesh, Puriflcation and characterization of a catalase-peroxidase
from the photosynthetic bacterium Rhodobacterium capsuate. J. Biol. Chem. 262 (1987)
6871-6876.
[109] C Jakopitsch, F R?uker, G Regelsberger, M Dockal, G Peschek, C Obinger, Catalase-
peroxidase from the cyanobacterium Synechocystis PCC 6803: cloning, overexpression
in Escherichia coli, and kinetic characterization. Biol. Chem. 380 (1999) 1087-1096.
[110] K Johnsson, W Froland, P Schultz, Overexpression, puriflcation, and characterization
of the catalase-peroxidase KatG from Mycobacterium tuberculosis. J. Biol. Inorg. Chem.
272 (1997) 2834-2840.
[111] SWM Kengen, FJ Bikker, WR Hagen, WM de Vos, J van der Oost, Characterization
of a catalase-peroxidase from the hyperthermophilic archaeon Archaeoglobus fulgidus.
Extremophiles 5 (2001) 323-332.
[112] J Marcinkeviciene, R Magliozzo, J Blanchard, Puriflcation and characterization of the
Mycobacterium smegmatis catalase-peroxidase involved in isoniazid activation. J. Biol.
Chem. 270 (1995) 22290-22295.
[113] P Pongpon, CR Cooper, N Vanittanakom, Isolation and characterization of a catalase-
peroxidase gene from the pathogenic fungus, Penicillium marnefiei. Med. Mycol. 43
(2005) 403-411.
[114] H Youn, Y Yim, K Kim, YC Hah, S Kang, Spectral characterization and chemical
modiflcation of catalase-peroxidase from Streptomyces sp. J. Biol. Chem. 270 (1995)
13740-13747.
[115] KG Welinder, Bacterial catalase-peroxidases are gene-duplicated members of the plant
peroxidase superfamily. Biochim. Biophys. Acta 1080 (1991) 215-220.
[116] Y Yamada, T Fujiwara, T Sato, N Igarashi, N Taanaka, The 2.0 ?A crystal structure of
catalase-peroxidase from Haloarcula marismortui. Nat. Struct. Biol. 9 (2002) 691-695.
156
[117] X Carpena, S Lopraset, S Mongkolsuk, J Switala, PC Loewen, I Fita, Catalase-
peroxidase KatG of Burkholderia pseudomallei at 1.7 ?A resolution . J. Mol. Biol. 327
(2003) 475-489.
[118] T Bertrand, NAJ Eady, JN Jones, Jesmin, JM Nahgy, B Jamart-Gregoire, EL Raven,
KA Brown, Crystal Structure of Mycobacterium tuberculosis Catalase-Peroxidase. J.
Biol. Chem. 279 (2004) 38991-38999.
[119] K Wada, T Tada, Y Nakamura, T Kinoshita, M Tamoi, S Shigeoka, K Nishimura,
Crystallization and preliminary x-ray difiraction studies of catalase-peroxidase from
Synechocystis PCC 7942. Acta Crystallogr. D Biol. Crystallogr. 58 (2002) 157-159.
[120] W Blodig, AT Smith, WA Doyle, K Piontek, Crystal structure of prostine and ox-
idatively processed lignin peroxidase expressed in Escherichia coli and of the W171F
variant that eliminates the redox active tryptophan 171. Implications for the reaction
mechanism. J. Mol. Biol. 305 (2001) 851-861.
[121] M Gajhede, DJ SChuller, A Henriksen, AT Smith, TL Poulos, Crystal structure of
horseradish peroxidase C at 2.15 ?A resolution. Nat. Struct. Biol. 4 (1997) 1032-1038.
[122] MG Klotz, PC Loewen, The molecular evolution of catalatic hydroperoxidases: Evi-
dence for multiple lateral transfer of genes between prokaryota and from bacteria into
eukaryota. Molec. Biol. Evol. 20 (2003) 1098-1112.
[123] Y Li, DC Goodwin, Vital roles of an interhelical insertion in catalase-peroxidase
bifunctionality. Biochem. Biophys. Res. Commun. 318 (2004) 970-976.
[124] RA Ghiladi, KF Medzihradszky, PR Ortiz de Montellano, Role of the Met-Tyr-Trp
cross-link in Mycobacterium tuberculosis catalase-peroxidase (KatG) as revealed by
KatG(M255I). Biochemistry 44 (2005) 15093-15105.
