THE ROLE OF THE N(5) INTERACTION AND ASSOCIATED CONFORMATIONAL CHANGES IN THE MODULATION OF THE REDOX PROPERTIES IN FLAVOPROTEINS
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the
Graduate School of The Ohio State University
By
Mumtaz Kasim, M.Sc.
* * * * *
The Ohio State University 2002
Dissertation Committee:
Dr. Richard P. Swenson, Advisor Approved by
Dr. Gary Means ______
Dr. Caroline Breitenberger Advisor
Dr. Mark Foster Department of Biochemistry
ABSTRACT
Many biochemical processes exploit the remarkable versatility of flavoenzymes
and their flavin cofactors by modulating the numerous interactions of the cofactor with
the apoflavoprotein. The interaction between the protein and N(5) of the flavin cofactor
has been of particular interest for this very reason. In the Clostridium beijerinckii
flavodoxin, the four residue reverse turn –Met56-Gly-Asp-Glu59- provides the majority of the critical interactions with the flavin mononucleotide cofactor (FMN) that contribute to the binding and differential stabilization of its three redox states. This turn undergoes a conversion from a mix of cis/trans peptide configurations that approximates a type II turn in the oxidized state to a type II′ turn upon reduction. This change results in the formation of a new hydrogen bond between the N(5)H of the reduced flavin and the carbonyl group of Gly57 of the central peptide bond of the turn, an interaction that contributes to the modulation of the oxidation-reduction potentials of the cofactor. Systematic replacement of the second and third residues of the turn (Gly57 and Asp58) with –Gly-Gly-, -Gly-
Ala-, -Ala-Gly- and –Ala-Ala- dipeptidyl sequences resulted in an altered stability of the
FMN semiquinone that was directly correlated to the conformational energetics of the turn. In addition, sequential elimination of all side chain interactions in various combinations through an alanine-scanning mutagenesis approach proved the overriding ii
importance of the main chain interactions with the N(5)H of the FMN and the associated
conformational change in this loop to be the primary determinant of the thermodynamic
stabilization of the FMN semiquinone.
In contrast, in the structurally homologous FMN-binding domain of cytochrome
P450 reductase from Bacillus megaterium, a main chain hydrogen bond to N(5) is present
in the oxidized state. 15N-NMR studies indicate that in this case, a conformational change
occurs in the flavin at the N(5) position. Sequence specificity of the type I′ turn adopted
by the residues –Tyr536-Asn-Gly-His539- was tested by replacement of the central residues
of the turn (Asn537 and Gly538) with –Gly-Gly-, -Gly-Ala-, -Ala-Gly- and –Ala-Ala-
dipeptidyl sequences. These mutations established the critical role of the position of the
glycine residue in maintaining turn stability. We conclude that the N(5) interaction and
the associated conformational change, as well as sequence specificity of the turn involved
in flavin binding, play a critical role in determining the redox properties of flavoproteins.
iii
To my dad, who always believed in me,
and to my husband, who taught me
to believe in myself.
iv
ACKNOWLEDGMENTS
I would like to thank my advisor, Dr. Richard P. Swenson, for his guidance, support and patience.
I would like to also thank my family who always had faith in me and who continue to understand my long absence away from home.
I would like to thank past and present members of the Swenson laboratory, especially, Dr Lawrence Druhan, Dr. Fu-Chung Chang, Dr. Yucheng Feng, Dr. Luke
Bradley, James Wu, Kun-Yun Yang, Tracey Murray and Michelle Nauerth, for valuable help and insight.
I would like to thank the people in Buckeye Tang Soo Do, whose support and friendship helped me through the years.
I would like to thank my friends: Joe Davis, Paul Erwin and especially Shane
Mellor, for being there when I needed them and for making these past few years more fun than I thought was possible.
Finally, I would like to thank my many friends in Graduate School, in particular:
Milan Jovanovic, Craig McElroy, Srisunder Subramanium, Ryan Pereira and Manoj Nair for providing a very interesting work environment.
v
VITA
April 6, 1972……………………………… Born – Bombay, India
1993……………………………………….. B. Sc. Life Sciences/Biochemistry, St. Xavier's College, University of Bombay
1995………………………………….……. M. Sc. Biochemistry, University of Bombay
1996 – present…………………………….. Graduate Teaching and Research Associate, The Ohio State University
PUBLICATIONS
1. Kasim, M. and Swenson, R. P. “Alanine-Scanning of the 50’s Loop in the Clostridium beijerinckii Flavodoxin: Evaluation of Additivity and the Importance of Interactions Provided by the Main Chain in the Modulation of the Oxidation- Reduction Potentials.” Biochemistry, 40, 13548–13555, (2001).
2. Kasim, M. and Swenson, R. P. “Conformational Energetics of a Reverse Turn in the Clostridium beijerinckii Flavodoxin is Directly Coupled to the Modulation of its Oxidation-Reduction Potentials.” Biochemistry, 39, 15322-15332, (2000).
3. Swenson, R. P., Kasim, M., Bradley, L. and Druhan, L. “Role of Conformational Dynamics and Associated Electrostatic and Hydrogen Bonding Interactions in the Regulation of Redox Potentials in the Clostridium beijerinckii Flavodoxin.” In Flavins and Flavoproteins 1999 (Ghisla, S., Kroneck, P., Macheroux, P., and Sund, H., Eds) Agency for Scientific Publishing, Berlin, 183-186, (2000).
vi
FIELDS OF STUDY
Major Field: Biochemistry
vii
TABLE OF CONTENTS
PAGE
Abstract……………………………………………………………………….. ii
Dedication…………………………………………………………………….. iv
Acknowledgments…………………………………………………………….. v
Vita……………………………………………………………………………. vi
List of Tables…………………………………………………………………. x
List of Figures………………………………………………………………… xii
List of Abbreviations…………………………………………………………. xvi
Chapters:
1. Introduction…………………………………………………………... 1
2. Materials and Methods………………………………………………. 34
3. Conformational Energetics of a Reverse Turn in the Clostridium beijerinckii Flavodoxin is Directly Coupled to the Modulation of its Oxidation- Reduction Potentials………………………..…………… 42
Introduction………………………………………………………….. 42
Materials and Methods………………………………………………. 49
Results……………………………………………………………….. 49
viii
Discussion…………………………………………………………… 71
4. Alanine-Scanning of the 50’s Loop in the Clostridium beijerinckii Flavodoxin: Evaluation of Additivity and the Importance of Interactions Provided by the Main Chain in the Modulation of the Oxidation-Reduction Potentials……………………………………… 84
Introduction………………………………………………………….. 84
Materials and Methods………………………………………………. 91
Results……………………………………………………………….. 91
Discussion…………………………………………………………… 101
5. Cloning and Characterization of the FMN-Binding Domain of Cytochrome P450 Reductase From Bacillus Megaterium…………… 113
Introduction………………………………………………………….. 113
Materials and Methods………………………………………………. 123
Results……………………………………………………………….. 123
Discussion…………………………………………………………… 153
6. The FMN-Binding Domain of P450BM-3: Investigation into the Possible Mechanisms of Redox Tuning……………………………... 159
Introduction………………………………………………………….. 159
Materials and Methods………………………………………………. 163
Results……………………………………………………………….. 164
Discussion…………………………………………………………… 184
7. General Conclusions and Future Directions…………………………. 190
List of references……………………………………………………………… 196
ix
LIST OF TABLES
TABLE PAGE
1. Turns in the flavodoxin from Clostridium beijerinckii………………… 15
2. Idealized dihedral angles of hydrogen bonded β turns………………… 17
3. Sequence and midpoint potential comparisons of several flavodoxins... 18
4. Oxidation-reduction midpoint potentials for the Clostridium beijerinckii flavodoxin…………………………………………………. 20
5. Energies of β turn formation for the Type II and Type II′ turns……….. 23
6. Relative free-energy changes for refolding from the Type II to the Type II′ turn……………………………………………………………. 24
7. Energies of β turn formation for the Type I′ turn……….…………….. 31
8. Relative free-energy changes for refolding from the extended to the Type I′ turn conformation……...………………………………………. 32
9. Oxidation-reduction midpoint potentials, FMN dissociation constants and Gibbs free energy changes of wild-type and mutant C. beijerinckii flavodoxins…………………………………………………………….. 57
10. 15N Chemical shifts for free and bound FMN in the oxidized state, pH 7.0, 300°K for C. beijerinckii………………………………………... 67
11. 1H-15N HSQC temperature coefficients for the Clostridium beijerinckii wild-type and mutant flavodoxins in the oxidized state….. 70
x
12. Oxidation-reduction midpoint potentials, FMN dissociation constants and Gibbs free energy of FMN binding for wild-type and mutant C. beijerinckii flavodoxins……………………………….……………….. 97
13. 15N Chemical shifts for free and bound FMN in the oxidized state, pH 7.0, 300°K for BM-3…………………………………………………… 145
14. 15N Chemical shifts for free and bound FMN in the reduced state, pH 7.0, 300°K for BM-3…………………………………………………… 149
15. Coupling constants for N(3)H and N(5)H in the oxidized and fully reduced states of BM-3…………………………………………….…. 150
16. 15N Chemical shifts for free and bound FMN in the oxidized state, pH 7.0, 300°K for –537Ala-Ala-…………………………………………… 172
xi
LIST OF FIGURES
FIGURE PAGE
1. The structures of riboflavin, flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD)……………………………………………. 2
2. The three oxidation states of the isoalloxazine ring…………………… 4
3. The UV-visible spectra of the flavin cofactor in different redox states... 5
4. Structure of the Clostridium beijerinckii flavodoxin…………………... 12
5. Comparison of the turn conformation in the oxidized and semiquinone states of the Clostridium beijerinckii flavodoxin………………………. 14
6. Structure of the FMN-binding site in BM-3 in the oxidized state showing the major interactions with the isoalloxazine ring…………… 27
7. Sequence alignment of the inner flavin binding loop of BM-3 and CPR…………………………………………………………………….. 28
8. Structure of the reverse turn involving residue 56-59 in the C. beijerinckii flavodoxin………………………………………………… 43
9. Representations of various turn conformations showing the orientation of the central carbonyl group…………………………………………... 46
10. UV-visible absorbance spectra for the C. beijerinckii –Gly-Ala- (solid lines) and the –Ala-Gly- (dashed lines) flavodoxin mutants in all three oxidation states………………………………………………………... 52
11. Representative titration for the determination of Eox/sq (panel A) and Esq/hq (panel B) for the for the –Ala-Gly- and the –Ala-Ala- mutants respectively…………………………………………………... 55
xii
12. Representative determination of the dissociation constant in the oxidized state shown for the –Ala-Gly- (circles) and the –Ala-Ala- (squares) mutants……………………………………………………… 61
13. Diagram depicting the free energy of binding of the FMN cofactor in all three oxidation states for wild type and mutant C. beijerinckii flavodoxins…………………………………………………………….. 64
14. Temperature dependencies of the proton chemical shift for the N(3)H oxidized of the bound 15N-labeled FMN for wild type and mutant C. beijerinckii flavodoxins………………………………………………... 69
15. Correlation between the midpoint potential values (in mV, pH 7) for the ox/sq couple (closed symbols) and for the sq/hq couple (open symbols) and the calculated conformational free energy differences between type II and type II′ turns (in kcal/mol)……………………….. 75
16. Thermodynamic cycles describing the free energies associated with the conformational changes and FMN binding in each oxidation state………………………………………………………… 77
17. Structure of the FMN binding site in wild-type Clostridium beijerinckii flavodoxin in the oxidized state showing the major interactions with the isoalloxazine ring……………………………….. 86
18. UV-visible absorbance spectra for wild-type (closed squeares), 56MGAA (open squares), 56AGDA (open circles) and 56AGAA (closed circles) flavodoxins in the oxidized (panel A) and fully reduced (panel B) states……………………………………………….. 93
19. Representative oxidation-reduction potential determinations for 56AGDA……………………………………………………………….. 96
20. Determination of the dissociation constant for the complex between oxidized FMN and mutant apoflavodoxins……………………………. 99
21. Histogram depicting the differences between the binding free energy changes for the FMN cofactor in the oxidized (filled bar), semiquinone (shaded bar) and hydroquinone (open bar) states for the single, double and triple mutants……………………………………… 103
xiii
22. Plot of the free energy changes predicted from the constituting single and double mutants versus the actual observed free energy changes for the multiple mutant in the oxidized (closed symbols), semiquinone (shaded symbols) and hydroquinone (open symbols) states…………… 107
23. Schematic of the domain organization in P450BM-3……………….… 115
24. Structural similarities between the bacterial flavodoxin (top panel) and the FMN-binding domain of P450BM-3 (bottom panel)…………. 116
25. Electron transfer mechanisms of CPR and P450BM-3………………... 119
26. Electrostatic potential around the flavin binding site………………….. 121
27. Expression of wild-type BM-3 under the control of the T7 promoter in BL21(DE3) cells……………………………………………………. 125
28. UV-visible absorbance spectra of the FMN-binding domain of P450BM-3 in the oxidized and fully reduced states…………………... 126
29. Representative titration ofr the determination of Eox/hq for BM-3 at different pH values…………………………………………………….. 129
30. Variation of Eox/hq with pH…………………………………………... 130
31. Schemes depicting the possible modes of electron and proton transfer…………………………………………………………. 133
32. Determination of Eox/sq and Esq/hq (panel A) and their pH dependency (panel B)……………………………………………… 134
33. Spectrophotometric titration of FMN to apoBM-3…………………… 138
34. Determination of the dissociation constant in the oxidized state of the FMN cofactor to apoBM-3……………………………………... 139
35. One-dimensional 15N-NMR spectra of BM-3 in the oxidized (panel A) and fully reduced (panel B) states…………………………... 141
36. One-dimensional 15N-NMR spectra obtained using the DEPT pulse sequence with (panel A) and without (panel B) broad-band decoupling in the oxidized state……………………………………….. 144 xiv
37. One-dimensional 15N-NMR spectra obtained using the DEPT pulse sequence without broad-band decoupling in the fully reduced state…………………………………………………………… 148
38. Temperature dependencies of the proton chemical shift for the N(3)H oxidized (closed circles), N(3)H reduced (open circles) and N(5)H reduced (closed squares) of the bound 15N-labeled FMN for BM-3…………………………………………………………………... 152
39. Diagrammatic depeiction of the N(10)-C(9a)-C(5a)-N(5) edge of the flavin ring detailing the events occurring upon reduction…………. 156
40. Representative UV-visible absorbance spectra of the N537A mutant in the oxidized and fully reduced states……………………………….. 165
41. Representative determinations of Eox/hq for the N537A (closed circles) and the N537G (open circles) mutants………………………... 166
42. Representative binding titrations for the determination of the dissociation constant of the FMN cofactor in the oxidized state for the N537A(closed circles) and N537G(open circles) mutants………………………………………………………………… 167
43. Determination of thedissociation constant for the –537Gly-Ala- (A) and the –537Ala-Ala- (B) mutants…………………………………….. 170
44. UV-visible absorbance spectra of the Y536H mutant in the oxidized and fully reduced states at pH 7.0…………………………… 176
45. Representative determinations of Eox/hq for the Y536H mutant……... 177
46. Determination of the dissociation constant for the complex between oxidized FMN and the mutant Y536H………………………………… 180
47. UV-visible absorbance spectra of the Y736H mutant in the oxidized and fully reduced states at pH 6.1…………………………… 181
48. Determination of the dissociation constant for the Y536R mutant……. 183
xv
LIST OF ABBREVIATIONS
1 D One-dimensional 2 D Two-dimensional BM-3 FMN-binding domain of P450BM-3 CPR Cytochrome P450 reductase from eukaryotes DEPT Distortionless enhancement of polarization transfer DSS Sodium-2,2-dimethyl-2-silapentane-5-sulfonate Eh System potential Eox/sq Midpoint potential of the oxidized-semiquinone couple Esq/hq Midpoint potential of the semiquinone-hydroquinone couple FAD Flavin adenine dinucleotide FMN Flavin mononucleotide FNR Ferredoxin NADP+ reductaset HQ Two-electron reduced hydroquinone state HSQC Heteronuclear single-quantum coherence NADPH Nicotinamide adenine dinucleotide phosphate reduced form NMR Nuclear magnetic resonance OX Oxidized state OX/SQ Oxidized-semiquionone couple OYE Old Yellow Enzyme P450BM-3 Cytochrome P450 reductase from Bacillus megaterium SQ One-electron reduced semiquinone state SQ/HQ Semiquinone-hydroquinone couple TARF Tetraacetylriboflavin
xvi
CHAPTER1
INTRODUCTION
Flavin was first identified in the late 19th century as a component of whey and was subsequently termed “lactochrome” (Blyth, 1879). Subsequently, this yellow pigment was isolated from a variety of biological sources and was named ovoflavin, hepatoflavin or lactoflavin depending on the source of isolation. It soon became apparent that all these compounds were essentially derivatives of riboflavin (vitamin B2), which is the most abundant flavin compound found in nature. Humans and other higher animals, unlike lower organisms, cannot synthesize the isoalloxazine component of flavins and depend upon their diet to obtain sufficient amounts to meet their metabolic requirements. The symptoms of riboflavin deficiency that result from general malnutrition or bizarre diets include an inflamed tongue, lesions in the corners of the mouth and dermatitis.
1
Flavin Adenine Dinucleotide lavin Mononucleotide F The structures of riboflavin, flavin mononucleotide (FMN) and adenine Riboflavin
Figure 1. dinucleotide (FAD). The numbering of the atoms in isoalloxazine ring is shown.
Ribityl chain Isoalloxazine Ring
2
The chemical structure of riboflavin was elucidated in the 1930’s (Karrer et al., 1935) and was found to consist of two main parts: the tricyclic ring component (isoalloxazine) and the ribityl tail end (Figure 1). The isoalloxazine ring is formed by a fusion of the dimethylbenzene, pyrazine and pyrimidine rings and is the redox-active component of the flavin cofactor. Several derivatives of riboflavin were later identified, the most common being flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD), the structures are shown in Figure 1.
3
The three oxidation states of the isoalloxazine ring. The physiologically relevant ionization of the Figure 2. semiquinone is shown, which occurs at a pKa of ~8.5.semiquinone is shown, which The hydroquinone is shown as the naturally occuring anionic form.
4
0.4 1
0.2 2b
Absorbance 2a
3
0.0 300 400 500 600 700
Wavelength (nm)
Figure 3. The UV-visible spectra of the flavin cofactor in different redox states. Spectrum 1 is oxidized FMN, spectrum 2a is the red anionic semiquinone, spectrum 2b is the blue neutral semiquinone and spectrum 3 is the anionic hydroquinone.
5
A distinctive chemical property of the flavin cofactor is its ability to participate in a variety of oxidation- reduction reactions. It has long since been realized that flavins are capable of accepting both one- and two-electrons and, as a result, can exist in three different oxidation states – the oxidized state, the one-electron reduced semiquinone state and the two-electron reduced hydroquinone state. Depending upon the pH (or protonation state), there are several forms of flavin for each oxidation state. These states and the most commonly occurring ionizations in nature are depicted in Figure 2. What is further
remarkable is that each of these states (and ionized species) has very distinct spectral
characteristics, making them relatively easy to monitor spectrophotometrically (Figure
3). Oxidized flavin has a visible absorption spectrum with maxima centered around 374
nm and 444 nm, the exact value of which is dependent upon the environment. Even
within a family of proteins, these values can change significantly (Mayhew & Tollin,
1992). Deprotonation at N(3)H occurs at a pH of 10 (Dudley et al., 1964). The flavin
semiquinone is protonated at neutral pH and exhibits a blue color. The pKa of N(5)H is ~
8.5 and deprotonation occurs above this pH to yield the red anionic semiquinone
(Ehrenberg et al., 1967). The hydroquinone state has a pKa at N(1) of 6.7 (Dudley et al.,
1964; van Schagen & Muller, 1981).
A family of proteins that bind these flavin compounds as cofactors are collectively termed flavoproteins. A large class of flavoproteins is now known to participate in a variety of biological reactions because of the inherent electron- transferring property of their cofactor. Some of these biological functions include aerobic metabolism, bioluminescence and detoxification via hydroxylation of aromatic pollutants.
6
They also participate in several light-dependent processes such as photosynthesis (Zanetti
& Aliverti, 1991), photoreduction of DNA dimers (Jorns et al., 1987) and plant
phototropism (Briggs et al., 1999). Recently, a novel mitochondrial flavoprotein was
isolated and found to play signal transduction roles in programmed cell death (Susin et
al., 1999). These proteins were all found to bind either the FMN or the FAD form of the
cofactor; riboflavin was not commonly found although a riboflavin-binding protein is
involved in the development of chicken and mammalian fetuses (Murty & Adiga, 1982;
White & Merrill, 1988).
