STRUCTURAL AND BIOCHEMICAL STUDIES OF Streptococcus sp.
α-GLYCEROPHOSPHATE OXIDASE
Timothy M. Colussi
A Dissertation Submitted to the Graduate Faculty of
WAKE FOREST UNIVERSITY SCHOOL OF ARTS AND SCIENCES
In Partial Fulfillment of the Requirement for the Degree of
DOCTOR OF PHILOSOPHY
In the Department of Biochemistry
of the Wake Forest University School of Medicine
December 2009
Winston-Salem, North Carolina
Approved By:
Al Claiborne, Ph.D., Advisor ______
Examining Committee:
S. Bruce King, Ph.D., Chair ______
Thomas Hollis, Ph.D. ______
Derek Parsonage, Ph.D. ______
W. Todd Lowther, Ph.D. ______
ACKNOWLEDGMENTS
First I would like to thank Dr. Al Claiborne for affording me the opportunity to
work in his lab. His guidance has been a tremendous help to me over the years and his
support has helped me grow as a scientific researcher. He has always been willing to give
me the opportunity to share my work with other members of the scientific committee both here at Wake Forest and at meetings elsewhere. He has always been willing to help and offer assistance whenever I might have needed it, both scientifically and otherwise.
I would also like to thank Dr. Conn Mallet for teaching me both the theory and practice of x-ray crystallography, and how to open a beer bottle using the desk handles at
NSLS. He was instrumental in getting me excited about x-ray crystallography. He has
also given me guidance throughout my graduate career, even after leaving Wake Forest
University.
An enormous thank you to Dr. Derek Parsonage, my molecular biology guru. He
patiently answered the same question three times on three different days many times
throughout my graduate career involving molecular biology and/or enzyme kinetics. I
could not have accomplished much without his continued help.
A huge thank you also to Dr. Tom Hollis for his guidance, not only with x-ray
crystallography, but other aspects of my graduate career. He never turned me away when
I had a question and was always willing to help me figure out the answer to the many
questions I had.
Thank you to Drs. Bruce King and Todd Lowther for serving on my committee
and the advice and guidance you have given me during my graduate career.
ii Dr. Andy Karplus was instrumental in the preparing Chapter II of this thesis. I am
grateful for the time he took helping me with structure refinement and figure preparation
even though he was three time zones away.
I would also like to thank the former members of the Claiborne lab: Conn Mallet,
Derek Parsonage, Jamie Wallen, Nate Nicely, Bill Boles, Carleitta Paige, and Sumana
Choudhury. You were always willing to help when I needed it.
I would also like to thank the many friends I have met through graduate school over the years. I will miss all the great times we have shared together having fun and complaining about the recent Wake Forest defeat in either football or basketball, we all know there were many. I will never forget the times you have helped keep me sane when
I was about to lose my mind. We all know those times were many, as well.
I would also like to thank my family and friends who have supported me through
this whole process. I would especially like to thank my mother, Norma Colussi. She is possibly the strongest woman I know and has helped me keep going during any of the tough times I’ve had over the years, even when she didn’t know it.
And last but not least, I would like to thank the man for whom I dedicate this to, my father, Larry Colussi. He is my hero and the reason I am the way I am today. This
one’s for you Dad.
iii TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS ...... ii
LIST OF ABBREVIATIONS ...... v
LIST OF ILLUSTRATIONS ...... vi
ABSTRACT ...... ix
Chapter I. INTRODUCTION: FLAVOPROTEIN OXIDASES, DEHYDROGENASES, AND α-GLYCEROPHOSPHATE BIOLOGY...... 1
II. STRUCTURE OF α-GLYCEROPHOSPHATE OXIDASE FROM Streptococcus sp.: A TEMPLATE FOR THE MITOCHONDRIAL α-GLYCEROPHOSPHATE DEHYDROGENASE ...... 31
Published in Biochemistry, January 2008
III. KINETIC ANALYSIS OF α-GLYCEROPHOSPHATE OXIDASE FROM Streptococcus sp.: IDENTIFICATION OF RESIDUES IMPORTANT IN GLP-SUBSTRATE BINDING AND CATALYSIS...... 78
IV. CONLUSIONS AND FUTURE DIRECTIONS ...... 112
SCHOLASTIC VITAE ...... 126
iv LIST OF ABBREVIATIONS
DAAO, D-amino acid oxidase; ThiO, glycine oxidase; MSOX, monomeric sarcosine
oxidase; FAD, flavin adenine dinucleotide; PutA, L-proline dehydrogenase; DHOD,
dihydroorotate dehydrogenase; P5C, Δ1-pyrroline-5-carboxylate; DHO, dihydroorotate;
OA, orotate; NAD, nicotinamide adenine dinucleotide; FMN, flavin mononucleotide;
GlpK, glycerol kinase; GlpO, α-glycerophosphate oxidase; GlpF, glycerol facilitator;
HPr, histidine phosphocarrier protein; FBP, fructose 1,6-bisphosphate; GlpD, α- glycerophosphate dehydrogenase; Glp, α-glycerophosphate; DHAP, dihydroxyacetone phosphate; GAP, glyceraldehyde phosphate; TIM, triosephosphate isomerase; ATP, adenosine triphosphate; GPD1, cytosolic α-glycerophosphate dehydrogenase; GPD1-L, cytosolic α-glycerophosphate dehydrogenase like protein; mGPD, mitochondrial α- glycerophosphate dehydrogenase; EMT, enzyme-monitored turnover; MAD, multiwavelength anomalous dispersion; SeMet, selenomethionine; MR, molecular replacement; GlpOΔ, α-glycerophosphate oxidase deletion mutant; mitoGlpD, mitochondrial α-glycerophosphate dehydrogenase; bactGlpD, bacterial α- glycerophosphate dehydrogenase; PCR, polymerase chain reaction; EDTA, ethylenediaminetetraacetic acid; PEG, polyethylene glycol; GR, glutathione reductase;
ADP, adenosine diphosphate; NAG, N-acetylglycine; D-Ala, D-alanine; Gut2P, yeast mitochondrial α-glycerophosphate dehydrogenase; Cox-1, prostaglandin H2 synthase;
HRP, horseradish peroxidase; DTT, dithiothreitol; β-OG, 1-O-n-octyl-β-D-
glucopyranoside; PMS, phenazine methosulfate; MTT, 2-(4,5-dimethyl-2-thiazolyl)-3,5-
diphenyl-2-H-tetrazolium bromide; SURE, subtilin-regulated gene expression; SDS-
PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis
v LIST OF ILLUSTRATIONS
Schemes Page
1-1 Kinetic mechanism for intact Streptococcus sp. GlpO 22
1-2 Kinetic mechanism for the [Glp] independent reaction seen in the nicked 23 Form of Streptococcus sp. GlpO
Tables
1-1 Chemical and spectral differences between flavoprotein oxidases and 3 dehydrogenases.
2-1 Data collection for GlpO and GlpOΔ and phasing statistics for GlpOΔ 40
2-2 Crystallographic refinement statistics for GlpO and GlpOΔ 41
2-3 Steady-state kinetic constants for intact and nicked forms of Streptococcus 44 sp. GlpO and for GlpOΔ
3-1 Results of sulfite titrations of wild-type and mutant GlpOΔ enzymes 93
3-2 Turnover numbers for wild-type and mutant GlpOΔ enzymes from HRP 96 Coupled assay
3-3 Steady-state kinetic constants for GlpOΔ and apparent turnover numbers for 101 GlpOΔ and mutants calculated from enzyme-monitored turnover experiments
Figures
1-1 Absorption spectra of glucose oxidase showing different redox states of FAD 5 cofactor
1-2 Sulfite titration from α-glycerophosphate oxidase from Enterococcus 6 casseliflavus
1-3 Spectra of the thiolate and benzoquinoid form of 8-mercaptoflavin 8
1-4 Schematic of a possible carbanion mechanism for the dehydration of 9 α-hyroxyacids
1-5 Depiction of bacterial glycerol metabolism 10
vi 1-6 α-Glycerophosphate shuttle 13
1-7 Sequence alignment of functional GlpOs and GlpDs from bacterial and 16 mitochondrial sources
1-8 Model for triggering host cell inflammation by M. mycoides subsp. mycoides 18
1-9 E. coli and B. halodurans GlpD dimers 20
1-10 Limited proteolysis of Streptococcus sp. and E. casseliflavus GlpO 21
2-1 Dithionite and sulfite titrations of GlpOΔ 45
2-2 Enzyme-monitored turnover of GlpOΔ 46
2-3 Final 2Fo-Fc maps for the FAD redox centers of GlpOΔ, intact GlpO chain A, 49 and intact GlpO chain B
2-4 The GlpOΔ dimer, and superposition of a GlpOΔ monomer with intact GlpO 50 chain A
2-5 Topology diagram for GlpOΔ, and superposition of GlpOΔ domains 1 and 2 52 with a ThiO monomer
2-6 Structure-based sequence alignment for Streptococcus sp. GlpO and four 56 known functional GlpOs and GlpDs from bacterial and mitochondrial sources
2-7 Active-site overlay, in stereo, for GlpOΔ and intact GlpO 60
2-8 Active-site overlay, in stereo, for GlpOΔ, the ThiO-N-acteylglycine (NAG) 62 complex (chain A), and the yeast DAAO-D-Ala complex
2-9 Stereoview of the proposed Glp-binding pocket, including N-acetylglycine 64 (NAG) from ThiO overlay in Figure 8
3-1 Active sites of two structural homologs of Streptococcus sp. GlpO 83
3-2 GlpOΔ active site 84
3-3 Proposed mechanism for Glp-binding, proton abstraction, and hydride 85 transfer for Streptococcus sp. GlpOΔ
3-4 Sulfite titration of the wild-type GlpOΔ 92
3-5 HRP coupled assay for wild-type and mutant GlpOΔ enzymes 95
vii 3-6 Active site overlays, in stereo, of GlpOΔ and E. coli GlpD in complex with 98 phosphate and DHAP
3-7 Enzyme-monitored turnover of wild-type GlpOΔ 100
3-8 Enzyme-monitored turnover of H65A 103
3-9 Refined mechanism for the oxidation of Glp to DHAP by GlpOΔ based on 107 the results of mutagenic and kinetic experiments
4-1 Expression, purification, and activity of E. coli GlpD 118
4-2 Dimer overlays of E. coli and B. halodurans GlpD with Streptococcus sp. 121 GlpOΔ
4-3 B. subtilis GlpD protein expression using the SURE system 123
viii ABSTRACT
The ability for bacteria to grow on glycerol has been linked to the genes
responsible for encoding the glycerol facilitator (GlpF), glycerol kinase (GlpK), and
either α-glycerophosphate oxidase (GlpO) or dehydrogenase (GlpD). GlpF aids in
bringing glycerol into the cell, while GlpK phosphorylates it, trapping it within the cell as
α-glycerophosphate (Glp). Glp is converted to dihydroxyacetone phosphate (DHAP) by
GlpO in certain bacteria and by GlpD in others. Heme deficient lactic acid bacteria, such
as Streptococcus sp. and Enterococcus casseliflavus, contain GlpO, whereas bacteria
such as Bacillus subtilis and Escherichia coli contain GlpD. Both GlpO and GlpD utilize
flavin adenine dinucleotide (FAD) as a cofactor, which is concomitantly reduced upon
substrate oxidation. Though sequentially similar, GlpOs and GlpDs are functionally
different based on the reoxidation of the cofactor; GlpO reacts with O2 to form H2O2, while GlpD transfers the reducing equivalents to ubiquinone. GlpO and GlpD also differ in that GlpDs are considered membrane associated, while most GlpOs are considered cytosolic. The GlpOs from heme deficient lactic acid bacteria also contain a ca. 50- residue insert that is not present in GlpDs. Limited proteolysis experiments with the
Streptococcus sp. GlpO enzyme have suggested that this insert region may be a flexible surface loop region.
A 2.4 Å resolution x-ray crystal structure of the streptococcal GlpO enzyme has been determined, along with a 2.3 Å resolution x-ray crystal structure of a deletion mutant (GlpOΔ) lacking the 50-residue insert region but retaining catalytic activity. A comparison of these two structures shows conformational changes involving an active- site histidine (H65) and the isoalloxazine ring of FAD. This has been interpreted as being
ix representative of two forms of the resting oxidized enzyme, seen biochemically with the
enterococcal GlpO enzyme.
Comparisons of GlpOΔ with two structural homologs, D-amino acid oxidase
(DAAO) and glycine oxidase (ThiO), showed the presence of a structurally conserved
arginine residue, R346 in GlpOΔ. Further analysis of the active site of GlpOΔ resulted in
the finding of other residues that are conserved among the GlpOs and GlpDs and are also
within proposed Glp-binding regions, namely H65, R69, Y70, and K429; however, these
residues are not present in DAAO or ThiO. This has led to the hypothesis that these
residues form a phosphate binding pocket. An active-site base is required for the
abstraction of a proton from the C(2)-OH of Glp and drives the hydride transfer of C(2)-
H to FAD N(5). Histidines are common residues that play this role in oxidoreductases. A
hypothesis was developed that H65 may play that role.
Mutagenic analyses show that R346 and K429 are essential for GlpO activity.
Both residues when mutated yielded an inactive enzyme based on the horseradish
peroxidase (HRP) assay and enzyme-monitored turnover (EMT) experiments. Sulfite
titration data for both mutants revealed that they both stabilized an FAD N(5)-sulfite
adduct, though at a higher Kd when compared to wild-type GlpOΔ. Analyzing the
relationship of Kd values for sulfite and redox potential, the increase in sulfite Kd seems to rule out the mutations affecting the redox potentials of the FAD cofactor for either of these mutants. Based on the HRP assay data, H65A is ~2% as active as wild type GlpOΔ, whereas the H65Q mutant is ~63% as active. This result contradicts the hypothesis that
H65 is an active site base. This result, however, does support the idea that this residue is important for activity but for substrate binding, possibly interacting with the Glp-
x phosphate moiety. The activity for the R69M is ~13% that of wild type, implicating it as an important residue in catalysis, also possibly interacting with the Glp-phosphate upon substrate binding. The Y70F mutant yields an activity of ~77% of wild type, indicating that it may be important for enzyme catalysis but not essential. The structural and biochemical experiments described in this dissertation have led to the formulation of a reaction scheme for binding and catalysis of GlpO.
.
xi Chapter I
INTRODUCTION: FLAVOPROTEIN OXIDASES, DEHYDROGENASES, AND
α-GLYCEROPHOSPHATE BIOLOGY
Timothy Colussi
1 Flavoprotein Oxidases and Dehydrogenases. Flavoprotein oxidases and
dehydrogenases can be distinguished based on certain chemical properties; these are
shown in Table 1 and will be described below. Oxidases characteristically react with O2 and form H2O2 as a catalytic product, whereas dehydrogenases react with O2 and form
- O2 ; however, the formation of superoxide is not part of the catalytic mechanism (1).
Examples of flavoprotein oxidases are D-amino acid oxidase (DAAO)1, glycine oxidase
(ThiO), and monomeric sarcosine oxidase (MSOX). DAAO has been considered the
prototypical flavoprotein oxidase and catalyzes the dehydrogenation of D-amino acids to
their corresponding α-imino acids, which self hydrolyze to their respective α-keto acids
and ammonia (2). ThiO catalyzes the deamination of primary and secondary amino acids
and also D-amino acids to form the corresponding α-keto acids similar to DAAO (3).
MSOX catalyzes the oxidation of sarcosine and other secondary amino acids to their respective imine acids, similar to ThiO (4). MSOX is different from the rest of these oxidases in that the flavin adenine dinucleotide (FAD) cofactor is covalently bound at the
8 position of the isoalloxazine ring to a cysteine (4). All three of the examples of oxidases react with O2 to form H2O2 (2, 4, 5). Examples of flavoprotein dehydrogenases are L-
proline dehydrogenase (PutA) and dihydroorotate dehydrogenase (DHOD). PutA
catalyzes the oxidation of L-proline to Δ1-pyrroline-5-carboxylate (P5C) that hydrolyzes
to glutamate-5-semialdehyde; and the FAD cofactor is reoxidized by transferring
electrons to an acceptor in the electron transfer chain (6). DHODs catalyze the oxidation
of dihydroorotate (DHO) to orotate (OA); however, they are divided into two different
classes based on several properties, one of which is how the enzyme is reoxidized (7).
1 Abbreviations: see page v.
2 Table 1: Chemical and Spectral Differences Between Flavoprotein Oxidases and Dehydrogenases. The data in this table was adapted from (1).
3 Class 2 DHODs are membrane-bound and are reoxidized by transferring electrons to
ubiquinone (7). Class 1 DHODs can be subdivided into Class 1A, which are homodimers reoxidized by fumarate, and Class 1B, which are α2β2 heterotetramers containing one
subunit similar to Class 1A and another subunit containing FAD and an iron-sulfur
cluster allowing reoxidation by nicotinamide adenine dinucleotide (NAD) (7).
Free flavin exists in rapid equilibrium with its oxidized and reduced forms (8-10); however, when bound to a protein, it is possible to stabilize a radical form. One distinction between oxidases and dehydrogenases is based on which radical form is stabilized by the protein, the red anionic radical for oxidases and the blue neutral for dehydrogenases (1). Glucose Oxidase, however, shows a stabilization of both semiquinones and the spectra for the oxidized, reduced, red anion, and blue neutral redox species are shown in Figure 1 (11).
Upon titrations with sulfite, flavoprotein oxidases generally stabilize a strong
FAD N(5)-sulfite adduct, evidenced by monitoring the bleaching of flavin absorption at
450 nm (1). This reaction is thought to parallel that of a hydride transfer mechanism.
Figure 2 shows the absorption spectra of α-glycerophosphate oxidase from Enterococcus
casseliflavus when reacted with sodium sulfite (Figure 2A) and the theorized reaction
scheme (Figure 2B) (12). Dehydrogenases do not appear to stabilize a sulfite adduct (1).
Another way to probe the active site of flavoproteins is with the use of modified
flavins (1). One of these modified flavins is 8-mercaptoflavin, which has a sulfur group in
place of the methyl group normally found at position 8 (Table 1). Oxidases tend to
stabilize the benzoquinoid form of this modified flavin depicted by the spectrum shown
4
Figure 1: Absorption spectra of Glucose Oxidase showing different redox states of FAD cofactor. The oxidized and reduced spectra are depicted by a solid line (____). The red anion semiquinone species is shown by dashed line (----) and the blue neutral semiquinone is shown by a dotted line (……). This figure is reprinted with permission from (11). Copyright © 1966, American Chemical Society.
5
Figure 2: Sulfite titration spectra of α-glycerophosphate oxidase from Enterococcus casseliflavus. (A) Titration of the enzyme with sodium sulfite showing decreasing absorption at 446 nm. (B) Reaction scheme showing the parallel between the formation of an FAD N(5)-Sulfite adduct and the hydride transfer from the substrate α- glycerophosphate. Both portions of this figure have been reprinted with permission from (12). Copyright © 1998, American Chemical Society.
6 in Figure 3 for D-amino acid oxidase, whereas dehydrogenases stabilize the thiolate form depicted by the spectrum shown in Figure 3 for flavodoxin (1).
Taken together, the stabilization of the red anion radical, the reactivity with sulfite, and the stabilization of the benzoquinoid form of 8-mercaptoflavin, implicates a stabilization of a negative charge at the N(1)-C(2)=O position of the isoalloxazine ring of
FAD or flavin mononucleotide (FMN) (1). This stabilization can be carried out by a positively charged residue or the dipole of the N-terminus of an α-helix near that position of the isoallozaxine ring, and is thought to be a trait present in flavoprotein oxidases (1).
This phenomenon is also thought to be indicative of a carbanion mechanism (1). A carbanion mechanism utilizes the abstraction of the α-proton of the substrate by a base present within the enzyme. The product can then be formed utilizing two pathways, adduct formation or two single electron transfer reactions, as depicted in Figure 4 (13).
This mechanism based on chemical experiments was developed prior to the determination of x-ray crystal structures of any flavoproteins. On the determination of the crystal structures of both pig kidney and yeast Rhodotorula gracilis DAAO, it was shown that a hydride transfer mechanism is possible, since an active-site base was not present
(14, 15). A hydride transfer mechanism transfers a proton and two electrons, known as a hydride, to the FAD cofactor. The hydride transfer mechanism is now thought to exist in other flavoprotein oxidases as well.
Glycerol metabolism. In lactic acid bacteria capable of growth on glycerol, the genes encoding glycerol kinase (GlpK), α-glycerophosphate oxidase (GlpO), and the glycerol facilitator (GlpF) are co-transcribed (16). The mode of uptake, phosphorylation, and shuttling into metabolic pathways of glycerol is depicted in Figure 5 along with the
7
Figure 3: Spectra of the thiolate and benzoquinoid form of 8-mercaptoflavin. Flavodoxin stabilizes the thiolate form of 8-mercaptoflavin, whereas D-amino acid oxidase stabilizes the benzoquinoid form. This figure is reprinted with permission from (1). Copyright © 1980, Portland Press, http://www.biochemsoctrans.org.
8
Figure 4: Schematic of a possible carbanion mechanism for the dehydrogenation of α- hydroxyacids. This figure has been reprinted with permission from (13). Copyright © 2000, Portland Press, http://www.biochemsoctrans.org.
9
Figure 5: Depiction of bacterial glycerol metabolism. Glycerol is taken into the cell by GlpF (1FX8) and phosphorylated by GlpK (1R59) to form Glp. GlpK is activated by HPr that has been phosphorylated (1FU0). Glp is oxidized to DHAP by GlpO/D. DHAP is converted to GAP by TIM (7TIM) and GAP enters into the glycolytic pathway resulting in pryruvate.