[125] G Smulevich, C Jakopitsch, E Droghetti, C Obinger, Probing the structure and bi-
functionality of catalase-peroxidases (KatG). J. Inorg. Biochem. 100 (2006) 568-585.
157
[126] C Jakopitsch, M Auer, A Ivanvich, F R?uker, PG Furtm?uller, C Obinger, Total con-
version of bifunctional catalase-peroxidase (KatG) to monofunctional peroxidase by
exchange of a conserved distal side tyrosine. J. Biol. Chem. 278 (2003) 20185-20191.
[127] X Carpena, B Wiseman, T Deemagarn, R Singh, J Switala, A Ivancich, I Fita, PC
Loewen, A molecular switch and electronic circuit modulate catalase activity in catalase
peroxidases. EMBO J. 6 (2005) 1156-1162.
[128] X Carpena, B Wiseman, T Deemagarn, B Herguedas, A Ivanvich, R Singh, PC
Loewen, I Fita, Roles for Arg426 and Trp111 in the modulation of NADH oxidase
activity of the catalase-peroxidase KatG from Burkholderia pseudomallei inferred from
pH-induced structural changes. Biochemistry 45 (2006) 5171-5179.
[129] M Zamocky, G Regelsberger, C Jakopitsch, C Obinger, The molecular peculiarities of
catalase-peroxidases. FEBS Lett. 492 (2001) 177-182.
[130] RD Baker, CO Cook, DC Goodwin, Properties of catalase-peroxidase lacking its C-
terminal domain. Biochem. Biophys. Res. Commun. 320 (2004) 833-839.
[131] C Jakopitsch, M Auer, G Regelsberger, W Jantschko, PG Furtm?uller, F R?uker, C Ob-
inger, Distal site aspartate is essential in the catalase activity of catalase-peroxidases.
Biochemistry 42 (2003) 5292-5300.
[132] C Jakopitsch, M Auer, G Regelsberger, W Jantschko, PG Furtm?uller, F R?uker, C
Obinger, The catalytic role of the distal site asparagine-histidine couple in catalase-
peroxidases. Eur. J. Biochem. 270 (2003) 1006-1013.
[133] C Jakopitsch, E Droghetti, F Schmuckenschlager, PG Furtm?uller, G Smulevich, C
Obinger, Role of the main access channel of catalase-peroxidase in catalysis. J. Biol.
Chem. 280 (2005) 42411-42422.
158
[134] E Santoni, C Jakopitsch, C Obinger, G Smulevich, Comparison between catalase-
peroxidase and cytochrome c peroxidase. The role of the hydrogen-bond networks for
protein stability and catalysis. Biochemistry 43 (2004) 5792-5802.
[135] Y Li, Ph.D. thesis (2004) Auburn University.
[136] C Jakopitsch, D Kolarich, G Petutschnig, PG Furtm?uller, C Obinger, Distal side
tryptophan, tyrosine, and methionine in catalase-peroxidases are covalently linked in
solution. FEBS Lett. 552 (2003) 135-140.
[137] C Varnado, Ph.D. thesis (2006) Auburn University.
[138] LJ Donald, OV Krokhin, HW Duckworth, B Wiseman, T Deemagarn, R Singh, J
Switala, X Carpena, I Fita, PC Loewen, Characterization of the catalase-peroxidase
KatG from Burkholderia pseudomallei by mass spectrometry. J. Biol. Chem. 278 (2003)
35687-35692.
[139] S Yu, S Girotto, X Zhao, R Magliozzo, Rapid formation of compound II and a tyrosyl
radicalin theY229F mutantofMycobacteriumtuberculosiscatalase-peroxidasedisrupts
catalase but not peroxidase function. J. Biol. Chem. 278 (2003) 44121-44127.
[140] C Jakopitsch, A Ivancich, F Schmuckenschlager, A Wanasinghe, G P?oltl, P
Furtm?uller, F R?uker, C Obinger, In uence of unusual covalent adduct on the kinetics
and formation of radical intermediates in Synechocystis catalase-peroxidase. A stopped-
 ow and EPR characterization of the met275, tyr249, and arg439 variants. J. Biol. Chem.