The flavin cofactor is bound in one of two different ways – covalently or non- covalently. Flavin-protein covalent linkages have been shown to exist through each of five known linkages: (a) 8 alpha-N(3)-histidyl, (b) 8 alpha-N(1)-histidyl, (c) 8 alpha-S- cysteinyl, (d) 8 alpha-O-tyrosyl, or (e) 6-S-cysteinyl, with the flavin existing as either the
FMN or the FAD form. The structural and mechanistic advantages of a covalent flavin linkage in flavoenzymes include: (a) the stabilization of the holoenzyme structure, (b) the steric alignment of the cofactor in the active site to facilitate catalysis, and (c) the modulation of the redox potential of the covalent flavin through electronic effects of the 8 alpha-substitution (Edmondson & Newton-Vinson, 2001). This class of enzymes is widely distributed in diverse biological systems and catalyzes a variety of enzymatic reactions. However, majority of the flavoproteins bind the flavin cofactor non-covalently with the flavin held in place by both hydrophobic as well as hydrogen bonding interactions. In either case, the flavin is never a substrate but always a true coenzyme and remains tightly bound to the protein during reactions.
7
Because flavins act very efficiently in a wide variety of enzymatic reactions, it was proposed that specific interactions between the flavin and apoflavoprotein play important roles in the stability of particular states and hence in determining the pathway of catalysis. Binding of the flavin to the apoprotein generally dramatically stabilizes the semiquinone state, and over the range of pH values where studies may be made, either the neutral or the anionic semiquinone is stabilized exclusively (Massey & Palmer, 1966).
The bacterial flavodoxins are renowned for their ability to thermodynamically stabilize the blue neutral form of the semiquinone state whereas the electron-transferring flavoproteins (ETF) are known for their ability to thermodynamically stabilize the red- anionic form of the semiquinone to such an extent that further reduction to the hydroquinone state is not possible or is very slow (Byron et al., 1989; Davidson et al.,
1986). A few exceptions are known where either ionic form may be detected if the pKa
of the semiquinone is in the observable range. Examples are glucose oxidase, with a pKa
of 7.5 (Massey & Palmer, 1966) and lysine monooxygenase, with pKa values between 7
and 8 (Flashner & Massey, 1974). Another notable exception is the flavoprotein
glutathione reductase that does not stabilize any semiquinone species and participates
only in two-electron transfer processes (Veine et al., 1998).
Given the vast majority of flavoproteins that have been identified, the choice of a
particular system to study depends entirely upon the aspect of flavoprotein structure-
function of interest. Our interests reside in addressing the more basic issues relevant to
function – what properties of the protein structure help modulate the oxidation-reduction
potential of the bound flavin? While it is possible to study that in more complex systems,
8
it has been worthwhile using a simpler system in which to address these issues. The
simple bacterial flavodoxins have served as an excellent model system in which to study
this important question. Flavodoxins are the simplest flavoproteins known, having
molecular weights ranging from 14 – 23 kDa (Mayhew & Tollin, 1992). These proteins
are soluble, highly stable and easily isolated from a variety of microbial sources. They
have been identified in prokaryotes and in red and green algae but not in higher plants
and animals. In E.coli, they act as the physiological electron donor for the reactivation or
priming of several enzymes including cobalamin dependent methionine synthase,
pyruvate formate lyase and anaerobic ribonucleotide reductase (Hall et al., 2001). In
cyanobacteria and algae, they substitute for the Fe-S protein ferredoxin as low-potential
electron carriers despite the difference in composition, structure and, size of the two
proteins (Ullmann et al., 2000). Although flavodoxins are not found as independent
proteins in mammalian systems, flavodoxin-like structural motifs and the flavin binding
site are conserved as domains in many complex proteins such as cytochrome P450
reductase from the Bacillus megaterium (Sevrioukova et al., 1999) as well as eukaryotic
systems (Wang et al., 1997; Zhao et al., 1999), nitric oxide synthase (Alderton et al.,
2001) and sulfite reductase from E. coli (Gruez et al., 2000).
Flavodoxins have long been used as a paradigm in which to study protein effects
on the oxidation-reduction potential of the bound cofactor. They typically
thermodynamically stabilize the neutral semiquinone state while preferentially
destabilizing the anionic hydroquinone (Ludwig & Luschinsky, 1992; Mayhew & Tollin,
1992). In this way, the oxidation-reduction potential of the functionally important sq/hq
9
couple is poised at very low values, in some cases approaching –500mV (pH 7 vs SHE).
The 200 – 300mV shift in the midpoint potentials for both couples from when free in
solution (Eox/sq = -314mV and Esq/hq = -124 mV) is indeed remarkable (Anderson,
1983). The factors that determine the magnitude of a redox protein’s reduction potential
have been explored for many years, and it is now clear that no single interaction is
dominant. A variety of flavin-protein interactions have been shown to play a role in
controlling the midpoint potential of the cofactor, including both short and long-range
electrostatic interactions (Chang & Swenson, 1997; Hoover et al., 1999; Swenson &
Krey, 1994; Zhou & Swenson, 1995), aromatic interactions (Lostao et al., 1997; Swenson
& Krey, 1994; Zhou & Swenson, 1996), sulfur (donor atom)-flavin interactions (Druhan
& Swenson, 1998) and hydrogen-bonding interactions at N(3)H (Bradley & Swenson,
1999; Bradley & Swenson, 2001) and N(5) (Chang & Swenson, 1999; Hoover et al.,
1999; Ludwig et al., 1997; O'Farrell et al., 1998). Another mechanism, suggested by
Massey and Hemmerich, by which the apoprotein could control the redox properties of the flavin coenzyme, is through distortion of the conformation of the flavin (Massey &
Hemmerich, 1980). The tertiary structures for several flavoproteins do indicate that conformation may play a role in flavin reactivity. Evidence for this is provided by the crystal structures of trimethylamine dehydrogenase and pyruvate oxidase, which show the oxidized cofactor to be severely distorted from its preferred planar geometry and more closely resemble the conformation of fully reduced flavin (Reibenspies et al.,
2000). Ab initio calculations using lumiflavin as a model indicate that the oxidized and flavin semiquinone radical is planar while the fully reduced flavin is bent with a ring
10
puckering angle of 27.3° along the N(5) and N(10) axis (Zheng & Ornstein, 1996).
Although the activation barrier for this ring inversion is low and the conformation of reduced flavin can be easily influenced upon binding to the apoflavoprotein (Moonen et al., 1984), the importance of this aspect in the regulation of the redox properties of the flavin has been a topic of debate for several years. However, conformational effects are not limited to flavin conformation alone and protein conformational changes also appear to play a crucial role. This phenomenon has been observed for the flavodoxins from
Clostridium beijerinckii (Burnett et al., 1974; Ludwig et al., 1997; Smith et al., 1977),
Desulfovibrio vulgaris (Watenpaugh et al., 1976; Watt et al., 1991) and Anacystis nidulans (Laudenbach et al., 1988; Luschinsky et al., 1991).
11
Figure 4. Ribbon diagram of the Clostridium beijerinckii flavodoxin structure. The α- helices are shown in red and the β-sheets in blue. The turn regions are shown in gray. The FMN shown in yellow sticks.
12
The crystal structures of several flavodoxins have been solved (Burnett et al.,
1974; Ludwig et al., 1997; Smith et al., 1977; Watenpaugh et al., 1976; Watt et al., 1991;
Laudenbach et al., 1988; Luschinsky et al., 1991) and, in some cases, in all three oxidation states, which have aided greatly in identifying protein conformational changes as a function of redox state. For the studies reported here, the flavodoxin from
Clostridium beijerinckii was used, the synthetic gene of which has been previously designed and expressed in Eschirichia coli (Eren & Swenson, 1989). Similar to other flavodoxins, the protein has a high degree of secondary structure - composed of a central region of five parallel β-sheets flanked on either side by α-helices (Figure 4). The isoalloxazine ring of the FMN cofactor is located at the periphery of the molecule and is bound noncovalently by a series of hydrogen bonds and other nonbonded contacts
(Ludwig & Luschinsky, 1992). Three polypeptide loops contribute to the flavin binding site – the teens loop (residues 8-11) that binds the ribityl chain end of the FMN; the 50’s loop (residues 56-59) that contacts the re face (inner) and N(5) edge of the flavin and the
90’s loop (residues 88-91) that contacts the si face of the flavin. The isoalloxazine ring is sandwiched between two hydrophobic residues, methionine on the re face and a tryptophan residue on the si face. The indole ring of the tryptophan (Trp 90) is not
perfectly stacked against the flavin ring, but rather forms an angle of 30°, in contrast with
the parallel stacking observed between homologous aromatic residues and the flavin ring
in other flavodoxins. The majority of these interactions remain unchanged upon a change
in redox state (Burnett et al., 1974; Ludwig et al., 1997; Smith et al., 1977).
13
Clostridium SEMIQUINONE
OXIDIZED Comparison of the turn conformation in the oxidized and semiquinone states of the Figure 5. beijerinckii flavodoxin.
14
Sequence Turn Type N-H…O Distance
34-35-36-37
Asn-Val-Ser-Asp III 2.7Å
35-36-37-38 Val-Ser-Asp-Val I 3.0 Å
39-40-41-42
Asn-Ile-Asp-Glu I 3.5 Å
43-44-45-46 Leu-Leu-Asn-Glu I 2.9 Å
56-57-58-59 Met-Gly-Asp-Glu II 2.5 Å
77-78-79-80
Ile-Ser-Gly-Lys II 3.5 Å
121-122-123-124 Pro-Asp-Glu-Ala I 2.25 Å
Table 1. Turns in the flavodoxin from Clostridium beijerinckii.
15
The conformation of the flavin-binding site between the oxidized and semiquinone states can be seen in Figure 5. Upon reduction, the conformation of
residues Gly57 and Asp58 changes with the carbonyl group of Gly57 moving from an
orientation that is pointing away from the flavin to one that is now pointing toward the
flavin and allows for the formation of a hydrogen bond between O57 and N(5)H . These
two residues lie on a turn in the flavodoxin comprised of residues -Met56-Gly-Asp-Glu59- that adopts a turn conformation that resembles a type II turn in the oxidized state. A turn is defined as a site where the polypeptide chain reverses its overall direction and can consist of three (γ-turn) or four (β-turn) residues that may be stabilized by an intra-turn hydrogen bond; in β-turns, the C=O of the first residue (i) may be hydrogen bonded to the NH of the fourth residue (i+3) (Rose et al., 1985), as seen in the flavodoxin. The flavodoxin has several such turn regions involving reasonable hydrogen bonds but only two of these are in the type II conformation (Table 1) which is in keeping with turn conformations commonly found in nature (Yang et al., 1996). Table 2 summarizes the idealized φ and ψ angles for the different turn types (Rose et al., 1985). The conformational change accompanying the change in redox state results in a conversion of this type II turn to a conformation resembling that of a type II′ turn (Figure 5). The glycine residue that is involved in the conformational change is frequently conserved throughout the flavodoxin family and there seems to be a tendency for those flavodoxins not having a glycine at that position to have a more negative value for the ox/sq couple – the redox states between which the conformational change occurs (Table 3).
16
i+1 i+2 Turn φ ψ φ ψ
Type I -60 -30 -90 0 Type I′ 60 30 90 0
Type II -60 120 80 0 Type II′ 60 -120 -80 0
Type III -60 -30 -60 -30 Type III′ 60 30 60 30
Type VIa(cis) -60 120 -90 0 Type VIb (cis) -120 120 -60 0
Table 2. Idealized dihedral angles of hydrogen bonded β-turns. The two central residues of the turn are labeled i+1 and i+2 respectively. Data is taken from Rose et al (Rose et al., 1985).
17
Eox/sq Esq/hq Flavodoxin Aligned sequences (mV) (mV)
Anacystis nidulans G C P T W N V G E L -220 -440
Anabaena 7120 G C P T W N I G E L -196 -425
Desulfovibrio vulgaris G C S T W G D D S I -185 -440
Klebsiella pneumoniae G T P T L G D G Q L -170 -422
Azotobacter vinelandii G T P T L G E G E L -165 -458
Clostridium pasteuranium G S P S M G S E V - -132 -419
Megasphaera elsdenii G C P A M G S E E L -115 -372
Clostridium beijerinckii G C S A M G D E V L -92 -399
Table 3. Sequence and midpoint potential comparisons of several flavodoxins. The conserved glycine residues are shown in red, where they have been replaced by another amino acid is shown in blue. The sequence for the Clostridium beijerinckii include the turn sequence of –Met56-Gly-Asp-Glu59-.
18
Another interesting aspect of this turn region is that the peptide bond between residues Gly57 and Asp58 adopts a mixture of cis and trans conformations, with the predominant conformer being the cis conformation, with the carbonyl group pointing away from the flavin (cis-O-down) in the oxidized state and the trans conformation with the carbonyl group pointing toward the flavin the reduced state (trans-O-up) (Ludwig et al., 1997). The unique structural properties of the glycine residue as well the role of this cis-trans isomerization event was investigated in a very elegant study that made several amino-acid substitutions at positions 57 and 58 in an effort to introduce different structural constraints (Gly57X mutants) as well as the more restricted X-Pro (Asp58Pro mutant) bond within this turn (Chang & Swenson, 1999; Ludwig et al., 1997).
Introduction of a side chain at position 57 resulted in increases in Eox/sq that were commensurate with the size of the side chain introduced (Table 4). Introduction of the β- branched side chain of threonine had much larger effects on the potentials as well as on the conformation. In this case, the peptide bond adopted a conformation that was all trans-O-down in the oxidized state and reverted to the trans-O-up conformation in the reduced states. The Asp58Pro mutant that has cis-trans conversion energies different from wild-type, was found to adopt an all cis-O-down conformation in the oxidized state, yet rearranged to the trans-O-up conformation upon reduction. While it was evident that these mutations affected the hydrogen bond strength at N(5)H in the reduced states, the effect of the turn conformation on these events was not investigated.
19
Flavodoxin Eox/sq Esq/hq Eox/hq
Wild-type -92 -399 -245 G57A -143 -373 -258 G57D -140 -378 -259 G57N -162 -372 -267 G57T -270 -320 -295 D58S -89 -381 -236 D58P -155 -360 -257
Midpoint potentials are reported in millivolts versus the standard hydrogen electrode at pH 7.0 and 25°C. Data is taken from Ludwig et al (Ludwig et al., 1997).
Table 4. Oxidation-reduction midpoint potentials for the Clostridium beijerinckii flavodoxin mutants.
20
Although the role of single amino acid residues within this loop in regulating the midpoint potentials of the cofactor has been thoroughly investigated in this laboratory
(Bradley & Swenson, 1999; Bradley & Swenson, 2001; Chang & Swenson, 1999;
Druhan & Swenson, 1998; Ludwig et al., 1997), several questions still remain
unanswered. Namely, what contributions does the stability of the various turn
conformations of the protein make to the observed midpoint potentials of the cofactor?
What role, if any, does the cis conformation of the central peptide bond of the turn have
on these events? A way to estimate the energetic contributions of turns in proteins is by
extrapolation from simpler systems i.e., with turns in peptides. Theoretical calculations of
preferred β-turn conformations in dipeptides led to a fundamental distinction between
turns based on sequences at positions i + 1 and i + 2. The primary origin of these turn-
sequence preferences is steric, arising from unfavorable nonbonded interactions between
the side chains of residues in the i + 1 and i + 2 positions and either the C=O of residue i
+ 1 or the NH of residue i + 2. The favored turns place the side chain of the i + 2 residue
on the opposite side of the turn from the i + 1 C=O. Thus, differences in turn stability
arise mainly from inter-residue interactions within the two central residues of the turn.
Quantitative computational free energy estimates for the various turns, for each
combination of the two central residues, were previously determined (Yan et al., 1995;
Yang et al., 1996) (Table 5 and Table 6). On the basis of their studies, an attempt was
made to correlate the changes in potentials of the FMN and the conformational energetics
of the loop. Mutations were designed at the two central positions of the turn (-57Gly-
58Asp-) to match their combinations of the two central residues that included -57Gly-
21
58Ala-, -57Gly-58Gly-, -57Ala-58Ala- and -57Ala-58Gly-. Each of these mutant flavodoxins was then analyzed by several different methods and the results of that study are included in the following chapters.
22
Type II Type II′ Sequence ∆Gc→t ∆Gc→t
-Gly-Ala- 6.3 2.3
-Gly-Gly- 3.0 2.9
-Ala-Ala- 5.6 3.6
-Ala-Gly- 2.8 4.3
∆Gc→t is in kcal/mol. Data is from Yang et al (Yang et al., 1996).
Table 5. Energies of β-turn formation for the TypeII and Type II′ turns.
23
Dipeptide sequence Refolding path ∆G kcal/mol
-Gly-Ala- Type II→Type II′ -1.82
-Gly-Gly- Type II→Type II′ 0
-Ala-Ala- Type II→Type II′ 1.01
-Ala-Gly- Type II→Type II′ 2.83
Table 6. Relative free-energy changes for refolding from the Type II to the Type II′ turn. Data is taken from Yan et al (Yan et al., 1995).
24
Clearly, the interaction with N(5) of the flavin is an important one and can be present in all three redox states. In some flavoproteins, a strong hydrogen bond can exist with N(5) of the oxidized flavin and either a backbone or side chain atom of the protein.
In human electron transfer flavoprotein (ETF), the hydrogen bond to the hydroxyl group of Thr266 was detected by nuclear magnetic resonance spectroscopy as well as by X-ray crystallographic studies (Griffin et al., 1998; Roberts et al., 1996). The loss of the
hydrogen-bond at N(5) results in decreases of both the ox/sq and sq/hq couples in the
human αThr266Met mutant (Salazar et al., 1997) indicating the importance of this
hydrogen bond in stabilizing the reduced states. Studies with old yellow enzyme from
brewer’s yeast are harder to interpret as the hydrogen bond exists with the backbone
amide proton of a threonine residue (Fox & Karplus, 1994). Although it is likely that this
hydrogen bond contributes to the strong binding of FMN to the apoenzyme (Kd = 1nM),
the exact role of this hydrogen bond still remains unclear. The flavodoxins on the other hand lack the hydrogen bond to N(5) in the oxidized state, rather provide a strong hydrogen bond only in the reduced states. How important is the hydrogen bond to N(5) in the flavodoxin? Recent evidence indicates that the 50’s turn modulates the redox properties of the cofactor though the interaction with N(5). Indeed mutations remote to this site also result in perturbation of the 50’s turn region and therefore the N(5) interaction. Crystallographic evidence for this was provided for mutations of Asp 95
(McCarthy et al., 2002) and Tyr 98 (Reynolds et al., 2001) in the Desulfovibrio vulgaris flavodoxin that are located in the 90’s loop region yet result in changes in conformation of the 50’s loop. Mutations designed within the 50’s loop of the C. beijerinckii
25
flavodoxin (Gly57X and Glu59X), all resulted in alterations of this N(5) interaction. On
the basis of these data, it was interesting to speculate whether the N(5) hydrogen bond
could compensate for all the other interactions provided by this loop. To address this
issue, all side-chain interactions in the loop were sequentially eliminated by alanine-
scanning mutagenesis, resulting in the minimalist sequence of the triple mutant, 56AGAA.
All of these mutant flavodoxins were characterized and the results and conclusions of this study can be found in the following chapters.
26
Tyr 536
Trp 574 Pro 541
Asn 537 Pro 540 Gly 538
His 539
Figure 6. Structure of the FMN-binding site in BM-3 in the oxidized state showing the major interactions with the isoalloxazine ring. Tyr 536 is situated on the re face with the coplanar Trp 574 on the si face of the ring. The dashed lines indicate the hydrogen bonding interactions that are present.
27
BM-3 534 A S Y N- G H P P D N A
CPR 138 A T Y GEG D P T D N A
Figure 7. Sequence alignment of the inner flavin binding loop of BM-3 and CPR. Additional residues on either end of the loop are included. Identical residues are shown in blue while similar residues are shown in red. The dash indicates the position where the BM-3 loop falls one residue short compared to CPR.
28
The flavodoxin-like fold, especially the flavin-binding site, is conserved among
the more complex flavin-binding systems. The FMN-binding domain of P450 reductase
from Bacillus megaterium (BM-3) shows significant homology to the flavodoxins.