10 activation and inhibition of the system. Glycerol is taken up into the cell across the cell
wall and cytoplasmic membrane by the aquaporin GlpF (17). GlpF is a membrane protein
that catalyzes the equilibration of glycerol inside and outside of the cell (18). This
protein, thought of as a channel, shows selectivity for small straight chain molecules,
such as glycerol, but not for four-, five-, and six-carbon sugars or α-glycerophosphate
(18). Charged residues within the cavity of the GlpF channel maintain the specificity for
uncharged molecules as well (17).
Once glycerol is passed through the cytoplasmic membrane by GlpF, it can be
phosphorylated by GlpK (19) to form α-glycerophosphate (Glp). Once glycerol is
phosphorylated, it is trapped in the cytoplasm and cannot leave the cell. This creates a
chemical potential for glycerol uptake into the cell as it creates a glycerol concentration
imbalance across the membrane (19). GlpK is activated by phosphorylation by histidine phosphocarrier protein (HPr), whereas Gram-negative GlpK is regulated allosterically by
enzyme IIAGlc (19). Both systems are inhibited by fructose 1,6-bisphosphate (FBP) (16).
Glp can be oxidized to dihydroxyacetone phosphate (DHAP). In some bacteria,
mainly heme-deficient lactic acid bacteria, α-glycerophosphate oxidase (GlpO) catalyzes
this reaction (20). In other bacteria, such as Bacillus subtilis and Escherichia coli, this
reaction is catalyzed by α-glycerophosphate dehydrogenase (GlpD), which is sequentially similar to GlpO (21). Both the oxidases and dehydrogenases are FAD- dependent enzymes. DHAP can be converted to glyceraldehyde-3-phosphate (GAP) by
triosephosphate isomerase (TIM) (22). GAP is part of stage 2 of glycolysis and is
converted into pyruvate.
11 Glycolysis results in the formation of the reduced form of nicotinamide adenine
dinucleotide (NADH) and the electrons from this reduced form need to be shuttled into
the mitochondria (23). This can be achieved via two shuttles, the malate-aspartate shuttle and the α-glycerophosphate shuttle (24). These redox shuttles work to help import equivalents from NADH into the mitochondrial respiratory chain resulting in the
generation of adenosine triphosphate (ATP), which is important for insulin secretion in
pancreatic β cells (24), among other biological processes. The Glp shuttle, which is
depicted in Figure 6, utilizes a cytosolic protein and mitochondrial membrane protein to transfer reducing equivalents from NADH into the electron transport chain (24). The cytosolic enzyme is L-α-glycerophosphate dehydrogenase 1 (GPD1) and is an NAD- dependent dehydrogenase that catalyzes the reversible redox conversion of Glp to DHAP
(25). The structure of human GPD1 has been solved in complex with NAD+ and DHAP
(25). Structural and biochemical analysis has shown that the protonation state of ε-NH2 of
a lysine residue controls the equilibrium position of the reversible reaction (25). A protein
that shows homology (~70% identical) to GPD1, GPD1-L, has been linked to the genetic
cardiac disease Brugada’s syndrome (26).
The second, membrane-associated enzyme of the Glp shuttle is the mitochondrial
α-glycerophosphate dehydrogenase (mGPD). This enzyme catalyzes the oxidation of Glp
back to DHAP with concomitant reduction of the FAD cofactor and is located on the
outer surface of the inner mitochondrial membrane (24). The reducing equivalents from
FADH2 are then transferred to ubiquinone and shuttled into the mitochondrial respiratory
chain (24). A structure of mGPD has not been solved; however, based on hydropathy plot
analysis, a topology diagram has been proposed that suggests the presence of three
12
Figure 6: α-Glycerophosphate shuttle. The cytosolic GPD1 converts DHAP to Glp using NADH as a cofactor, whereas the mitochondrial GlpD converts Glp to DHAP, transferring the equivalents originating from NADH to quinones within the inner membrane. This figure was adapted from (27).
13 transmembrane segments (28). The sequence of mGPD shows high sequence similarity to the GlpOs and GlpDs described above for bacterial glycerol metabolism with a major difference being the existence of a C-terminal calcium binding domain. Calcium has been shown to regulate mGPD activity (28). An mGPD gene knockout in mice shows the ability to secret insulin upon the addition of glucose; however, when the malate-aspartate shuttle was inhibited, insulin secretion was nearly abolished upon the addition of glucose
(23). Inhibition of the malate-aspartate shuttle in normal mice showed slightly suppressed insulin secretion, implicating the α-glycerophosphate shuttle as a key component of the redox shuttles needed for glucose-derived insulin secretion, and the impairment of this type of insulin secretion has been seen in non-insulin-dependent diabetes mellitus (23).
α-Glycerophosphate Oxidases and Dehydrogenases. As mentioned above, GlpO catalyzes the oxidation of Glp to DHAP, like GlpDs; however, the FAD cofactor is reoxidized by reacting with O2 to form H2O2 rather than reacting with ubiquinone (29).
This allows heme-deficient lactic acid bacteria, such as Streptococcus and Enterococcus, to be able to grow on glycerol as the main carbon source (16). Rather than showing sequence homology with flavoprotein oxidases, GlpOs exhibit high sequence homology to the bacterial and mitochondrial GlpDs (Figure 7).
GlpO is also present in Mycoplasma mycoides subspecies mycoides (30). This
GlpO is membrane associated and has been shown to be a Mycoplasma virulence factor by generating large amounts of H2O2 that is taken up by host cells and leads to cell death
(30). Mycoplasma take up glycerol from the interstitial fluid via the ABC transporter or the putative glycerol facilitator, GlpF, and phosphorylated by GlpK, as described
14 Figure 7: Sequence alignment of functional GlpOs and GlpDs from bacterial and mitochondrial sources. Residue numbering is based on the Streptococcus sp. GlpO. Residues 1-51 and 690-805 of the human mitochondrial GlpD (hMito; full-length precursor sequence) have been omitted for clarity. Red and yellow blocks indicate residues that are conserved and conservative substitutions, respectively. The blue boxes indicate residues that are contained within the proposed Glp-binding regions (21). The sand colored box indicates residues that make up the major motif for the FAD-binding domain of GlpO.
15
16
previously, and depicted in Figure 5 (30). Figure 8 shows a model of how this is believed
to occur (30)
The glpF, glpK, and glpD genes are constitutively expressed in Mycoplasma
pneumoniae (31). The enzyme encoded by the glpD gene in M. pneumoniae is an oxidase
rather than a dehydrogenase and is predominantly cytosolic, unlike the M. mycoides
GlpO (31). M. pneumoniae, along with the weakly pathogenic M. mycoides, lacks the
ABC transporter found in the highly pathogenic M. mycoides strain (31). Glyecrol is only
taken into the cell in this case by GlpF, making glycerol metabolism less efficient in M. pneumoniae than in M. mycoides subspecies mycoides (31)
The E. coli GlpD is thought to be membrane associated. It was first thought to be an integral membrane protein based on the need for detergent for solubilization (32).
Later, a His-tagged version was purified without the need of detergent, only high salt, which is indicative of a peripheral membrane protein (33). Recently it was said to contain
“six transmembrane spanning” regions (34). The E. coli GlpD has been shown to participate in the formation of persister cells (35). Persister cells exhibit multidrug tolerance, an ability to survive when exposed to antibiotics that act against unrelated targets (35). A strain lacking the glpD gene was produced and the persister level was decreased in this cell line when compared to the wild-type cell line (35).
GlpD has also been implicated in the production of PrfA, a transcription activator of virulence factors, in Listeria monocytogenes (36). The genes that code for GlpK and
GlpD were knocked out and the resultant mutants lacked the ability to activate PrfA when
17
Figure 8: Model for triggering host cell inflammation by M. mycoides subsp. mycoides. GtsABC: glycerol import and phosphorylation system. GlpF: glycerol facilitator. GlpK: glycerol kinase. Bold arrows indicate the main virulence pathway. This figure is reprinted with permission from (30). Copyright © 2005, American Society for Microbiology.
18 grown in medium that contained glycerol as the main carbon source (36). In both bacteria
described above, GlpD is essential for the bacteria to grow on glycerol as well (35, 36).
The x-ray crystal structure of the E. coli GlpD was determined in 2008 (37). The
x-ray crystal structure of GlpD from Bacillus halodurans was also determined in 2008
(PDB code 3DA1). The E. coli and B. halodurans GlpDs are ~32% identical based on
amino acid sequence. Analysis of these structures shows a different dimer interface
(Figure 9). The portion of the protein that forms the dimer of the B. halodurans GlpD is
said to be the portion of the protein embedded in the membrane for the E. coli GlpD (37).
Enterococcus casseliflavus and Streptococcus sp. α-Glycerophosphate Oxidase.
Enterococcus casseliflavus and Streptocococus sp. GlpOs are cytosolic flavoprotein
oxidases that show 30-43% identity by sequence with the bacterial GlpDs and the
mitochondrial GlpD (mGPD), with one of the main differences being that a ca. 50-
residue insert is present in these GlpOs and not in GlpDs (Figure 10B) (12). Flanking
these inserts are two regions that contain residues predicted to be part of a Glp-binding
motif based on sequence similarity with the phosphate binding site of E. coli
triosephosphate isomerase (21), and are boxed in blue in Figure 7 and 10B. Limited
proteolysis with trypsin has been performed on both the enterococcal and streptococcal enzymes with the resultant gel presented in Figure 10A (29). The arrows in Figure 10B show the trypsin cleavage sites, K368 and K404. These sites fall within the insert region found in these GlpOs which has led to the thought that this region could be a flexible surface loop region like the one seen for pig kidney DAAO (14).
The intact and nicked streptococcal GlpOs and the enterococcal GlpO were all
subjected to a series of spectral and biochemical experiments (29). They all showed the
19
Figure 9: E. coli and B. halodurans GlpD dimers. A superposition of monomers A of both GlpDs gives an RMSD ~ 2.3 Å. The E. coli GlpD (PDB code 2QCU) is colored deep purple and the B. halodurans GlpD (PDB code 3DA1) is colored forest green.
20
Figure 10: Limited proteolysis of Streptococcus sp. and E. casseliflavus GlpO. (A) A gel image shows the results of incubating E. casseliflavus GlpO (lanes 2-5) and Streptococcus sp. GlpO (lanes 7-9) with trypsin under native conditions. (B) A sequence alignment of two GlpOs and three GlpDs from bacterial and mitochondrial sources showing the presence of a ca. 50-residue insert found in GlpOs and not GlpDs. The blue boxes indicate residues within the proposed Glp-binding regions (21). The arrows show the trypsin cleavage sites for streptococcal GlpO (K368 and K404). Part (A) of this figure is reprinted with permission from (29). Copyright © 2000, American Chemical Society.
21 ability to stabilize a strong FAD N(5)-sulfite adduct (Kd = 1.5 mM for intact Strep. sp.,
1.3 mM for nicked Strep. sp., and 0.81 mM for E. casseliflavus) (29). Anaerobic titrations
with dithionite for the enterococcal enzyme resulted in the formation of small amounts of
both the anionic and neutral semiquinone; however, no semiquinone formation was seen
with either form of the streptococcal enzyme (29). Enzyme-monitored turnover (EMT)
experiments were carried out by a stopped-flow spectrophotometer for all three enzymes
(12, 29). EMT experiments are utilized to determine the steady-state kinetic properties of
enzymes. The stopped-flow spectrophotometer instrument mixes the contents of two
syringes, one consisting of the enzyme and the other consisting of the reaction substrates,
and, in this case, monitors the absorbance change of FAD. This technique is useful for
monitoring rapid reactions. The intact streptococcal GlpO differs from the enterococcal
enzyme in that the Km(O2) and kcat are ~2-fold lower and the Km(Glp) is 13-fold lower
(29). The nicked streptococcal GlpO shows values similar to those of the intact GlpO for
Km(O2) and kcat, but shows a 19-fold increase in Km(Glp) (29).
Kinetic analysis of the intact streptococcal GlpO has led to the following kinetic
mechanism:
k1 k3 E-FAD + Glp E-FADH2 DHAP E-FADH2 + DHAP k2 k4
k5 E-FADH2 + O2 E-FAD + H2O2
Scheme 1: Kinetic mechanism for intact Streptococcus sp. GlpO (29).
Kinetic analysis of the reductive half reaction, an anaerobic experiment only looking at the reduction of FAD and not the reoxidation, of the nicked form of the streptococcal
22 GlpO showed a kslow that is independent of Glp concentration (29). This has led to the thought that there are two forms of the resting oxidized enzyme of the nicked, but not intact, GlpO and the following mechanism:
k k [Glp] * 10 1 E-FAD E-FAD E-FADH2 DHAP k11 k2
Scheme 2: Kinetic mechanism for the [Glp] independent reaction seen in the nicked form of Streptococcus sp. GlpO (29).
This species was also seen for the E. casseliflavus GlpO (29).
Initial attempts at solving the crystal structure of the streptococcal GlpO led to a data set collected (38), but attempts to solve the phase problem by multiwavelength anomalous dispersion (MAD) (39) using a selenomethionine (SeMet) derivative (40) were unsuccessful and there were no viable structures to use as a search model for molecular replacement (MR) (41). Using the limited proteolysis as a guide (29), a deletion mutant, GlpOΔ, was genetically engineered, which lacked the insert region. This region is believed to be a flexible surface loop region, similar to that seen for pig kidney
DAAO (14), and could be hindering further crystal formation. Biochemical analysis of the GlpOΔ mutant shows that it binds FAD and is active, with increased Km(O2) and
Km(Glp) and similar kcat. This makes this enzyme a good candidate for crystallization experiments since it is catalytically active and lacks the flexible region believed to be hindering further crystallization studies.
Statement of Purpose. A major distinction between flavoproteins deals with the classification of a flavoenzyme as an oxidase or dehydrogenase. This distinction deals with the reactivity FAD has with O2 (1) and the resultant product. Glycerol metabolism
23 makes it possible for bacteria to grow on glycerol (16). An integral step in this process involves the oxidation of Glp to DHAP, which can enter into the glycolytic pathway. This reaction is catalyzed by GlpD, a dehydrogenase, in certain bacteria; however, heme- deficient lactic acid bacteria, and other bacteria such as M. mycoides and M. pneumoniae, utilize GlpO, an oxidase, to catalyze this reaction (16, 30, 31). Mitochondrial GlpDs aid in shuttling reducing equivalents from NADH into the mitochondrial respiratory chain
(24). GlpDs and some GlpOs have been shown to be membrane associated, but to different extents (28, 31-34, 42). The experiments described in this dissertation are designed to investigate both the structural and catalytic properties of the Streptococcus sp. GlpO enzyme that has high sequence identity to both bacterial and also mitochondrial
GlpDs that are membrane associated. The x-ray crystal structure of this enzyme will lead to the identification of residues important in substrate binding and catalysis. Chapter II of this dissertation describes the 2.3 Å resolution crystal structure determined for GlpOΔ and the 2.4 Å resolution crystal structure of the intact GlpO, from the data previously collected (38).
Analysis of these structures led to the identification of an arginine residue within the active site that is not only conserved in GlpOs, but is also conserved in two structural homologs, ThiO and DAAO (5, 15). Chapter III of this dissertation describes mutational and kinetic experiments designed to test the hypothesis that this arginine residue (R346) aids in binding and orienting the Glp substrate through interactions with the C(1)-OH, similar to the way this residue aids in binding and orienting the substrates of both structural homologs through interactions with the carboxylate moieties of their respective substrates (5, 15). Structural comparisons also led to the discovery of residues within the
24 active site that are conserved in GlpOs and GlpDs but not present in the structural homologs. Chapter III of this dissertation also describes mutational and kinetic experiments designed to test the hypothesis that these residues (H65, R69, Y70, and
K429) are involved substrate binding through interactions with the Glp-phosphate moiety. An active-site base must be present to abstract a proton from the C(2)-OH position of the Glp substrate to drive the hydride transfer reaction. Often times a histidine residue facilitates this mechanism (43). By mutating H65 to a glutamine and alanine, the hypothesis that H65 could be the active-site base present in GlpO was tested. The structural and biochemical experiments described in this dissertation have led to the formulation of a reaction scheme for binding and catalysis of GlpO.
25 REFERENCES
1. Massey, V., and Hemmerich, P. (1980) Active-site probes of flavoproteins, Biochem Soc Trans 8, 246-257.
2. Molla, G., Porrini, D., Job, V., Motteran, L., Vegezzi, C., Campaner, S., Pilone, M. S., and Pollegioni, L. (2000) Role of arginine 285 in the active site of Rhodotorula gracilis D-amino acid oxidase. A site-directed mutagenesis study, J Biol Chem 275, 24715-24721.
3. Molla, G., Motteran, L., Job, V., Pilone, M. S., and Pollegioni, L. (2003) Kinetic mechanisms of glycine oxidase from Bacillus subtilis, Eur J Biochem 270, 1474- 1482.
4. Zhao, G., Bruckner, R. C., and Jorns, M. S. (2008) Identification of the oxygen activation site in monomeric sarcosine oxidase: role of Lys265 in catalysis, Biochemistry 47, 9124-9135.
5. Settembre, E. C., Dorrestein, P. C., Park, J. H., Augustine, A. M., Begley, T. P., and Ealick, S. E. (2003) Structural and mechanistic studies on ThiO, a glycine oxidase essential for thiamin biosynthesis in Bacillus subtilis, Biochemistry 42, 2971-2981.
6. Monaghan, P. J., Leys, D., and Scrutton, N. S. (2007) Mechanistic aspects and redox properties of hyperthermophilic L-proline dehydrogenase from Pyrococcus furiosus related to dimethylglycine dehydrogenase/oxidase, Febs J 274, 2070- 2087.
7. Fagan, R. L., and Palfey, B. A. (2009) Roles in binding and chemistry for conserved active site residues in the class 2 dihydroorotate dehydrogenase from Escherichia coli, Biochemistry 48, 7169-7178.
8. Gibson, Q. H., Massey, V., and Atherton, N. M. (1962) The nature of compounds present in mixtures of oxidized and reduced flavin mononucleotides, Biochem J 85, 369-383.
9. Swinehart, J. H. (1966) Electron transfer in the flavin mononucleotide system, J Am Chem Soc 88, 1056-1058.
26 10. Ehrenberg, A., Muller, F., and Hemmerich, P. (1967) Basicity, visible spectra, and electron spin resonance of flavosemiquinone anions, Eur J Biochem 2, 286- 293.
11. Massey, V., and Palmer, G. (1966) On the existence of spectrally distinct classes of flavoprotein semiquinones. A new method for the quantitative production of flavoprotein semiquinones, Biochemistry 5, 3181-3189.
12. Parsonage, D., Luba, J., Mallett, T. C., and Claiborne, A. (1998) The soluble α- glycerophosphate oxidase from Enterococcus casseliflavus. Sequence homology with the membrane-associated dehydrogenase and kinetic analysis of the recombinant enzyme, J Biol Chem 273, 23812-23822.
13. Massey, V. (2000) The chemical and biological versatility of riboflavin., Biochem. Soc. Trans. 28, 283-296.
14. Mattevi, A., Vanoni, M. A., Todone, F., Rizzi, M., Teplyakov, A., Coda, A., Bolognesi, M., and Curti, B. (1996) Crystal structure of D-amino acid oxidase: a case of active site mirror-image convergent evolution with flavocytochrome b2, Proc Natl Acad Sci U S A 93, 7496-7501.
15. Umhau, S., Pollegioni, L., Molla, G., Diederichs, K., Welte, W., Pilone, M. S., and Ghisla, S. (2000) The x-ray structure of D-amino acid oxidase at very high resolution identifies the chemical mechanism of flavin-dependent substrate dehydrogenation, Proc Natl Acad Sci U S A 97, 12463-12468.
16. Lin, E. C. (1976) Glycerol dissimilation and its regulation in bacteria, Annu Rev Microbiol 30, 535-578.
17. Fu, D., Libson, A., Miercke, L. J., Weitzman, C., Nollert, P., Krucinski, J., and Stroud, R. M. (2000) Structure of a glycerol-conducting channel and the basis for its selectivity, Science 290, 481-486.
18. Heller, K. B., Lin, E. C., and Wilson, T. H. (1980) Substrate specificity and transport properties of the glycerol facilitator of Escherichia coli, J Bacteriol 144, 274-278.
19. Yeh, J. I., Charrier, V., Paulo, J., Hou, L., Darbon, E., Claiborne, A., Hol, W. G., and Deutscher, J. (2004) Structures of enterococcal glycerol kinase in the absence
27 and presence of glycerol: correlation of conformation to substrate binding and a mechanism of activation by phosphorylation, Biochemistry 43, 362-373.
20. Esders, T. W., and Michrina, C. A. (1979) Purification and properties of L-α- glycerophosphate oxidase from Streptococcus faecium ATCC 12755, J Biol Chem 254, 2710-2715.
21. Austin, D., and Larson, T. J. (1991) Nucleotide sequence of the glpD gene encoding aerobic sn-glycerol 3-phosphate dehydrogenase of Escherichia coli K- 12, J Bacteriol 173, 101-107.
22. Davenport, R. C., Bash, P. A., Seaton, B. A., Karplus, M., Petsko, G. A., and Ringe, D. (1991) Structure of the triosephosphate isomerase- phosphoglycolohydroxamate complex: an analogue of the intermediate on the reaction pathway, Biochemistry 30, 5821-5826.
23. Eto, K., Tsubamoto, Y., Terauchi, Y., Sugiyama, T., Kishimoto, T., Takahashi, N., Yamauchi, N., Kubota, N., Murayama, S., Aizawa, T., Akanuma, Y., Aizawa, S., Kasai, H., Yazaki, Y., and Kadowaki, T. (1999) Role of NADH shuttle system in glucose-induced activation of mitochondrial metabolism and insulin secretion, Science 283, 981-985.