279 (2004) 46082-46095.
[141] RA Ghiladi, GM Knudsen, KF Medzihradszky, PR Montellano, The Met-Tyr-Trp
cross-link in Mycobacterium tuberculosis catalase-peroxidase (KatG): autocatalytic for-
mation and efiect on enzyme catalysis and spectroscopic properties. J. Biol. Chem. 280
(2005) 22651-22663.
159
[142] K Ranguelova, S Girotto, G Gerfen, S Yu, J Suarez, L Metlitsky, R Maggliozzo,
Radical sites in Mycobacterium tuberculosis KatG identifled using electron paramgnetic
resonance spectroscopy, the three-dimensional crystal structure, and electron transfer
couplings. J. Biol. Chem. 282 (2007) 6255-6264.
[143] J Suarez, K Ranguelova, A Jarzecki, J Manzerova, V Krymov, X Zhao, S Yu, L Metlit-
sky, G Gerfen, R Magliozzo, An oxyferrous heme/protein-based radical intermediate is
catalytically competent in the catalase reaction of M. tuberculosis catalase-peroxidase
(KatG). J. Biol. Chem. 284 (2009) 7017-7029.
[144] R Singh, J Switala, PC Loewen, A Ivancich, Two [Fe(IV)=O Trp*] intermediates in
M. tuberculosis catalase-peroxidase discriminated by multifrequency (9-285 GHz) EPR
spectroscopy:reactivity toward isoniazid. J. Am. Chem. Soc. 129 (2007) 15954-15963.
[145] A Invancich, C Jakopitsch, M Auer, S Un, C Obinger, Protein-based radicals in the
catalase-peroxidase of Synechocystis PCC 6803: a multifrequency EPR investigation of
wild-type and variants on the envirnoment of the heme active site. J. Am. Chem. Soc.
125 (2003) 14093-14102.
[146] R Singh, B Wiseman, T Deemagarn, LJ Donald, HW Duckworth, X Carpena, I
Fita, PC Loewen, Catalase-peroxidases (KatG) exhibit NADH oxidase activity. J. Biol.
Chem. 279 (2004) 43098-43106.
[147] C Jakopitsch, G Regelsberger, P Furtm?uller, F R?uker, G Peschek, C Obinger, Engi-
neering the proximal heme cavity of catalase-peroxidase. J. Inorg. Biochem. 91 (2002)
78-86.
[148] SL Scott, WJ Chen, A Bakac, JH Espenson, Spectroscopic parameters, electrode
potentials, acid ionization constants, and electron exchange rates of the 2,2?-azinobis(3-
ethylbenzothiazoline-6-sulfonate) radicals and ions. J. Phys. Chem. 97 (1993) 6710-
6714.
160
[149] DP Nelson, LA Kiesow, Enthalpy of decomposition of hydrogen peroxide by catalase
at 25?. (with molar extinction coe?cients of H2O2 solutions in the uv). Anal. Biochem.
49 (1972) 474-478.
[150] HB Dunford, Heme Peroxidases, John Wiley & Sons, Inc., New York, 1999, pp. 73-85.
[151] C Jakopitsch, J Vlastis, B Wiseman, PC Loewen, C Obinger, Redox intermediates in
the catalase cycle of catalase-peroxidases from Synechocystis PCC 6803, Burkholderia
pseudomallei, and Mycobacterium tuberculosis. Biochemistry 46 (2007) 1183-1193.
[152] JN Rodr??guez-L?opez, MA Gilabert, J Tudela, RNF Thorneley, F Garc??a-C?anovas,
Reactivity of horseradish peroxidase compound II towards substrates: kinetic evidence
for a two-step mechanism. Biochemistry 39 (2000) 13201-13209.
[153] A Claiborne, I Fridovich, Chemical and enzymatic intermediates in the peroxidation
of o-dianisidine by horseradish peroxidase. 1. Spectral properties of the products of
dianisidine oxidation. Biochemistry 18 (1979) 2324-2329.
[154] F Bosco, A Capolongo, B Ruggeri, Efiect of temperature, pH, ionic strength, and
sodium nitrate on activity of LiPs: Implications for bioremediation. Biorem. J. 6 (2002)
65-76.