However, in contrast to the type II turn conformation of the 50’s loop region seen in the
C. beijerinckii flavodoxin formed by the residues 56Met-Gly-Asp-Glu-, the homologous
loop in BM-3 adopts a type I′ turn conformation and is comprised of residues 536Tyr-Asn-
Gly-His- (Figure 6). Sequence alignment of this loop region with that of flavodoxin and
CPR indicates that the BM-3 loop is one residue shorter at the position where the loop makes the sharp turn (Figure 7). Therefore, while the flavodoxin loop encompasses the
N(5) edge of the flavin, the loop of BM-3 changes direction abruptly after Asn537 and moves away from the flavin. This enables the formation of a potentially strong hydrogen bond between the amide proton of Asn537 and N(5) in the oxidized state (2.65 Å) and is a unique feature. In the FMN binding domain of the mammalian cytochrome P450 reductase (CPR), the nearest atoms to N(5) are the Cα and amide N of Gly 81 which are
3.6 and 3.8Å respectively (Zhao et al., 1999). In flavodoxins, protein atoms or water
molecules are not present within 3.5 Å of N(5) in the oxidized state with the exception of
γ the flavodoxin from C. crispus where the O H of Thr58 is 3.07 Å from N(5) (Fukuyama et al., 1992). The exact contribution of this putative hydrogen bond to either binding of
FMN or to Eox/sq is still unclear. Additionally, the BM-3 loop is stabilized by two hydrogen bonds between main-chain atoms and ends in two proline residues that further help rigidify this loop and inhibit conformational flexibility (Sevrioukova et al., 1999)
(Figure 6). Another notable feature is the coplanar tryptophan that flanks the si face of
29
the flavin ring (Figure 6). This contrasts with the clostridial flavodoxin where the indole ring of the homologous tryptophan is stacked at an angle of 30° to the isoalloxazine ring.
This allows for efficient electron sharing between the electron-rich indole ring and the electron deficient flavin ring in the oxidized state. Interestingly, CPR has a tyrosine residue in place of the tryptophan and is also positioned coplanar to the isoalloxazine ring. The presence of this coplanar tryptophan is correlated with the formation of a long- wavelength charge-transfer band that is seen in BM-3 but is absent in CPR. The
Desulfovibrio vulgaris flavodoxin mutant Y98W, that positions the indole ring coplanar to the flavin, also results in the formation of a long-wavelength charge-transfer band in the oxidized state that is absent in wild-type (Swenson & Krey, 1994).
30
Type I′
Sequence ∆Gc→t
-Gly-Gly- 4.6
-Ala-Gly- 5.0
-Gly-Ala- 7.5
-Ala-Ala- 7.7
∆Gc→t is in kcal/mol. Data is from Yang et al (Yang et
al., 1996).
Table 7. Energies of β-turn formation for the Type I′ turn.
31
Dipeptide sequence Refolding path ∆G kcal/mol
-Gly-Gly- extended→Type I′ 0
-Ala-Gly- extended→Type I′ 0.06
-Gly-Ala- extended→Type I′ 0.82
-Ala-Ala- extended→Type I′ 0.88
Table 8. Relative free-energy changes for refolding from the extended to the Type I′ turn conformation. Data is taken from Yan et al (Yan et al., 1995).
32
Similar to the flavodoxins that have a conserved glycine residue in the loop region, the BM-3 loop also has a glycine residue, although the position of this glycine within the loop is different. It is found at the third position (second in the flavodoxins) and is likely correlated with the formation of a type I′ turn conformation of this loop.
Sequence preferences for the type I′ turn conformation indeed indicate that the glycine residue is an important determinant for this turn type (Table 7 and Table 8) (Yan et al.,
1995; Yang et al., 1996). To test this hypothesis, the two central residues of the turn were mutated in various Gly/Ala combinations in an attempt to alter the turn preferences.
Although the residue at the second position of the turn (Asn537) does not dictate the turn type in this case, it was also mutated to the respective Gly/Ala residue in order to directly compare the results to the previously established free energy estimates for the turn having these sequences. In addition, these mutations would also indirectly assess the role of the
N(5) hydrogen bond found in this flavoprotein. A direct approach of eliminating the N(5) hydrogen was also taken and the details of these studies can be found in later chapters.
33
CHAPTER 2
MATERIALS AND METHODS
Materials. Sodium dithionite and benzyl viologen was from Aldrich Chemical
Company. Indigodisulfonate, anthraquinone-2-sulfonate and anthraquinone-2,6-
disulfonate were purchased from Fluka Chemicals. Safranin T was also from Fluka
Chemicals but was recrystallized from ethanol before use. Phenosafranin was obtained
from Allied Chemicals. Flavin mononucleotide was extracted from recombinant wild-
type Clostridium beijerinckii flavodoxin and purified by anion-exchange
15 chromatography. Sodium 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) and NH4Cl
(99%) was from Cambridge Isotope Laboratories. All other chemicals were of analytical reagent grade.
Bacterial Strains and Plasmids: The chemically synthesized artificial gene encoding the Clostridium beijerinckii flavodoxin has been previously constructed and cloned into both the phagemid pBluescript and the pKK223-3 expression vector (Eren &
Swenson, 1989). The CJ236 strain of E.coli was used in the production of uracil-enriched single-stranded DNA for mutagenesis. The E.coli strain XL-1 was used for expression of the flavodoxin proteins. Bacillus megaterium ATCC 14581 was purchased from ATCC.
34
BL21(DE3) competent cells was from Novagen and the expression vector pT7-7 was
from Worthington Biochemical Corporation.
Oligonucleotide-Directed Mutagenesis and Protein Expression and Purification of the flavodoxin mutants. The synthetic gene for the flavodoxin from C. beijerinckii has
previously been prepared and cloned into the phagemid pBluescript (pBSFlasy).
Mutagenesis was carried out using the Kunkel method (Kunkel, 1985). The mutagenic
oligonucleotides that were synthesized for the generation of the different Gly/Ala
dipeptidyl sequence combinations at positions 57 and 58 in the Clostridium beijerinckii
flavodoxin are as follows:
-Gly-Ala- 5′ - GCCATGGGCGMTGMAGT(A)CTCGAGG - 3′
-Ala-Ala- 5′ - GCTCTGCCATGGCCGCGGAAGTTCTCG - 3′
-Gly-Gly- and -Ala-Gly- 5′ - GCTCTGCCATGGSCGGCGAAGT(A)CTCG - 3′
(where M = A and C, and S = G and C). The nucleotide in parentheses is a silent mutation that introduces a unique restriction site (ScaI) for screening purposes. The underlined positions are the altered sites that introduce the appropriate amino acid replacement. Mutagenesis generating the -Gly-Ala- sequence introduces a unique HaeII site; a SstII site for the -Ala-Ala- sequence; and a NaeI site for -Ala-Gly- that was absent in the -Gly-Gly- mutant. The sequence of the oligonucleotides that were synthesized to introduce multiple alanines at positions 56 through 59 is as follows:
56MGAA 5′ - GCCATGGGCGCTGCAGT(A)CTCGAGG - 3′
56AGDA 5′ - GGTTGCTCTGCCGCGGGCGATGCAGTTCTCGAGG - 3′
56AGAA 5′ - GCTCTGCCGCGGGCGCTGCAGT(A)CTCG - 3′
35
The underlined bases represent the mutations required for the appropriate amino acid
replacements. Screening was facilitated in some cases by the introduction of a unique
ScaI site by the silent mutation shown in parentheses, by the introduction of a unique PstI
site (for the 56MGAA mutant), a SacII site along with the elimination of an NcoI site
(56AGDA), and PstI and SacII sites with the elimination of an NcoI site (56AGAA).
In this way, all of the mutants were initially screened and identified by restriction
mapping; however, all mutations and the sequence integrity of the entire flavodoxin gene,
including within the following expression constructions, were confirmed by automated
DNA sequence analysis. All mutants were subcloned into the EcoRI and HindIII sites of
the pKK223-3 expression vector for heterologous expression in the XL-1 Blue strain of
E. coli. Protein purification was carried out by established protocols (Swenson & Krey,
1994) except that in some cases the proteins were expressed at reduced temperatures to
increase the yields of soluble holoproteins. Expression levels for all mutant proteins were
comparable to wild-type flavodoxin. All proteins were >95 % pure as determined by
SDS-PAGE.
Cloning, expression and purification of the FMN binding domain of cytochrome P450
reductase from Bacillus megaterium: The FMN binding domain of cytochrome P450
reductase from BM-3 (ATCC 14581) was cloned by the polymerase chain reaction using
the isolated genomic DNA as a template. The sequences of the primers used for PCR are
as follows:
5′ primer: 5′ GCTGCTAGAATTCGCAAAAAGGCAGAAAACGCTC 3′
3′ primer: 5′ ACTCAAGAGGATCCCTACTAGCTGTCGACAAATTGAAGTG 3′
36
These introduced the requisite restriction sites (EcoRI and BamHI) that were used for cloning. The 569 base pair PCR product was digested with EcoRI and BamHI and ligated into the pT7-7 vector similarly digested with EcoRI and BamHI. The ligation product was used to transform BL21(DE3) cells for expression of the protein. Prior to protein purification, the integrity of the construct was verified by automated DNA sequencing.
A similar approach was followed for all the mutants except that a mutagenic primer was used to introduce the desired mutation. Thus a two-step “megaprimer” PCR approach was used where in the first round, the mutagenic primer and the 3′ primer were used to amplify the 3′ end of the gene, while in the second round, the PCR product of the first round (megaprimer) and the 5′ primer were used to amplify the full length product.
The sequences of the mutagenic primers for the various mutants are as follows:
N537G 5′ - GGCGTCTTATGSCGG(C)CATCCGCC - 3′
N537A 5′ - GTAACGGCGTCTTATGCCGG(C)CATCCG - 3′
N537P 5′ - GTAACGGCGTCTTATCCCGG(G)CATCCG - 3′
59Gly/Ala-Ala 5′ - GTAACGGCGTCTTATGSAGCTCATCCG - 3′
Y536R 5′ - GTAACGGCGTCTAGAAACGGTCATCCGCC - 3′
Y536H 5′ - GTAACGGCGTC(A)CATAACGGTCATCCGCC - 3′
(where S = G and C). The nucleotides in parentheses introduce a silent mutation that generate a unique restriction in each case. The underlined positions are the altered sites that introduce the appropriate amino acid replacement. The following restriction sites were generated: NaeI and NgOMIV for N537A, SmaI for N537P, XbaI for Y536R,
37
Tsp45I for Y536H and SacI/SstI for 59Ala-Ala- and not for 59Gly-Ala. Automated DNA
sequencing was used to identify those mutations where a unique restriction site was not
generated and for the verification of the sequence integrity of all the constructs.
The procedure for protein purification was as detailed by Govindaraj et al
(Govindaraj et al., 1997) with the following modifications. The cell pellet suspended in
50 mM Tris buffer, pH 7.4 was lysed using a French press cell and centrifuged at
16,000rpm for >60 mins. The supernatant was loaded directly onto a DEAE Sephacel column equilibrated with 50 mM Tris, pH 7.4. After washing with 500 mls of the same buffer, the column was washed with 50 mM Tris, pH 7.4 containing 125 mM NaCl.
When the absorbance at 280 nm of the wash fractions reached a value of <0.1, the protein was eluted with 50 mM Tris, pH 7.4 containing 190 mM NaCl. Fractions having A274:
A468 ratios <10 were pooled, concentrated by ultrafiltration and the protein further purified to homogeneity using a Sephacryl S-100HR gel filtration column. The purity of the protein was confirmed by SDS-polyacrylamide gel electrophoresis.
UV-Visible Spectroscopy and Determination of One- and Two-Electron
Oxidation-Reduction Potentials. All UV-visible spectra were recorded on a Hewlett-
Packard HP8452A diode array spectrophotometer. The one-electron oxidation-reduction potentials for the flavodoxin mutants were determined as described elsewhere (Chang &
Swenson, 1997; Zhou & Swenson, 1995). All measurements were performed at 25°C in
50mM sodium phosphate buffer, pH 7.0. The indicator dyes used throughout the flavodoxin study were indigodisulfonate (-116 mV), anthraquinone-2,6-disulfonate (-184 mV), phenosafranin (-244 mV), safranin T (-280 mV) and benzyl viologen (-359 mV).
38
The mid-point potentials for the dyes indicated in parentheses are at 25°C, pH 7.0 versus
the standard hydrogen electrode (Clark, 1972). The estimated error of these values is within + 5 mV.
The indicator dyes used for BM-3 to obtain the two-electron midpoint potential
were anthraquinone-2,6-disulfonate (Em,7: -184 mV) and anthraquinone-2-sulfonate (Em,7:
-226 mV). While anthraquinone-2,6-disulfonate was used over most of the pH range of
6.0-8.0, anthraquinone-2-sulfonate was used at lower pH values. The following buffer systems were used: 50mM sodium acetate (pH 6.0), 50mM sodium phosphate (pH 6.0-
7.5) and 12 mM sodium pyrophosphate (pH 8.0). Similar ionic strength conditions were maintained throughout the pH range studied. The pH of the solution was determined at the end of the titration. The error in Eox/hq values is estimated to be + 5mV.
A slightly different approach was adopted to estimate the reduction potentials for the two single-electron transfers for BM-3. This was accomplished by fitting the changes in the extinction at 468 nm to the following equation:
468 (Eh-Eox/sq)/59 (Esq/hq-Eh)/59 E = a10 + b + c10
1 + 10(Eh-Eox/sq)/59 + 10(Esq/hq-Eh)/59 where a, b and c are the extinction coefficients for OX, SQ and HQ respectively, and Eh is the system potential (Daff et al., 1997). As noted earlier, Eox/sq is more negative than
Esq/hq, causing all three species to coexist in solution. As a result, the inherent error in the midpoint potential values derived for the individual steps is quite high.
Fluorescence Spectroscopy and Determination of FMN Dissociation Constants in the Oxidized State. Flavin fluorescence was measured either on a Perkin-Elmer LS50B
39
spectrophotometer or a Jobin Yvon FluorMax-3 fluorometer. Fluorescence emission was
recorded at 522 nm with excitation at 445 nm after allowing sufficient time for
equilibration. Experimental conditions were the same as used for the oxidation-reduction
potential determinations. The concentration of the FMN solutions was determined using
the published extinction coefficient of 12,500 M-1cm-1 (Whitby, 1953). Apoflavodoxin and apoBM-3 were prepared by the trichloroacetic acid precipitation technique as previously described (Wassink & Mayhew, 1975). However, it was observed that only 50
– 70 % of the apoBM-3 prepared in this manner was active in binding FMN. Presumably, some of the protein was inactivated by the formation of disulfide bonds via the single cysteine residue located in this domain or through partial denaturation of the apoprotein
(Haines et al., 2000). Despite incubation of apoBM-3 with β-mercaptoethanol, no increase in the yield of active protein was noted. This is in contrast to the results obtained by Haines et al (Haines et al., 2000) who obtained >90% active apoprotein. The concentration of apoBM-3 was calculated using the extinction coefficient of 28,700 M-
1cm-1 which was determined by the method of Pace et al (Pace et al., 1995) while that of the apoflavodoxin was calculated using an extinction coefficient of 25,000 M-1cm-1. The
Kd for the complex with oxidized FMN was determined by non-linear regression analysis
of the fluorescence emission plotted as a function of added apoprotein (Druhan &
Swenson, 1998). The dissociation constants for the SQ and HQ states, which cannot be
determined directly, were calculated using a thermodynamic cycle linking the Kd for oxidized FMN complex and the one-electron reduction potentials for the various redox states for both the bound and free FMN. The one-electron reduction potentials for free
40
FMN used in this study were those determined by Anderson (Anderson, 1983), which
have recently been reported (Mayhew, 1999) to be more reliable than those of Draper and
Ingraham (Draper & Ingraham, 1968) previously used.
In some instances, the dissociation constant for the binding of the FMN cofactor to the apoprotein was determined by monitoring the spectral changes accompanying binding as described elsewhere (Bradley & Swenson, 1999).
Preparation of 15N-Enriched FMN, Reconstitution with Apoflavodoxin and 1- and
2-Dimensional NMR Spectroscopy. FMN uniformly enriched at all four nitrogen atoms
with >95% 15N was prepared and used to reconstitute the apoprotein according to the
procedure detailed by Chang et al. (Chang & Swenson, 1999). 1D 15N-NMR spectra were recorded of approximately 1-2 mM solutions of the oxidized and reduced protein samples in 50-100 mM phosphate buffer, pH 7, containing 10% D2O as described previously
(Chang & Swenson, 1999). Identical conditions were maintained for the 1D 15N-NMR spectra recorded under non-decoupled conditions and applying the distortionless enhancement by polarization transfer pulse sequence (DEPT). Where applicable, these samples were also used for the 1H-15N HSQC NMR experiments. Reduction was achieved by the addition of freshly prepared sodium dithionite under anaerobic conditions. Proton chemical shifts were referenced to an internal standard of sodium-2,2- dimethyl-2-silapentane-5-sulfonate (DSS) set at 0.0 ppm and the nitrogen chemical shifts were referenced to an external standard of 15N-urea set at 76.0 ppm. The details of the
method have been described elsewhere (Chang & Swenson, 1999).
41
CHAPTER 3
CONFORMATIONAL ENERGETICS OF A REVERSE TURN IN THE Clostridium
beijerinckii FLAVODOXIN IS DIRECTLY COUPLED TO THE MODULATION OF
ITS OXIDATION-REDUCTION POTENTIALS
INTRODUCTION
Conformational changes appear to play a prominent role in protein function. For the flavodoxins, conformational changes provide additional flavin-protein interactions that play a crucial role in controlling the midpoint potentials of the cofactor. Such changes have been observed for the flavodoxins from Clostridium beijerinckii (Burnett et
al., 1974; Mayhew & Tollin, 1992; Smith et al., 1977), Desulfovibrio vulgaris
(Watenpaugh et al., 1976; Watt et al., 1991), and Anacystis nidulans (Laudenbach et al.,
1988; Luschinsky et al., 1991).
42
t f flavodoxin. The structures C. beijerinckii O-up" represents the configurationreduced states observed in both of the trans O-down" represent the predominant conformations of this turn in the oxidized state o state oxidized in the this turn of conformations predominant the represent O-down" trans Structure of the reverse turn involving residues 56-59 in the 8.
O-down" and " IGURE cis " of the FMN in which carbonyl oxygen Gly57 forms a hydrogen bond to the flavin N(5)H. The arrow points toward the carbonyl group at which the rearrangement occurs. The orientation of each figure is slightly differen F the FMN cofactor. The structure " in an effort to optimize the reader's view of the peptide bond between residues 57 and 58. The hydrogen atoms clarity. omitted been for FMN have the chain of side ribityl the and
43
For the C. beijerinckii flavodoxin, this change occurs in a surface loop comprised of residues -Met56-Gly-Asp-Glu59, which forms a reverse turn flanking the
C(6)/N(5)/C(4)O edge of the FMN (Figure 8) (for a complete structural description see ref. Ludwig et al., 1997). In the oxidized state, the peptide bond between Gly57 and
Asp58 adopts a mixture of trans and cis configurations that primarily point the carbonyl group away from the flavin (“O-down” configuration). The individual contributions of the various conformers in the oxidized state is estimated to be 50% and 20% for the cis
O-down and the trans O-down forms respectively, with the remainder being the trans O- up species. Reduction of the cofactor to the SQ state results in a structural rearrangement involving a rotation or “flipping” of the carbonyl group so that it now points toward the flavin (“O-up” configuration). In this way, the carbonyl group can form a new hydrogen bond with N(5)H of the SQ (Smith et al., 1977) which apparently shifts the equilibrium of the conformers to that favoring the trans O-up species alone. The thermodynamic stabilization of the SQ as a result of hydrogen bond formation contributes significantly to the large separation between the two redox couples and to the relatively high midpoint potential of -92 mV for the ox/sq couple in this protein. No significant structural differences between the SQ and HQ state are apparent (Ludwig et al., 1997; Smith et al.,
1977).
The glycine residue frequently found at the second position of the equivalent turn in several flavodoxins has long been thought to be of functional importance. The conformational change associated with the change in redox state of the FMN approximates the inter-conversion of a type II reverse- or β-turn (resembling the O-down
44
configuration found in the oxidized state of the flavodoxin) and the type II' turn (the O-up
configuration in the reduced states) (see Figure 9 for the idealized structures of several
common β-turn types). Glycine, lacking many of the steric constraints imposed by the
presence of a side chain, would appear to favor the formation of the type II' turn (O-up)
configuration, thus contributing to the thermodynamic stabilization of the FMNSQ through the hydrogen-bonding interaction at N(5)H. In fact, the replacement of Gly57 by a series of amino acids in the C. beijerinckii flavodoxin results in a more negative midpoint potential for the ox/sq couple (Ludwig et al., 1997). The crystal structures of the mutants revealed no major structural changes, with all except the G57T adopting in part the unusual cis O-down configuration as in wild type. Furthermore, all of the mutants are capable of undergoing the O-down to O-up transition upon reduction. This was a particularly surprising observation for the D58P mutant for which this transition represents the cis to trans isomerization of an X-Pro bond, thought to be separated by a substantial free energy barrier. These results indicate that most replacements at the two central positions of this turn are accommodated without significantly disrupting the conformation and changes associated with the change in the redox state of the protein, yet the Eox/sq for all the mutants were lower than wild type. These observations support the
view that glycine is favored at the second position in the type II′ turn configuration found
in the reduced state and that this preference is critical in the modulation of the redox
potentials of this couple (Ludwig et al., 1997).