24. Bender, K., Newsholme, P., Brennan, L., and Maechler, P. (2006) The importance of redox shuttles to pancreatic beta-cell energy metabolism and function, Biochem Soc Trans 34, 811-814.
25. Ou, X., Ji, C., Han, X., Zhao, X., Li, X., Mao, Y., Wong, L. L., Bartlam, M., and Rao, Z. (2006) Crystal structures of human glycerol 3-phosphate dehydrogenase 1 (GPD1), J Mol Biol 357, 858-869.
26. Hedley, P. L., Jorgensen, P., Schlamowitz, S., Moolman-Smook, J., Kanters, J. K., Corfield, V. A., and Christiansen, M. (2009) The genetic basis of Brugada syndrome: A mutation update, Hum Mutat 30, 1256-1266.
27. Berg, J. M., Tymoczko, J. L., and Stryer, L. (2002) Biochemistry, 5 ed., W. H. Freeman and Company, New York, NY.
28. MacDonald, M. J., and Brown, L. J. (1996) Calcium activation of mitochondrial glycerol phosphate dehydrogenase restudied, Arch Biochem Biophys 326, 79-84.
28 29. Charrier, V., Luba, J., Parsonage, D., and Claiborne, A. (2000) Limited proteolysis as a structural probe of the soluble α-glycerophosphate oxidase from Streptococcus sp, Biochemistry 39, 5035-5044.
30. Pilo, P., Vilei, E. M., Peterhans, E., Bonvin-Klotz, L., Stoffel, M. H., Dobbelaere, D., and Frey, J. (2005) A metabolic enzyme as a primary virulence factor of Mycoplasma mycoides subsp. mycoides small colony, J Bacteriol 187, 6824-6831.
31. Hames, C., Halbedel, S., Hoppert, M., Frey, J., and Stulke, J. (2009) Glycerol metabolism is important for cytotoxicity of Mycoplasma pneumoniae, J Bacteriol 191, 747-753.
32. Schryvers, A., Lohmeier, E., and Weiner, J. H. (1978) Chemical and functional properties of the native and reconstituted forms of the membrane-bound, aerobic glycerol-3-phosphate dehydrogenase of Escherichia coli, J Biol Chem 253, 783- 788.
33. Walz, A. C., Demel, R. A., de Kruijff, B., and Mutzel, R. (2002) Aerobic sn- glycerol-3-phosphate dehydrogenase from Escherichia coli binds to the cytoplasmic membrane through an amphipathic α-helix, Biochem J 365, 471-479.
34. Yeh, J. I., Du, S., Tortajada, A., Paulo, J., and Zhang, S. (2005) Peptergents: peptide detergents that improve stability and functionality of a membrane protein, glycerol-3-phosphate dehydrogenase, Biochemistry 44, 16912-16919.
35. Spoering, A. L., Vulic, M., and Lewis, K. (2006) GlpD and PlsB participate in persister cell formation in Escherichia coli, J Bacteriol 188, 5136-5144.
36. Joseph, B., Mertins, S., Stoll, R., Schar, J., Umesha, K. R., Luo, Q., Muller- Altrock, S., and Goebel, W. (2008) Glycerol metabolism and PrfA activity in Listeria monocytogenes, J Bacteriol 190, 5412-5430.
37. Yeh, J. I., Chinte, U., and Du, S. (2008) Structure of glycerol-3-phosphate dehydrogenase, an essential monotopic membrane enzyme involved in respiration and metabolism, Proc Natl Acad Sci U S A 105, 3280-3285.
38. Finnerty, C. M., Charrier, V., Claiborne, A., and Karplus, P. A. (2002) Crystallization and preliminary crystallographic analysis of the soluble α- glycerophosphate oxidase from Streptococcus sp, Acta Crystallogr D Biol Crystallogr 58, 165-166.
29 39. Hendrickson, W. A. (1991) Determination of macromolecular structures from anomalous diffraction of synchrotron radiation, Science 254, 51-58.
40. Doublie, S. (1997) Preparation of selenomethionyl proteins for phase determination, Methods Enzymol 276, 523-530.
41. Rossmann, M. G. (1990) The molecular replacement method, Acta Crystallogr A 46 ( Pt 2), 73-82.
42. Klingenberg, M. (1970) Localization of the glycerol-phosphate dehydrogenase in the outer phase of the mitochondrial inner membrane, Eur J Biochem 13, 247- 252.
43. Holliday, G. L., Mitchell, J. B., and Thornton, J. M. (2009) Understanding the functional roles of amino acid residues in enzyme catalysis, J Mol Biol 390, 560- 577.
30 Chapter II
STRUCTURE OF α-GLYCEROPHOSPHATE OXIDASE FROM Streptococcus sp.: A
TEMPLATE FOR THE MITOCHONDRIAL α-GLYCEROPHOSPHATE
DEHYDROGENASE
Timothy Colussi, Derek Parsonage, William Boles, Takeshi Matsuoka, T. Conn Mallett,
P. Andrew Karplus, and Al Claiborne
The following manuscript is reproduced with permission from Biochemistry, volume 47, number 3, pages 965-977, 2008. Copyright 2008, American Chemical Society. Stylistic variations are due to the requirements of the journal.
Coordinates have been deposited with the Protein Data Bank under the file names 2RGH,
2RGO.
31 ABSTRACT
The FAD-dependent α-glycerophosphate oxidase (GlpO) from Enterococcus casseliflavus and Streptococcus sp. was originally studied as a soluble flavoprotein oxidase; surprisingly, the GlpO sequence is 30-43% identical to those of the α- glycerophosphate dehydrogenases (GlpDs) from mitochondrial and bacterial sources.
The structure of a deletion mutant of Streptococcus sp. GlpO (GlpOΔ, lacking a 50- residue insert that includes a flexible surface region) has been solved using multiwavelength anomalous dispersion data and refined at 2.3 Å resolution. Using the
GlpOΔ structure as a search model, we have also solved the intact GlpO structure, as refined at 2.4 Å resolution. The first two domains of the GlpO fold are most closely related to those of the flavoprotein glycine oxidase, where they function in FAD-binding and in substrate-binding, respectively; the GlpO C-terminal domain consists of two helix bundles and is not closely related to any known structure. The flexible surface region in intact GlpO corresponds to a segment of missing electron density that links the substrate- binding domain to a ββα element of the FAD-binding domain. In accordance with earlier biochemical studies (stabilizations of covalent FAD-N5-sulfite adduct and p-quinonoid form of 8-mercapto-FAD), Ile430-N, Thr431-N, and Thr431-OG are hydrogen-bonded to
FAD-O2α in GlpOΔ, stabilizing the negative charge in these two modified flavins and facilitating hydride transfer to FAD-N5 (from Glp) as well. Active-site overlays with the glycine oxidase-N-acetylglycine and D-amino acid oxidase-D-alanine complexes demonstrate that GlpOΔ Arg346 is structurally equivalent to Arg302 and Arg285, respectively; in both cases these residues interact directly with the amino acid substrate or
32 inhibitor carboxylate. The structural and functional divergence between GlpO and the bacterial and mitochondrial GlpDs is also discussed.
33 The FAD-linked oxidation of α-glycerophosphate (Glp)1 to DHAP occurs in at least three contexts. The first two are α-glycerophosphate dehydrogenases (GlpDs) from mitochondria [mitoGlpD (1-4)] and from bacteria [bactGlpD (5, 6)], clearly homologous enzymes having ca. 30-33% sequence identity. The third FAD-dependent Glp-oxidizing enzyme is α-glycerophosphate oxidase (GlpO), found predominantly (if not exclusively) in heme-deficient lactic acid bacteria such as Enterococcus casseliflavus [formerly
Streptococcus faecium (7, 8)]. GlpOs are homologous with the bact- and mitoGlpDs, having 30-43% sequence identity with them but being distinguished by a ca. 50-residue insert falling between two segments proposed by sequence to interact with the Glp substrate (9). The mitoGlpD is an important component of the Glp shuttle that serves to transfer NADH-derived reducing equivalents from the cytosol to the mitochondrion (10).
Mammalian mitoGlpD has also been implicated in insulin signaling (11, 12) and differs from yeast mitoGlpD and bactGlpDs in that it has a C-terminal ca. 100-residue extension that is homologous to calmodulin (3). Little is known of the structure and properties of the mammalian mitoGlpD because of its extreme insolubility and the difficulty of purifying it in good yield and in an unmodified form (2, 13). It has been localized to the outer surface of the inner mitochondrial membrane, with both the Glp-substrate and
3- Fe(CN)6 -acceptor sites facing the intermembrane space (14).
While all GlpDs are membrane associated, it is unclear in some cases whether they are peripheral or integral membrane proteins. For instance, the mammalian mitoGlpD has been localized to the outer surface of the inner membrane (14), but another analysis predicts it to be an integral membrane protein with three transmembrane helices
(4). Similarly, Escherichia coli GlpD was reported to require detergent solubilization,
1 Abbreviations: see page v.
34 leading to the conclusion that it was an integral membrane protein (15), but later a His- tagged version of E. coli GlpD was purified without detergent solubilization (16). Still another recent report referred to the E. coli GlpD as an integral membrane flavoenzyme with "six transmembrane spanning" regions (17). The Bacillus subtilis GlpD is found predominantly (~75%) in the cytosolic fraction following ultracentrifugation of an extract from glycerol-grown cells (18).
Functionally, GlpO differs from GlpD in its efficient catalytic reduction of O2 →
6 -1 -1 H2O2, with the E. casseliflavus enzyme having kcat/Km(O2) ~10 M s (9); also, GlpO is clearly a dimeric (8), cytosolic protein (8, 19). Because GlpO also shows
"dehydrogenase" activities that are similar to those of E. coli GlpD (8, 15), with acceptor
3- substrates such as Fe(CN)6 , there is great potential for using the better-behaved GlpO as a structural and even functional template for the GlpDs. Studies of E. casseliflavus GlpO revealed an active-site environment similar to those of D-amino acid oxidase (DAAO) and other flavoprotein oxidases: the enzyme stabilizes the blue p-quinonoid form of 8- mercapto-FAD (8), it reacts reversibly with sulfite to form a covalent FAD-N5-adduct
(8), and the apoprotein raises the redox potential of bound FAD by 100 mV relative to that of the free coenzyme (9). Stopped-flow analyses of the reductive half-reaction also indicate that the resting oxidized enzyme exists in two states (E and E*), only one of which productively binds Glp for reduction (9). A limited proteolysis study with
Streptococcus sp. GlpO (20) demonstrated that the ca. 50-residue insert unique to the
GlpOs was largely excised, with the nicked enzyme (40 kDa and 23 kDa fragments remain associated) retaining catalytic activity with an ~20-fold increase in Km(Glp) and
35 an eight-fold decrease in kred. Interestingly, whereas intact Streptococcus sp. GlpO has no E* form, for the nicked enzyme E* is near 50%.
The relative paucity of structural information for Glp recognition and binding together with its central role in an important mitochondrial shuttle and electron transport
(10) led us to initiate a structural study of GlpO from Streptococcus sp. (21). Here, we report the crystal structures of both GlpO∆, a GlpO deletion mutant lacking the 50- residue insert, and of intact GlpO, refined at 2.3 and 2.4 Å resolution, respectively. Of particular interest are the two respective active-site conformations, one of which appears to be unproductive for catalysis and may represent the E* state.
36 EXPERIMENTAL PROCEDURES
Expression and Purification of GlpO∆. The Streptococcus sp. glpO gene (20), contained in a plasmid generously provided by Asahi Kasei Pharma, was amplified by
PCR and cloned into pQE30 (Qiagen) to express the N-terminal His-tagged protein. The
3'-end of the gene was PCR-amplified using a primer bridging the required deletion in combination with a downstream primer. The resulting PCR product was purified and served as a primer, along with a primer from the 5'-end of the gene, to PCR the entire glpOΔ coding sequence, which was then cloned into pQE30. After sequencing the gene to confirm both the required deletion and the absence of other mutations, the plasmid was transformed into E. coli B834 cells containing pREP4 (Qiagen) for expression of GlpOΔ.
These cells were grown at 37°C in a 3-L culture of MOPS-based defined medium (EZ
Rich Defined Medium, Teknova) containing 25 μg/mL kanamycin and 100 μg/mL ampicillin, with GlpOΔ expression being induced by adding 0.5 mM isopropyl-β-D- thiogalactopyranoside at an A600 of 0.9. After induction the temperature was lowered to
25°C, and growth was continued overnight. GlpOΔ was purified using a combination of
Q-Sepharose and Ni-NTA columns as described by Nicely et al. (22), except that cells were lysed using an Aminco French Press, and β-mercaptoethanol was omitted from the column buffers. Pure protein was buffer-exchanged into 50 mM potassium phosphate, pH 7.0, containing 1 mM EDTA, and concentrated to 10 mg/mL before freezing in aliquots at -70 °C. The SeMet protein was expressed and purified in the same way, but with SeMet replacing Met in the growth medium.
Spectral Analyses of GlpOΔ and Stopped-Flow Kinetics. Titrations with dithionite and sulfite and other static spectral measurements followed established
37 -1 -1 protocols (9, 20), using an experimentally-determined ε450 = 11,600 M cm for purified
GlpOΔ. Steady-state kinetic parameters for GlpOΔ were determined in enzyme- monitored turnover experiments, carried out with the Applied Photophysics DX.17 MV stopped-flow spectrophotometer as described earlier (9, 20). Kinetic data were analyzed as described by Cornish-Bowden (23), as follows.
ν = Vab/(KiAKmB + KmBa + KmAb + ab) where A and B represent Glp and O2, respectively. Primary plots of b/v versus b intersect to the left of the y-axis in the case of an ordered ternary-complex mechanism. A secondary plot of (slope x a) versus a gives a y-axis intercept of KmA/Vmax and has a slope equal to 1/Vmax. A secondary plot of (intercept x a) versus a gives a y-axis intercept of
KiAKmB/Vmax and has a slope equal to KmB/Vmax.
Crystallization and Data Collection. Purification, and crystallization and data collection for native intact Streptococcus sp. GlpO have been described in earlier reports
(20, 21). Thawed aliquots of the SeMet GlpO∆ protein were buffer-exchanged into 1 mM sodium N-(2-hydroxyethyl)piperazine-N′-2-ethanesulfonic acid, pH 7.2, to give a final protein concentration of 10 mg/mL. Crystals were grown in 24-well sitting-drop plates over reservoirs of 0.5 mL, using drop sizes of 4 + 4 μL. Large single crystals of
SeMet GlpO∆ grew best using a reservoir solution of 0.2 M Na2SO4, 17-22% PEG3350, and 10 mM β-mercaptoethanol, with glycerol (20-35%) present as the cryoprotectant.
Crystals appeared within one day, growing to full size in two-three days, and were flash- frozen in a nitrogen stream at 100 K directly after being removed from the drop. Data sets for MAD phasing were collected at National Synchrotron Light Source beamline
X26C. Although we have no evidence that any flavin reduction occurred during
38 synchrotron data collection with either intact GlpO or GlpO∆ crystals, we cannot eliminate this possibility. Crystals of native GlpO∆ were obtained under very similar conditions, omitting glycerol and β-mercaptoethanol; cryoprotectant soaks using a solution of 0.3 M Na2SO4, 25% PEG3350, and 15% glycerol were optimized for these
GlpO∆ crystals.
GlpO∆ Phasing and Structure Refinement. Images in each MAD data set were indexed, integrated, and scaled in d*TREK (24); data collection and phasing statistics are summarized in Table 1. SOLVE (25) was used to locate all 12 of the Se sites; density modification was performed with RESOLVE (26, 27). The resulting electron density map led to an initial model which was subjected to refinement with REFMAC5 (28).
Water molecules, identified with COOT (29) in the Fo-Fc map using a 3σ difference
Fourier cutoff, were visually confirmed and added at this stage. Also, one clear sulfate ion was added to the model. As refinement progressed, the electron density indicated that a bent conformation for the flavin isoalloxazine, along the FAD-N5-N10 axis, was required. The butterfly conformation was added and refined in CNS (30); final refinement of the full model was also performed with CNS, and refinement statistics for the GlpO∆ structure are summarized in Table 2.
Intact GlpO Structure Solution and Refinement. Since the intact GlpO differs from GlpO∆ only by the absence of the His-tag and the presence of Asp356-Ala405
(essentially, the 50-residue insert sequence), the GlpO∆ coordinates (minus the FAD, sulfate ion, and solvent waters) were used as the search model for the intact GlpO molecular replacement solution [using PHASER (31, 32)] to locate two molecules in the asymmetric unit, using data from 30-2.50 Å resolution. Phasing and density modification
39 Table 1: Data Collection for GlpO and GlpO∆ and Phasing Statistics for GlpO∆
GlpO∆b GlpOa edge peak remote Wavelength (Å) 0.9296 0.9790 0.9783 0.960
c Space group P21 P21212
Cell dimensions a, b, c (Å) 128.36, 106.78, 58.89 86.21, 95.48, 77.61 α, β, γ (˚) 90, 99.18, 90 90, 90, 90
Resolution (Å) 2.40 2.10 2.10 2.10
Reflections 141,187 248,145 238,765 233,374
Unique reflections 57,965 37,395 36,799 37,759
Completeness (%) 94.8 (97.4)d 97.7 (83.7) 96.5 (85.4) 98.7 (89.6)
e Rsym (%) 6.4 (29.8) 7.4 (32.4)
I/σ 9.9 (2.8) 12.9 (4.0)
Phasing powerf 0.9 1.26
Resolution range (Å) 41.78-2.10 41.76 – 2.10 41.79-2.10
Figure of meritg 0.60
aCollected with crystals of native intact GlpO at the Cornell High Energy Synchrotron Source F-1 station using an ADSC Quantum-4 CCD detector (21). bCollected with crystals of SeMet GlpO∆ at beamline X26C of the National Synchrotron Light Source using an ADSC Quantum-4 CCD detector. cThe space group assignment given in Ref 21 is incorrect; the correct space group is d e P21. Numbers in parentheses represent data for the highest resolution shell. Rmeas for native intact GlpO as taken from Ref 21. fCalculated to 2.9 Å resolution. gComputed after solvent flattening using RESOLVE.
40 Table 2: Crystallographic Refinement Statistics for GlpO and GlpO∆
GlpO a GlpO∆ Resolution range (Å) 12.0-2.40 41.76-2.30 Amplitude cutoff none none Number of amino acid residues 1087 546 Number of waters 232 271 Number of sulfate ions 1 Number of total non-hydrogen atoms 8870 4618 R-factor (%) 23.8 21.0
R-free (%) 24.6 25.3 Stereochemical ideality bond length rmsd (Å) 0.012 0.015 bond angle rmsd (deg) 1.7 1.3 φ,ψ most favored (%) 85.0 89.1 φ,ψ additional allowed (%) 14.6 10.5 φ,ψ generously allowed (%) 0.3 0.4 aIntact GlpO dimer.
41 were also performed with PHASER. Early rounds of refinement extending out to the limit of resolution (2.4 Å) were performed with CNS, with solvent waters being added as for GlpOΔ. In accordance with the electron density appearance, a butterfly flavin conformation was added for chain A, but the chain B flavin was left planar. The latter stages of refinement were carried out with REFMAC5.
42 RESULTS AND DISCUSSION
Kinetic and Redox Properties of GlpO∆. The GlpO∆ deletion mutant, lacking residues Asp356-Ala405, was designed to test the role of the ca. 50-residue insert that is present in GlpO but not GlpD enzymes (9). GlpO∆ differs from nicked GlpO (20) in that the limited proteolysis cleavage sites correspond to Lys368 and Lys404; nicked GlpO therefore includes 14 additional residues (Asp356-Lys368 plus Ala405) not present in
GlpO∆. In dithionite and sulfite titrations (Figure 1), GlpO∆ behaves similarly to both intact and nicked enzymes (20). The tightly-bound flavin is reduced directly on titration with 1.1 equiv of dithionite, and there is no evidence for any stable semiquinone intermediate. Sulfite titration of GlpO∆ gives the stable flavin N(5)-sulfite adduct, with a
Kd of 0.92 mM at pH 7.0, 25˚C. Enzyme-monitored turnover analyses with GlpO∆
(Figure 2) [10 μM enzyme and 0.77 mM O2 over the Glp concentration range of 2.5-50 mM (pH 7.0, 5˚C)] demonstrate that His-tagged GlpO∆ is catalytically active, with kcat =
-1 11.2 + 0.5 s and Km(Glp) = 6.6 + 1.0 mM (Table 3). Interestingly, whereas the kcat is similar to that of nicked GlpO, the Km(Glp) is 5.5-fold lower, leading to a 4.4-fold increase in the GlpO∆ specificity constant (Table 3). The higher Km(Glp) of nicked
GlpO compared with GlpO∆, even though nicked GlpO retains more residues, is surprising and suggests that the 14 additional residues in nicked GlpO do not contribute substantially to catalysis, but actually appear to somehow interfere with Glp-binding and/or catalysis. One possible interpretation, for example, would invoke the larger E* component (9, 20) in the resting nicked enzyme population.
Structure Solutions of GlpO∆ and Intact GlpO. Given the absence of suitable search models for a molecular replacement solution with the 2.4 Å data set collected with
43 Table 3: Steady-State Kinetic Constants for Intact and Nicked Forms of Streptococcus sp. GlpO and for GlpO∆ term intacta nickeda GlpO∆b Km(Glp) (mM) 1.9 36.2 6.6 Km(O2) (μM) 52 69 460 -1 kcat (s ) 17.9 14.1 11.2 -1 -1 3 3 kcat/Km(Glp) (M s ) 9.4 x 10 390 1.7 x 10 aRef 20. bCalculated from the data in Figure 2.