[155] DC Goodwin, T Grover, SD Aust, Roles of e?cient substrates in enhancement of
peroxidase-catalyzed oxidations. Biochemistry 36 (1997) 139-147.
[156] E Derat, S Shaik, An e?cient proton-coupled electron-transfer process during oxida-
tion of ferulic acid by horseradish peroxidase: coming full cycle. J. Am. Chem. Soc.
128 (2006) 13940-13949.
[157] R Marjaana, A Lambeir, N Ellfolk, B Dunford, A rapid-scan spectrometric and
stopped- ow study of Compound I and Compound II of Pseudomonas cytochrome
c peroxidase. Arch. Biochem. Biophys. 236 (1985) 714-719.
161
[158] V Ximenes, L Catalani, A Campa, Oxidation of melatonin and tryptophan by an HRP
cycle involving compound III. Biochem. Biophys. Res. Commun. 287 (2001) 130-134.
[159] R Singh, B Wiseman, T Deemagarn, V Jha, J Switala, PC Loewen, Comparative
study of catalase-peroxidases (KatGs). Arch. Biochem. Biophys. 471 (2008) 207-214.
[160] C Jakopitsch, C Obinger, S Un, A Ivancich, Identiflcation of Trp106 as the trypto-
phanyl radical intermediate in Synechocystis PCC 6803 catalase-peroxidase by multifre-
quency Electron Paramagnetic Resonance spectroscopy. J. Inorg. Biochem. 100 (2006)
1091-1099.
[161] JK Falk, in KM Smith (Ed.), Porphyrins and Metalloporphyrins, Elsevier Publishing,
New York, 1964, pp. 804-807.
[162] S Chouchane, S Girotto, S Kapetanaki, JP Schelvis, S Yu, RS Magliozzo, Analysis
of Heme Structural Heterogeneity in Mycobacterium tuberculosis Catalase-Peroxidase
(KatG). J. Biol. Chem. 278 (2003) 8154-8162.
[163] EW Svastits, JH Dawson, Models for ferrous cytochrome b5: Sign inversions in the
magnetic circular dichroism spectra of bis-imidazole ferrous porphyrin systems. Inorg.
Chim. Acta 123 (1986) 83-86.
[164] J Cheek, D Mendelman, TL Poulos, JH Dawson, A study of the K+-site mutant of
ascorbate peroxidase: mutations of protein residues on the proximal side of the heme
cause changes in iron ligation on the distal side. J. Biol. Inorg. Chem. 4 (1999) 64-72.
[165] AE Pond, M Sono, EA Elenkova, DB Goodin, AM English, JH Dawson, In uence of
protein environment on magnetic circular dichroism spectral properties of ferric and
ferrous ligand complexes of yeast cytochrome c peroxidase. Biospectroscopy 5 (1999)
S42-S52.
162
[166] L Vickery, T Nozawa, K Sauer, Magnetic circular dichroism studies of low-spin cy-
tochromes. Temperature dependence and efiects of axial coordination on the spectra
of cytochrome c and cytochrome b5. J. Am. Chem. Soc. 98 (1976) 351-357.
[167] MK Johnson, in L Que (Ed.), Physical Methods in Bioinorganic Chemistry: Spec-
troscopy and Magnetism, University Science Books, Sausolito, 2000, pp. 233-285.
[168] W Hagen, TU Delft { Biomolecular EPR Spectroscopy. http://www.bt.tudelft.nl/
biomolecularEPRspectroscopy.
[169] T Deemagarn, B Wiseman, X Carpena, A Ivancich, I Fita, PC Loewen, Two alter-
native substrate paths for compound I formation and reduction in catalase-peroxidase
KatG from Burkholderia pseudomallei. PROTEINS: Struct. Funct. Bioinf. 66 (2007)
219-228.
[170] S Yu, S Chouchane, R Magliozzo, Characterization of the W321F mutant of Mycobac-
terium tuberculosis catalase peroxidase KatG. Prot. Sci. 11 (2002) 58-64.
[171] H Heering, C Indiani, G Regelsberger, C Jakopitsch, C Obinger, G Smulevich, New
insights into the heme cavity structure of catalase-peroxidase: A spectroscopic ap-
proach to the recombinant Synechocystis enzyme and selected distal cavity mutants.
Biochemistry 41 (2002) 9237-9247.
163