45
3 n φ = -30, 2 ψ = -60, 2 φ = 0 and type I 3 ψ - being -Ala-Gly, -Gly-Ala- or -Ala-Ala- (3) -X = 90 and (2) 3 φ = 120, with -X with 3 2 ψ = -60, -NH-CH 2 (3) φ -X (2) turn are inverse of the type II turn. The orientation of each structure is ′ CO-X 3 angles for the type II atom with residues 2 and 3 being the two central residues of the turn. The figure was generated i ψ α , φ and type I turns respectively, representing the favored sequences for these turn types. The following ′ = 0. The 3 angles for these turns were used: for type II Representations of various turn conformations showing the orientation of the central carbonylRepresentations group. The residues of various turn conformations showing the orientation ψ ψ , 9. φ
IGURE are numbered near their C typical F the dipeptide CH HyperChem Pro 5.1 using = -90 and slightly different in order to optimize the reader's view of the central peptide bond. for type II, type II
46
A similar conformational change occurs in the A. nidulans flavodoxin; however,
in this case, the second residue in the turn is an asparagine residue, which based on the
above study and sequence preferences should less favorably adopt the type II' (O-up)
conformation in the SQ state. This situation may contribute to the lower Eox/sq of –221 mV in this flavodoxin. In fact, replacement of this asparagine by a glycine resulted in an increase in Eox/sq. The proposed increased stability of the type II' conformation in this
mutant appears to be confirmed by the observation that this configuration is retained in
the hydroquinone state, whereas in wild type where the turn rearranges back to the
apparently more stable type II or O-down conformation as seen in the oxidized state
(Drennan et al., 1999; Hoover et al., 1999). A series of glycine mutants at the equivalent
position (Gly61) in D. vulgaris were also found to modulate Eox/sq in a manner resembling that of the C. beijerinckii flavodoxin although the structural consequences of this replacement is quite different and may not directly apply to the arguments presented here (O'Farrell et al., 1998; Watenpaugh et al., 1976; Watt et al., 1991).
Thus, the evidence so far seems to point to the importance in the flavodoxin of the sequence specificity of the reverse turn near N(5) of the FMN cofactor in the modulation of its reduction potentials, particularly the ox/sq couple. But, how is this specificity translated to the cofactor? Can it be correlated directly to the energetic differences between each turn type? If so, in structural terms, how does this modulate the stability of each redox state of the cofactor? Only recently has a more direct correlation been found between the lower Eox/sq values of the Gly57 mutants of the C. beijerinckii flavodoxin and the strength of the hydrogen bond at N(5)H. This was accomplished by measuring the
47
temperature dependency of the chemical shift of the proton on N(5) of the fully reduced
FMN by 1H-15N HSQC nuclear magnetic resonance spectroscopy (Chang & Swenson,
1999). The wild-type situation having a glycine residue at position 57 offered the
strongest hydrogen bond while substitution with increasingly bulky side chains decreased
the hydrogen bond strength in accordance with their size. Although the conformational
energetics of this turn was an important determinant of Eox/sq, no evidence was sought for the direct role of the stability of the various conformations on these events.
The study presented here was initiated to more systematically investigate the contribution of the position specific preferences for residues with and without side chains at the 2nd and 3rd position of this turn, and to more solidly establish a thermodynamic link between the conformational energetics of this loop and the modulation of the midpoint potential of the FMN cofactor. Based on quantitative computational estimates of the energy differences between type II and type II' turns (Yan et al., 1995; Yang et al., 1996),
an attempt was made to correlate the changes in the potentials of the FMN and the
conformational energetics of the loop. Alanine was used to represent all residues with
side chains with the exception of proline. Using site-directed mutagenesis, the 2nd and 3rd residues of the turn (Gly57 and Asp58) were systematically replaced to -Gly-Ala-, -Gly-
Gly-, -Ala-Ala- and -Ala-Gly in order to alter the local conformational energies, thus energetically favoring a particular conformation. The results provide conclusive evidence for a direct correlation between Eox/sq and the conformational stability of the turn and
underlines the role of this turn in the modulation of redox potentials in this flavodoxin.
48
MATERIALS AND METHODS
All information regarding materials and methods can be found in Chapter 2.
RESULTS
UV-Visible Spectral Characteristics of the Mutant Flavodoxins. Four mutants of the C. beijerinckii flavodoxin were generated with all possible dipeptidyl sequence combinations containing glycine and alanine residues at positions 57 and 58, the central residues within the targeted reverse turn in this protein. All proteins were heterologously expressed in E. coli at high yields comparable to wild type. The purified holoproteins all exhibited a A274/A446 ratio of 4.4 + 0.1, a value similar to wild-type and characteristic of a
1:1 cofactor-protein complex. The spectral characteristics of all three redox states were recorded during a reductive titration with sodium dithionite under anaerobic conditions for each mutant. The visible absorbance spectrum for the -Gly-Ala- mutant in the oxidized state is very similar to the wild-type flavodoxin (Figure 10, solid line). The -
Gly-Gly-, -Ala-Ala- and -Ala-Gly- mutants display very similar spectral properties as a group, displaying small spectral shifts in the 450nm region compared to wild type and the
-Gly-Ala- mutant (see Figure 10). The extinction coefficients for the 450nm transition were determined to be within experimental error about the same for all the holoproteins.
It is difficult to interpret these small spectral changes with certainty because many factors
49
are influential. They could be the result of a slight increase in solvent exposure of the
flavin ring or slight changes in the flavin-protein interactions (Müller, 1991).
The neutral form of the FMN semiquinone accumulates in all the mutants during the anaerobic reduction with dithionite. Except for the -Ala-Gly- protein, the SQ is generated in direct proportion to the loss of the OX state until all the OX species has been consumed. Results shown for the -Gly-Ala- in the inset in Figure 10 (closed circles) are
representative. Note the linearity of the absorbance changes at 578nm due to the SQ
versus those at 450nm primarily due to the oxidized state. The second step also proceeds
with the proportional conversion of the SQ to the HQ. The two redox couples are
obviously well separated in these proteins. For the -Ala-Gly- mutant, however, the HQ
began to form before all of the oxidized species has been reduced to the SQ (Figure 10
inset, open circles). In all cases, the spectral characteristics of the semiquinone are
similar to wild type, with no obvious trends in the difference spectra. On the other hand,
distinct spectral differences for the FMN hydroquinone are noted among this group of
proteins. Two peaks at 315nm and 367nm and a shoulder at 450nm characterize wild-
type hydroquinone spectrum. The -Gly-Ala- and -Gly-Gly mutants are the most similar to
the wild-type spectrum, displaying spectral perturbations that can be generally
characterized as a slight decrease in extinction for all three transitions. The spectral
characteristics of the -Ala-Ala- and -Ala-Gly- mutant are comparable and the most
perturbed, lacking the distinguishing features of the wild type. Most notable is the
marked reduction in the shoulders at 367nm and 450 nm, particularly for the -Ala-Ala-
mutant. Although the differences are small, it is perhaps of interest that the spectral
50
perturbations for the HQ seem to segregate with the presence of either a glycine or an
alanine at position 57. On the basis of model compounds, the spectral features of the
wild-type flavodoxin have been suggested as representing the planar, anionic form of the
flavin (Ghisla et al., 1974); however, it is possible that the spectrum of the hydroquinone is sensitive to other changes in its environment (Yalloway et al., 1999).
51
12
4 ) ox -1 10 3 cm -1 8 2
Extinction (578nm) Extinction 1 6 0 0246810 4 Extinction (450nm) sq Extinction (mM 2 hq
300 400 500 600 700 800 Wavelength (nm)
FIGURE 10. UV-visible absorbance spectra for the C. beijerinckii -Gly-Ala- (solid lines) and the -Ala-Gly- (dashed lines) flavodoxin mutants in all three oxidation states obtained during a reductive titration with sodium dithionite in 50mM sodium phosphate buffer, pH 7.0 at 25°C. The inset shows a plot of the absorbance changes at 450nm (primarily due to the oxidized FMN) versus 580nm (flavin semiquinone) during the course of the titration for the -Gly-Ala- mutant (closed circles) and the -Ala-Gly- mutant (open circles).
52
One-Electron Oxidation-Reduction Potentials. The one-electron oxidation- reduction potentials of the mutants were determined at 25oC in 50mM sodium phosphate buffer pH 7.0 during reduction with sodium dithionite under anaerobic conditions in the presence of an indicator dye with an established mid-point potential (Clark, 1972). The titration data sets for this group of proteins could be fit to the linear version of the Nernst equation with a slope of 59 + 5, indicative of a transfer of a single electron accompanied by the uptake of a proton generating the observed neutral form of the semiquinone. A representative plot of the titration for the -Ala-Gly- mutant is shown in Figure 11A. It is
quite apparent that the midpoint potentials of the FMN are very sensitive to the dipeptidyl
sequence of the two central residues in the turn, particularly for the ox/sq couple. With
the exception of the -Gly-Ala- mutant, which displays an Eox/sq value very similar to wild type, the midpoint potentials for these proteins were substantially more negative than for wild type (Table 9). Both the -Gly-Gly- and -Ala-Ala- mutants had similar potential values of –157mV and –158mV, respectively. The -Ala-Gly- mutant exhibited a significantly more negative value of –241mV. This represents a rather remarkable decrease of nearly 150 mV relative to wild type and the -Gly-Ala- mutant, given that this difference is, in its simplest terms, the consequence of the movement of a single methyl group by one amino acid position within this protein! The midpoint potentials for the sq/hq couple were determined by equilibration with benzyl viologen as the indicator dye by methods described previously (Zhou & Swenson, 1995). A representative plot of such a titration is shown in Figure 11B. All mutants exhibited Esq/hq values that were less negative than wild type (Table 9); however, the changes are much less pronounced than
53
for the ox/sq couple. The increase of 19mV for the -Gly-Ala- mutant is identical to that observed for the D58S mutant previously characterized (Ludwig et al., 1997) and is consistent with the loss of the electrostatic effect of removal of a single negatively charged residue near the FMN (Zhou & Swenson, 1995).
54
FIGURE 11. Representative titrations for the determination of Eox/sq (panel A) and Esq/hq (panel B) for the -Ala-Gly- and the -Ala-Ala- mutants respectively. The data for Eox/sq are plotted as the linear version of the Nernst equation with the open circles representing the system potential (Eh) based on the equilibrium concentration of the oxidized and reduced dye species. The closed circles are for the -Ala-Gly- mutant plotted using the calculated Eh. The Esq/hq values were calculated from the equilibrium mixture of the flavodoxin mutant and the monomeric and dimeric forms of benzyl viologen as documented previously (Zhou & Swenson, 1995).
55
-120 A -160
-200 (mV)
h E -240
-280
-1.0 0.0 1.0 log [OX/RED]
20 B
16
12
[BVr]+2[dBVr] 8
4 0246810 [BVr][SQ]/[HQ]
Figure 11
56
Kd OX SQ HQ Flavodoxin Eox/sq Esq/hq OX(µM) SQ(nM) HQ(µM) ∆G ∆G ∆G (mV) (mV)
wtd -92 -399 0.018 0.0032 0.14 -10.6 -15.7 -9.3
-Gly-Ala- -93e -380h 0.042+0.02 0.0077 0.16 -10.1 -15.2 -9.3
-Gly-Gly- -157f -367h 0.021+0.009 0.046 0.60 -10.5 -14.1 -8.5
-Ala-Ala- -158f -345h 0.15 +0.01 0.34 1.9 -9.3 -12.9 -7.8
-Ala-Gly- -241g -337h 0.015+0.005 0.87 3.5 -10.7 -12.4 -7.4
a Values are in millivolts at pH 7.0 and 25°C. b The dissociation constants in the oxidized state were measured by fluorescence spectroscopy and those for the reduced states calculated as described in Experimental Procedures. Values are an average of two independent titrations each done in duplicate. c Values are in kcal/mol. d From Ludwig et al. (Ludwig et al., 1997) and Druhan et al. (Druhan & Swenson, 1998) e, f, g, h Midpoint potential was determined using indigo disulfonate, anthraquinone-2,6-disulfonate, phenosafranin, and benzyl viologen as the indicator dyes, respectively. Values are an average of at least two independent titrations.
Table 9 : Oxidation-Reduction Midpoint Potentialsa, FMN Dissociation Constantsb and Gibbs Free Energyc changes of Wild-Type and Mutant C. beijerinckii Flavodoxins
57
It is essential for the interpretation of the results of this study to establish that endogenous side chain interactions in the wild-type protein by themselves would not affect the potentials, particularly for the ox/sq couple. The only relevant residue in this case is Asp58 as a glycine is normally found at position 57. For this, the -Gly-Ala- mutant serves as a good control. The Eox/sq for this mutant (-93 mV) does not differ from
wild type (-Gly-Asp-) (-92 mV). Furthermore, the previously characterized D58S mutant
(i.e. -Gly-Ser-) has an Eox/sq value of -89 mV (Ludwig et al., 1997). Similarly, the Eox/sq for the -Ala-Ala- mutant (-158 mV) is comparable to that for the G57A mutant (i.e. -Ala-
Asp-) (-143mV) previously characterized (Ludwig et al., 1997). Thus, these examples
serve to establish that it is not the physico-chemical nature of the particular side chain
that is important in establishing the potential changes observed in the group of mutants
reported here but rather the local effects of the various combinations of glycine and
alanine on the conformational energetics of the turn as will be argued in depth later. Also
of significance, these data indicate that an alanine residue can be representative of non-
prolyl residues with side chains in establishing the potentials of this couple. This
assumption is also crucial in the computational studies evaluating the energetics of
reverse turns (Yan et al., 1995; Yang et al., 1996). Direct evidence to validate the above assumption comes from the free energy landscapes of these various dipeptides in water that were computed by the enhanced conformational sampling method (Nakajima et al.,
2000). The free energy landscapes of the -Gly-Ser- and -Gly-Ala- dipeptides were identical, as were those of the -Asn-Gly- and -Ala-Gly- dipeptides, conclusively proving
58
that the identity of a particular amino acid side-chain has little to do with the overall free
energy of the peptide.
The Dissociation Constants and Binding Free Energy Changes for the FMN. The dissociation constants were measured in the oxidized state, making use of the quenching of the FMN fluorescence upon binding to the apoprotein. A representative plot of the fluorescence changes as a function of added apoprotein for the -Ala-Gly- and -Ala-Ala- mutants are shown in Figure 12. The Kd values in each case were obtained from the fit of
these data to a binding isotherm involving a 1:1 complex as described in Experimental
Procedures. With the exception of the -Ala-Ala- mutant, which increased about 10-fold,
the mutants bound FMN in the oxidized state with values comparable to wild type (Table
9). The tight binding observed for most of the mutants is consistent with the absence of
significant perturbations in flavin-protein contacts in the oxidized state.
The dissociation constants for the FMN cofactor in its two reduced states cannot
be established directly due to their instability in solution. However, because the free
energy changes associated with binding or a change in redox state are pathway
independent, these values can be calculated using the midpoint potentials for free and
bound FMN together with the Kd for oxidized FMN via the appropriate thermodynamic cycle (Dubourdieu et al., 1975). With the exception of the -Gly-Ala- mutant, this group of mutants formed significantly weaker complexes with both reduced states of the FMN than for wild type. The -Ala-Gly- mutant displays a Kd value for the FMNSQ that is about
100-fold higher than for the -Gly-Ala- sequence despite the identity of the amino acids involved. The Kd for the FMNSQ for the -Ala-Ala- mutant was increased 45-fold
59
compared to the oxidized state, which had increased by only 4-fold. Despite these
increases, it is clear that the SQ is still the most thermodynamically favorable oxidation
state just as for wild type. Alterations in the dissociation constants for the FMNHQ followed a trend similar to that observed for the FMNSQ; however, the relative increase is of an order of magnitude less than for the FMNSQ.
60
100
50
Fluorescence Emission (522 nm) 0 0.0 0.3 0.6 0.9 1.2 µ [Apoprotein] M
FIGURE 12. Representative determination of the dissociation constant in the oxidized state shown for the -Ala-Gly- (circles) and -Ala-Ala- (squares) mutants. In each case, a FMN solution (~0.1 µM) in 50 mM sodium phosphate buffer, pH 7.0 at 25°C was titrated with increasing amounts of a 50-60 µM apoprotein solution having an identical buffer composition. Data were corrected for dilution. The solid lines represent the best non- linear regression fit of the data to a binding isotherm for a 1:1 complex, also from which the dissociation constants was derived.
61
The effects of these mutations can be better visualized when one compares the changes in the free energy of binding of the FMN cofactor in each of the three redox states (Figure 13 and Table 9). With exception of the -Ala-Ala- mutant, all display binding free energy values that are within 0.5 kcal/mol of wild type in the oxidized state.
The SQ and HQ complexes are all substantially less stable than wild type except for the -
Gly-Ala- mutant, which more closely resembles wild type in all three oxidation states. As indicated earlier, the -Gly-Ala- mutant will be used for comparison instead of wild type in order to avoid electrostatic contributions from the aspartate side chain, which has been uniformly removed in all cases, although either comparison generates qualitatively similar conclusions. The -Ala-Gly- mutant displays the largest changes in the free energy of binding in the reduced states. The SQ complex has been destabilized by about 2.8 kcal/mol and the HQ complex by about 1.8 kcal/mol. These increases fall to more intermediate corresponding values of about 1.1 and 0.8 kcal/mol for the -Gly-Gly- mutant. The -Ala-Ala- mutant exhibits increases for the SQ and HQ complexes that are similar to the -Ala-Gly- mutant; however, these changes must be viewed in the context of the significantly less stable complex in the initial oxidized state. If one normalizes or corrects for this initial instability (by assuming that the interaction responsible uniformly affects all three redox states) then the relative increase in the free energy of binding is similar to the -Gly-Gly- mutant, lying intermediate between the -Gly-Ala- (and wild type) and -Ala-Gly- proteins. Despite this, it is quite obvious that the stability of the SQ complex is principally affected by these alterations in the loop sequence. It will be argued below that the destabilization of the reduced states is likely to be the result of differences
62
in conformational free energy between the type II configuration of the turn in the oxidized state and the type II' conformer in the reduced state which primarily affects the stability of the SQ complex.
63
4
2
0
-2 G (kcal/mol) ∆∆ -4
-6 OX SQ HQ
FIGURE 13. Diagram depicting the free energy of binding of the FMN cofactor in all three oxidation states for wild type (open circles) and mutant C. beijerinckii flavodoxins: -Gly-Ala- (closed circles), -Gly-Gly- (squares), -Ala-Ala- (triangles), and -Ala-Gly- (diamonds). All values are relative to the binding free energy for the oxidized cofactor to wild-type flavodoxin.
64
15N- and 1H-Nuclear Magnetic Resonance Spectroscopy. The effects of the various Gly/Ala combinations on the interactions between the protein and the cofactor were evaluated by 1D-15N- and 2D-15N-1H HSQC-nuclear magnetic resonance spectroscopic analyses of the flavodoxins reconstituted with uniformly 15N-enriched
FMN. Emphasis was placed on evaluating the effects of the amino acid replacements on hydrogen bonding interactions at N(1), N(3), and N(5), particularly in the oxidized state.
The N(1) and N(5) atoms in the oxidized flavin represent the pyridine-like or β-type nitrogen atoms. The chemical shifts of such atoms are quite sensitive to hydrogen bonding, displaying large upfield shifts. Both N(3) and N(10) in oxidized FMN as well as all four nitrogen atoms in the reduced state are pyrrole-like or α-type nitrogen atoms. The
chemical shifts of such atoms show only small downfield shift on hydrogen bonding
(Witanowski et al., 1981). For comparison, the 15N chemical shift values for unbound
FMN and the wild-type C. beijerinckii flavodoxin were taken from Vervoort et al.
(Vervoort et al., 1986) and Chang et al. (Chang & Swenson, 1999).
The N(5) chemical shift of the -Gly-Ala- mutant is slightly upfield of wild type but remains significantly downfield from TARF in chloroform for which hydrogen bonding is not possible. This suggests that the N(5) atom is not hydrogen bonded and is in a relatively apolar environment in the oxidized state for this mutant (Table 10). All the other mutants displayed chemical shifts for N(5) that were significantly upfield shifted relative to the -Gly-Ala- mutant. However, even though the chemical shifts were slightly more upfield than for TARF in chloroform, they remained significantly downfield relative to FMN in aqueous solutions, suggesting that these upfield shifts are unlikely to
65
be the result of strong hydrogen bonding interactions with solvent. As the apoprotein is incapable of forming a hydrogen bond with N(5) of the flavin in the oxidized state, the upfield shifts are more likely a result of changes in the local environment and/or to differences in the conformer equilibrium of the loop in this group of mutants (Chang &
Swenson, 1999). The N(1) chemical shifts of the mutants are all similar to wild type, being upfield of that of FMN in aqueous solution, conforming to the previous observation of a strong hydrogen bond at this position in this flavodoxin (Vervoort et al., 1986).