44
Figure 1: Dithionite and sulfite titrations of GlpO∆. (A) The enzyme (40.8 μM, in 0.8 mL of 50 mM potassium phosphate, pH 7.0, plus 0.5 mM EDTA) was titrated with a 3.9 mM solution of dithionite. Spectra shown, in order of decreasing absorbance at 450 nm, correspond to the addition of 0 (―), 0.16 (–·–), 0.48 (– –), 0.8 (- - -), and 1.28 (···) equiv of dithionite/FAD. Inset: Absorbance change at 450 nm versus added dithionite. The end point corresponds to 1.14 equiv of dithionite/FAD. (B) The enzyme (39.1 μM, prepared as above) was titrated with sodium sulfite solutions of 0.1 and 1.0 M. Spectra shown, in order of decreasing absorbance at 450 nm, correspond to the addition of 0 (―), - 2- 0.25 (–·–), 0.75 (– –), 2 (- - -), and 17.5 mM (···) total sulfite (HSO3 + SO3 ). Inset: Double-reciprocal plot of absorbance change at 450 nm versus added sulfite concentration.
45
Figure 2: Enzyme-monitored turnover of GlpO∆. The enzyme (9.9 μM after mixing, in air-equilibrated 0.1 M phosphate, pH 7.0, plus 0.5 mM EDTA) was mixed at 5˚C with different concentrations of Glp [2.5 (―), 5 (–·–), 10 (– –), and 50 mM (- - -)] that had been equilibrated with 100% O2 in the same buffer. The reactions were followed at 450 nm.
46 the native Streptococcus sp. GlpO (21), we prepared the SeMet derivatives of N-terminal
His-tagged and wild-type (non His-tagged) intact GlpO, but neither of these allowed us to reproduce the earlier-reported (21) crystal form. However, with the SeMet derivative of
His-tagged GlpO∆, novel well-diffracting crystals were grown, and a three-wavelength
MAD data set was collected to solve the structure (Table 1). The crystals have a single molecule in the asymmetric unit, and the final GlpO∆ model (refined using data to 2.30 Å resolution) comprises residues 0-350 (one residue from the His-tag precedes the wild- type initiating Met) and 412-606, the FAD cofactor, and ordered solvent; it has an R- factor of 21% (R-free = 25.3%) with good geometry (Table 2). Missing from the model and presumed mobile are the first 13 residues of the N-terminal His-tag, residues 351-355 and 406-411 bordering the two ends of the deletion, and the C-terminal residue Lys607.
The crystal structure of intact GlpO, solved by molecular replacement using
GlpO∆, has two chains in the asymmetric unit. Refinement at 2.4 Å resolution led to a final model with reasonable statistics (Table 2). Due to lack of electron density, chain A is missing residues 361-407, 518, and the C-terminal residues 606-607; chain B, apparently less well-tethered, is missing electron density for residues 16-17, 111-112,
203, 245-246, 361-408, 476-494, and 516-518. The common major disordered segment
(residues ~361-407) corresponds well with the region deleted in GlpO∆. All seven short segments missing in chain B are involved in crystal contacts in chain A, explaining their better definition. The two subunits have an overall Cα rmsd of just 0.5 Å; this agrees well with the expected coordinate errors (33) of 0.33 and 0.38 Å for the well-ordered parts of the GlpO∆ and intact GlpO models, respectively. Detailed descriptions of the intact structure will focus on chain A, with residues from chain B being designated by a prime
47 symbol (e.g., His65 = chain A; His65′ = chain B). The 2Fo-Fc maps for the respective
FAD coenzymes illustrate the quality of electron density for well-ordered regions of both structures (Figure 3).
Overall Structure. Comparison of the molecular packing in the GlpO and
GlpO∆ crystal forms reveals one common packing interaction that buries the most surface, and we interpret this to represent the physiologically relevant dimer (Figure 4A); both refined structures show the same overall chain-fold (Figure 4B). Each chain (taking the GlpO∆ structure as the point of reference) is organized as three compact domains
[Figure 5A, as assigned with DSSP (34)]: domain 1 – an FAD-binding domain that consists of three chain segments (residues 1-58, 173-254, and 412-452) and has a central six-stranded β-sheet; domain 2 – a dimerization and putative substrate-binding domain that consists of two chain segments (residues 59-172 and 255-350) and has a mixed, eight-stranded β-sheet, and domain 3 – a C-terminal domain that is a single chain segment (residues 453-606) and is largely α-helical. The proteolytically labile ~50- residue insert unique to GlpO enzymes (9) and deleted in GlpO∆ is, as expected, not present as an integral part of the structure, but is a largely disordered segment that bridges the second segment of domain 2 to the third segment of domain 1 (Figure 5A). The distance (Cα-to-Cα) between residues 349 and 413 that must be bridged by the segment is only 7.4 Å. This short distance leads us to suspect that the insert may be a segment that is a small folded domain on its own, but has few interactions with the rest of the protein and thus is not fixed in the crystal structure. In terms of the overall structure, there are two related differences between intact GlpO and GlpO∆. The first is that GlpO has additional ordered residues (356-360) corresponding to positions 2-6 of the 50-residue
48
Figure 3: Final 2Fo-Fc maps for the FAD redox centers of GlpO∆ (A), intact GlpO chain A (B), and intact GlpO chain B (C), together with the refined models. Carbon atoms are colored yellow; others are color-coded by atom type. A portion of the FAD in (A) has been omitted for clarity. The depicted contour level is 1σ.
49
Figure 4: The GlpO∆ dimer, and superposition of a GlpO∆ monomer with intact GlpO chain A. (A) One subunit (top) of the GlpO∆ dimer is color-coded by domain: FAD- binding, sand; dimerization/substrate-binding, blue; and C-terminal, red. Both FAD are color-coded as in Figure 3, and helical elements providing the main interactions in the dimer interface are labeled. (B) Superposition, in stereo, based on the FAD-binding domain gives rmsd = 1.1 Å for a GlpO∆ monomer (color-coded as in A) and intact GlpO chain A (green). Both FAD are color-coded as in A. An ~3 Å shift in a portion of domain 2 (helical elements indicated by arrows) accompanies insertion of the additional loop and short helix (circled) near the flavin of intact GlpO.
50 Figure 5: Topology diagram for GlpO∆, and superposition of GlpO∆ domains 1 and 2 with a ThiO monomer. (A) Each secondary structural element, as assigned by DSSP (34), is color-coded by domain, as in Figure 4, and the respective first and last residue numbers are labeled. The single 310-helix shown corresponds to that in Figure 4. The location of the 50-residue insert, deleted in GlpO∆ and largely disordered in the intact GlpO structure, is also indicated (···). (B) Superposition, in stereo, of GlpO∆ domains 1 and 2 (color-coded as in Figure 4) and a ThiO monomer (cyan). DALILITE gives rmsd = 3.0 Å for 330 Cα atoms with 19% sequence identity (Z-score = 28.2). Both FAD are color-coded as in Figure 3; secondary structural elements described in the text (with the exception of α1) are labeled. (C) Stereoview of the GlpO∆ C-terminal domain. The orientation, relative to that of domains 1 and 2 in B, is given in Figure 4B. Cα positions corresponding to the N- and C-termini (Gly468 and Glu606, respectively) are labeled. This domain, which consists of eight α-helices organized into two bundles, is absent in ThiO and other DAO pfam members.
51
52 insert, and these residues (plus residues 351-355) form a loop and short helix that sits just above the flavin in the cleft between domains 1 and 2 (Figure 4B) and represents one significant difference. Associated with the insertion of these 10 residues into the cleft is the shift by about 3 Å of a portion of domain 2 that includes helices 3 and 4. As discussed in detail in a later section, we interpret the more "open" GlpO∆ conformation as equivalent to the fully active resting oxidized enzyme (E) form; we interpret the more
“closed” intact GlpO structure as representative of the E* form.
In a search for structurally similar proteins (35), the top four scores correspond to
B. subtilis glycine oxidase [ThiO; rmsd = 3.0 Å, 19% identity, Z-score 28.2 (36)], the
Pyrococcus horikoshii L-proline dehydrogenase β-subunit [rmsd = 3.1 Å, 19% identity,
Z-score 27.9 (37)], the β-subunit of Pseudomonas maltophilia heterotetrameric sarcosine oxidase [rmsd = 3.3 Å, 14% identity, Z-score 25.3 (38)], and Bacillus sp. monomeric sarcosine oxidase [MSOX; rmsd = 3.4 Å, 14% identity, Z-score 24.1 (39)]. Pig kidney
D-amino acid oxidase [DAAO (40)] yields rmsd = 3.3 Å and 16% identity (Z-score 22.7).
All five of these proteins belong to the GR2 subfamily of FAD-dependent enzymes (41) and consist of two domains (FAD-binding and substrate-binding, or interface) that are similar to the first two domains of GlpO. In terms of backbone rmsd and catalytic mechanism, ThiO (36) appears to be the structurally-known homolog most similar to
GlpO and will be used for most comparisons below.
A topology diagram of the first two domains of GlpO (Figure 5A) reveals a fold that is remarkably similar to that of ThiO, in fact conserving every element of secondary structure (see Figure 5 of Ref 36). The spatial arrangement of the two domains is also quite similar (Figure 5B), suggesting that there is little new to describe about these two
53 domains, compared with ThiO. The one new and unique feature is the dimerization interface of GlpO. Although ThiO and DAAO both have quaternary structure, neither has the same interface as GlpO. The GlpO dimer interface is rather small, burying a total surface area of 640 Å2 per monomer (~3% of each monomer surface), with residues from the substrate-binding domain providing the bulk of the contact area. The main interactions involve residues from helix α4 and the loop connecting helices 310 and α3, which pack across the two-fold symmetry axis with the same structural elements of the other chain (Figure 4A).
In contrast to the first two domains, the structural similarity search also shows that the GlpO∆ C-terminal domain (Gly453-Glu606) is a novel fold. This domain is an aggregation of eight α-helices (Figure 5C) which are organized into a three-helix bundle
(α10-α11-α12) followed by a four-helix bundle (α13-α14/α15-α16-α17), where the two short bent helices 14 and 15 take the place of one longer helix. Both helical bundles have simple "up-down-up" topology; when viewed down α1 (from the N-terminus) helices
α17-α16-α13-α12-α10 form a surface-exposed helical skirt that covers the back of the
FAD-binding domain. Although the functional importance of this domain is unknown
(see below), it appears to be conserved in all known GlpO and GlpD sequences (Figure
6).
FAD-Protein Interactions. During refinement of the GlpO∆ structure, we noticed that the FAD isoalloxazines in both GlpO∆ and in intact GlpO chain A appear to be bent along the N5-N10 axis, with angles of ~12˚ (Figure 3). In contrast, the electron density for the GlpO chain B flavin, which has B-factors similar to those for the chain A flavin, could be well-fit with a planar flavin (Figure 3). At the 2.4 Å resolution of this analysis,
54 Figure 6: Structure-based sequence alignment for Streptococcus sp. GlpO and four known functional GlpOs and GlpDs from bacterial and mitochondrial sources (references given in the text). Residue numbering corresponds to both intact and deletion mutant forms of Streptococcus sp. GlpO, where GlpO∆ lacks residues 356-405. Residues 1-51 and 690-805 of the human mitochondrial GlpD (H.sapi.mt; full-length precursor sequence) have been omitted for clarity. Red and yellow blocks correspond to conserved residues and conservative substitutions, respectively. Boxes represent FAD-binding (sand) and putative Glp-binding (blue) segments, as defined earlier (20). Secondary structure assignments (color-coded by domain) correspond to GlpO∆. Green triangles correspond to those active-site residues shown in Figure 7 and to residues involved in both hydrogen-bonding and hydrophobic interactions with the FAD isoalloxazine in GlpO∆. The position of Tyr357 in the intact enzyme structure suggests that this active- site conformation represents the E* state.
55
56 we cannot conclude that the chain B flavin is perfectly planar, but we do conclude that the chain B flavin is much less bent than the others. Interestingly, none of ThiO, MSOX, or DAAO show bent flavins, so this appears to be an unusual feature of GlpO.
While similar bent butterfly flavin conformations are characteristic of reduced flavins and flavoproteins [e.g., reduced thioredoxin reductase (42) and the reduced PutA proline dehydrogenase domain (43)], there are fewer examples for oxidized flavoenzymes. Lennon et al. (42) identified three cases in which the oxidized isoalloxazine bending angle is greater than 18˚ – pyruvate oxidase (PDB code 1POW),
NADH oxidase (PDB code 1NOX), and trimethylamine dehydrogenase (PDB code
2TMD). Most recently Schiffer et al. (44) have identified a butterfly conformation for the oxidized FAD in adenosine-5′-phosphosulfate reductase; as with trimethylamine dehydrogenase, this has also been considered as an energetically favorable factor in catalysis. The bent oxidized flavin in GlpO may have a similar redox impact, as the
FAD/FADH2 redox potential for the E. casseliflavus enzyme has been measured at -118 mV (9), roughly 100 mV higher than that for free FAD (-219 mV). Since the redox potential for the Glp/DHAP couple is -190 mV (45), the effect of the GlpO apoprotein in providing a thermodynamically favorable system for Glp dehydrogenation includes an environment that favors the FAD-N1/O2α-anionic form of FADH2 as well as the butterfly conformation of oxidized FAD. Also, as studies of ThiO, MSOX, and DAAO indicate that the substrate binds at the flavin re-face (36, 39, 40), this “butterfly” bending of the flavin pyrimidine and dimethylbenzene rings away from the re-face would direct
FAD-N5 toward the Glp substrate and promote reactivity (see below).
57 The hydrogen-bonding interactions of GlpO∆ with FAD are rather similar to those seen for other GR2-subfamily flavoenzymes (41). The ADP portion of the FAD interacts, as for the classic nucleotide-binding fold (46), mostly via residues from the first
βαβ portion of the fold, including the pyrophosphate-interacting GxGxxG fingerprint
(here GGGITG) and a carboxylic acid side chain (here Glu48) at the end of the second β- strand (47). The adenine is bound in a pocket surrounded by Ile24, Met49, Ser232,
Trp235, and Val239. All of the interactions shown with the ADP portion of FAD are also seen in both chains A and B of intact GlpO, and the side chains involved are well- conserved in the bacterial GlpOs and GlpDs given in Figure 6.
In terms of interactions with the flavin itself, in GlpO∆ the main chain Ile430-N and Thr431-N hydrogen-bond with FAD-O2α as does Thr431-OG, consistent with the expectation of the stabilization of a negative charge at this locus (8, 9). Lys429-NZ sits just 3.5 Å above the FAD-C2 atom, so its positive charge will further stabilize anionic
FAD-N1/O2α forms. The FAD-N3 pKa for intact E. casseliflavus GlpO is decreased to
8.5 (from 10.4 for free FAD), entirely consistent with these observations (8). We have reported (20) that this pKa value appears to be further decreased for intact Streptococcus sp. GlpO, since very little visible absorbance change was observed during a spectral pH titration of the latter enzyme between pH 8.5 and 10. There is no significant change in this active-site pKa with the nicked Streptococcus sp. GlpO, and we similarly expect no significant change for GlpO∆. Other notable interactions with the flavin include hydrogen bonds with FAD-N5 by Thr61-OG (which in turn receives a hydrogen bond from His259-NE2) and with FAD-N3 by Leu63-O. In terms of nonpolar packing interactions, Ile430 (from the loop preceding helix α9) packs against the pyrimidine ring,
58 while Gly257, Gly344, and Leu345 (from strands β11 and β15) have similar interactions with the dimethylbenzene ring, showing that residues from domain 2 block solvent access in this region. The MSOX equivalent of β15 carries the Cys315 covalent attachment
(through FAD-C8α) site (39). As will be discussed later in the context of substrate binding, the side chains of Ser60, His65, Arg69, Tyr70, His259, Thr298, Arg346, and
Lys429 line the pocket just above the re-face of the flavin. Also in this region are three backbone amides: Gly67-N forms a hydrogen bond with His65-ND1, and both Ile68-N and Arg69-N hydrogen bond with waters. Interestingly, interactions between Ser60 and
Lys62-N stabilize a backbone conformation that allows Thr61-N to point directly at
FAD-C4a. ThiO (Ala47-N and Gly48-N) and pig kidney DAAO (Trp52-N) have similar interactions with the respective FAD-C4a; the high-resolution crystal structure of the yeast DAAO complex with D-Ala (48) has provided evidence for a covalent hydroperoxide, most likely bonded via FAD-C4a.
In intact GlpO (both chains A and B), the flavin environment is rather different because the insertion of residues 351-360 into the active-site pocket is associated with a shift in the flavin and the reorganization of surrounding residues (Figures 4B and 7). The bent oxidized flavins (intact chain A and GlpO∆) superimpose well in the vicinity of
FAD-C8, and they diverge to give separations of 2 to 3 Å of the pyrazine and pyrimidine moieties. This shift contributes to the different protein interactions with the latter, with the Ile430-N, Thr431-N, and Thr431-OG interactions in GlpO∆ being replaced by a
Lys429-NZ:FAD-O2α interaction in intact GlpO. Also, the Leu63-O:FAD-N3 hydrogen bond is not seen in intact enzyme. GlpO Tyr357 stacks neatly above the re-face of the central pyrazine ring, with a Tyr357-CE1:FAD-N10 distance of 3.4 Å. Tyr357 comes at
59
Figure 7: Active-site overlay, in stereo, for GlpO∆ and intact GlpO. The superposition was performed as described in Figure 4B. Intact GlpO protein and FAD are rendered in green (residues 351-360, including the short helical insert, are unique to the intact GlpO structure). GlpO∆ protein is color-coded by domain, as in Figure 4; the GlpO∆ FAD is color-coded by atom type. In this view, the 310 helices of both proteins (as shown in Figure 4B) are also depicted.
60 the N-terminus of the 50-residue insert (20), it is present in nicked GlpO, and it is deleted in GlpO∆. Intact GlpO chain A also reveals direct solvent water interactions with both
FAD-O4α and -N5; the latter interaction replaces that observed between Thr61-OG and
FAD-N5 in GlpO∆. Similarly, GlpO∆ lacks this direct solvent water interaction.
Glp-Binding and the Active Site. Austin and Larson (5) were the first to deduce sequence motifs for the "Glp-binding domain" in E. coli GlpD. These were based on limited sequence similarity between GlpD and E. coli triosephosphate isomerase, focused on the residues of the latter enzyme that interact with the phosphate of the bound DHAP substrate (49) [and with Glp (50)]. Interestingly, two of these sequence motifs, Ser341-
Ile350 and Gly427-Ile430, closely flank the 50-residue insert in GlpO. The GlpO∆ structure also shows that residues from these segments [Gly344, Leu345, Arg346, Ile430,
(and Thr431)] actually are part of the active site, interacting with or lining the pocket above the flavin.
Attempts to soak Glp into GlpO∆ crystals have led to cracked crystals that do not diffract, implying that some conformational change occurs during Glp binding and/or turnover. In the absence of a ligand-bound structure, and to gain insight into the substrate-binding mode and catalytic mechanism, we carried out overlays of GlpO∆
(representing the "open" conformation of GlpO) with the ThiO-N-acetylglycine and yeast
DAAO-D-Ala complexes. As seen in Figure 8, GlpO∆ Arg346, conserved in the GlpO and GlpD sequences, is remarkably structurally similar with the key residues ThiO
Arg302 (36) and yeast DAAO Arg285 (48), providing a direct link to these active sites.
Further active-site similarities include the main chain hydrogen-bonding interactions with the respective FAD-O2α (equivalent to GlpO∆ Ile430 and Thr431) and FAD-N3
61
Figure 8: Active-site overlay, in stereo, for GlpO∆, the ThiO-N-acetylglycine (NAG) complex (chain A), and the yeast DAAO-D-Ala complex. The superposition was performed based on the AMP moieties of the respective FAD. GlpO∆ side chains and FAD are color-coded by atom type, the ThiO-NAG complex is rendered in cyan, and the DAAO-D-Ala complex is colored magenta. GlpO∆ His65 and Lys429 have no structural equivalents in either ThiO or DAAO.
62 (equivalent to GlpO∆ Leu63). Differences are that a Tyr residue interacting with the substrate carboxylate in yeast DAAO (Tyr223), which is structurally similar to ThiO
Tyr246, is replaced by Phe286 in GlpO∆, leaving more space in this region. At the other end of the substrate-binding pocket, GlpO∆ residues His65, Arg69, Tyr70, and Lys429 have no equivalents in ThiO or DAAO but offer positively-charged/hydrogen-bonding side chains, well-positioned to interact with substrate and conserved in GlpO and GlpD sequences. Accordingly, based on the "open" GlpO∆ structure, we propose the following
Glp recognition and dehydrogenation scheme (Figure 9). As the Glp substrate approaches the flavin re-face, the Glp-phosphate is attracted to the pocket surrounded by
Lys429, His65, Arg69, and Tyr70 with Glp-C1-OH anchored by interaction with Arg346.
Early work with crude preparations of the mitoGlpD from insect flight muscle (51) showed that substitution of Glp-C1-OH with either -H (1-deoxy-Glp) or -F (1-deoxy-1- fluoro-Glp) led to a 60-80-fold decrease in kcat/Km for the respective substrate, demonstrating that Glp-C1-OH is indeed an important determinant for substrate recognition and catalysis.
The oxidation of the secondary alcohol function in Glp can be compared with the mechanisms for amino acid oxidation as catalyzed by other members of the DAO pfam
(52), such as ThiO, MSOX, and DAAO. For yeast DAAO, the structure of the complex with D-Ala at 1.20 Å resolution (48) clearly shows the substrate Cα-H pointing toward
FAD-N5, entirely consistent with the hydride transfer mechanism proposed by Mattevi et al. (40). There is no acid-base catalyst; the rate of D-Ala oxidation is seven orders of
+ magnitude greater than that for D-lactate, reflecting the relative pKa values of α-NH3 versus α-OH. A very similar mechanism has been suggested for ThiO (36), as the
63
Figure 9: Stereoview of the proposed Glp-binding pocket, including N-acetylglycine (NAG) from the ThiO overlay in Figure 8. GlpO∆ residues and FAD (yellow carbon atoms) are color-coded by atom type; NAG (cyan carbon atoms) is rendered as a 50% transparency. Three active-site WAT are colored red; hydrogen-bonding interactions and a well-aligned close approach (3.8 Å) of Thr61-N to FAD-C4a are described in the text. The respective first and last residues of each peptide segment are labeled, with the exception of Gly344.
64 structure of the N-acetylglycine complex positions Cα-H almost directly above FAD-N5 at a distance of 3.5 Å. No acid-base catalyst is present, and in both DAAO and ThiO, the anionic reduced FAD is stabilized by protein interactions with the -N1/O2α locus.
While earlier evidence implicated His269 (39, 53) and/or Tyr317 (54) as potential acid- base catalysts in the MSOX-substrate complex, more recent mutagenesis studies have demonstrated that ionization of the complex, as required for enzyme reduction, is independent of either residue. A hydride transfer mechanism has been considered for
MSOX and would seem to be preferred in light of these mutagenesis studies (54). While the hydride transfer mechanism preferred for GlpO (8, 9, 20) resembles those discussed above, the GlpO mechanism is unique in requiring ionization of the Glp-C2-OH. Recent structure-based mechanisms for the NAD+-linked (flavin-independent) Glp dehydrogenases from Leishmania mexicana (55) and from human liver (56) both implicate a Lys residue in ionization of Glp-C2-OH, thus promoting hydride transfer to the NAD+ acceptor. As to the identity of the proposed active-site base in GlpO, His65 is reasonably well-positioned to interact with and abstract the Glp-C2-OH proton, but the hydrogen-bonding interaction Gly67-N:His65-ND1 observed in the GlpO∆ structure implies that His65 is monoprotonated on -NE2. In this sense His65 could not serve as an acid-base catalyst, unless a conformational change disrupts that interaction.
GlpO∆ as a Structural Template for the GlpDs. Sequence alignments for GlpO∆ with the GlpDs from E. coli (5) and human mitochondria (57) give identities of ~30% over the entire GlpO∆ sequence; with B. subtilis GlpD (6) the identity is 45%. Now, with the GlpO∆ structure in hand, we can assess the level of active-site conservation and evaluate proposals for how mitoGlpD, in particular, is anchored to the membrane. As
65 seen in Figure 6, the FAD-binding motif and nearly all of the key active-site residues described above are conserved. The structure also reveals that the 50-residue insert unique to the GlpOs (and deleted in GlpO∆) does not contribute in a major way to the active site, and the chains at the two ends of the deleted region are just 7.4 Å apart, close enough to be bridged by the 3-4 residues in GlpD (B. subtilis residues 357-360) without altering the chain path.
The structure now also reveals clearly that there are no portions of the GlpD sequence that could serve as transmembrane helices, as all parts (except for the mammalian mitoGlpD calmodulin-homology domain) are incorporated into the three domains seen in the GlpO structure. In this light, we have reevaluated the predicted membrane topology of the human mitochondrial, E. coli, and B. subtilis GlpDs with five contemporary transmembrane helix prediction protocols: PHOBIUS (58), THUMBUP
(59), TUPS (59), the SPLIT 4.0 SERVER (60), and UMDHMMTMHP (59). Apart from the N-terminal mitochondrial signal peptide predicted by PHOBIUS and the SPLIT 4.0
SERVER [Brown et al. (3) reported that the rat mitoGlpD precursor is cleaved following
Ala42], the consensus prediction by all five algorithms is that there are no transmembrane helices. For the human mitoGlpD sequence, UMDHMMTMHP and PHOBIUS identify two possible weak transmembrane helices, but these are not supported by any of the other algorithms. We conclude, as has been proposed for the human mitochondrial protoporphyrinogen oxidase (61), that human mitoGlpD is strongly associated with the outer surface of the inner membrane (14), but is attached as a peripheral rather than integral membrane protein. A very recent report (62) of similar analyses for the yeast mitoGlpD (Gut2p) also concludes that this protein lacks transmembrane helices. Yeast
66 Gut2p aligns with residues 17-599 of the human mitoGlpD (full-length precursor sequences; 43% identity), but Gut2p lacks the C-terminal calmodulin-homology domain.
Janssen et al. (63) demonstrated that Gut2p is a peripheral membrane protein that interacts tightly with the mitochondrial inner membrane, as is likely required for its respiratory function. This peripheral localization has been suggested to involve partial insertion of a hydrophobic region as the photoactivatable probe 3-(trifluoromethyl)-3- phenyldiazirine-phosphatidylcholine specifically and efficiently labels Gut2p in the mitochondrial inner membrane (63).
The properties of the E. coli (16) and B. subtilis (18) bactGlpDs indicate that their respective membrane associations are considerably weaker, and analysis of the bactGlpD sequences with any of the five algorithms described above also fails to identify transmembrane helices. A proposal by Walz et al. (16) suggests that the E. coli GlpD is a peripheral membrane protein, with residues 355-370 serving as an amphipathic helix that penetrates and stably anchors the protein to the membrane. We now see that the corresponding segment in GlpO∆ is the surface-exposed α9 helix in the FAD-binding domain (Figure 6) that is preceded closely by the loop bearing FAD-binding residues
Ile430 and Thr431. This is indeed an amphipathic helix, but the nonpolar side of the helix is buried within the protein core and can in no way interact with the membrane without the domain structure being disrupted. So where might the membrane-anchoring segment reside? We propose that the C-terminal helical domain 3 of the GlpDs, which is apparently uninvolved in catalysis, conveys the peripheral membrane-binding function.
The alignment in Figure 6 demonstrates that the respective sequences for the GlpO versus
GlpD C-terminal domains are poorly conserved: GlpO∆ aligns with the mature Gut2p
67 (63) with 33% identity over the first two domains, but with only 14% identity for the C- terminal segment (GlpO∆ Glu438-Lys607 and Gut2p Glu433-Val612). Also interestingly, the helical structure of the C-terminal domain shows certain similarities with the membrane-binding motif demonstrated for prostaglandin H2 synthase-1 [Cox-1
(64)], a monotopic integral membrane protein. The bulk of the Cox-1 protein faces the lumen of the endoplasmic reticulum, and integration into one leaflet of the membrane is via a set of surface helices (the membrane-binding motif). These helices also form the entrance to the narrow channel that leads to the center of the Cox-1 monomer; this arrangement may have functional significance in the delivery of the hydrophobic arachidonic acid substrate from the membrane to the Cox-1 active site. As the mitoGlpDs reduce membrane-associated quinone substrates; the C-terminal domain in yeast Gut2p might play a role similar to that of Cox-1 in providing access for the hydrophobic GlpD acceptor substrate to the active site. The variability in sequences for the respective C-terminal domains, from the cytosolic GlpO to the tightly-bound yeast mitoGlpD and mammalian enzymes, is expected to provide the structural differentiation required for these distinct electron-transfer functions and membrane associations.
Glp Oxidase versus Dehydrogenase Reactivity. Although the GlpDs reduce quinone acceptors (65-67) instead of molecular oxygen, Mattevi (68) has cautioned that there are no structural rules for predicting flavoprotein oxidase versus dehydrogenase function. For example, while the active-site structures of the FMN-containing glycolate oxidase and flavocytochrome b2 are highly conserved [the two sequences are 37% identical, and a Cα superposition gives rmsd = 0.93 Å over 311 atoms (69)], the relative
4 5 O2 reactivities vary by a factor of 10 -10 , and the function of the latter enzyme involves
68 one-electron transfers from reduced flavin to heme. Subtle differences in FMN environments (both enzymes also stabilize the FMN-N5-sulfite adduct) have been implicated in this dramatic difference in electron-transfer paths from the respective
FMNH2, and the specific molecular factors involved in this divergence remain obscure.
Similarly, for GlpO versus GlpD, it appears that all of the key residues in the GlpO active site are conserved in GlpD; the goal of gleaning the structural basis for oxidase (O2) versus dehydrogenase (quinone) functions for GlpO and GlpD remains an elusive one at present.
In E. casseliflavus and other lactic acid bacteria capable of growth on glycerol, the genes encoding glycerol kinase [GlpK (70)], GlpO, and the glycerol facilitator [GlpF
(71)] are cotranscribed.2 An elaborate network of transcriptional as well as post- translational regulatory controls has been described (72) which ensures efficient glycerol uptake and catabolism in the absence of phosphotransferase system sugars, and vice versa. The crystal structure of Streptococcus sp. GlpO complements and in fact completes the structural picture of this important metabolic network in these heme- deficient bacteria.
69 ACKNOWLEDGMENT
We thank Dr. Tom Hollis for helpful discussions during final refinements, and Dr.
Todd Lowther and Dr. Thomas Jonsson for help in data collection.
70 REFERENCES
1. Green, D. E. (1936) alpha-Glycerophosphate dehydrogenase, Biochem. J. 30, 629- 644.
2. Garrib, A., and McMurray, W. C. (1986) Purification and characterization of glycerol-3-phosphate dehydrogenase (flavin-linked) from rat liver mitochondria, J. Biol. Chem. 261, 8042-8048.
3. Brown, L. J., MacDonald, M. J., Lehn, D. A., and Moran, S. M. (1994) Sequence of rat mitochondrial glycerol-3-phosphate dehydrogenase cDNA. Evidence for EF-hand calcium-binding domains, J. Biol. Chem. 269, 14363-14366.
4. MacDonald, M. J., and Brown, L. J. (1996) Calcium activation of mitochondrial glycerol phosphate dehydrogenase restudied, Arch. Biochem. Biophys. 326, 79-84.
5. Austin, D., and Larson, T. J. (1991) Nucleotide sequence of the glpD gene encoding aerobic sn-glycerol 3-phosphate dehydrogenase of Escherichia coli K- 12, J. Bacteriol. 173, 101-107.
6. Holmberg, C., Beijer, L., Rutberg, B., and Rutberg, L. (1990) Glycerol catabolism in Bacillus subtilis : nucleotide sequence of the genes encoding glycerol kinase (glpK) and glycerol-3-phosphate dehydrogenase (glpD), J. Gen. Microbiol. 136, 2367-2375.
7. Esders, T. W., and Michrina, C. A. (1979) Purification and properties of L-alpha- glycerophosphate oxidase from Streptococcus faecium ATCC 12755, J. Biol. Chem. 254, 2710-2715.
8. Claiborne, A. (1986) Studies on the structure and mechanism of Streptococcus faecium L-alpha-glycerophosphate oxidase, J. Biol. Chem. 261, 14398-14407.
9. Parsonage, D., Luba, J., Mallett, T. C., and Claiborne, A. (1998) The soluble alpha-glycerophosphate oxidase from Enterococcus casseliflavus. Sequence homology with the membrane-associated dehydrogenase and kinetic analysis of the recombinant enzyme, J. Biol. Chem. 273, 23812-23822.
10. Bender, K., Newsholme, P., Brennan, L., and Maechler, P. (2006) The importance of redox shuttles to pancreatic beta-cell energy metabolism and function, Biochem. Soc. Trans. 34, 811-814.
11. Ravier, M. A., Eto, K., Jonkers, F. C., Nenquin, M., Kadowaki, T., and Henquin, J. C. (2000) The oscillatory behavior of pancreatic islets from mice with mitochondrial glycerol-3-phosphate dehydrogenase knockout, J. Biol. Chem. 275, 1587-1593.
12. Eto, K., Tsubamoto, Y., Terauchi, Y., Sugiyama, T., Kishimoto, T., Takahashi, N., Yamauchi, N., Kubota, N., Murayama, S., Aizawa, T., Akanuma, Y., Aizawa,
71 S., Kasai, H., Yazaki, Y., and Kadowaki, T. (1999) Role of NADH shuttle system in glucose-induced activation of mitochondrial metabolism and insulin secretion, Science 283, 981-985.
13. Cole, E. S., Lepp, C. A., Holohan, P. D., and Fondy, T. P. (1978) Isolation and characterization of flavin-linked glycerol-3-phosphate dehydrogenase from rabbit skeletal muscle mitochondria and comparison with the enzyme from rabbit brain, J. Biol. Chem. 253, 7952-7959.
14. Klingenberg, M. (1970) Localization of the glycerol-phosphate dehydrogenase in the outer phase of the mitochondrial inner membrane, Eur. J. Biochem. 13, 247- 252.
15. Schryvers, A., Lohmeier, E., and Weiner, J. H. (1978) Chemical and functional properties of the native and reconstituted forms of the membrane-bound, aerobic glycerol-3-phosphate dehydrogenase of Escherichia coli, J. Biol. Chem. 253, 783- 788.
16. Walz, A. C., Demel, R. A., de Kruijff, B., and Mutzel, R. (2002) Aerobic sn- glycerol-3-phosphate dehydrogenase from Escherichia coli binds to the cytoplasmic membrane through an amphipathic alpha-helix, Biochem. J. 365, 471-479.
17. Yeh, J. I., Du, S., Tortajada, A., Paulo, J., and Zhang, S. (2005) Peptergents: peptide detergents that improve stability and functionality of a membrane protein, glycerol-3-phosphate dehydrogenase, Biochemistry 44, 16912-16919.
18. Hedin, T. (1991) Glycerol-3-phosphate dehydrogenase of Bacillus subtilis BR95: cellular localisation, biochemical properties and reconstitution into natural and artificial membranes, University of Lund, Lund, Sweden.
19. Koditschek, L. K., and Umbreit, W. W. (1969) Alpha-glycerophosphate oxidase in Streptococcus faecium F 24, J. Bacteriol. 98, 1063-1068.
20. Charrier, V., Luba, J., Parsonage, D., and Claiborne, A. (2000) Limited proteolysis as a structural probe of the soluble alpha-glycerophosphate oxidase from Streptococcus sp., Biochemistry 39, 5035-5044.
21. Finnerty, C. M., Charrier, V., Claiborne, A., and Karplus, P. A. (2002) Crystallization and preliminary crystallographic analysis of the soluble alpha- glycerophosphate oxidase from Streptococcus sp., Acta Crystallogr. D Biol. Crystallogr. 58, 165-166.
22. Nicely, N. I., Parsonage, D., Paige, C., Newton, G. L., Fahey, R. C., Leonardi, R., Jackowski, S., Mallett, T. C., and Claiborne, A. (2007) Structure of the type III pantothenate kinase from Bacillus anthracis at 2.0 A resolution: implications for coenzyme A-dependent redox biology, Biochemistry 46, 3234-3245.
72 23. Cornish-Bowden, A. (1995) Fundamentals of Enzyme Kinetics, Portland Press Ltd., London.
24. Pflugrath, J. W. (1999) The finer things in X-ray diffraction data collection, Acta Crystallogr. D Biol. Crystallogr. 55, 1718-1725.
25. Terwilliger, T. C., and Berendzen, J. (1999) Automated MAD and MIR structure solution, Acta Crystallogr. D Biol. Crystallogr. 55, 849-861.
26. Terwilliger, T. C. (2000) Maximum-likelihood density modification, Acta Crystallogr. D Biol. Crystallogr. 56, 965-972.
27. Terwilliger, T. C. (2003) Automated main-chain model building by template matching and iterative fragment extension, Acta Crystallogr. D Biol. Crystallogr. 59, 38-44.
28. Murshudov, G. N., Vagin, A. A., and Dodson, E. J. (1997) Refinement of macromolecular structures by the maximum-likelihood method, Acta Crystallogr. D Biol. Crystallogr. 53, 240-255.
29. Emsley, P., and Cowtan, K. (2004) Coot: model-building tools for molecular graphics, Acta Crystallogr. D Biol. Crystallogr. 60, 2126-2132.
30. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse- Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Crystallography & NMR system: a new software suite for macromolecular structure determination, Acta Crystallogr. D Biol. Crystallogr. 54, 905-921.
31. Storoni, L. C., McCoy, A. J., and Read, R. J. (2004) Likelihood-enhanced fast rotation functions, Acta Crystallogr. D Biol. Crystallogr. 60, 432-438.
32. Read, R. J. (2001) Pushing the boundaries of molecular replacement with maximum likelihood, Acta Crystallogr. D Biol. Crystallogr. 57, 1373-1382.
33. Luzzati, V. (1952) Traitement statistique des erreurs dans la determination des structures cristallines, Acta Cryst. 5, 802-810.
34. Kabsch, W., and Sander, C. (1983) Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features, Biopolymers 22, 2577-2637.
35. Holm, L., and Sander, C. (1993) Protein structure comparison by alignment of distance matrices, J. Mol. Biol. 233, 123-138.
36. Settembre, E. C., Dorrestein, P. C., Park, J. H., Augustine, A. M., Begley, T. P., and Ealick, S. E. (2003) Structural and mechanistic studies on ThiO, a glycine
73 oxidase essential for thiamin biosynthesis in Bacillus subtilis, Biochemistry 42, 2971-2981.
37. Tsuge, H., Kawakami, R., Sakuraba, H., Ago, H., Miyano, M., Aki, K., Katunuma, N., and Ohshima, T. (2005) Crystal structure of a novel FAD-, FMN-, and ATP-containing L-proline dehydrogenase complex from Pyrococcus horikoshii, J. Biol. Chem. 280, 31045-31049.
38. Chen, Z. W., Hassan-Abdulah, A., Zhao, G., Jorns, M. S., and Mathews, F. S. (2006) Heterotetrameric sarcosine oxidase: structure of a diflavin metalloenzyme at 1.85 A resolution, J. Mol. Biol. 360, 1000-1018.
39. Trickey, P., Wagner, M. A., Jorns, M. S., and Mathews, F. S. (1999) Monomeric sarcosine oxidase: structure of a covalently flavinylated amine oxidizing enzyme, Structure 7, 331-345.
40. Mattevi, A., Vanoni, M. A., Todone, F., Rizzi, M., Teplyakov, A., Coda, A., Bolognesi, M., and Curti, B. (1996) Crystal structure of D-amino acid oxidase: a case of active site mirror-image convergent evolution with flavocytochrome b2, Proc. Natl. Acad. Sci. USA 93, 7496-7501.
41. Dym, O., and Eisenberg, D. (2001) Sequence-structure analysis of FAD- containing proteins, Protein Sci. 10, 1712-1728.
42. Lennon, B. W., Williams, C. H., Jr., and Ludwig, M. L. (1999) Crystal structure of reduced thioredoxin reductase from Escherichia coli : structural flexibility in the isoalloxazine ring of the flavin adenine dinucleotide cofactor, Protein Sci. 8, 2366-2379.
43. Zhang, W., Zhang, M., Zhu, W., Zhou, Y., Wanduragala, S., Rewinkel, D., Tanner, J. J., and Becker, D. F. (2007) Redox-induced changes in flavin structure and roles of flavin N(5) and the ribityl 2'-OH group in regulating PutA-membrane binding, Biochemistry 46, 483-491.
44. Schiffer, A., Fritz, G., Kroneck, P. M., and Ermler, U. (2006) Reaction mechanism of the iron-sulfur flavoenzyme adenosine-5'-phosphosulfate reductase based on the structural characterization of different enzymatic states, Biochemistry 45, 2960-2967.
45. Loach, P. A. (1976) Oxidation-reduction potentials, absorbance bands and molar absorbance of compounds used in biochemical studies, in Handbook of Biochemistry and Molecular Biology (Fasman, G. D., Ed.) 3rd ed., pp 122-130, CRC Press, Boca Raton, FL.
46. Rossmann, M. G., Moras, D., and Olsen, K. W. (1974) Chemical and biological evolution of nucleotide-binding protein, Nature 250, 194-199.
74 47. Wierenga, R. K., Terpstra, P., and Hol, W. G. J. (1986) Prediction of the occurrence of the ADP-binding beta-alpha-beta-fold in proteins, using an amino acid sequence fingerprint, J. Mol. Biol. 187, 101-107.
48. Umhau, S., Pollegioni, L., Molla, G., Diederichs, K., Welte, W., Pilone, M. S., and Ghisla, S. (2000) The x-ray structure of D-amino acid oxidase at very high resolution identifies the chemical mechanism of flavin-dependent substrate dehydrogenation, Proc. Natl. Acad. Sci. USA 97, 12463-12468.
49. Alber, T. C., Davenport, R. C., Jr., Giammona, D. A., Lolis, E., Petsko, G. A., and Ringe, D. (1987) Crystallography and site-directed mutagenesis of yeast triosephosphate isomerase: what can we learn about catalysis from a "simple" enzyme?, Cold Spring Harbor Symp. Quant. Biol. 52, 603-613.
50. Noble, M. E. M., Wierenga, R. K., Lambeir, A.-M., Opperdoes, F. R., Thunnissen, A.-M. W. H., Kalk, K. H., Groendijk, H., and Hol, W. G. J. (1991) The adaptability of the active site of trypanosomal triosephosphate isomerase as observed in the crystal structures of three different complexes, Proteins 10, 50-69.
51. Lloyd, W. J., and Harrison, R. (1974) Interaction of deoxy and deoxyfluoro- substrate analogs and related compounds with sn-glycerol-3-phosphate dehydrogenase of locust flight muscle mitochondria, Arch. Biochem. Biophys. 163, 185-190.
52. Finn, R. D., Mistry, J., Schuster-Bockler, B., Griffiths-Jones, S., Hollich, V., Lassmann, T., Moxon, S., Marshall, M., Khanna, A., Durbin, R., Eddy, S. R., Sonnhammer, E. L., and Bateman, A. (2006) Pfam: clans, web tools and services, Nucleic Acids Res. 34, D247-251.