66
15N Chemical Shifts (ppm)
N(1) N(3) N(5) N(10)
FMNa 190.8 160.5 334.7 164.6
TARFa 199.9 159.8 344.3 150.2
CMPa 184.5 161.1 351.5 164.8
rCBb 183.7 160.3 350.9 163.9
-Gly-Ala- 182.9 159.1 348.5 162.3
-Gly-Gly- 184.2 159.1 341.9 161.5
-Ala-Ala- 184.3 158.9 341.7 161.4
-Ala-Gly- 184.1 158.9 341.2 161.2
a From Vervoort et al. (Vervoort et al., 1986). b From Chang et al. (Chang & Swenson, 1999). Abbreviations: TARF: tetraacetyl-riboflavin in CHCl3, CMP: Clostridium MP (beijerinckii) flavodoxin, rCB: recombinant Clostridium beijerinckii flavodoxin.
Table 10. 15N Chemical Shifts for Free and Bound FMN in the oxidized state at pH 7.0, 300°K.
67
The N(3) chemical shifts of the mutants are similar and upfield of that of the isoalloxazine ring in polar and apolar solutions. It is likely that the N(3) hydrogen bonding interaction is similar to wild type in all of these mutants, a conclusion that is supported by strong correlation peaks for the N(3)H in the 15N-1H HSQC NMR spectra in the oxidized state that exhibit temperature coefficients that are comparable to wild type
(Figure 14, Table 11). Because the N(10) atom cannot form hydrogen bonds, the slight upfield shift for this atom in all mutants is probably due to a decrease in sp2 hybridization as a result of the decrease in polarization of the isoalloxazine ring caused by a weakened hydrogen bond with C(4)O and/or C(2)O (Vervoort et al., 1986). As these mutations are located near the N(5)/C(4)O flavin edge, a weakened interaction with C(4)O is conceivable. Changes in the chemical shift for N(5) in the mutants are in agreement with the NMR data obtained for several glycine 57 mutants previously prepared and characterized (Chang & Swenson, 1999). Unfortunately, 15N-1H HSQC signals were not observed for either the N(3)H and N(5)H in the reduced state for all the mutants at the field strengths provided by either 600 MHz or 800 MHz spectrometers. The absence of either signal as has been noted in several other flavodoxin mutants and may be the consequence of local conformational dynamics affecting the T2 relaxation rate or possible the solvent exchange rates [unpublished results]. Had these experiments been feasible, the temperature dependency of the chemical shift of particularly the N(5)H would have provided a measure of any alterations in the hydrogen bonding strength in this group of mutants, which would have contributed significantly to our overall conclusions as with other Gly57 mutants previously characterized (Chang & Swenson, 1999).
68
11.9
11.8 Chemical Shift (ppm) Shift Chemical 11.7
270 280 290 300 310 320 Temperature (OK)
Figure 14: Temperature dependencies of the proton chemical shift for the N(3)H oxidized of the bound 15N-labeled FMN for the wild-type (squares), -Gly-Ala- (closed circles), -Gly-Gly- (open triangles), -Ala-Ala- (closed triangles) and -Ala-Gly- (open circles) mutant flavodoxins.
69
∆δ/∆T ppb/°K
Flavodoxin N(3)Hox
Wild-type -0.889a -Gly-Ala- -0.201 -Gly-Gly- -0.222 -Ala-Ala- -0.021 -Ala-Gly- -0.026
Table 11: 1H-15N HSQC Temperature Coefficients for the Clostridium beijerinckii Wild-Type and mutant flavodoxins in the oxidized state. All experiments were performed on the Bruker 600 MHz spectrometer at pH 7.0 in phosphate buffer. aFrom (Bradley & Swenson, 2001).
70
DISCUSSION
The conformational change associated with the reduction of the FMN cofactor in the C. beijerinckii flavodoxin can be adequately described as the conversion of
predominantly a type II reverse- or β-turn, albeit in a mix of cis and trans configurations for the central peptide bond, in the oxidized state to that of a type II′ turn in the reduced states (see Figures 8 and 9). This change appears to be facilitated by the presence of a glycine residue at the second position of the turn [Ludwig, 1997 #4] as residues other than glycine rarely populate that position in type II' turns [Smith, 1980 #57; Wilmot,
1988 #71]. Glycine lacks the Cβ atom that would otherwise be involved in steric clashes with the NH of the adjacent residue (Richardson, 1981). Furthermore, type II′ turns
require the second residue to adopt positive φ values for which glycine and asparagine
have the highest propensities. It is not surprising then that the residue that replaces this
glycine in many long-chain flavodoxins is an asparagine (Ludwig & Luschinsky, 1992).
Questions are raised as to whether the relative sequence-derived stability of type II versus
type II′ turns has a role in this conformational change and whether the conformational
energetics associated with this transition are directly linked to the modulation of the
redox potentials of the FMN cofactor in these proteins.
Several computational studies have attempted to quantify the conformational
energetics associated with the sequence specificity of various reverse turns. Their results
prompted a possible means to gain greater insight into these questions. Using either a
molecular dynamics simulation approach or a Monte Carlo sampling technique as well as
71
molecular mechanical force fields in combination with a solvation model, Yan et al. (Yan et al., 1995) and Yang et al. (Yang et al., 1996) independently calculated the free energy differences between various types of β-turn conformers for a series of blocked dipeptides representing the central residues in such turns. Because the conformational energies are thought to be primarily a function of local interactions with the backbone atoms and to simplify the calculations, alanine was used to represent all non-prolyl amino acid residues having side chains and, of course, glycine without. Sequence preferences for a particular turn type were found to be a direct result of the intrinsic conformational preferences of the different residues occupying each of the specific central positions of the turn.
On the basis of their calculations, Yan et al. obtained free energy differences between the type II' and type II turn for the sequences -Gly-Ala-, -Ala-Ala- and -Ala-Gly- of -1.82 kcal/mole, 1.01 kcal/mole and 2.83 kcal/mole, respectively. The corresponding values derived by Yang et al. are -4.0 kcal/mole, -2.0 kcal/mole and 1.5 kcal/mole. The free energy difference for the -Gly-Gly- sequence is negligible due to its symmetry. The inconsistencies in results among various computational studies have recently been shown to be the direct result of the different force fields used (Nakajima et al., 2000). However, it is the relative differences in the conformational free energies among the sequences that are of concern in our evaluations and irrespective of the parameters used, the same conclusions are reached. From these data, one can unambiguously conclude that the -Gly-
Ala- sequence strongly favors the type II′ turn while the -Ala-Gly- sequence prefers the type II turn, the energy difference between both situations being >4.5 kcal/mol. The calculated relative stability of each of the β-turn conformers is also consistent with their
72
frequency of occurrence in the crystal structures of native proteins (Nakajima et al.,
2000; Yan et al., 1995; Yang et al., 1996). The stability of the -Ala-Ala- sequence
relative to the -Gly-Gly- sequence is somewhat ambiguous, preferring a type II'
configuration in the Yang et al. analysis and showing a slight preference for the type II
turn in the Yan et al. study. Furthermore, the -Ala-Ala- dipeptidyl sequence may actually
have a greater preference for the type I turn configuration (see Figure 9) (Möhle et al.,
1997; Nakajima et al., 2000; Yang et al., 1996). As will be addressed further below, these
ambiguities do not substantially affect the conclusions of this study, however.
Midpoint Potentials for the Flavodoxin Correlate with Conformational Free
Energy Differences of the Turn. Within the context of these computational studies, the
experimentally derived midpoint potential values for both flavin couples for the group of
flavodoxin mutants studied here were observed to correlate remarkably well with both
sets of the calculated conformational free energy differences between type II' and type II
turns (Figure 15). It is quite evident for both sets of energy values that the -Gly-Ala-
sequence, which favors the type II' turn, like the wild-type flavodoxin has the least
negative Eox/sq value, while the -Ala-Gly- sequence, which substantially favors the type II
turn, displays the most negative reduction potential. The decrease in Eox/sq for the -Ala-
Gly- mutant by nearly 150 mV is a particularly extraordinary consequence of what on the face of it represents a "simplistic" reversal of the Gly/Ala positioning in this turn (i.e. -
Gly-Ala- versus -Ala-Gly-). It is difficult to interpret this effect in any other way than as an effect on the conformational energetics/dynamics of the turn in relation to the conformational changes associated with the reduction of the FMN cofactor. On the basis
73
of the crystal structures of the various redox states of this flavodoxin, the tendency for a particular dipeptidyl sequence to prefer a type II turn should favor the oxidized state, making reduction more difficult, while conversely the type II′ turn should favor the reduced states. Accordingly, the -Gly-Gly- and -Ala-Ala- mutants, containing dipeptidyl sequences that exhibit only relatively small free energy differences between the two turn types, should have potential values that would lie between these two extremes. It is quite apparent that the experimental results are completely consistent with these preferences.
74
0 AB
-80
-160 (mV) h
E -240
-320
-400 -2 0 2 4 -4 -2 0 2 ∆∆G (Type II' - Type II)
FIGURE 15. Correlation between the midpoint potential values (in mV, pH 7) for the ox/sq couple (closed symbols) and for the sq/hq couple (open symbols) and the calculated conformational free energy differences between type II′ and type II turns (in kcal/mol) obtained from Yan et al. (Panel A) (Yan et al., 1995) and Yang et al. (Panel B) (Yang et al., 1996). Symbols are as follows: -Gly-Ala-, circles; -Gly-Gly-, inverted triangles; - Ala-Ala-, squares; and -Ala-Gly-, diamonds.
75
The similarity in the Eox/sq values for the –Gly-Gly- and -Ala-Ala- mutants was not predicted from the computational studies. It is possible that the loop in the -Ala-Ala- mutant alternatively adopts a type I turn as is preferred for alanine-containing turns
(Möhle et al., 1997; Nakajima et al., 2000; Yang et al., 1996). However, the molecular modeling of a type I turn into the loop suggests that this configuration is poorly accommodated and geometry optimization calculations using the AMBER molecular mechanical force filed resulted in the turn reverting to conformational parameters more like that of a type II′ turn (results not shown). In fact, the loop structure of the reduced
G57D mutant previously characterized (which is equivalent to the -Ala-Ala- mutant in terms of the presence of a side chain at both positions), while similar to wild-type, displays slight differences in the torsion angles at positions 56 and 57 to reduce the overlap of the β-carbon with the amide NH of residue 58, a major unfavorable feature of the type II′ turn (Ludwig et al., 1997). This suboptimal situation may be reflected in the
higher Kd for the oxidized FMN in the -Ala-Ala- mutant. In either configuration, a
conformational change would not be required to form the hydrogen bond at N(5)H;
therefore, one might predict that there would not be a free-energy difference between the
oxidized and reduced states as should also be the case for the -Gly-Gly- sequence.
76
ox (ox) ∆Gi PROTEIN + FMNox FLVox
∆Gc ox/sq
∆Gi sq PROTEIN(sq) + FMNsq FLVsq
∆Gc sq/hq
hq (hq) ∆Gi PROTEIN + FMNhq FLVhq
Figure 16. Thermodynamic cycles describing the free energies associated with the conformational changes and FMN binding in each oxidation state. PROTEIN(ox,sq,hq) represents “virtual” forms of the apoflavodoxin having the same structure as the holoflavodoxin (FLV) in each oxidation state but without the bound cofactor. These forms are incorporated in the figure in order to separate the free energies associated with the conformational changes (∆Gc) from those associated with the direct protein-flavin interactions (∆Gi).
77
Conformational Energetics of the Reverse Turn is Coupled to the Modulation of the Oxidation-Reduction Potentials of the FMN Cofactor. The data persuasively associate
a significant role of the conformational energetics of the reverse turn to the regulation of
the midpoint potentials of the flavin cofactor in this flavoprotein. But, how are the
changes in the protein conformational energy translated to the modulation of the flavin
potentials? A thermodynamic scheme has been proposed previously that separates energy
changes as a result of protein conformational changes (∆Gc) and changes in FMN-protein interactions (∆Gi) (Figure 16) (Druhan & Swenson, 1998; Ludwig et al., 1997). The reverse turn in question can be considered to exist in an equilibrium between the O-down
(resembling a type II turn) and the O-up (type II' turn) conformers, with the two states separated by the free energy change ∆Gc (depending on the couple involved, this change
ox/sq sq/hq is designated ∆Gc or ∆Gc in Scheme I). Upon reduction, the protonation of N(5) of the reduced FMN results in the formation of a new N(5)H•••O(57) interaction that contributes to the stability of the reduced states by contributing favorably to the free energy difference ∆Gi. The formation of this new hydrogen-bonding interaction greatly
favors the O-up conformer and shifts the overall equilibrium in favor of this conformer
and towards the reduced states. The midpoint potential of the bound FMN is a function of
the free energy difference between the two oxidation states concerned. For the ox/sq
ox/sq sq ox couple, the free energy difference would be equal to ∆Gc + (∆Gi - ∆Gi ). The amino acid substitutions could either affect the protein conformational energy directly or affect the energy of the FMN-protein interaction or some combination of the two.
78
Can the observed changes be accounted for by the conformational energetics alone? A difference of approximately 3.4 kcal/mol was noted between the -Gly-Ala- and
-Ala-Gly- mutants in the change in the binding free energy in going form the OX to the
SQ OX SQ SQ state of the FMN (i.e., from Table 1; [∆G (-Ala-Gly-) - ∆G (-Ala-Gly-)] - [∆G (-Gly-Ala-)
OX - ∆G (-Gly-Ala-)] = [-12.4 - (-10.7)] - [-15.2 - (-10.1)] = 3.4 kcal/mol). This value
represents about 75% of the free energy difference between the two turn types, i.e.∆∆G
(II'-II) in Figure 7. Therefore, it is quite possible to account for all of the observed free-
energy changes by the effects of the sequence changes on ∆Gc if one reasonably assigns
∆∆G (II'-II) as the primary component of the ∆Gc term. However, do the amino acid
substitutions have any effect on the strength of the N(5)H•••O(57) interaction itself, i.e.
∆Gi? This is a difficult question to answer. It is not possible to adequately separate experimentally the ∆Gc and ∆Gi terms. Only small changes in the geometry and
interatomic distances were noted from most of the G57 mutants previously characterized,
suggesting that the hydrogen-bonding interaction, which contributes to ∆Gi, should remain substantially unaltered (Ludwig et al., 1997). However, a link between amino acid replacements in this region and changes in the interactions at N(5)h has been established through the determination of the temperature dependency of the chemical shift for the
N(5)H of the FMNHQ, which serves as an indicator of the relative strength of the hydrogen bond(Chang & Swenson, 1999). Thus, it is quite likely that a similar situation occurs with these substitutions. Unfortunately, an HSQC NMR signal for the N(5)H in the reduced state was not detected for any of the mutants, an unexpected observation but one also noted previously for the D58P mutant (Chang & Swenson, 1999). Interestingly,
79
based on these and other unpublished observations, the side chain properties of the
aspartate residue seem to be critical for the generation of this signal, perhaps influencing
local conformational dynamics that affect the T2 relaxation rate and/or the proton exchange rate.
Some additional insights are provided by the fact that the midpoint potentials for the sq/hq couple displayed a more modest and opposite correlation with the free-energy differences between the type II and type II′ turns that that for the ox/sq couple (Figure
15). Because of the similar conformation of both reduced states, one might predict that
sq/hq the contributions of ∆Gc and thus changes to this term as the result of the amino acid
replacements should be very small. It is evident in Figure 13 that the increases in the
binding free energy for the HQ states in response to the amino acid replacements were
smaller than for the SQ. This phenomenon was observed previously for the Gly57
mutants and is interpreted as reflecting differing hydrogen-bonding strengths at N(5)H
between the SQ and HQ states, i.e., differences in ∆Gi, with any modification in this
interaction having a greater affect on the stability of the SQ(Chang & Swenson, 1999).
This is reasonable if you take into account the weaker N(5)H interaction in the fully
reduced state due to the decrease in charge at this position when the flavin becomes more
electron rich (Hall et al., 1987; Ludwig et al., 1997). Therefore, given the strong
correlation between the midpoint potentials for both couples and the energy differences
between the turn types and, within the assumptions and limitations of the computational
studies, the similarity between the computed ∆∆G (II'-II) values and the observed changes in the the binding free energies principally for the OX and SQ states between
80
which the conformation change occurs, we conclude that the amino acid replacements
ox/sq introduced at the two central positions of the β-turn primarily alter ∆Gc in Scheme I, although the linkage of this effect to changes in ∆Gi can not be entirely excluded.
Contributions of the cis Conformer: The central peptide bond in the reverse turn
adopts three different conformations in the oxidized state of the C. beijerinckii flavodoxin; the trans O-up, the cis O-down, and trans O-down (Ludwig et al., 1997).
These ambiguities complicate our analysis and this issue needs to be addressed more thoroughly. The conformational energies for the two turn types were calculated assuming a trans configuration of the peptide bond. However, to what extent would the cis configuration affect these values? More importantly, would the contributions of the cis conformer affect all the mutations equally and in so doing be factored out for this analysis?
Computational studies have predicted a change in the conformational energy of
1.8 kcal/mol for the cis-trans isomerization that occurs on going from the oxidized to the semiquinone state when a cis Gly-Asp (wild type) sequence has been replaced by a cis
Gly-Pro sequence (Ludwig et al., 1997). The observed change of 1.4 kcal/mol for the
Gly-Pro mutant being primarily due to differences in conformational energies would also reflect its altered conformer equilibrium, where wild type consists of a mixture of cis O- down, trans O-down and trans O-up species, the Gly-Pro mutant is all cis O-down. Yet the difference between the observed and predicted values differs by only 0.4 kcal/mol.
(Ludwig et al., 1997). Also, a distribution of both the cis and trans O-down conformers is observed in the oxidized state (Ludwig et al., 1997). Both observations suggest that
81
differences in cis and trans conformer equilibrium are not a major contributing factor as they may have relatively similar conformational energies.
A detailed analysis of these unusual Xaa-Xaa (where Xaa is any amino acid except proline) cis peptide bonds revealed a strong preference for the φ, ψ angles of the
residues involved (Jabs et al., 1999). All residues N-terminal and most residues C-
terminal (except glycine) to a non-proline cis peptide bond are found in the β region of
the Ramachandran plot. Presumably, any combination of residues is capable of adopting
a cis peptide bond. Furthermore, it was observed that ψ1 and φ2 of the residues involved are restricted to only two energetically allowed conformations irrespective of φ1 and ψ2.
The lack of significant structural differences on mutation of either of these residues in the
C. beijerinckii flavodoxin is then not surprising (Ludwig et al., 1997). A later study calculated the preferences of amino acid residues to be involved in or adjacent to a cis peptide bond (Pal & Chakrabarti, 1999). The lack of bulky side chains on the two residues involved supported the inhibitory role these would have on the isomerization process. In keeping with this theory, small residues (glycine and alanine) which offer minimal steric resistance should favor the cis form and, therefore, could be accommodated at both positions, as was observed. A strong preference for polar residues
C-terminal to the cis peptide bond was also noted and it was believed that hydrogen bonding of these polar side chains with its main-chain amide would lower the activation energy and thus facilitates isomerization. Such a phenomenon may actually be occurring in wild type flavodoxin, which has an aspartate residue C-terminal to the cis bond. But as this polar side chain has been uniformly removed in all mutants, such energetic
82
contributions are not possible. This would strongly suggest that the contribution of the cis
conformer would be similar for all these mutants. As the interpretations are based more
on comparisons among the various mutants rather than the actual values, we believe that
the analysis is valid.
Conclusions: For the first time to our knowledge, this study more conclusively demonstrates a direct linkage between the conformational energetics of the protein induced by the sequence specificity of a β-turn and the modulation of the redox potentials
of the flavin cofactor as previously postulated (Ludwig & Luschinsky, 1992; Ludwig et
al., 1997). This study also further illuminates the critical structural role of the often
conserved glycine residue in the peptide loop adjacent to the N(5)/C(4)O edge of the
flavin particularly in the thermodynamic stabilization of the neutral semiquinone state of
the flavodoxin (Bradley & Swenson, 1999; Chang & Swenson, 1999). These results
could have more general implications regarding the role of protein conformation
energetics/dynamics in the modulation of the activities of other proteins as well. For
example, the interconversion of a β turn in the HIV-1 protease occurs in a loop region
upon inhibitor binding and is believed to facilitate function by allowing substrate access
to and product release from the active site (Gunasekaran et al., 1998; Nicholson et al.,
1995).