53. Zhao, G., Song, H., Chen, Z. W., Mathews, F. S., and Jorns, M. S. (2002) Monomeric sarcosine oxidase: role of histidine 269 in catalysis, Biochemistry 41, 9751-9764.
54. Zhao, G., and Jorns, M. S. (2005) Ionization of zwitterionic amine substrates bound to monomeric sarcosine oxidase, Biochemistry 44, 16866-16874.
55. Choe, J., Guerra, D., Michels, P. A., and Hol, W. G. (2003) Leishmania mexicana glycerol-3-phosphate dehydrogenase showed conformational changes upon binding a bi-substrate adduct, J. Mol. Biol. 329, 335-349.
56. Ou, X., Ji, C., Han, X., Zhao, X., Li, X., Mao, Y., Wong, L. L., Bartlam, M., and Rao, Z. (2006) Crystal structures of human glycerol 3-phosphate dehydrogenase 1 (GPD1), J. Mol. Biol. 357, 858-869.
57. Lehn, D. A., Brown, L. J., Simonson, G. D., Moran, S. M., and MacDonald, M. J. (1994) The sequence of a human mitochondrial glycerol-3-phosphate dehydrogenase-encoding cDNA, Gene 150, 417-418.
75 58. Kall, L., Krogh, A., and Sonnhammer, E. L. (2004) A combined transmembrane topology and signal peptide prediction method, J. Mol. Biol. 338, 1027-1036.
59. Zhou, H., Zhang, C., Liu, S., and Zhou, Y. (2005) Web-based toolkits for topology prediction of transmembrane helical proteins, fold recognition, structure and binding scoring, folding-kinetics analysis and comparative analysis of domain combinations, Nucleic Acids Res. 33, W193-197.
60. Juretic, D., Zoranic, L., and Zucic, D. (2002) Basic charge clusters and predictions of membrane protein topology, J. Chem. Inf. Comput. Sci. 42, 620- 632.
61. Morgan, R. R., Errington, R., and Elder, G. H. (2004) Identification of sequences required for the import of human protoporphyrinogen oxidase to mitochondria, Biochem. J. 377, 281-287.
62. Rijken, P. J., De Kruijff, B., and De Kroon, A. I. (2007) Phosphatidylcholine is essential for efficient functioning of the mitochondrial glycerol-3-phosphate dehydrogenase Gut2 in Saccharomyces cerevisiae, Mol. Membr. Biol. 24, 269- 281.
63. Janssen, M. J., van Voorst, F., Ploeger, G. E., Larsen, P. M., Larsen, M. R., de Kroon, A. I., and de Kruijff, B. (2002) Photolabeling identifies an interaction between phosphatidylcholine and glycerol-3-phosphate dehydrogenase (Gut2p) in yeast mitochondria, Biochemistry 41, 5702-5711.
64. Picot, D., Loll, P. J., and Garavito, R. M. (1994) The X-ray crystal structure of the membrane protein prostaglandin H2 synthase-1, Nature 367, 243-249.
65. Nelson, D. L., and Cox, M. M. (2005) Lehninger Principles of Biochemistry, 4th ed., W.H. Freeman and Company, New York.
66. Bunoust, O., Devin, A., Averet, N., Camougrand, N., and Rigoulet, M. (2005) Competition of electrons to enter the respiratory chain: a new regulatory mechanism of oxidative metabolism in Saccharomyces cerevisiae, J. Biol. Chem. 280, 3407-3413.
67. Spoering, A. L., Vulic, M., and Lewis, K. (2006) GlpD and PlsB participate in persister cell formation in Escherichia coli, J. Bacteriol. 188, 5136-5144.
68. Mattevi, A. (2006) To be or not to be an oxidase: challenging the oxygen reactivity of flavoenzymes, Trends Biochem. Sci. 31, 276-283.
69. Lindqvist, Y., Branden, C.-I., Mathews, F. S., and Lederer, F. (1991) Spinach glycolate oxidase and yeast flavocytochrome b2 are structurally homologous and evolutionarily related enzymes with distinctly different function and flavin mononucleotide binding, J. Biol. Chem. 266, 3198-3207.
76 70. Yeh, J. I., Charrier, V., Paulo, J., Hou, L., Darbon, E., Claiborne, A., Hol, W. G., and Deutscher, J. (2004) Structures of enterococcal glycerol kinase in the absence and presence of glycerol: correlation of conformation to substrate binding and a mechanism of activation by phosphorylation, Biochemistry 43, 362-373.
71. Fu, D., Libson, A., Miercke, L. J., Weitzman, C., Nollert, P., Krucinski, J., and Stroud, R. M. (2000) Structure of a glycerol-conducting channel and the basis for its selectivity, Science 290, 481-486.
72. Deutscher, J., Francke, C., and Postma, P. W. (2006) How phosphotransferase system-related protein phosphorylation regulates carbohydrate metabolism in bacteria, Microbiol. Mol. Biol. Rev. 70, 939-1031.
77 Chapter III
KINETIC ANALYSIS OF α-GLYCEROPHOSPHATE OXIDASE FROM
Streptococcus sp.: IDENTIFICATION OF RESIDUES IMPORTANT IN GLP-
SUBSTRATE BINDING AND CATALYSIS
Timothy Colussi
78 ABSTRACT
Streptococcus sp. α-glycerophosphate oxidase (GlpO) catalyzes the oxidation of
α-glycerophosphate (Glp) to dihydroxyacetone phosphate (DHAP) with concomitant reduction of the flavin adenine dinucleotide (FAD) cofactor. The x-ray crystal structures of streptococcal GlpO and a deletion mutant, lacking a 50-residue flexible surface region
(GlpOΔ), have previously been determined [Colussi, T., Parsonage, D., Boles W.,
Matsuoka, T., Mallett, T. C., Karplus, P. A., and Claiborne, A. (2008) Biochemistry 47,
965-977]. Analysis of the active sites of these structures revealed an arginine (R346) residue that is structurally conserved in two structural homologs of GlpO, Bacillus subtilis glycine oxidase (ThiO) and the yeast Rhodotorula gracilis D-amino acid oxidase
(DAAO), and is implicated in substrate binding. A mutation of R346M in GlpOΔ was generated. Comparisons of the GlpOΔ structure with the structures of both DAAO and
ThiO led to the discovery of residues lining the active site that are present in GlpOΔ but not the structural homologs. These residues (H65, R69, Y70, and K429) are conserved by
GlpOs and GlpDs. Mutations to these residues were also generated testing the hypothesis that they play a role in Glp-binding through interactions with the Glp-phosphate moiety.
The effect of each respective mutation on GlpOΔ catalysis was ascertained through kinetic experiments with the purified enzyme including both a horseradish peroxidase
(HRP) coupled assay adapted for use with a microplate reader and enzyme-monitored turnover (EMT) experiments by stopped-flow using a spectrophotometer. K429A and
R346M appear to have no activity relative to wild-type based on either kinetic experiment leading to the thought that these residues are essential for Glp-substrate binding and/or catalysis. Mutating H65 to a glutamine resulted in activity that is ~60%
79 that of wild-type, but mutation to an alanine yielded an activity that was ~2% that of wild-type based on the HRP coupled assay. Y70F and R69M are ~80% and ~10%, respectively, as active when compared to wild-type based on the HRP coupled assay. A
Glp-binding/dehydrogenation mechanism has been proposed based on the results of these experiments; this mechanism is discussed in light of similar studies with ThiO and
DAAO.
80 α-Glycerophosphate oxidase (GlpO)1 is a flavoprotein oxidase that catalyzes the oxidation of α-glycerophosphate (Glp) to dihydroxyacetone phosphate (DHAP), which results in the simultaneous reduction of the flavin adenine dinucleotide (FAD) cofactor
(1). GlpO is part of a glycerol metabolic system that affords many bacteria, especially heme-defficient lactic acid bacteria such as Streptococcus sp. and Enterococcus casseliflavus, the ability to grow on glycerol as the main carbon source (2). The
Streptococcus sp. GlpO is similar to bacterial and mitochondrial GlpDs based on amino acid sequence, with a major difference being the presence of a ca. 50-residue flexible surface region in GlpO but not the GlpDs (3). Functionally, GlpO and GlpD both oxidize
Glp to DHAP; however, GlpDs transfer the reducing equivalents from the FAD cofactor to ubiquinone, whereas GlpOs transfer the reducing equivalents to O2 resulting in the formation of H2O2 (3-6).GlpDs exhibit varying degrees of membrane association (7-11), while GlpOs are generally thought to be cytosolic (12, 13).
The x-ray crystal structures of streptococcal GlpO and GlpOΔ, a mutant in which the ca. 50-residue flexible surface region has been deleted, were determined (13).
Analyses of these structures led to the discovery of two structural homologs of GlpO, glycine oxidase (ThiO) from Bacillus subtilis (14) and D-amino acid oxidase (DAAO) from the yeast Rhodotorula gracilis (15), both of which are members of the DAO pfam
(16). A structurally conserved arginine is present in GlpO (R346), ThiO (R302), and
DAAO (R285). Based on the x-ray crystal structures of ThiO and DAAO in complex with the substrate analog N-acetylglycine (14)) and substrate D-alanine (15)), respectively, it was determined that these arginine residues interact with the carboxylate
1 Abbreviations: see page v.
81 moieties of the respective ligand (Figure 1). Kinetic analysis of wild-type and R285 mutant DAAOs showed that R285 is essential for orienting the substrate for catalysis
(17).
Figure 2 shows the active site of GlpOΔ. Comparisons of this active site to that of
DAAO and ThiO led to the discovery of four residues that are conserved in GlpOs and
GlpDs but not present in the structural homologs. These residues are H65, R69, Y70, and
K429. DAAO catalyzes the dehydrogenation of D-amino acids (17) and ThiO catalyzes the deamination of primary and secondary amino acids and also D-amino acids (18); whereas GlpO catalyzes the oxidation of α-glycerophosphate, which structurally replaces a hydroxyl from glycerol with a phosphate (13). This led to the hypothesis that these residues present in GlpO are involved in Glp-substrate binding through interactions with the phosphate moiety. An active-site base is required in GlpO to abstract a proton from Glp to drive the hydride transfer reaction and, often times, histidine residues act in this role (19). In GlpO, this residue could be H65. Mutagenic and kinetic experiments were conducted to test the hypothesis that is depicted by the scheme in Figure 3.
Due to the fact that the GlpOΔ enzyme is catalytically active, easy to purify, and attempts to reproduce crystals of intact GlpO have been unsuccessful, five residues within the active site proposed to be involved in the Glp binding pocket have been analyzed by mutagenesis using the GlpOΔ enzyme as a template. Kinetic experiments using both horseradish peroxidase (HRP) coupled assays in a microplate reader and enzyme-monitored turnover (EMT) experiments with a stopped-flow spectrophotometer were employed to determine kinetic parameters. Sulfite titrations were also conducted to ascertain the effect, if any, on the ability of the GlpOΔ mutants to stabilize an FAD N(5)-
82
Figure 1: Active sites of two structural homologs of Streptococcus sp. GlpO. (A) The active site of ThiO from B. subtilis (PDB code 1NG3) showing the interaction of a structurally conserved arginine (R302) with the substrate analog N-acetylglycine (14). (B) The active site of DAAO from R. gracilis (PDB code 1C0P) showing the interaction of a structurally conserved arginine (R285) with the substrate D-alanine (15).
83
Figure 2: GlpOΔ active site. Stereo figure showing the active site of GlpOΔ, highlighting the residues that have been mutated to elucidate their possible roles in binding and/or catalysis.
84 Thr431
OON R H H O N N H Ile430 N H Lys429 HN N NH2
O Arg69 H NH2 HN O HN Arg346 H HN H OPO O H HN C NH O 2 H OH H Tyr70 O N N
His65
Figure 3: Proposed mechanism for Glp-binding, proton abstraction, and hydride transfer for Streptococcus sp. GlpOΔ.
85 sulfite complex, a common characteristic of flavin oxidases (20) that has also been linked quantitatively to redox potential (21) of the FAD cofactor. The results of these experiments are presented here, and a refined scheme for Glp binding and catalysis is proposed.
86 EXPERIMENTAL PROCEDURES
Expression and Purification of GlpOΔ mutants. Mutant (H65A, H65Q, R69M,
Y70F, R346M, and K429A) alleles of the streptococcal glpOΔ gene (13) were generated using the PCR method with appropriate forward and reverse primers (Operon
Biotechnologies for H65A, H65Q, R346M, and K429A mutants and Sigma-Aldrich for
R69M and Y70F mutants; sequences available on request) for each mutant to generate the two halves. The halves were annealed and amplified by PCR using the sequencing primers for the 3’ and 5’ end of the glpoΔ gene. The presence of only the desired mutations was confirmed by sequence analysis. The mutant genes were amplified by
PCR and cloned into pQE30 (Qiagen) to express the N-terminal His-tagged protein. The expression plasmid was transformed into Escherichia coli B834 cells containing pREP4
(Qiagen) for expression of each mutant. These cells were grown in a 3-L culture (4 x 750 mL in 2.8-L Fernbach flasks) of tryptone-yeast extract-phosphate medium supplemented with 2.25 mM glucose, 30 μg/mL kanamycin, and 100 μg/mL ampicillin at 37 °C.
Cultures were induced with 0.5 mM isopropyl-β-D-thiogalactopyranoside at an A600 of
1.0 and allowed to grow overnight at 16 °C. Cells were centrifuged at 5000 x g for 15 min. Pellets were resuspended/washed with a total volume of ~700 mL of 50 mM potassium phosphate, pH 7.0, containing 0.5 mM EDTA and spun again at 5000 x g for
10 min. The washed cell pellet was resuspended in a minimal amount of 50 mM sodium phosphate, pH 7.0, containing 2 mM EDTA and 2 mM DTT. The cells were lysed via disruption with a pneumatic cell homogenizer (Avestin EmulsiFlex-C5). Lysate was centrifuged at 40,000 x g for 45 min. and 2% (w/v) streptomycin sulfate was added to the supernatant with stirring and centrifuged again at 27,000 x g for 15 min. Supernatant was
87 filtered and loaded onto a Q-Sepharose HP column equilibrated in 25 mM sodium phosphate, pH 7.0. The protein was eluted using a 0→1 M NaCl gradient in the loading buffer. Yellow fractions were pooled, filtered, and loaded onto a Ni-NTA Superflow
(Qiagen) column equilibrated in 50 mM sodium phosphate, pH 8.0, containing 0.5 M
NaCl and 20 mM imidazole. The His-tagged protein was eluted by increasing the imidazole concentration to 100 mM. The protein pool was concentrated and buffer exchanged into a buffer of 50 mM potassium phosphate, pH 7.0, containing 0.5 mM
EDTA. All mutants were then reconstituted by incubating with 5 molar excess of FAD at room temperature for 5 min.; excess FAD removed by ultracentrifugation using a
Vivaspin 2 microconcentrator (30,000 MWCO PES, Sartorious Stedim). The protein was concentrated to 10 mg/mL and 20% glycerol (v/v) was then added prior to flash freezing in 0.5 mL aliquots.
Horseradish Peroxidase Activity Assay. The activity of wild-type and mutant
GlpOΔs was determined by monitoring the coupled oxidation of o-dianisidine by horseradish peroxidase (HRP) (1). Absorbance changes were monitored using a Tecan
Safire2 microplate reader. The experiments were performed as described previously (1) with the following modifications made for use with the microplate reader. 120 μL of air- equilibrated reaction buffer containing all substrates was added to each well of a Corning
96-well flat bottom UV-transparent plate. 10 μL of 0.3 mg/mL HRP was then added to each well before initiating reactions with 2 μL of GlpOΔ enzymes at final concentrations of 0.01, 0.05, 0.1, 0.5, and 1 μM. Enzyme concentrations of up to 10 μM were utilized for the R346M and K429A mutants. The reactions were followed at 460 nm; and one unit of α-glycerophosphate oxidase activity was defined as the amount of enzyme required to
88 generate 1 μmol of H2O2 per minute at 25 °C, using the extinction coefficient of 11,300
M-1cm-1 for the dianisidine oxidation product at 460 nm (22) and a path length of 0.417 cm for the assay solution in the microplate well. Reactions were monitored for 5 min., and data were analyzed using SigmaPlot 8.0.
Sulfite Titrations. Spectral measurements were made with an Agilent 8453E UV- visible spectrophotometer. The proteins were first buffer exchanged into 50 mM potassium phosphate, pH 7.0, buffer containing 0.5 mM EDTA, and titrations followed the protocol described in (13). The change in absorbance with respect to initial
2- absorbance at 450 nm was plotted versus the concentration of SO3 , and the data were fit to the equation:
− 2− Δ max [ 3 + SOHSOA 3 ] 0 AA =− − 2− d [ 3 ++ SOHSOK 3 ]
using SigmaPlot 8.0.
Stopped-Flow Kinetics. Prior to each experiment, the respective enzyme was buffer-exchanged into 50 mM potassium phosphate, pH 7.0, buffer containing 0.5 mM
EDTA. Steady-state kinetic parameters for mutant and wild-type GlpOΔ were determined in enzyme monitored turnover (EMT) experiments as previously described (13) with differences in reaction temperature and [O2]. The experiments were done at 25 °C, and the final concentration of O2 after mixing was 0.78 mM. Kinetic data were analyzed as described by Cornish-Bowden (23), as follows:
ν = Vab/(KiAKmB + KmBa + KmAb + ab) where A and B represent Glp and O2, respectively. Primary plots of b/v versus b intersect to the left of the y-axis in the case of an ordered ternary complex mechanism, but on the y-axis in the case of a ping-pong mechanism. A secondary plot of slope x a versus a gives
89 a y-axis intercept of KmA/Vmax and has a slope equal to 1/Vmax, in either case. A secondary plot of intercept x a versus a gives a y-axis intercept of KiAKmB/Vmax and has a slope equal to KmB/Vmax, only in the case of an ordered ternary complex mechanism. Data were analyzed using KaleidaGraph 4.0, and figures were prepared using SigmaPlot 8.0.
90 RESULTS AND DISCUSSION
SulfiteTtitration. GlpOΔ, like other flavoprotein oxidases (20), stabilizes a flavin
N(5)-sulfite adduct. This stabilization of the negative charge produced from binding sulfite has been attributed to the presence of a positively charged residue or N-terminus of an α-helix near the N(1)-C(2)=O position of the isoalloxazine ring (20) and is thought to mimic the hydride transfer reaction (See Figure 2B of Chapter I of this dissertation).
Sulfite titrations were performed to determine if the active site structural integrity of the mutants is maintained and if the redox chemistry of the isoalloxazine ring has been perturbed. Figure 4 shows the titration of wild-type GlpOΔ with sulfite, which results in the bleaching of flavin absorbance attributed to the formation of a flavin N(5)-sulfite adduct. Similar titrations were performed for each mutant as well (data not shown) and the results are summarized in Table 1. H65Q has a Kd value of 0.18 mM, R69M has a Kd value of 0.23 mM, and Y70F has a Kd value of 0.37 mM, which are about 5-, 5-, and 2- fold lower than that for wild-type GlpOΔ (Kd=0.81 mM), respectively. R346M has a Kd value of 2.3 mM, and K429A has a Kd value of 3.1 mM, which are 3- and 4-fold higher than wild-type, respectively. Using Müller and Massey’s (21) analysis of the relationship between Kd for sulfite and redox potential, the ~3- and 4-fold increase in Kd only relates to ~20 mV decrease in redox potential toward that for the Glp/DHAP couple, which is -
2- 190 mV (24). Taking the redox potential for the E. casseliflavus GlpO enzyme [Kd(SO3
)=0.81 mM] of -118 mV (12), this change would equate to a redox potential of -138 mV, which is closer to that of the Glp/DHAP couple, but does not seem to be enough to cause inactivation of the enzyme due to perturbed redox potential. Even though K429 is near the N(1)-C(2)=O position of the isoalloxazine ring (Figure 2), it does not appear to be
91
Figure 4: Sulfite titration of the wild-type GlpOΔ. The enzyme (33 μM, in 0.8 mL of 50 mM potassium phosphate, pH 7.0, and 0.5 mM EDTA) was titrated with Na2SO3 solutions of 0.02, 0.1, and 1.0 M. The legend shows the final amount of Na2SO3 present after each addition. The inset shows the graph of the change in absorbance at 450 nm - 2- versus [HSO3 + SO3 ]. Titrations were performed for each mutant and similar plots and calculations were done (data not shown).
92 Table 1: Results of Sulfite Titrations of Wild-Type and Mutant GlpOΔ Enzymes.
Enzyme Kd (SO3) (mM) Wild Type GlpOΔ 0.81 H65Q 0.18 R69M 0.23 Y70F 0.37 H65A 1.24 R346M 2.3 K429A 3.1
93 solely responsible, if at all, for the stabilization of the negative charge in this position during sulfite adduct formation. GlpOΔ also has the N-terminus of an α-helix near that position as well, which could be responsible for this stabilization.
Activity of Wild-Type and Mutant GlpOΔ. Wild-type and mutant GlpOΔs were assayed for enzyme activity using the standard spectrophotometric assay (1), in which the catalytic production of H2O2 is coupled with horseradish peroxidase, and has been adapted for use with a microplate reader. Figure 5 depicts the average of three experiments and standard error for each enzyme. The turnover number, which is defined as the concentration of enzyme needed to produce 1 μmol of product per minute, for each enzyme is given in Table 2.
H65 was hypothesized to act as an active-site base that would abstract a proton from the C(2)-OH of Glp, driving the hydride transfer reaction of the C(2)-H to the N(5) position of the isoalloxazine ring of the FAD cofactor (1, 12) as shown in Figure 3.
However, mutation of this histidine to glutamine does not totally disrupt activity, which would be expected if it were an essential active-site base. The activity for the H65Q mutant is 63% that of the wild-type enzyme. Mutating this residue to alanine; however, results in a catalytic activity of 2% that of wild-type, implicating this residue as important for Glp-binding and/or catalysis. The R69M mutant is 13% as active as the wild-type, whereas Y70F is 77% as active. Both the R346M and K429A mutants had activities that could not be determined, implicating them as essential for Glp-binding and/or catalysis.
94
Figure 5: HRP coupled assay for wild-type and mutant GlpOΔ enzymes. The enzymes were added to a final concentration of 100 nM for wild type, H65Q, R69M, and Y70F; 1 μM for H65A; and 10 μM for R346M and K429A. The air-equilibrated ([O2] = 0.26 mM at 25 °C) assay buffer contained 0.2 M D,L-α-glycerophosphate, 0.2 mM o-dianisodine, and 0.023 mg/mL HRP in 100 mM potassium phosphate buffer, pH 7.0, plus 0.3 mM EDTA. The assay temperature was 25 °C. The results are an average of 3 experiments done for each enzyme and error bars show standard error.
95
Table 2: Turnover Numbers for Wild-Type and Mutant GlpOΔ Enzymes from HRP Coupled Assay.
Enzyme Turnover (s-1) % Wild Type Wild Type GlpOΔ 1.70 100 Y70F 1.31 77 H65Q 1.07 63 R69M 0.22 13 H65A 0.04 2 R346M ND ND K429A ND ND
96 The x-ray crystal structure of the Escherichia coli GlpD has been determined with phosphate bound in the active site (25). Figure 6A shows an overlay of the GlpOΔ active site with that of the GlpD-phosphate complex structure (PDB code 2QCU). This structure shows R317, R54, Y55, and K354 (R346, R69, Y70, and K429 for GlpOΔ, respectively) interacting with the phosphate; however, it does not show H50 (H65 in GlpOΔ) interacting with the phosphate. Based on the HRP coupled assay data, H65 in GlpOΔ plays an important role in Glp-substrate binding, but is not the active-site base. This result is interpreted to mean that it plays an important role by interacting with the phosphate moiety of Glp, unlike what is shown in the E. coli GlpD complex structures.
The x-ray crystal structure of the E. coli GlpD has also been determined with
DHAP bound in the active site (25). Figure 6B shows a similar overlay as shown in part
A only with the GlpD-DHAP complex structure (PDB code 2R4E). In this structure, the phosphate moiety of DHAP is in a similar position as the phosphate in the GlpD- phosphate complex, making similar interactions with the GlpD active-site residues, with the exception of K354; however, the C(2) of DHAP is ~8.5 Å away from the N(5) of the isoalloxazine ring, not in a favorable position for hydride transfer. We believe, based on our mutagenesis data, that, rather than binding as shown in the E. coli GlpD complex structures, the phosphate binds in a pocket that is carved out by H65, R69, Y70, and
K429 as shown in the previous scheme (Figure 3). However, based on the HRP coupled assay data, we believe that H65 no longer can be the active-site base.
Enzyme-Monitored Turnover. The HRP coupled assay adapted for use in the microplate reader is advantageous in that several experiments can be done at the same
97
Figure 6: Active site overlays, in stereo, of GlpOΔ and E. coli GlpD in complex with phosphate (A) and DHAP (B). (A) An active site overlay of streptococcal GlpOΔ and E. coli GlpD-phosphate complex (PDB code 2QCU). The residues shown are conserved in GlpOs and GlpDs; mutants of these residues have been generated for the streptococcal GlpOΔ. (B) An active site overlay of streptococcal GlpOΔ with E. coli GlpD-DHAP complex (PDB code 2R4E). DHAP is the product of the GlpO and GlpD catalyzed oxidation of Glp. The residues shown are the same shown in (A).
98 time under the same conditions; however, there are disadvantages. Even though control experiments were performed to validate the assay (data not shown), the measurement is based on a secondary product formed by a coupled enzyme. Also, the reaction buffers were air-equilibrated, which has an O2 concentration of 260 μM at 25 °C. The Km(O2) reported for GlpOΔ at 5 °C is 460 μM (13), well above the concentration of O2 present in the HRP assay buffers. To determine the steady state kinetic parameters of wild-type and mutant GlpOΔs, enzyme-monitored turnover (EMT) experiments were performed using a stopped-flow spectrophotometer. The stopped-flow spectrophotometer can measure the absorbance change of FAD, in this case, under rapid reaction conditions, allowing for steady-state measurements to be taken and kinetic constants determined.
Figure 7A represents the stopped-flow traces of wild-type GlpOΔ. Immediately after mixing, a pre-steady-state species of the enzyme is seen meaning that the majority of the enzyme is in the oxidized state. A steady-state is achieved shortly after mixing, where the majority of the enzyme is in the enzyme:substrate complex. After all the O2 was consumed, the enzyme went to full reduction. Analysis of the traces as described by
Gibson et al. (26) gives rise to the plots in Figure 7B where [O2]/ν is plotted versus [O2]
(23). The intersection of these lines occurs to the left of the y-axis, which is indicative of a ternary complex mechanism. Two of the lines, however, do not intersect at the same point as the other three. These lines correspond to the lowest and highest [Glp] and is a phenomenon that has been seen previously2. Secondary plots of [Glp]/Vapp versus [Glp]
app app (Figure 7C) and [Glp]Km /V versus [Glp] (Figure 7D) yield the kinetic parameters
2 Unpublished data from Dr. Derek Parsonage.
99
Figure 7: Enzyme-monitored turnover of wild-type GlpOΔ. The enzyme (15 μM after mixing, in 50 mM phosphate, pH 7.0, and 0.5 mM EDTA air equilibrated buffer) was mixed at 25 °C with different concentrations of Glp that had been equilibrated with 100% O2 in the same buffer. The reactions were followed at 450 nm. (A) The stopped-flow traces of the time course reduction of FAD at 450 nm by Glp. (B) shows the primary plots of [O2]/ν versus [O2] for each concentration of Glp. Wild type GlpOΔ exhibited behavior indicative of an ordered ternary complex mechanism (23). Graphs of (C) app app app [Glp]/V versus [Glp] and (D) [Glp]Km /V versus [Glp] were plotted to determine kinetic parameters described in the text.
100
Table 3: Steady-State Kinetic Constants for GlpOΔ and Apparent Turnover Numbers for GlpOΔ and Mutants Calculated from Enzyme-Monitored Turnover Experiments.
Wild Type EMT Resultsa Turnoverb (s-1) Enzyme Conc. (μM) 15 Wild Type 98 Km(Glp) (mM) 8.3 Y70F 8.0 Km(O2) (μM) 720 H65Q 2.8 -1 kcat(s ) 110 R69M 0.8 Vmax (μM/s) 1760 H65A 0.3 aCalculated from data in Fig. 7. bApparent turnover numbers calculated by dividing the Vapp at 50 mM Glp by the concentration of enzyme.
101 Km(Glp), Km(O2), Vmax, and kcat. The slope of the secondary plot in Figure 7C is equal to
1/Vmax, and the y-intercept is equal to Km(O2)/Vmax. The slope of the secondary plot in
Figure 7D is equal to Km(Glp)/Vmax. The analyses of these plots yield a Km(Glp) = 8.3
-1 mM, Km(O2) = 720 μM, kcat = 110 s , and Vmax = 1760 (μM/s) at pH 7.0 and 25 °C (left side of Table 3).
The qualitative analysis of the steady-state kinetic data obtained for the mutants, however, differs from that obtained for wild-type. On analysis of the stopped-flow traces for H65A, H65Q, R69M, and Y70F (only the data for H65A is shown here but the traces for H65Q, R69M, and Y70F are all very similar), one distinct difference is very noticeable. Figure 8A shows the normalized absorbance traces for both wild-type and
H65A at 50 mM Glp. In the time it takes the wild-type GlpOΔ to achieve full reduction,
H65A has shown little reduction with ~3.5 fold higher enzyme concentration than wild- type. There is also no evidence of a pre-steady-state species that was seen in the traces for the wild-type enzyme. Analyses of the H65A traces (26) and plots of [O2]/rate versus
[O2] (23) resulted in the plot shown in Figure 8B. The intersection of these lines occurs on the y-axis very close to the origin. The interpretation is that for these experiments
[Glp] << Km(Glp). For this reason, Km(Glp), Km(O2), kcat, and Vmax were not calculated for the mutants. What was calculated, as a way to quantitatively compare the different species of enzyme, was apparent turnover at the highest Glp concentration (50 mM) and is presented in the right half of Table 3. This is an apparent turnover due to the fact that only one substrate concentration is varied, O2. The data in Table 3 seems to mirror the data presented in Table 2 for the HRP coupled assay in that the Y70F mutant gives the highest turnover number and most HRP coupled assay activity, followed by H65Q,
102
Figure 8: Enzyme-monitored turnover of H65A. The enzyme (54.9 μM after mixing, in 50 mM phosphate, pH 7.0, and 0.5 mM EDTA air equilibrated buffer) was mixed at 25 °C with different concentrations of Glp that had been equilibrated with 100% O2 in the same buffer. The reactions were followed at 450 nm. (A) The stopped-flow traces of H65A (black) and wild-type GlpOΔ (red) mixed with 50 mM L-α-Glp. (B) The primary plots of [O2]/ν versus [O2] for each concentration of Glp with H65A. Similar behavior was seen for the H65Q, R69M, and Y70F mutants (data not shown).
103 R69M, and with the H65A mutant having the lowest turnover number and being the least active in the HRP assay. The difference in the turnover numbers calculated from the
EMT data compared to that calculated from the HRP coupled assay data is that the [O2] was much higher for these experiments and would not be as much a limiting factor as it was for the HRP coupled assay.
No absorbance change was detected by stopped-flow for the R346M mutant (data not shown) at 58 μM enzyme and 50 mM Glp in 100% O2 saturated buffer after mixing after 1000 s (the limitations of the stopped-flow instrument). Aerobic additions of 50 mM
Glp in 100% O2 saturated buffer were added to both R346M and K429A and spectra were taken at 5, 10, and 15 minutes. No absorbance change was noticed (data not shown).
These mutants were not active in the HRP coupled assay (Table 2) and seem to not be active by EMT analysis as well, strengthening the thought that these residues are essential for substrate binding and/or catalysis.
Glp-Substrate Binding. Comparisons with the structure of GlpOΔ and two structural homologs, ThiO and DAAO, led to the discovery of a conserved arginine residue (302 and 285, respectively). This residue has been shown to interact with the carboxylate moiety of each homologs’ respective substrate (14, 17). It would be reasonable to think that R346 in GlpOΔ could interact with the phosphate moiety of Glp; however, based on analysis of both of the E. coli GlpD complex structures shows that this interaction would place the phosphate above the N(5) position of the FAD, not C(2) of
Glp. This interaction appears to not be favorable in promoting catalysis. Both the HRP coupled assay and EMT experiments show that this residue, R346, is essential for Glp binding and/or catalysis. As was proposed in Figure 3, this residue could alternatively
104 interact with the C(1)-OH of Glp, and this interaction could lock Glp in a position that would place the C(2) above N(5) of FAD, which is favorable for hydride transfer.
K429 also seems to be essential for Glp binding and/or catalysis for GlpOΔ based on the HRP assay data. Like the R346M mutant, K429A forms an FAD N(5)-sulfite adduct, though with a ~4-fold higher Kd than wild-type. This is being interpreted to mean that this residue is not likely to be the sole reason for stabilizing an negative charge on the N(1) position of FAD, rather it is interacting with the phosphate moiety of the Glp substrate on binding.
The original hypothesis consisted of H65 being the active-site base necessary to extract a proton from Glp and drive the hydride transfer reaction. Glutamine cannot act as an active-site base but can interact with the Glp-phosphate. If H65 was the active-site base essential for catalysis, it would be expected that a mutation to glutamine would render the enzyme inactive. That is not seen in either the HRP coupled assay or EMT experiments. However, a mutation of this H65 to alanine severely affects the ability of
GlpOΔ to oxidize Glp. R69 also seems important for catalysis. A mutation of this residue to alanine severely affects catalysis, though not as much as the H65A mutant. A mutation of Y70 to phenylalanine yielded an enzyme that was 77% as active as wild-type. All this data taken together suggests that these residues could also interact with the Glp- phosphate moiety on binding.
Mutational analysis of Streptococcus sp. GlpOΔ was performed to elucidate residues important for substrate binding and catalysis. The mutations were made based on structural overlays (13) with B. subtilis ThiO (14) and yeast R. gracilis DAAO (15), and analysis of the active site and residues around the isoalloaxazine ring of FAD. Mutations
105 were made to five residues and all were found to be important for binding and/or catalysis. At this time, the initial hypothesis depicted in Figure 3 has been revised to exclude H65 as a possible active-site base, leading to the scheme shown in Figure 9.
106 Thr431
OON R H H O N N H Ile430 N H Lys429 HN N NH2
O Arg69 H NH2 HN O HN Arg346 H HN H OPO O H HN C NH O 2 H OH H Tyr70 O N NH B
His65 Figure 9: Refined mechanism for the oxidation of Glp to DHAP by GlpOΔ based on the results of mutagenic and kinetic experiments.
107 ACKNOWLEDGMENTS
I would like to thank Dr. Derek Parsonage for his help and guidance in using the stopped-flow spectrophotometer and analyzing the enzyme-monitored turnover data. I would also like to thank Mr. Takeshi Matsuoka of Asahi Pharma, Shizuoka, Japan, for supplying the glpO gene from Streptococcus sp. I would also like to thank Dr. Al
Claiborne for advising and editing this chapter.
108 REFERENCES
1. Claiborne, A. (1986) Studies on the structure and mechanism of Streptococcus faecium L-α-glycerophosphate oxidase, J Biol Chem 261, 14398-14407.
2. Lin, E. C. (1976) Glycerol dissimilation and its regulation in bacteria, Annu Rev Microbiol 30, 535-578.
3. Charrier, V., Luba, J., Parsonage, D., and Claiborne, A. (2000) Limited proteolysis as a structural probe of the soluble α-glycerophosphate oxidase from Streptococcus sp., Biochemistry 39, 5035-5044.
4. Austin, D., and Larson, T. J. (1991) Nucleotide sequence of the glpD gene encoding aerobic sn-glycerol 3-phosphate dehydrogenase of Escherichia coli K- 12, J Bacteriol 173, 101-107.
5. Brown, L. J., MacDonald, M. J., Lehn, D. A., and Moran, S. M. (1994) Sequence of rat mitochondrial glycerol-3-phosphate dehydrogenase cDNA. Evidence for EF-hand calcium-binding domains, J Biol Chem 269, 14363-14366.
6. Holmberg, C., Beijer, L., Rutberg, B., and Rutberg, L. (1990) Glycerol catabolism in Bacillus subtilis: nucleotide sequence of the genes encoding glycerol kinase (glpK) and glycerol-3-phosphate dehydrogenase (glpD), J Gen Microbiol 136, 2367-2375.
7. Klingenberg, M. (1970) Localization of the glycerol-phosphate dehydrogenase in the outer phase of the mitochondrial inner membrane, Eur J Biochem 13, 247- 252.
8. MacDonald, M. J., and Brown, L. J. (1996) Calcium activation of mitochondrial glycerol phosphate dehydrogenase restudied, Arch Biochem Biophys 326, 79-84.
9. Schryvers, A., Lohmeier, E., and Weiner, J. H. (1978) Chemical and functional properties of the native and reconstituted forms of the membrane-bound, aerobic glycerol-3-phosphate dehydrogenase of Escherichia coli, J Biol Chem 253, 783- 788.
10. Walz, A. C., Demel, R. A., de Kruijff, B., and Mutzel, R. (2002) Aerobic sn- glycerol-3-phosphate dehydrogenase from Escherichia coli binds to the cytoplasmic membrane through an amphipathic α-helix, Biochem J 365, 471-479.
109 11. Yeh, J. I., Du, S., Tortajada, A., Paulo, J., and Zhang, S. (2005) Peptergents: peptide detergents that improve stability and functionality of a membrane protein, glycerol-3-phosphate dehydrogenase, Biochemistry 44, 16912-16919.
12. Parsonage, D., Luba, J., Mallett, T. C., and Claiborne, A. (1998) The soluble alpha-glycerophosphate oxidase from Enterococcus casseliflavus. Sequence homology with the membrane-associated dehydrogenase and kinetic analysis of the recombinant enzyme, J Biol Chem 273, 23812-23822.
13. Colussi, T., Parsonage, D., Boles, W., Matsuoka, T., Mallett, T. C., Karplus, P. A., and Claiborne, A. (2008) Structure of α-glycerophosphate oxidase from Streptococcus sp.: a template for the mitochondrial α-glycerophosphate dehydrogenase, Biochemistry 47, 965-977.
14. Settembre, E. C., Dorrestein, P. C., Park, J. H., Augustine, A. M., Begley, T. P., and Ealick, S. E. (2003) Structural and mechanistic studies on ThiO, a glycine oxidase essential for thiamin biosynthesis in Bacillus subtilis, Biochemistry 42, 2971-2981.
15. Umhau, S., Pollegioni, L., Molla, G., Diederichs, K., Welte, W., Pilone, M. S., and Ghisla, S. (2000) The x-ray structure of D-amino acid oxidase at very high resolution identifies the chemical mechanism of flavin-dependent substrate dehydrogenation, Proc Natl Acad Sci U S A 97, 12463-12468.
16. Finn, R. D., Mistry, J., Schuster-Bockler, B., Griffiths-Jones, S., Hollich, V., Lassmann, T., Moxon, S., Marshall, M., Khanna, A., Durbin, R., Eddy, S. R., Sonnhammer, E. L., and Bateman, A. (2006) Pfam: clans, web tools and services, Nucleic Acids Res 34, D247-251.
17. Molla, G., Porrini, D., Job, V., Motteran, L., Vegezzi, C., Campaner, S., Pilone, M. S., and Pollegioni, L. (2000) Role of arginine 285 in the active site of Rhodotorula gracilis D-amino acid oxidase. A site-directed mutagenesis study, J Biol Chem 275, 24715-24721.
18. Molla, G., Motteran, L., Job, V., Pilone, M. S., and Pollegioni, L. (2003) Kinetic mechanisms of glycine oxidase from Bacillus subtilis, Eur J Biochem 270, 1474- 1482.
110 19. Holliday, G. L., Mitchell, J. B., and Thornton, J. M. (2009) Understanding the functional roles of amino acid residues in enzyme catalysis, J Mol Biol 390, 560- 577.
20. Massey, V., Muller, F., Feldberg, R., Schuman, M., Sullivan, P. A., Howell, L. G., Mayhew, S. G., Matthews, R. G., and Foust, G. P. (1969) The reactivity of flavoproteins with sulfite. Possible relevance to the problem of oxygen reactivity, J Biol Chem 244, 3999-4006.
21. Muller, F., and Massey, V. (1968) Flavin-Sulfite Complexes and Their Structures, J Biol Chem 244, 4007-4016.
22. Worthington Enzyme Manual (1972), pp. 43-45, Worthington Biochemical Co., Freehold, NJ.
23. Cornish-Bowden, A. (1995) Fundamentals of Enzyme Kinetics, Portland Press Ltd., London.
24. Loach, P. A. (1976) Handbook of Biochemistry and Molecular Biology, Vol. I, 3 ed., CRC Press, Boca Raton, FL.
25. Yeh, J. I., Chinte, U., and Du, S. (2008) Structure of glycerol-3-phosphate dehydrogenase, an essential monotopic membrane enzyme involved in respiration and metabolism, Proc Natl Acad Sci U S A 105, 3280-3285.
26. Gibson, Q. H., Swoboda, B. E., and Massey, V. (1964) Kinetics and Mechanism of Action of Glucose Oxidase, J Biol Chem 239, 3927-3934.
111 Chapter IV
CONCLUSIONS AND FUTURE DIRECTIONS
Timothy Colussi
112 The flavin adenine dinucleotide (FAD)1 dependent α-glycerophosphate oxidases
(GlpOs) and dehydrogenases (GlpDs) catalyze the oxidation of α-glycerophosphate (Glp) to dihydroxyacetone phosphate (DHAP) with concomitant reduction of the FAD cofactor
(1, 2). GlpDs are mostly considered membrane-associated; however, GlpOs from heme- deficient lactic acid bacteria, such as Streptococcus sp. and Enterococcus casseliflavus, are cytosolic (1). GlpOs and GlpDs are homologous based on sequence with one major difference being the presence of a ca. 50-residue insert that falls between two regions proposed earlier to be Glp-binding regions (2). The work described in this dissertation provides both structural and kinetic evidence for how the Glp substrate may bind and how catalysis may occur. Analysis of the Streptococcus sp. GlpO and GlpOΔ, a mutant of
GlpO lacking the ca. 50-residue insert, x-ray crystal structures show different conformations for key players within the active site and give insight into two possible forms of the resting oxidized enzyme as seen kinetically (3). Comparisons of GlpOΔ with two structural homologs show the presence of a conserved arginine that is implicated in substrate binding and orientation for both proteins, D-amino acid oxidase (DAAO) from the yeast Rhodotorula gracilis (4) and glycine oxidase (ThiO) from Bacillus subtilis (5).
Analysis of the structures also led to the generation of active-site mutants that were analyzed by both static and kinetic experiments. The results of these experiments give insight into how substrate binding and catalysis may occur in GlpO and can be extended to the GlpDs.
Streptococcus sp. α-Glycerophosphate Oxidase. Chapter II of this dissertation describes the genetically engineered deletion mutant of Streptococcus sp. GlpO, GlpOΔ.