83
CHAPTER 4
ALANINE-SCANNING OF THE 50’S LOOP IN THE Clostridium beijerinckii
FLAVODOXIN: EVALUATION OF ADDITIVITY AND THE IMPORTANCE OF
INTERACTIONS PROVIDED BY THE MAIN CHAIN IN THE MODULATION OF
THE OXIDATION-REDUCTION POTENTIALS
INTRODUCTION
The remarkable biochemical versatility of the riboflavin-based cofactor lies in part in the ability of its oxidation-reduction properties to be modulated by the numerous interactions made with the apoflavoprotein. The flavodoxin represents one important class of flavoprotein electron transferase in which this phenomenon has been extensively investigated. These small (≤ 20 kDa), soluble proteins have been isolated from a variety of microorganisms and eukaryotic algae and contain a single non-covalently bound flavin mononucleotide (FMN) cofactor as their only redox-active component. They share significant structural homology with a variety of more complex flavoproteins that function in a myriad of biological reactions, such as cytochrome P450 reductase, cytochrome P450BM-3, and perhaps nitric oxide synthase. Although notable differences in the detailed amino acid sequences occur, even among flavodoxins, many of the 84
interactions made between the apoprotein and the cofactor are generally conserved. Thus,
they represent a simplified model in which to investigate structure-function properties
that can then be applied to other more complex systems.
A wealth of information on the role played by the protein in perturbation of the potentials of the bound cofactor exists. This includes short and long range electrostatic interactions (Chang & Swenson, 1997; Hoover et al., 1999; Swenson & Krey, 1994;
Zhou & Swenson, 1995), aromatic interactions (Lostao et al., 1997; Swenson & Krey,
1994; Zhou & Swenson, 1996), sulfur-flavin interactions (Druhan & Swenson, 1998) and
hydrogen bonding interactions at both N(3) (Bradley & Swenson, 1999; Bradley &
Swenson, 2001) and N(5) (Chang & Swenson, 1999; Hoover et al., 1999; Ludwig et al.,
1997; O'Farrell et al., 1998). Conformational changes that alter flavin protein contacts also make significant contributions as seen in the flavodoxins from Clostridium beijerinckii (Burnett et al., 1974; Ludwig et al., 1997; Smith et al., 1977), Desulfovibrio
vulgaris (Watenpaugh et al., 1976; Watt et al., 1991) and Anacystis nidulans
(Laudenbach et al., 1988; Luschinsky et al., 1991). Many of these interactions have been
corroborated and extensively investigated within novel flavin-host model systems
(Breinlinger & Rotello, 1997; Cuello et al., 2000; Goodman et al., 2001).
85
FIGURE 17: Structure of the FMN binding site in wild-type Clostridium beijerinckii flavodoxin in the oxidized state showing the major interactions with the isoalloxazine ring. Of note is the re-face interaction with the side chain of Met56 as well as the si-face interaction with Trp90. The dashed lines indicate the hydrogen bonding interactions made with the backbone amide and side chain of Glu59. Overlaid in green is the energy- minimized structure of the triple mutant 56AGAA (the loop only). The residues are numbered at the Cα position. The ribityl side chain of FMN, all hydrogen atoms with the exception of N(3)H as well as the side chains of residues 55 and 60 have been omitted for clarity. The wild-type main chain atoms for residue 60 are obscured by the mutant structure due to nearly complete overlap.
86
High-resolution crystal structures have been solved for all three oxidations states for the flavodoxin that is isolated from Clostridium beijerinckii (MP) and its recombinant form (Burnett et al., 1974; Ludwig et al., 1997; Smith et al., 1977). Of note here is a four-residue surface reverse-turn, the so called “50’s loop”, comprised of residues
56MGDE, that contributes many of the crucial interactions made with the isoalloxazine
ring of the FMN cofactor (Figure 17). The first residue in the turn, Met 56, flanks the
inner or re face of the flavin ring, as also seen in the flavodoxins from Clostridium pasteurianum and Megasphaera elsdenii, which is replaced by either a tryptophan or a
leucine in other flavodoxins (Druhan & Swenson, 1998; Ludwig et al., 1997).
Mutagenesis studies revealed the importance of sulfur-flavin interactions in the
stabilization of the oxidized and destabilization of the hydroquinone states, thus pointing
to the functional importance of this residue in maintaining the redox properties of the
cofactor (Druhan & Swenson, 1998).
Perhaps the most significant distinction of this loop lies in the unusual
configuration of the central Gly57-Asp58 peptide bond of the turn. In the oxidized state,
this peptide bond exists in the unusual cis conformation with its carbonyl group oriented
away from the flavin ring. This conformation, referred to as the “cis O-down”
conformation, along with the trans-O-down form constitutes about 70% of the observed conformers, with the remainder as the trans-O-up form with the orientation of the
carbonyl group “flipped” over to point towards the flavin ring (Ludwig et al., 1997).
Reduction of the cofactor results in a structural rearrangement to predominantly the
trans-O-up conformer. This new orientation brings the carbonyl group in close proximity
87
to the flavin ring such that it can serve as a hydrogen bond acceptor to N(5)H of the
reduced FMN. This contributes significantly to the thermodynamic stabilization of the
SQ and to a lesser extent the HQ (Ludwig et al., 1997). Also unique to this flavodoxin is that this structural rearrangement approximates a conversion of a type II β-turn in the OX to a type II′ turn in the reduced states. The functional importance of the Gly57 residue in favoring the type II′ turn as well as the importance of a side chain at position 58 (or position three in the turn) has been established (Kasim & Swenson, 2000; Ludwig et al.,
1997). The stability of the turn conformation, which is a direct function of the sequence of the two central residues with the positioning of the Gly and Ala residues, where Ala represents any amino acid with the exception of glycine and proline, is a prominent feature in the modulation of Eox/sq [Kasim, 2000 #73]. The fourth and final residue of the
turn, Glu59, also plays a critical role. Its side chain carboxylate group forms a hydrogen
bond “bridge” between the N(3)H of the flavin and the backbone amide of Trp95 and
serves to anchor the loop in an optimal orientation in this flavodoxin (Bradley &
Swenson, 1999; Bradley & Swenson, 2001).
The importance of this loop in the modulation of the redox properties of the
cofactor is through its direct interaction with the N(5). Substitutions within this loop
region have differing effects on structure depending upon the source of the flavodoxin.
X-ray crystal structure analysis of the oxidized forms of the Gly61 mutants of
Desulfovibrio vulgaris flavodoxin, a position homologous to Gly57 of C .beijerinckii
flavodoxin, revealed that this loop is moved away from the flavin by 5-6 Å (O'Farrell et
al., 1998). On the other hand, substitution at the equivalent Asn58 position of Anacytis
88
nidulans [Drennan, 1999 #51] and Gly57 and Asp58 of C. beijerinckii (Ludwig et al.,
1997) did not result in any significant structural changes. The only difference noted in the clostridial flavodoxin mutants was the alteration of the conformation of the central peptide bond and the orientation of its carbonyl group, being all cis-O-down for
Asp58Pro and all trans-O-down for Gly57Thr in the oxidized state. While it is understandable for mutations involving this loop region, it is harder to rationalize the effects of more remote mutations when they seem to perturb this loop conformation and therefore the N(5) hydrogen bond strength. Recently the crystal structures of Y98H and
Y98W mutants of the flavodoxin from Desulfovibrio vulgaris were solved (Reynolds et al., 2001). Tyr98 is located in the 90’s loop and flanks the si face of the flavin, analogous to Trp90 in the clostridial flavodoxins (see Figure 17). Yet the only structural difference noted in these mutants was the conformation of the 60’s loop that had altered resulting in a new orientation of the central carbonyl group. Although the number of flavodoxin mutants has been extensive, the lack of availability of crystal structure information limits our understanding of their effects on the conformation of this loop.
It is therefore of importance to more thoroughly understand all the interactions made by this loop that is adjacent to N(5). However, to date, most of the substitutions involved single amino acid changes. In order to elucidate any cooperative interactions, especially involving Glu59 since that mutation alone resulted in a disproportionate destabilization of all redox states, a more detailed study involving multiple substitutions within this loop is required. Therefore, this study was initiated in order to more completely understand the role of this loop and any synergistic relationships among its
89
contributing amino acid residues. All side chain interactions in the loop were sequentially
eliminated by alanine-scanning mutagenesis (with exception of Gly57), ultimately
resulting in the minimalist sequence, 56AGAA (Figure 17). The use of alanine ensures against large-scale disruptions in main-chain configurations while eliminating the effects of the flavin interactions with each side chain. Gly57 was left as is because of the special structural requirements at this position (Kasim & Swenson, 2000; Ludwig et al., 1997).
The results demonstrate the general additivity of mutant effects and also further emphasize the roles of each of the amino acid residues involved. But, perhaps the most striking observation is that the mutant lacking all of the side chain interactions provided by this loop was still capable of binding FMN and thermodynamically stabilizing the neutral semiquinone state, a feature characteristic of flavodoxins.
90
MATERIALS AND METHODS
All information regarding materials and methods can be found in Chapter 2.
RESULTS
Characterization of the Double and Triple Mutants: The yields of soluble
holoprotein were low compared to wild type for the 56MGAA, 56AGDA and 56AGAA mutants. Some of the FMN was found to dissociate during purification, resulting in
A274/A446 ratios that were higher than wild type (5 to 6 versus 4.4, respectively). This
phenomenon is reflective of the significantly higher experimental dissociation constants
reported below. The spectral characteristics in all three redox states were recorded during
reductive titrations with sodium dithionite under anaerobic conditions. Because of the
relatively high dissociation constants of these mutants, at least a 5-fold excess of
apoprotein was included in these titrations in order to insure that >95% of the FMN was
bound. The spectral features of the oxidized species for all mutants differed from wild
type with bathochromic shifts of 10, 8 and 8 nm observed for the first transition for
56MGAA, 56AGDA and 56AGAA respectively, with concomitant changes in intensity
(Figure 18A). An almost complete absence of the characteristic shoulder at 450 nm was striking. The spectral characteristics of the second transition are known to be more sensitive to changes in polarity (Müller, 1991). Minor differences in the intensity of the second transition with a slight blue shift in the λmax were also noted, perhaps reflecting only small changes in the solvent exposure of the flavin ring in these mutants. The
91
extinction coefficients for the 450 nm transition were determined to be 9.95 + 0.25, 10.94
+ 0.6 and 11.45 + 0.8 mM-1 cm-1 for 56MGAA, 56AGDA and 56AGAA, respectively, which are similar to the recombinant wild-type value of 10.6 mM-1 cm-1 (Eren, 1990).
The neutral semiquinone species accumulated during the reductive titration of all of the mutant flavodoxins, but to a lesser extent than the stoichiometric levels observed for the wild type, reflecting the less stable FMN semiquinone. The spectra of the semiquinone displayed small changes in λmax and intensity, particularly in the 400-500 nm region, but were overall similar to wild type. The differences between the hydroquinone spectra of the mutants and also when compared to wild type were more dramatic (Figure 18B).
Changes in intensity of the absorbance at 315 nm, 380 nm, and the shoulder at 450 nm are clearly evident. Based on model compounds, the wild-type anionic hydroquinone spectrum is characterized by a relatively strong absorbance peak at 380 nm with the flavin in a planar conformation (Müller, 1991). The presence of this peak in all of these mutants suggests that a change in ionization state of N(1) is unlikely. Rather, these changes mirror those observed for other flavodoxin mutants and emphasize the sensitivity of the HQ spectrum to slight changes in its environment (Bradley & Swenson, 1999;
Bradley & Swenson, 2001; Kasim & Swenson, 2000; Lostao et al., 2000). Taken together, all these changes do indicate some degree of structural perturbation in the 50′s binding loop in these mutants.
92
56 FIGURE 18: UV-vis absorbance spectra for wild-type (closed squares), MGAA (open squares), 56AGDA (open circles) and 56AGAA (closed circles) flavodoxins in the oxidized (panel A) and fully reduced (panel B) states. The spectra were obtained during a reductive titration with sodium dithionite in a 50mM sodium phosphate buffer, pH 7.0 at 25°C. The inset in panel B depicts the spectral changes during the course of the reduction and illustrates the substoichiometric accumulation of the semiquinone species (580nm) for 56AGDA (open circles) and 56AGAA (closed circles) flavodoxin mutants.
93
12 A
10 ) -1
cm 8 -1
6
4 Extinction (mM Extinction 2
0 300 400 500 B 5 4 ) 3 -1 cm 2 -1 Extinction (580 nm) Extinction
1
0 024681012 Extinction (454nm) Extinction (mM Extinction
300 400 500 600
Wavelength (nm)
Figure 18.
94
One-Electron Oxidation-Reduction Potentials: The one-electron midpoint
potentials were determined in 50mM sodium phosphate buffer pH 7.0, at 25OC in the presence of at least 5-fold excess apoprotein to insure that all of the FMN remains bound.
The system potential at each point in the anaerobic titration was established using suitable redox indicator dyes with established midpoint potential values. All of the titrations were fit to the non-linear version of the Nernst equation for a single electron reduction. Representative plots of the titration for the OX/SQ and the SQ/HQ couples are shown in Figures 19. The midpoint potential for the OX/SQ couple (Eox/sq) for all mutants was more negative than wild type by approximately 70 – 100 mV (Table 12).
The increase for the SQ/HQ couple (Esq/hq) was also quite large, being 94mV for
56MGAA, 135 mV for 56AGDA and 145mV for 56AGAA (Table 12). The smaller separation in the potentials for each couple is consistent with the sub-stoichiometric accumulation and the lower stability of the semiquinone species observed during reductive titrations. The role of Met56 in destabilizing the hydroquinone species as reported previously (Druhan & Swenson, 1998) was clearly evident on comparison of the
56 56 Esq/hq values of the various mutants, with AGDA and AGAA displaying the least negative values, primarily reflective of the more stable hydroquinone species in these mutants.
95
100
80
60
40 % Semiquinone% 20
0 -400 -300 -200 -100 0 E h (mV)
56 FIGURE 19: Representative oxidation-reduction potential determinations for AGDA. The OX/SQ couple (open circles) and the SQ/HQ couple (closed circles) were determined using the indicator dyes anthraquinone-2,6-disulfonate and phenosafranin to establish the system potential (Eh), respectively. In each case, the Eh values are limited by the potential range of each dye. The amount of semiquinone formed is expressed as a percent to facilitate comparison between the two couples. The solid lines are the best fit of the data by non-linear regression analyses to the Nernst equation, generating midpoint potential values of -165 and -264 mV, respectively.
96
Kd
Eox/sq Esq/hq OX SQ HQ Flavodoxin ∆GOX ∆GSQ ∆GHQ (mV) (mV) (µM) (nM) (µM)
WTd -92 -399 0.018+0.002 0.0032 0.142 -10.6 -15.7 -9.3
M56Ad -72 -331 0.044+0.008 0.0036 0.011 -10.0 -15.6 -10.8
D58Ae -93 -380 0.042+0.02 0.0077 0.164 -10.1 -15.2 -9.3
E59Af -186 -298 0.58+0.07 3.97 3.48 -8.5 -11.5 -7.4
MGAA -189g -305h 2.74+0.12 21.0 24.3 -7.6 -10.5 -6.3 AGDA -165g -264h,I 2.84+0.1 8.57 2.0 -7.6 -11.0 -7.8
AGAA -177g -254h,I 2.79+0.11 13.44 2.12 -7.6 -10.7 -7.7
a Values are in millivolts at pH 7.0 and 25°C. b The dissociation constants in the oxidized state were measured either by fluorescence or visible spectroscopy and those for the reduced states calculated as described in the text. c Values are in kilocalories per mole. d From Druhan et al (Druhan & Swenson, 1998). e From Kasim et al (Kasim & Swenson, 2000). f From Bradley et al (Bradley & Swenson, 2001). g Midpoint potential was determined using anthraquinone-2,6-disulfonate as the indicator dye. h Midpoint potential was determined using Safranin T. i Midpoint potential was determined using phenosafranin. All values are an average of at least two independent experiments.
Table 12. Oxidation - Reduction Midpoint Potentialsa, FMN Dissociation Constantsb and Gibbs Free Energyc of FMN Binding for Wild-Type and Mutant C.beijerinckii Flavodoxins
97
These results demonstrate that the one-electron reduction potentials of both couples are highly dependent upon the side chain interactions made by both Met56 and
Glu59. Particularly, the 56MGAA mutant has midpoint potentials for both couples that are similar to those of the E59A mutant. Within this context, when the side chain at position
56 56 is also deleted to yield the AGAA mutant, both Eox/sq and Esq/hq shift to less negative values as seen with the M56A substitution relative to wild type (Druhan & Swenson,
1998). Similar conclusions can be drawn for the 56AGDA substitution as well. It is worth noting that despite these dramatic shifts in midpoint potentials, the two couples still remain reasonably well separated with Esq/hq being the more negative, resulting from the thermodynamic stabilization of the SQ state.
98
100
80
Wavelength (nm)
60 400 500 600
) 4 -1 cm -1 40 2
0 20 Absorbance (498 nm - 440 nm) nm - 440 (498 Absorbance Extinction (mM Extinction ∆ ∆ ∆ ∆ -2 % ∆ % ∆ % ∆ % ∆ 0 0 5 10 15 20 25 30 µ Apoprotein ( M)
FIGURE 20: Determination of the dissociation constant for the complex between oxidized FMN and mutant apoflavodoxins are shown as follows: 56AGDA (closed circles), 56MGAA (open squares) and 56AGAA (closed triangles). In each case, an FMN solution (~5 µM) in 50 mM sodium phosphate buffer pH 7.0 at 25°C was titrated with substoichiometric amounts of freshly prepared apoflavodoxin. After correction for dilution, the changes in extinction coefficients at 440nm and 498nm (inset) associated with complex formation were plotted as a function of the added apoprotein. The data are shown as a percent change in absorbance to facilitate comparison amongst the various mutants. The solid lines show the best fit to the data to a single-site binding isotherm.
99
Changes in FMN Binding. The dissociation constants in the oxidized state were measured
by monitoring the spectral changes associated with titrating a known concentration of
FMN with increasing amounts of freshly prepared apoprotein under conditions identical
to the redox potential determinations (Figure 20). While the Kd values for the semiquinone and hydroquinone forms of the cofactor cannot be measured directly, they can be easily calculated from the linked equilibria that connect the dissociation constants to the determined midpoint potentials as well as to the midpoint potentials of free FMN in solution (Dubourdieu et al., 1975). All of the mutants showed an approximately 150-fold increase in the dissociation constants for the oxidized state (Table 12). The largest changes in Kd for the single substitutions were observed for E59A, which increased by
56 56 56 32-fold (Bradley & Swenson, 2001). The Kd values for MGAA, AGDA, and AGAA are all nearly identical and about 5-fold higher than that for E59A. Thus, the removal of the Met56 and/or the Asp58 side chains in this context resulted in a similar loss of binding of the oxidized cofactor as was observed for each individual replacement. It is clear from these results that the hydrogen bonding interaction provided by Glu59 is the most critical interaction in all of these contexts. The semiquinone state was by far the most destabilized by the alanine replacements, with 6570-, 2680- and 4200-fold increases observed for 56MGAA, 56AGDA and 56AGAA, respectively. The substitutions in
56 MGAA also cause the largest loss in the stability of the HQ, with the Kd increasing 171- fold, whereas the increases for 56AGDA and 56AGAA were 14- and 15-fold, respectively.
One of the important observations made for the individual Met56 substitutions previously characterized was that complete or nearly complete elimination of the side chain at this
100
position, represented by the M56G and M56A mutants, bound the HQ state with
approximately equal affinity to that of the oxidized state emphasizing the unfavorable
sulfur-aromatic interactions that are present in the fully-reduced state (Druhan &
Swenson, 1998). This was accomplished in the M56A mutant by a 14-fold improvement
in the binding of the HQ state relative to wild type. It was therefore interesting to note
that both 56AGDA and 56AGAA had an approximately 12-fold improvement in the
binding of the HQ relative to 56MGAA such that they again bound the HQ with an affinity nearly equal to that for the OX state. Thus, the loss of the interactions provided by the side chains for Asp58 and Glu59 did not significantly affect the sulfur-flavin interaction provided by Met56. This was somewhat surprising, particularly for the loss of the anchoring effect of Glu59. Finally, while all the three redox states have been significantly affected by these mutations, it is obvious that it is the stability of the SQ state that has been affected to the greatest extent by the multiple replacements.