1 Abbreviations: see page v.
113 The x-ray crystal structure of this deletion mutant was determined at 2.3 Å resolution.
This structure was then used as a search model in order to determine a 2.4 Å resolution data set of the intact GlpO (6) by molecular replacement (7). An overlay of the intact
GlpO and GlpOΔ shows a conformational change of H65, in which Nε2 shifts ~6 Å towards the active site from intact GlpO to GlpOΔ. There is also a shift of the isoalloxazine ring; the pyrazine and pyrimidine moieties of the FAD cofactor for intact
GlpO are shifted 2-3 Å toward the re face with respect to the same moieties in GlpOΔ, and Y357 of intact GlpO is positioned ~3.5 Å above the central ring of the isoalloxazine.
Y357 is contained within the 50-residue insert that is deleted in GlpOΔ. These conformational changes could represent two forms of the resting oxidized enzyme as described for E. casseliflavus GlpO (3). On searching the PDB with the DALI server (8), two structural homologs of GlpO were discovered, DAAO (4) and ThiO (5); both are members of the DAO pfam of which GlpO is a member. Overlays of GlpO, DAAO, and
ThiO show the presence of a structurally conserved arginine (GlpO R346, DAAO R285, and ThiO R302) within the active site. In DAAO and ThiO, this residue interacts with the carboxylate moieties of their respective substrates on binding and helps to orient the C(2) position of the respective substrates above the N(5) of the FAD cofactor (4, 5, 9). This led to the hypothesis that R346 in GlpO may interact with Glp on substrate binding, and aid in orienting the C(2) position above the N(5) of FAD. These overlays also showed residues within the active site that are conserved in GlpOs and GlpDs but are not present in DAAO or ThiO. Further analysis has led to the hypothesis that these residues, H65,
R69, Y70, and K429, provide a phosphate binding pocket. Along with possibly interacting with the phosphate moiety of Glp, H65 could, in principle, serve as an active-
114 site base that abstracts a proton from the C(2)-OH of Glp, driving the hydride transfer reaction necessary for catalysis.
Kinetic Analysis of Streptococcus sp. GlpOΔ Mutants. Analysis of the x-ray crystal structure of GlpOΔ led to the discovery of residues conserved in the GlpO and
GlpD active site. Chapter III of this dissertation describes the generation of mutations of these residues (H65A, H65Q, R69M, Y70F, R346M, and K429A) using PCR and their subsequent expression and kinetic analysis. The kinetic analysis employed the horseradish peroxidase (HRP) coupled assay (10) and enzyme-monitored turnover (EMT) analysis using the stopped-flow spectrophotometer. Sulfite titrations were also conducted in order to evaluate the ability of each mutant to stabilize an FAD N(5)-sulfite adduct, which mimics a hydride transfer reaction mechanism. All the mutants were able to stabilize sulfite adducts with a range of dissociation constants from 0.18 to 3.1 mM. The results of the kinetic experiments showed that R346 is an essential residue for substrate binding and/or catalysis, as it showed little or no activity by either the HRP coupled assay or EMT analysis. This, along with comparisons to the Escherichia coli GlpD-phosphate complex structure (11), has led to the proposition that this residue interacts with the C(1)-
OH of Glp on binding and aids in orienting the C(2) above the N(5) of FAD; similar to the way R302 and R285 for ThiO (5) and DAAO (9), respectively, interact with the carboxylate moieties of their respective substrates. Studies of the K429A mutant also support an essential role for K429 in substrate binding and/or catalysis, with the proposition being that it interacts with the Glp-phosphate.
H65 was hypothesized to be the active-site base required for proton abstraction, resulting in the oxidation of substrate and reduction of FAD. The results presented in
115 Chapter III of this dissertation seem to contradict that proposal. The H65Q mutant cannot serve as a base abstracting the C(2)-OH proton from Glp, but this enzyme is 63% as active as wild-type GlpOΔ by HRP analysis; however, mutation of this residue to alanine which is incapable of H-bonding (unlike glutamine), renders the enzyme ~2% as active as wild-type. This is in support of an important role for this residue in substrate binding via interaction with the phosphate moiety of Glp.
Other mutations to the proposed phosphate binding pocket include the R69M and
Y70F mutations. The results presented in Chapter III show that R69 seems to be important for substrate binding and/or catalysis due to the fact that it is 13% as active as wild-type by HRP coupled assay. The Y70F mutation had the least affect on activity of all the mutations (77% as active as wild-type) based on the HRP coupled assay. Based on these kinetic analyses, a reaction mechanism is proposed in Chapter III.
Future Kinetic Analysis of Streptococcus sp. GlpOΔ. The kinetic experiments described in Chapter III of this dissertation have paved the way for more experiments to be done. A comparison could not be made between the wild-type and mutant enzymes based on the kinetic parameters of Km and kcat due to the fact that for the EMT experiments with the mutants [Glp] << Km(Glp). Anaerobic stopped-flow experiments can be performed to determine the kred for each enzyme species, along with the Kd for
Glp. Reduction of flavin can be analyzed by monitoring the change in absorbance at 450 nm over a concentration range of Glp as described in Chapter III of this dissertation for enzyme-monitored turnover. In the simplest case, analysis of the absorbance change versus time will yield monophasic behavior, as has been seen for the intact streptococcal
GlpO (3) at a series of Glp concentrations. The pseudo-first order rate in such a case,
116 kobs, can be determined directly from single-exponential fits and a second-order rate constant can be determined from a plot of kobs versus [Glp] in that specific mechanism. In the case of wild-type streptococcal GlpOΔ and E. casseliflavus GlpO, biphasic behavior has been seen (3, 12). The kslow for both was [Glp] independent and was attributed to the presence of a second form, which is supported by the intact GlpO structure, of the resting oxidized enzyme and a second-order rate constant was calculated by plotting kfast versus
[Glp] (3, 12). Analyzing the reductive half reactions of the GlpOΔ mutants will add more to the kinetic parameters and shed more light on the reaction mechanism for GlpOΔ.
GlpD Membrane Association. The E. coli GlpD has been described as an integral membrane protein requiring detergent for purification (13), but a His-tagged version was purified without detergent solubilization (14); and has also been described as an integral membrane protein containing “six transmembrane-spanning” regions (15). A soluble form of the His-tagged E. coli GlpD has been purified in this laboratory without the use of detergent (Figure 1A) consistent with the report in (14). To determine whether this enzyme is active, a ferricyanide [Fe(CN)6] reduction assay was employed. Fe(CN)6 acts as a two-electron acceptor that can accept the reducing equivalents from FAD of GlpD after substrate reduction. The absorbance change at 420 nm was monitored on a microplate reader (Figure 1B) and the turnover number at 0.2 M L-α-Glp and 1 mM
-1 Fe(CN)6 was determined to be 10.7 s .
The x-ray crystal structure of E. coli GlpD (PDB code 1QCU) was determined using an enzyme preparation purified using 1-O-n-octyl-β-D-glucopyranoside (β-OG)
(11). This enzyme was also reported to lose 40% activity after 1 day in the presence of
1% β-OG, utilizing a phenazine methosulfate (PMS)-coupled reduction of 2-(4,5-
117
Figure 1: Expression, purification, and activity of E. coli GlpD. (A) A gel showing the expression and purification of E. coli GlpD. Fractions 8-13 from the Ni-NTA column were pooled and buffer exchanged/concentrated to a final concentration of 10 mg/mL in 50 mM potassium phosphate buffer, pH 7.0, with 0.5 mM EDTA and 20 % glycerol. Aliquots of 500 μL were flash frozen in liquid nitrogen and stored at -80 °C. (B) Fe(CN)6 activity assay showing that the purified E. coli GlpD is active. Varied concentrations of GlpD enzyme (10-100 nM) were added to a reaction buffer containing 0.2 M D,L-α- glycerophosphate, 0.1 M potassium phosphate, pH 7.0, 0.3 mM EDTA, and 1 mM Fe(CN)6 to a final reaction volume of 132 μL (2 μL of enzyme plus 130 μL of reaction buffer). The absorbance change at 420 nm was monitored for 300 s on a Tecan Safire2 microplate reader.
118 dimethyl-2-thiazolyl)-3,5-diphenyl-2-H-tetrazolium bromide (MTT) (11). A monomer- monomer superposition with GlpOΔ gives an RMSD of 3.0 Å. Dimer formation, however, is quite different. The E. coli GlpD dimer interface utilizes portions of the
FAD-binding domain and C-terminal domain (11), whereas the dimer interface of GlpOΔ is formed by portions of the dimerization/substrate-binding domain (Figure 2A). The portion of GlpOΔ that forms the dimer interface has been described as being the portion of E. coli GlpD that imbeds into the membrane bilayer (11). In 2008, the x-ray crystal structure of Bacillus halodurans GlpD was solved and deposited to the PDB (PDB code
3DA1). Monomer-monomer superpositions of the B. halodurans GlpD with GlpOΔ and
E. coli GlpD give an RMSD of 1.3 Å and 2.3 Å, respectively. B. halodurans is 44% and
32% identical by sequence to GlpOΔ and E. coli GlpD, respectively. Upon analysis of crystal symmetry with COOT (16), a symmetry mate was discovered that gave the best dimer for the B. halodurans GlpD. This dimer utilizes regions of the dimerization/substrate-binding domain, similar to that of GlpOΔ (Figure 2B), but not identical.
Based on the structure of GlpO presented in Chapter II of this dissertation, we hypothesize that the C-terminal domain in GlpD provides the scaffold for membrane association. The structure of E. coli GlpD (11) would seem to contradict that hypothesis; however, the structure of B. halodurans GlpD gives some support. To test this hypothesis, the Bacillus subtilis GlpD enzyme will be utilized. Previous reports have stated that expression of the B. subtilis GlpD in E. coli did not yield protein (17). A subtilin-regulated gene expression (SURE) system has been developed utilizing the
Gram-positive bacterium B. subtilis (18). The glpD gene has been amplified by PCR in
119 Figure 2: Dimer overlays of E. coli and B. halodurans GlpD with Streptococcus sp. GlpOΔ. (A) Superposition of monomer A of the E. coli GlpD (PDB code 1QCU) dimer with monomer A of the GlpOΔ (PDB code 2RGH) dimer. The RMSD for the monomers is 2.3 Å. E. coli GlpD is colored in deep purple while GlpOΔ is color-coded by domain: sand for FAD-binding, blue for dimerization/substrate-binding, and red for C-terminal. (B) Superposition of monomer A of the B. halodurans GlpD dimer (PDB code 3DA1) with monomer A of the GlpOΔ (PDB code 2RGH) dimer. The RMSD for the monomers is 1.3 Å. B. halodurans GlpD is colored in forest green while GlpOΔ is color-coded by domain as described above.
120
121
this laboratory and sent to Dr. Oscar Kuipers’ laboratory at the University of Groningen.
Dr. Hein Trip of Dr. Kuipers’ laboratory has kindly cloned the glpD gene into two SURE system expression vectors, pNZ8901 (chloramphenicol resistant) and pNZ8902
(erythromycin resistant). Initial expression trials, performed by Dr. Trip, using both expression plasmids show a moderate level of subtilin-induced B. subtilis GlpD protein expression (Figure 3). Figure 3 also shows that after an OD600 of 1.0 is reached, the B. subtilis NZ8900 cells do not continue to grow coinciding with subtilin induction and protein expression. The data presented in Figure 3 is the only representative of B. subtilis
GlpD protein expression using the SURE system, optimization will undoubtedly have to be done to maximize expression. After protein expression has been optimized, mutations deleting portions of the C-terminal domain will be generated. GlpD is necessary for B. subtilis to grow on glycerol (17). The ability of cells containing these mutant GlpDs to grow on glycerol will be analyzed.
The future experiments described in this chapter pertaining to the GlpOΔ mutants will provide a means to help further elucidate the role each of these key residues play in the active site of GlpOs, and based on sequential similarity, GlpDs. The experiments described for the B. subtilis GlpD will test the hypothesis presented for the role the C- terminal domain plays in membrane association.
122
Figure 3: B. subtilis GlpD protein expression using the SURE system. Growth curves demonstrate the presence of the GlpD gene in either expression vector, pNZ8901 (pNZ01, chloramphenicol resistance) or pNZ8902 (pNZ02, erythromycin resistance), does not have an effect on B. subtilis growth; however, upon induction with subtilin and expression, the growth stops. The resulting SDS-PAGE gel shows that a moderate amount of GlpD protein is expressed using either vector. This figure was kindly provided by Dr. Hein Trip, University of Groningen, The Netherlands.
123 REFERENCES
1. Claiborne, A. (1986) Studies on the structure and mechanism of Streptococcus faecium L-α-glycerophosphate oxidase, J Biol Chem 261, 14398-14407.
2. Austin, D., and Larson, T. J. (1991) Nucleotide sequence of the glpD gene encoding aerobic sn-glycerol 3-phosphate dehydrogenase of Escherichia coli K- 12, J Bacteriol 173, 101-107.
3. Charrier, V., Luba, J., Parsonage, D., and Claiborne, A. (2000) Limited proteolysis as a structural probe of the soluble α-glycerophosphate oxidase from Streptococcus sp, Biochemistry 39, 5035-5044.
4. Umhau, S., Pollegioni, L., Molla, G., Diederichs, K., Welte, W., Pilone, M. S., and Ghisla, S. (2000) The x-ray structure of D-amino acid oxidase at very high resolution identifies the chemical mechanism of flavin-dependent substrate dehydrogenation, Proc Natl Acad Sci U S A 97, 12463-12468.
5. Settembre, E. C., Dorrestein, P. C., Park, J. H., Augustine, A. M., Begley, T. P., and Ealick, S. E. (2003) Structural and mechanistic studies on ThiO, a glycine oxidase essential for thiamin biosynthesis in Bacillus subtilis, Biochemistry 42, 2971-2981.
6. Finnerty, C. M., Charrier, V., Claiborne, A., and Karplus, P. A. (2002) Crystallization and preliminary crystallographic analysis of the soluble α- glycerophosphate oxidase from Streptococcus sp, Acta Crystallogr D Biol Crystallogr 58, 165-166.
7. Rossmann, M. G. (1990) The molecular replacement method, Acta Crystallogr A 46 ( Pt 2), 73-82.
8. Holm, L., Kaariainen, S., Rosenstrom, P., and Schenkel, A. (2008) Searching protein structure databases with DaliLite v.3, Bioinformatics 24, 2780-2781.
9. Molla, G., Porrini, D., Job, V., Motteran, L., Vegezzi, C., Campaner, S., Pilone, M. S., and Pollegioni, L. (2000) Role of arginine 285 in the active site of Rhodotorula gracilis D-amino acid oxidase. A site-directed mutagenesis study, J Biol Chem 275, 24715-24721.
124 10. Worthington Enzyme Manual (1972), pp. 43-45, Worthington Biochemical Co., Freehold, NJ.
11. Yeh, J. I., Chinte, U., and Du, S. (2008) Structure of glycerol-3-phosphate dehydrogenase, an essential monotopic membrane enzyme involved in respiration and metabolism, Proc Natl Acad Sci U S A 105, 3280-3285.
12. Parsonage, D., Luba, J., Mallett, T. C., and Claiborne, A. (1998) The soluble α- glycerophosphate oxidase from Enterococcus casseliflavus. Sequence homology with the membrane-associated dehydrogenase and kinetic analysis of the recombinant enzyme, J Biol Chem 273, 23812-23822.
13. Schryvers, A., Lohmeier, E., and Weiner, J. H. (1978) Chemical and functional properties of the native and reconstituted forms of the membrane-bound, aerobic glycerol-3-phosphate dehydrogenase of Escherichia coli, J Biol Chem 253, 783- 788.
14. Walz, A. C., Demel, R. A., de Kruijff, B., and Mutzel, R. (2002) Aerobic sn- glycerol-3-phosphate dehydrogenase from Escherichia coli binds to the cytoplasmic membrane through an amphipathic α-helix, Biochem J 365, 471-479.
15. Yeh, J. I., Du, S., Tortajada, A., Paulo, J., and Zhang, S. (2005) Peptergents: peptide detergents that improve stability and functionality of a membrane protein, glycerol-3-phosphate dehydrogenase, Biochemistry 44, 16912-16919.
16. Emsley, P., and Cowtan, K. (2004) Coot: model-building tools for molecular graphics, Acta Crystallogr D Biol Crystallogr 60, 2126-2132.
17. Holmberg, C., Beijer, L., Rutberg, B., and Rutberg, L. (1990) Glycerol catabolism in Bacillus subtilis: nucleotide sequence of the genes encoding glycerol kinase (glpK) and glycerol-3-phosphate dehydrogenase (glpD), J Gen Microbiol 136, 2367-2375.
18. Bongers, R. S., Veening, J. W., Van Wieringen, M., Kuipers, O. P., and Kleerebezem, M. (2005) Development and characterization of a subtilin-regulated expression system in Bacillus subtilis: strict control of gene expression by addition of subtilin, Appl Environ Microbiol 71, 8818-8824.
125 SCHOLASTIC VITAE
Timothy M. Colussi
Wake Forest University School of Medicine Center for Structural Biology Department of Biochemistry Winston-Salem, NC 27157 (336) 716-6050, E-mail: [email protected]
Education:
Ph.D., Biochemistry and Molecular Biology, Wake Forest University School of Medicine, Winston-Salem, NC. Degree anticipated October, 2009 Advisor: Dr. Al Claiborne, Professor of Biochemistry
Bachelor of Science, Chemistry, Wright State University, Dayton, OH. June, 2001
Doctoral Research: Wake Forest University School of Medicine, Winston-Salem, NC. 08/2004-present. Advisor: Dr. Al Claiborne
My dissertation work has involved structural studies of flavoproteins involved in redox biology. The specific flavoprotein I have been studying is α- glycerophosphate oxidase (GlpO) from Streptococcus sp. This enzyme oxidizes α-glycerophosphate (Glp) to dihydroxyacetone phosphate (DHAP) with concomitant reduction of the flavin adenine dinucleotide (FAD) cofactor, which is reoxidized by reacting with O2 and forming H2O2. GlpO has high sequence identity with the membrane associated Glp-dehydrogenases (GlpD) from bacterial sources, such as Escherichiae coli and Bacillus subtilis, and with mitochondrial GlpDs. Highlights include:
1. We have published the structure of the Streptococcus sp. GlpO enzyme along with a deletion mutant that lacks a 50-residue insert found in GlpOs but not GlpDs (Colussi, T., Parsonage, D., Boles, W., Matsuoka, T., Karplus, P. A., Mallett, T. C., and Claiborne, A. (2008) Biochemistry 47, 956-977). Analysis of this structure has led to hypotheses involving residues conserved in GlpOs and GlpDs that we believe are important for substrate binding and catalysis.
2. Mutations have been made to residues we hypothesize to be essential for substrate binding and catalysis. Kinetic analysis has been performed on these mutants. The results of these experiments have led to the development of a proposed mechanism for these enzymes found in both bacteria and the mitochondrion.
126 Academic Awards
2007 Selected to attend National Synchrotron Light Source (NSLS) RapiData course in Macromolecular X-ray Diffraction Measurement
2007 Travel Grant from the American Crystallographic Association (ACA) to attend the 2007 ACA meeting in Salt Lake City, UT
Publications
Colussi, T., Parsonage, D., Boles, W., Matsuoka, T., Karplus, P. A., Mallett, T. C., and Claiborne, A. (2008) Structure of α-Glycerophosphate Oxidase from Streptococcus sp.: A Template for the Mitochondrial α-Glycerophosphate Dehydrogenase. Biochemistry 47, 956-977.
Poster Presentations
Al Claiborne, Derek Parsonage, William Boles, T. Conn Mallett, Timothy Colussi, and Takeshi Matsuoka. Crystallographic Analyses of the Soluble α-Glycerophosphate Oxidase from Streptococcus sp.: A Structural Template for the Mitochondrial α- Glycerophosphate Dehydrogenase. 15th International Symposium on Flavins and Flavoproteins, Shonan Village Center, Japan (2005).
Timothy Colussi, T. Conn Mallett, William Boles, Derek Parsonage, P. Andrew Karplus, Takeshi Matsuoka, and Al Claiborne. Streptococcus sp. α-Glycerophosphate Oxidase: A structural Template for the Mitochondrial α-Glycerophosphate Dehydrogenase. 36th Annual Mid-Atlantic Macromolecular Crystallography Meeting, Wake Forest University, Winston-Salem, NC (2006).
Timothy Colussi, Derek Parsonage, William Boles, Takeshi Matsuoka, P. Andrew Karplus, T. Conn Mallett, and Al Claiborne. Crystallographic Analysis of the Flavin- Linked α-Glycerophosphate Oxidase from Streptococcus sp.: Structural Insight for the Respiratory α-Glycerophosphate Dehydrogenase from Bacillus subtilis. 4th Conference on Functional Genomics of Gram-Positive Microorganisms, Tirrenia, Pisa, Italy (2007).
Timothy Colussi, Derek Parsonage, William Boles, Takeshi Matsuoka, P. Andrew Karplus, T. Conn Mallett, and Al Claiborne. Structure of α-Glycerophosphate Oxidase from Streptococcus sp.: A Structural Template for the Mitochondrial α- Glycerophosphate Dehydrogenase. The 2007 Meeting of the American Crystallographic Association, Salt Lake City, UT (2007).
Timothy Colussi, Derek Parsonage, William Boles, Takeshi Matsuoka, P. Andrew Karplus, T. Conn Mallett, and Al Claiborne. Structure of α-Glycerophosphate Oxidase from Streptococcus sp.: A Structural Template for the Mitochondrial α- Glycerophosphate Dehydrogenase. The 8th Annual Wake Forest University Graduate School of Arts and Sciences Graduate Student Research Day (2008).
127 References
Dr. Al Claiborne (Advisor) Wake Forest University [email protected]
Dr. Thomas Hollis Wake Forest University [email protected]
Dr. Conn Mallett Rigaku Europe [email protected]
128