DISCUSSION
Three surface loops provide the majority of interactions with the FMN cofactor in the flavodoxin and related flavoproteins. The “50’s loop” in the C. beijerinckii flavodoxin provides the largest number of direct interactions with the isoalloxazine ring. The principal interactions include re-face sulfur-flavin contacts involving Met56, hydrogen bonding interaction to N(5)H in the reduced states with the carbonyl group between residues Gly57 and Asp58, hydrogen bonding interaction to C(4)=O with the backbone
101
amide group of Glu59 and the hydrogen bond formed with the side chain of Glu59 to
N(3)H. A significant goal has been to thoroughly appreciate the role of this loop in establishing many of the properties of the FMN. The importance of each individual amino acid has been demonstrated during the course of the study of various amino acid replacements at each of these positions (Bradley & Swenson, 1999; Chang & Swenson,
1999; Druhan & Swenson, 1998; Kasim & Swenson, 2000; Ludwig et al., 1997). It was noted that, with the exception of Glu59, removal of a single molecular contact by a point mutation causes relatively small reductions in the free energy of binding of the cofactor in all oxidation states (Table 12). The fundamental question arises as to whether these interactions act independently or whether synergistic effects are evident. The double mutant analysis approach, which involves pair-wise substitutions with alanine at multiple sites, would be useful here because the measured energetic changes can be compared with those of the corresponding single mutations. This should enable the identification of interactions between residues that would normally have been overlooked by single mutations alone. Also, central to this issue is whether it is possible to generate a flavodoxin mutant that can bind flavin despite having none of the side-chain interactions made with the 50’s loop. If so, what are its properties?
102
5
4
3 (kcal/mole)
wt 2 G ∆∆ ∆∆ ∆∆ ∆∆ 1 -
0 mut G ∆∆ ∆∆ ∆∆ ∆∆ -1
A A A A 9 A A 6 8 D 5 A A A D G G +A59 + G G M5 D5 E59A A 6 AGAA A M 58 +M 6 8+ *A58+A59 *A5 A 5 6+ A5 A 5 * * *A
FIGURE 21: Histogram depicting the differences between the binding free energy changes for the FMN cofactor in the oxidized (filled bar), semiquinone (shaded bar) and hydroquinone (open bar) states for the single, double and triple mutants. All values are relative to the binding free energy for the oxidized cofactor to wild-type flavodoxin. Also shown are the various ways (denoted by a *) in which the individual and double mutants can be added to predict the free energy changes that are observed in the actual multiple mutant.
103
Effects of the single substitution within the context of multiple mutations-
Methionine 56: The sulfur atom of the methionine side chain makes a weak attractive electrostatic interaction with the electron deficient isoalloxazine ring of the flavin in the oxidized state. As the FMN becomes more electron rich in the fully reduced state, this attractive interaction turns repulsive and contributes to the destabilization of the HQ
(Druhan & Swenson, 1998). The elimination of the Met56 side chain in the context of the multiple alanine replacements studied here resulted in free energy changes that are entirely consistent with this role of the sulfur atom (Figure 21). Comparison of the free
energy changes for the HQ state in 56AGDA and 56AGAA relative to that of 56MGAA reveal that there was an improvement in the binding by 1.5 and 1.4 kcal/mole respectively. This is comparable to the stability of the HQ state of M56A that is 1.5 kcal/mole greater than wild type. Evidently, the effects of removal of the methionine side chain are noticeable in these multiple mutants and counter the destabilizing effect of the
Glu59 mutation. This suggests that the flavin-sulfur interaction was operative even within an altered electrostatic environment and altered flavin binding.
Glutamate 59: Previous studies have demonstrated the importance of Glu59 in the stabilization of the FMN complex in all redox states (Bradley & Swenson, 1999; Bradley
& Swenson, 2001). It is remarkable that the majority of the characteristics of those multiple replacement mutants lacking the Glu59 side chain reflect those of the E59A containing the individual substitution. The λmax for the first transition for all three mutants shows identical shifts to longer wavelengths (454 – 456nm) as that of the E59A mutant (Bradley & Swenson, 2001). The binding free energy histogram also reveals that
104
the free energy changes for these mutants are dominated by the loss of the interaction(s) provided by Glu59 (Figure 21). These large changes are immediately attributable to the elimination of the donor-acceptor interactions provided by the side chain carboxylate group of Glu59 and possibly to slight alterations between its main chain amide group and the FMN C(4)=O (Bradley & Swenson, 2001). Furthermore, the stabilities of the OX and
SQ states are very similar among the mutants which is not surprising if you take into account that the individual Met56 and Asp58 mutations had negligible effects on these oxidation states (Figure 21) (Druhan & Swenson, 1998; Kasim & Swenson, 2000).
Additivity of mutational effects: Our complete understanding of the function of a given amino acid requires knowledge beyond its singular role because its function may also depend upon interactions with other residues. The existence of such functionally important interactions between side chains is most convincingly demonstrated by non- additivity in double-mutant thermodynamic cycles. The sum of the free energy changes of each double mutant are compared with that predicted from the sum of the constituting single mutants in Figure 21. Also shown in the figure are the various ways in which the stabilities of the individual and double mutants can be combined to give the same substitutions as in the triple mutant. The values were additive within experimental uncertainty for all redox states except for the HQ state in some instances. This situation is referred to as simple additivity (Wells, 1990) wherein the sum of the free energy changes derived from the single mutations is nearly equal to the free energy change measured in the multiple mutants. What is extraordinary is that it is evident from the free energy histogram (Figure 21) and the correlation plot (Figure 22) that of all redox states, it is
105
the SQ that is the most affected, yet based on the single replacements, the prediction of
the free energy changes of the SQ for the multiple mutants is remarkably accurate.
Figure 22 plots the free energy changes predicted using various single and double mutant
combinations versus the observed free energy changes in the mutants. Additivity would
be reflected in data points that lie on the diagonal. A discrepancy of approximately 1.1 +
0.1 kcal/mole in the estimation of the free energy change for the HQ state was observed, suggesting that perturbations here are not totally additive. Generally, for simple additivity, an error as large as + 25% has been observed (Wells, 1990). Part of this can be attributed to the compounding of errors when summing the single mutants and the rest is probably due to a weak interaction energy term between the residues that is normally neglected in the summation. A difference of 1.0 kcal/mol translates into a midpoint potential difference of 43mV, which is well outside the error range of determination. Is it coincidental that the predicted values for the free energy change for all mutants all fall short by a similar amount despite the obvious differences in stability of the HQ state, at least among the double mutants? Given the strong related correlation for these values (the dashed double dot line), we suggest that there is a common reason behind this non- additivity amongst all these mutants.
106
FIGURE 22: Plot of the free energy changes predicted from the constituting single and double mutants versus the actual observed free energy changes for the multiple mutant in the oxidized (closed symbols), semiquinone (shaded symbols) and hydroquinone (open symbols) states. The data for each mutant are represented as follows: 56AGDA (circles), 56MGAA (squares), 56AGAA (diamonds) (all versus the sum of the appropriate single replacements), and 56AGAA (inverted triangle relative to the sum of 56MGAA + M56A) or 56AGAA (hexagons relative to the sum of 56AGDA + D58A). The solid diagonal line represents a totally additive correlation between the values for mutants containing multiple replacements and the summation of the appropriate mutants containing the individual replacements. The dashed lines on either side indicate the limit of uncertainty in the data. The dashed double dot line indicates the observed correlation for the hydroquinone state for which the values fall outside the limit for simple additivity (see text). The values for the triple mutant and the two different summations involving the double mutants (i.e. 56MGAA + M56A and 56AGDA + D58A) fall on the diagonal. Where error bars are not indicated, the errors are commensurate with the relative symbol size.
107
6
5
4
G (kcal/mole) G 3 ∆∆ ∆∆ ∆∆ ∆∆
2
Observed Observed 1
0 0123456 ∆∆ Predicted G (kcal/mole)
Figure 22
108
The quantitative effect of a second mutation may be of several types - no effect, antagonistic, partially additive, additive or synergistic with respect to the first mutation and this can be extrapolated to higher order mutants as well (Mildvan et al., 1992). In this particular case, the effects are clearly synergistic because the free energy changes observed in the multiple mutants exceed the sum of the changes in the single mutants.
Synergy has been attributed to extensive unfolding of the enzyme and/or to the presence of strain that is introduced by the individual residues with each single mutation being less debilitating as the loss in free energy of binding is compensated partially by the loss in strain (Mildvan et al., 1992). There is little evidence of extensive unfolding of these mutant structures. Cofactor binding is retained in all cases, and although its affinity is reduced, there was no pattern to this loss among the multiple mutants. In fact, Kd’s are all quite similar. The changes observed in the visible spectrum are rather modest and are not consistent with large changes in the flavin environment. Circular dichroism spectroscopy does not reveal any global disruption of secondary structure (Bradley & Swenson, 1999).
In terms of strain, it is difficult to rationalize how the surface exposed side chain of
Asp58 can introduce any significant strain in the protein. The methionine side chain is also relatively flexible and packs against the isoalloxazine ring with minimal steric interference, ruling out any possibility of strain being introduced here (Druhan &
Swenson, 1998). Perhaps the anchoring role played by the side chain carboxylate of
Glu59 may introduce some degree of strain in the protein but this alone does not provide an explanation.
109
The most probable cause for the breakdown of additivity is one that is cited repeatedly in literature – the mutated residues interact with each other in some manner.
However, direct contact among the residues can be ruled out as the sulfur atom of Met56 is situated at a distance that is >10Å from the carboxyl side chain of Glu59, while the closest approach of the Asp58 side chain to Glu59 is 5.2Å (Ludwig et al., 1997). An
indirect interaction either through electrostatic interactions or structural perturbations
such that the residues no longer behave independently is likely. Since all mutations that
showed non-additivity involved the negatively charged side chain of Glu59, an
electrostatic interaction as being the probable cause is conceivable. Non-additivity in the
HQ state alone is not so easily rationalized, but perhaps this interaction exists or is
enhanced to a detectable level in the fully reduced anionic state when the N(3)H
hydrogen bond interaction is known to be relatively unimportant (Bradley & Swenson,
1999). The fact that summation using 56MGAA+M56A and 56AGDA+D58A results in simple additivity for the HQ state for the triple mutant 56AGAA does indeed indicate that this interaction is already accounted for in the double mutants with either Met56 or
Asp58 (Figure 21 and Figure 22 open inverted triangle and hexagon). In addition, it also provides evidence against any interaction between Met56 and Asp58, which supports the conclusion that the lack of additivity results only from the side chain interaction of
Glu59. It should be noted that the lack of the anchoring effect of Glu59 is believed to cause the flavin ring to sit differently in its binding pocket such that the interactions made with the loop are now sufficiently altered. This phenomenon that is present in all mutants could explain the differences in the visible spectra and binding of FMN when compared
110
to wild type. Whether the underlying nature of the interaction responsible for non-
additivity lies in electrostatics or slight structural perturbations is not clear, however it
does seem to involve the Glu59 residue. Further support for this theory comes when
substitutions at position 57 and 58 were considered (data not shown, see also refs (Chang
& Swenson, 1999; Kasim & Swenson, 2000; Ludwig et al., 1997), where the effects were
additive for all three redox states. There are numerous examples of non-additive effects
of double mutations in literature stressing again that not all residues in proteins act
independently and interpretation of the results of single mutations must be done with
caution (Mildvan et al., 1992; Ohmae et al., 1998; Zhang et al., 1991).
56AGAA: Special mention must be made of the triple mutant, 56AGAA. In this mutant, all the side chain interactions provided by this loop are essentially eliminated through their replacement with alanine while maintaining the main chain configuration through the retention of the crucial structural characteristics provided by Gly57 (Kasim &
Swenson, 2000; Ludwig et al., 1997). Surprisingly, these rather drastic alterations did not preclude FMN binding. Furthermore, despite the 3-kcal/mol loss in binding free energy for the oxidized cofactor and the elimination of nearly all the side chain interactions with the isoalloxazine ring, the neutral FMN SQ is still substantially thermodynamically stabilized and the two redox couples remain reasonably well separated. As the site of the conformational change is relatively intact in 56AGAA (and all the other mutants), the changes in midpoint potentials observed are the direct result of the loss of side chain interactions with Met56 and Glu59. A distortion in the N(5) interaction can not be ruled out due to the removal of the Glu59 side chain, however, the extent of disruption would
111
likely be uniform for all these mutants, including the E59A mutant. Main chain
configurations for the turn likely have not been substantially altered given the ability of
the protein to bind FMN, which displays some of the characteristic spectral qualities as
the wild type. Molecular modeling and computational studies also support this conclusion
(Figure 17) [data not shown].
The one remaining isoalloxazine ring/side chain interaction in the 56AGAA mutant is with the si-face aromatic residue, Trp90 (Figure 17). For another study, this residue was independently replaced with alanine. It was observed that despite the approximately 30-fold increase in the dissociation constant for the oxidized FMN, the midpoint potential for the OX/SQ couple for W90A was only marginally affected by this replacement, again demonstrating that aromatic-flavin interactions do not seem to play a major role in the preferential stabilization of the SQ [unpublished results and (Swenson &
Krey, 1994)]. In fact, we entertained the notion of generating the 56AGAA/W90A quadruple mutant for which all side chain interactions with the flavin ring are eliminated.
However, based on the observed dissociation constants for 56AGAA and W90A, it was thought that this protein would bind FMN too weakly for meaningful biochemical characterization, so it was not produced. Nonetheless, collectively, these observations emphasize the overriding importance of the main chain interactions with the N(5)H of the
FMN and the associated conformational change in this loop that occurs during the reduction, especially in the thermodynamic stabilization of the neutral SQ state. These studies can now leave little doubt of this fact, which was a very early and insightful observation for this group of proteins (Burnett et al., 1974).
112
CHAPTER 5
CLONING AND CHARACTERIZATION OF THE FMN-BINDING DOMAIN OF
CYTOCHROME P450 REDUCTASE FROM Bacillus Megaterium
INTRODUCTION
Cytochrome P450BM-3, a fatty acid monooxygenase from Bacillus megaterium
(ATCC 14581), is the first known bacterial P450 that belongs to the microsomal P450 class, resembling the eukaryotic P450s in both structure and function. The 119 kDa holoenzyme consists of three functionally independent domains; the heme domain and the diflavin domain containing distinct FAD and FMN binding subdomains, all of which are part of a single polypeptide chain (Figure 23). The heme domain of P450BM-3 exhibits about 25% sequence identity with mammalian fatty acid ω-hydroxylases, and the diflavin domain shows about 35% identity and 56% similarity with mammalian NADPH- cytochrome P450 reductase (CPR), but the local regions involved in binding the flavins are notably conserved (Miles et al., 2000; Ruettinger et al., 1989). There exists significant homology between the NADPH/FAD-binding domain and ferredoxin NADP+ reductase (FNR), whereas the FMN-binding domain is homologous to the FMN- containing bacterial flavodoxins (Figure 24). The expression of the holoenzyme, the
113
individual functional domains; the heme (BMP), the FMN-binding domain, and the FAD- binding domain as soluble proteins in E. coli has made this system a very appealing model for studying microsomal P450s. Various combinations of these domains, such as the heme and FMN (BMP/FMN-binding domain) and the reductase domain (BMR) has also been successfully expressed as soluble proteins and greatly facilitate studies on the mechanism of electron transfer.
114
1 471 654 1048
Heme Domain FMN Domain FAD/NADPH Domain
Structural homology with Structural homology bacterial flavodoxins with ferredoxin NADP+ reductase (FNR)
Figure 23. Schematic of the domain organization in P450BM-3.
115
Figure 24. Structural similarities between the bacterial flavodoxin (top panel) and the FMN-binding domain of P450BM-3 (bottom panel). The flavin-binding loops are colored red and purple respectively. The orientation of the structures are slightly different to optimize the view.
116
From the detailed studies of the kinetics of electron transfer and of reduction of the flavins of BMR and CPR, it was concluded that under physiological conditions, CPR cycles between the 1-3-2-1 reduced states whereas BM-3 cycles between the 0-2-1-0 reduced states, where the numbers represent the total number of electrons present. These differences appear to be related to the flavin potentials of these enzymes. Figure 25
details the differences in the electron transfer mechanisms of CPR and P450BM-3. For
CPR, the enzyme exists in the 1-e- reduced state with FMN as the semiquinone species.
Upon binding to NADPH, the 2-e-s are accepted by FAD to form the 3-e- reduced species. Inter-flavin electron transfer results in the formation of the FAD semiquinone and the FMN hydroquine that subsequently transfers its electron to the P450 and returns to the neutral semiquinone state (2-e- reduced enzyme). Another round of inter-flavin electron transfer occurs resulting in the formation of the oxidized FAD and the FMN hydroquinone that can once again reduce the P450 molecule. This returns the enzyme to the resting 1-e- reduced state (Sevrioukova & Peterson, 1995). In contrast, for P450BM-3,
the enzyme exists in the fully oxidized state. Upon binding to NADPH, the 2-e-s are accepted by FAD to form the 2-e- reduced species with the FAD existing as the hydroquinone. Inter-flavin electron transfer results in the formation of two semiquinone species, the neutral FAD semiquinone and the anionic FMN semiquinone. The latter transfers its electron to the P450, when substrate bound, returning the enzyme to the 1-e- reduced state (FADsq/FMNox). Another round of inter-flavin electron transfer generates the FMN semiquinone that can once again reduce the P450 and returns the enzyme to the fully oxidized state (Daff et al., 1997; Sevrioukova et al., 1996a). Thus, for CPR, the
117
electron donor to the P450 is the FMN hydroquinone, the FMN cycling between the
sq/hq states. While for P450BM-3, the electron donor is the FMN semiquinone, the FMN
cycling between the ox/sq states. It is therefore believed that the oxidative-reductive
properties of most likely the FMN moiety of P450BM-3 are different from those of CPR,
despite the functional analogy. Unlike CPR and flavodoxins that stabilize the blue neutral
form of the FMNsq, the FMN-binding domain of cytochrome P450BM-3 (henceforth
called BM-3) utilizes the anionic FMN SQ as the catalytically relevant species (Hazzard
et al., 1997; Sevrioukova et al., 1996a; Sevrioukova & Peterson, 1995). This anionic SQ is only transiently formed under equilibrium conditions due to the reversal of the midpoint potentials in BM-3 where it is the Eox/sq (-206mV) that is more negative than the Esq/hq (-177mV) causing the SQ to be the thermodynamically unfavorable species.
While the two-electron HQ species was thermodynamically more stable, it was incapable of reducing the heme, a function performed by only the low-potential anionic SQ.
Interestingly, the flavin in the isolated FMN-binding domain was shown to retain the redox and electron-accepting properties of the holoenzyme bound FMN as well as in the isolated diflavin domain suggesting that the structure of BM-3 is conserved upon separation from the holoenzyme (Daff et al., 1997), thus making it amenable for
structure-function studies.
118
¯
are the blue the are ! andFMNH !
is the red anionic semiquinone form and FADH¯ and FMNH ¯ ! Electron transfer mechanisms of CPR and P450BM-3. Top panel shows the direction of electron flow. The
Figure 25. notations are as follows: FAD are the and FMN oxidized forms FADH of the cofactor, are the anionic hydroquinone forms of the flavin cofactors. neutral semiquinone forms of each cofactor, FMN
119
The crystal structure of a putative electron-transfer complex between the FMN- binding domain and the heme domain was recently solved at 2.03Å resolution, providing for a more detailed look at the structure as well as apoprotein-FMN interactions. Mainly neutral and hydrophobic residues were found to surround the flavin-binding site unlike the clusters of conserved negatively charged residues seen not only in flavodoxins but also in CPR (Figure 26)(Wang et al., 1997; Zhao et al., 1999). The asymmetric charge
distribution results in a dipole moment along the axis passing through the FMN, which is
thought to be important in the formation of electron-transfer complexes between
flavodoxins and CPR and their respective redox partners (Feng & Swenson, 1997;
Jenkins et al., 1997; Matthew et al., 1983; Weber & Tollin, 1985). This element of redox
partner recognition is not required for BM-3 where the heme, present on the same
polypeptide chain, is always in proximity with the FMN. In addition, BM-3 has an
ionizable histidine residue (His539) adjacent to the isoalloxazine ring. Protonation of this
histidine during the catalytic cycle could raise the total positive charge near the flavin.
Redox-linked ionization of a histidine residue was shown to stabilize the FMN anionic
hydroquinone in the D. vulgaris flavodoxin by the favorable through-space electrostatic
interaction from the positive charge on the imidazole ring (Chang & Swenson, 1997).
This residue, along with the presence of two lysine residues within 13 Å of N(1), might
be responsible in part for the high Esq/hq and the formation of the anionic SQ species.
120
Figure 26. Electrostatic potential around the flavin-binding site. The proteins shown are (a) FMN domain of P450BM-3, (b) D. vulgaris flavodoxin, and (c) FMN domain of microsomal P450 reductase. All three are oriented approximately the same with the FMN-binding site facing the viewer. Figure is taken from Sevrioukova et al (Sevrioukova et al., 1999).
121
As was discussed in detail in Chapter 1, the unique conformation of the loop region that is homologous to the 50’s loop region in the flavodoxins is likely to have a direct effect on the redox properties of the FMN cofactor. The potentially strong hydrogen bond to N(5) in the oxidized state may play a role in stabilizing that state and could be the reason why the flavin cycles between the ox/sq states rather than the sq/hq states, as seen for CPR and the flavodoxins. Extrapolation from previous studies of flavoproteins can help predict some of the factors that might modulate the redox potentials of this flavoprotein. However, a better understanding of the structure in the reduced state becomes essential in order to identify any structural changes that might be occurring on reduction, the influence of which on redox potentials have already been characterized in depth in several flavodoxins (Chang & Swenson, 1999; Drennan et al.,
1999; Hoover et al., 1999; Kasim & Swenson, 2000; Ludwig et al., 1997; O'Farrell et al.,
1998). While the protein has been partially characterized with the determination of its midpoint potentials at pH 7.0 (Daff et al., 1997) and dissociation constant for binding of
FMN and several other flavin analogues (Haines et al., 2000), many unresolved questions remain unanswered. This work was undertaken in an attempt to more completely characterize the protein and to try and understand the events that occur on reduction of the flavin. The study focused primarily on the N(5) interaction as some change at this position is a prerequisite for protonation to occur. Another long-term goal of this project is to identify the factors that, in contrast to the flavodoxin and CPR, destabilize the anionic FMN semiquinone state in this unique flavoprotein. Equilibration with indicator
122
dyes was employed for the determination of the midpoint potentials unlike the previous
method of potentiometry.
EXPERIMENTAL METHODS
All information regarding materials and methods can be found in Chapter 2.
RESULTS
Cloning and expression of BM-3
The gene fragment encoding the FMN-binding domain of cytochrome P450BM-3
(BM-3) was cloned using the polymerase chain reaction (PCR) approach. Genomic DNA from Bacillus megaterium was used as a template and oligonucleotide primers (sense and antisense) to the known termini of the discrete FMN-binding domain. The 5′ and 3′
oligonucleotide primers were designed based on the known sequence of the FMN domain
of P450BM-3. Arg471 and Ser649 were chosen as the amino and carboxy terminal ends
(Sevrioukova et al., 1996b) as this isolated region has already been shown to be capable of binding FMN. The 5′ primer was synthesized with an overhang that introduced an
EcoRI site while the overhang of the 3′ primer introduced a BamHI site in addition to two
stop codons. The 569 base pair PCR product was cloned into the pT7-7 expression vector
after digestion with the appropriate restriction enzymes. The use of the start codon of the
vector slightly upstream of the EcoRI site introduced an additional 3 residues (Ala, Arg,
Ile) which are inconsequential as not only are they remote from the FMN binding site but
123
are part of the linker region to the heme domain and thus solvent exposed (Sevrioukova
et al., 1999). Therefore, they are unlikely to affect the redox properties of the protein.
The soluble protein was expressed in BL21(DE3) cells at high yields under the control of the T7 promoter, the time induced expression of which can be seen in Figure
27. The oxidized BM-3 has absorbance maxima at longer wavelengths (388 and 468 nm)
than is observed for the highly homologous flavodoxins, with A278/468 and A468/388 values of 5.18 and 1.19, respectively (Sevrioukova et al., 1996b). The extinction coefficient for
BM-3ox at 468nm was determined to be 9.9 + 0.3 mM-1cm-1 that is similar to the value of
9.8 + 0.2 mM-1cm-1 obtained by Sevriuokova et al (Sevrioukova et al., 1996b). The
presence of a long shoulder between 550 and 700 nm is a unique spectral feature of this
FMN domain. It is likely due to the formation of a charge transfer complex between
Trp574 and the flavin ring, the coplanarity of which would facilitate the sharing of
electrons between the electron rich indole ring and the electron deficient flavin ring in the
oxidized state.
124
1 2 3 4 5 6 7 8 9
43,250
28,300
18,350
14,025 5,500 2,750
Figure 27. Expression of wild-type BM-3 under the control of the T7 promoter in BL21DE3 cells. Lanes 1 and 9: prestained protein ladder, lane 2: uninduced; lane 3: 4 hour induction, lane 4: 8 hour induction, lane 5: 16 hour induction, lane 6: 24 hour induction, lane 7: 48 hour induction and lane 8: purified protein. The arrow marks the position of the induced protein. The sizes of the protein ladder are in Daltons.
125
0.3
0.2
Absorbance 0.1
0.0 300 400 500 600 700 800 Wavelength (nm)
Figure 28. UV-visible absorbance spectra of the FMN-binding domain of P450BM-3 in the oxidized and fully reduced states. Data were collected during a reductive titration with sodium dithionite under anaerobic conditions in 50mM sodium phosphate buffer, pH 7.0 at 25°C.
126
Oxidation-reduction potentials of BM-3 and their pH dependency
The ability of sulfite ions to form an adduct at N(5) with the flavin prosthetic group has
been documented for a number of flavoproteins. This results in a bleaching of the visible
absorption spectrum with concomitant production of a spectrum similar to, but not
identical with, that of the fully reduced flavoprotein (Müller & Massey, 1969). It was of
obvious interest to determine if BM-3 reacted with sulfite ions with significant avidity
and if so, the use of sodium dithionite as a reductant would be rendered unfeasible, sulfite
being a product of its anaerobic oxidation. Spectral changes indicative of a flavin-sulfite
adduct formation were not observed on addition of a concentrated solution of freshly
prepared sodium sulfite. This is in keeping with the crystal structure of the protein where
the N(5) of the flavin ring is appears to be involved in a strong hydrogen bond with the
amide proton of Asn537 and hence not accessible to solvent (Sevrioukova et al., 1999).
The presence of this hydrogen bond is also apparent in the one-dimensional NMR studies
detailed later. Lack of the flavin-sulfite adduct formation enabled us to use sodium
dithionite as a reductant for reductive titrations.
The spectral characteristics of BM-3 for the OX and HQ redox states were recorded during a reductive titration with sodium dithionite under anaerobic conditions
(Figure 28). Reduction did not occur through two discrete one-electron steps, but rather by a single two-electron reduction to the HQ. The reoxidation pathway was identical to the reduction. It has previously been noted that this protein does not thermodynamically stabilize the SQ species as was evidenced here by the presence of an isosbestic point at
350 nm which was maintained throughout the titration until the addition of sodium
127
dithionite was in excess. However, a transient increase in the absorbance near 380 nm
does indicate that the the red anionic SQ species does form momentarily, especially
during the early portion of the titration.
The two-electron oxidation-reduction potentials of BM-3 were determined at
25°C in either 50 mM sodium acetate, 50 mM sodium phosphate or 12 mM sodium pyrophosphate over the pH range of 6 – 8 during the reductive titration with sodium dithionite in the presence of redox indicator dyes with established mid-point potentials.
The titration at each of the pH values tested could be fit to the linear version of the Nernst equation with a slope of 29.5 + 5 mV (Figure 29). Titrations at some pH values have
been omitted for clarity. The Em value at pH 7.0 was determined to be –194 + 5 mV.
This is equivalent to the value of –192 + 13 mV, which was determined by electrochemical methods (Daff et al., 1997). The value of Em was found to vary by 35.5 mV/pH unit (Figure 30) indicating that the reaction involves the transfer of only one proton per two electrons. This data provides strong evidence that the HQ remains in the anionic state. Because the plot is essentially linear over the pH range tested, it was not possible to determine a pKa value of the HQ but suggests that it must be <6.0. This should be compared to a pKa value of 6.7 for FMN in solution (Dudley et al., 1964; van
Schagen & Muller, 1981).
128
-140
-160
-180 h E -200
-220
-240 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
log[OX/HQ]
Figure 29. Representative titrations for the determination of Eox/hq for BM-3 at different pH values. The data are plotted as the linear version of the Nernst equation with pH 6.0 (closed circles), pH 6.5 (open circles), pH 7.0 (closed squares), pH 7.5 (open squares) and pH 7.75 (closed hexagons).
129
-140
-160
-180 ox/hq
E -200
-220
-240 6.06.57.07.58.0
pH
Figure 30. Variation of Eox/hq with pH. The slope of the line is 35.5mV/pH indicating the transfer of one proton per two electrons. Each point is the average of at least two independent titrations.
130
The transfer of a single proton raises the question whether it is the OX/SQ or the
SQ/HQ couple that is linked to the proton transfer. If proton transfer is linked to the
OX/SQ couple, the Em value of the SQ/HQ couple will be independent of pH while that of the OX/SQ couple will vary by 60mV/pH unit and indicate the formation of the blue neutral SQ (Figure 31, scheme 1). On the other hand, if the red anionic SQ were the
species formed then it would be the SQ/HQ couple that is pH dependent (Figure 31,
scheme 2). For BM-3, it is not possible to experimentally obtain the values for Eox/sq
and Esq/hq as these two processes appear to occur simultaneously and the data is best fit
by a two-electron reduction function. However, this two-electron process can be
separated into two one-electron reduction processes. Figure 32A shows the fit of the
changes in extinction at 468 nm at pH 7 to an equation (equation 1, Chapter 2)
comprising the sum of two one-electron redox functions designed to model the
absorbance of the flavin passing through three different oxidation states (Daff et al.,
1997). The fit to the data yields the midpoint potentials for the individual couples. The
extinction coefficient for OX was determined experimentally while that of the HQ was
calculated from the standard spectra. The extinction coefficient for the SQ was assumed
to be 0 as no SQ was seen to accumulate. The fits were less accurate at higher pH
possibly due to accumulation of small amounts of the SQ that would contribute to the
extinction at 468 nm. Nevertheless, the Em values for the OX/SQ couple were found to
be clearly independent of pH while that of the SQ/HQ couple varied by 68mV/pH unit
(Figure 32B). From these data, it is evident that proton transfer is linked to the SQ/HQ couple and consequently, the SQ species must be the red anionic form. Therefore, both
131
the UV-visible data as well as the pH dependency data corroborate one another and provide indirect evidence to the formation of the anionic SQ.
132
SCHEME 1
Eox/sq + ! OX + e¯ + H SQH!
Esq/hq ! SQH! + e¯ HQ¯
SCHEME 2
Eox/sq OX + e¯ SQ¯
+ Esq/hq SQ¯+ e¯ + H HQ¯
Figure 31. Schemes depicting the possible modes of electron and proton transfer. Scheme 1 indicates a scenario where Eox/sq is pH dependent with the formation of the blue neutral semiquinone species whereas Scheme 2 shows the formation of the anionic semiquinone with Esq/hq being pH dependent. Eox/sq and Esq/hq are the midpoint potentials associated with reduction from the oxidized to the semiquinone state and the semiquinone to the hydroquinone state, respectively. SQH" and SQ- represent the neutral and anionic forms of the semiquinone and HQ- is the anionic hydroquinone formed.
133
Figure 32. Determination of Eox/sq and Esq/hq and their pH dependency. Fig (A). The data points show the changes in the extinction at 468nm with the system potential at pH 7.0. A fit of the data to equation 1 (chapter 2) yields the midpoint potentials for the individual couples (solid line). Similarly, the values at the different pH values were obtained. Fig (B).pH dependency of Eox/sq (closed circles) and Esq/hq (open circles). The linear regression through each of the data sets is shown in the figure in solid lines. A slope near zero for Eox/sq indicates that proton transfer is not linked with the transfer of the first electron and thus, the semiquinone species is the anionic form.
134
10 A ) -1 8 cm -1 6
4
Extinction (mM Extinction 2
0 -400 -200 0
Eh
-80 B
-120
m -160
E m = 68
-200 m = -0.5 -240
678 pH
Figure 32.
135
Determination of the dissociation constant for FMN in the oxidized state.
It was observed that only 50 – 70 % of the apoprotein of BM-3 prepared by the TCA
precipitation method was able to bind FMN. Incubation of the apoprotein in the presence
of β-mercaptoethanol did not increase the yield of active protein suggesting that
dimerization via the single cysteine residue was not a probable cause (Haines et al.,
2000). It should be noted that the A278/468 ratio of 5.18 is indicative of a 1:1 complex
formation and therefore excess apoprotein that is not bound to FMN is not present in the
holoprotein sample. This would indicatate that the isolated domain is susceptible to
denaturation under the acidic conditions utilized to make the apoprotein. The spectral
changes on binding of active apoBM-3 to FMN were identical to those previously
observed with distinct troughs and peaks at 358 nm, 442 nm and 502 nm, respectively
(Haines et al., 2000). The appearance of a long-wavelength band (500-700nm)
characteristic of this flavoprotein was also present (Figure 33, inset). The linearity of the
absorbance changes for most of the titration curve suggests stoichiometric binding and a
low Kd vaule for FMN for this flavoprotein. Indeed, a fit to the data yielded a Kd in the
nanomolar range. In order to more accurately determine the Kd, the quenching of FMN
fluorescence on binding to the protein was monitored which is a more sensitive technique
and allows the use of lower concentrations. Again, the linearity of the fluorescence
changes over a significant region of the titration curve indicated very tight binding
(Figure 34). An average Kd value of 6 + 4 nM was obtained from eight independent experiments. Such tight binding is not surprising given that the binding site and flavin- protein interactions are similar to that of the flavodoxins which also bind FMN in the
136
nanomolar range. In addition, unlike most flavodoxins, a strong hydrogen bond with N(5) is also present which may likely contribute to the binding. Old Yellow Enzyme (OYE) similarly binds FMN very tightly and also possesses the hydrogen bond to N(5) in the oxidized state (Beinert et al., 1985; Fox & Karplus, 1994). However, a measure of the dissociation constant of OYE for 5-deaza-FMN which would lack the N(5) hydrogen bond resulted in a Kd value that was 5 orders of magnitude higher (Abramovitz &
Massey, 1976). Such a significant increase is unlikely solely due to the lack of hydrogen bond formation at the N(5) position and may possibly reflect the altered electronic structure of 5-deaza-FMN and/or structural changes that are necessary to accommodate the extra proton. To date, the exact contribution of this hydrogen bond is still unknown.
137
0.10
0.08
0.06 0.10 442-502nm
0.04 0.05
0.00 Difference Absorbance δ δ δ δ 0.02 -0.05 Wavelength (nm)
400 600 800 0.00 02468101214 Apoprotein (µM)
Figure 33. Spectrophotometric titration of FMN to apoBM-3. The data show the changes in absorbance upon binding with the solid line showing the best fit to the data. Inset shows the difference absorbance spectra upon binding, recorded after addition of increasing amounts of apoBM-3. For clarity, not all difference spectra are shown.
138
160
120
80
40 Fluorescence Intensity
0 0.0 0.2 0.4 0.6 0.8 1.0 µ Apoprotein ( M)
Figure 34. Determination of the dissociation constant in the oxidized state of the FMN cofactor to apoBM-3. A FMN solution (~0.1µM) in 50 mM sodium phosphate buffer, pH 7.0 at 25°C was titrated with increasing amounts of a 50–60 µM apoBM-3 solution having an identical buffer composition. The data were corrected for dilution. The solid line represents the best nonlinear regression fit of the data to a binding isotherm for a 1:1 complex, from which the dissociation constant was derived.
139
One- and two-dimensional NMR spectroscopy in the oxidized and fully reduced state
Lack of crystal structure data in the reduced state hinders the interpretation of biochemical results. This is particularly crucial when trying to understand why this flavoprotein favors the OX and/or the HQ state over the anionic SQ which differs significantly from the flavodoxins and CPR, that tend to stabilize the blue neutral SQ state over the other redox states. To gain some insight into the interactions that might be occurring in the fully reduced state, one-dimensional 15N-NMR spectroscopic analysis of
BM-3 reconstituted with 15N-enriched FMN was performed in both OX and HQ states
(Figure35 A and B). Such data would yield information on the hydrogen bonding environment and/or the hybridization state at the particular flavin nitrogen atom. In the oxidized flavin, N(1) and N(5) atoms represent pyridine-like or β-type nitrogen atoms that are extremely sensitive to hydrogen bonding, which results in large upfield shifts.
N(3) and N(10) atoms on the other hand are pyrrole-like or α-type nitrogen atoms that are relatively insensitive to hydrogen bonding, showing small downfield shifts (Witanowski et al., 1981). The results for BM-3 are compared to the 15N chemical shift values for
FMN in water and for TARF in chloroform, in both the oxidized and fully reduced states
(Vervoort et al., 1986).
140
A N(1) N(10)
N(5) N(3)
B
N(3) N(10) N(1) N(5)
Figure 35. One-dimensional 15N-NMR spectra of BM-3. A. The oxidized state. B. The fully reduced state.
141
In the oxidized state, the most dramatic shift was observed in the chemical shift
value of N(5) has shifted more than 10ppm upfield from the value of FMN in a polar
solution (Table 13). This is consistent with the presence of a very strong hydrogen bond
at N(5) in the oxidized state and corroborates the crystal structure data where a hydrogen
bond with the amide proton of Asn 537 was noted (Sevrioukova et al., 1999). The upfield
shift of N(1) compared to FMN in a polar solution also indicates a strong hydrogen bond
at this position. Again, a hydrogen bond with a backbone amide was also observed in the
crystal structure (Sevrioukova et al., 1999). As the chemical shifts of N(3) and N(10) are
very similar, they cannot be assigned purely by analogy to the values of FMN in a polar
solution. In this case, the N(3) chemical shift was unambiguously assigned by repeating
the 1D 15N-NMR experiment with and without broad band decoupling and applying the
DEPT (distortionless enhancement of polarization transfer) pulse sequence (Doddrell et
al., 1982). The 1H-15N DEPT experiment makes different nitrogen atoms respond in different fashions based upon the number of protons attached. N(10) does not give a signal because it has no attached protons. The chemical shift value of N(3) is indicative of a hydrogen bond, whereas for the N(10) atom that cannot participate in hydrogen bonding, the downfield shift on going to a polar solution has been explained to be due to an increase in sp2 hybridization that occurs when the polarization of the flavin ring is stabilized by hydrogen bonding at C(2)O and C(4)O. For BM-3, the upfield shift for
N(10) suggests that the N(10) atom is not fully sp2 hybridized but that the pyrazine ring
has a slightly bent conformation at N(10). However, as the hybridization state at N(10)
depends on the polarization of the ring, this result is also probably indicative of a
142
weakened hydrogen bonding interaction at C(2)O and/or C(4)O. Both C(4)O and C(2)O are within hydrogen bonding distance with the side chain of Thr577 and backbone amide protons respectively. The latter might contribute to the upfield shift of N(10), being weaker in strength than a hydrogen bond to solvent.
143
A 160.5
B 161.2 159.7
Figure 36. One-dimensional 15N-NMR data obtained using the DEPT pulse sequence with (panel A) and without (panel B) broad-band decoupling in the oxidized state.
144
15N Chemical Shifts (ppm) Atom FMNa TARFa BM-3 N(1) 190.8 199.9 189.0 N(3) 160.5 159.8 160.5 N(5) 334.7 344.3 321.5 N(10) 164.6 150.2 162.6 a From Vervoort et al (Vervoort et al., 1986).
Table 13. 15N Chemical Shifts for Free and Bound FMN in the Oxidized State at pH 7.0, 300°K
145
The chemical shifts in the fully reduced state were particularly revealing. The chemical shifts of both N(1) and N(10) clearly prove that the HQ is ionized which was also seen from the pH dependency data (Table 14). What is of even greater significance
- is the chemical shift value of N(5) which is far upfield of that of FMNH2 and FMNH .
This suggests that the N(5) atom is more sp3-hybridized and that now in the fully reduced state the flavin ring is bent out of plane at N(5). This effect at N(5) was unexpected and therefore, the change in hybridization state was verified by measuring the coupling constants obtained by running the one-dimensional 15N-NMR spectroscopy data under proton non-decoupling conditions applying the DEPT pulse sequence (Figure 37, Table
15). The semi empirical relationships between the experimentally observed 15N-1H coupling constants and the hybridization state of the nucleus under investigation yielded values of about 72 Hz for sp3 hybridized nitrogen atoms and values of about 93 Hz for sp2 hybridized nitrogen atoms (Binsch et al., 1964; Bourn & Randall, 1964). For BM-3 in the oxidized state, N(3) exhibits a coupling constant of 90.2 Hz indicative of its sp2 hybridization state and the planarity of the ring at this nitrogen atom. This hybridization state is maintained in the reduced state. However, N(5) in the reduced state exhibits a coupling constant of 70.3 Hz which provides clear evidence of its almost complete sp3 character. This is indeed remarkable because this suggests that in the reduced state, only the N(5) atom is lifted out of the plane of the ring. This result is most likely correlated to the nearly complete absence of an absorption band in the visible region of the reduced protein as it has been suggested that the absorption coefficient at 450 nm of a reduced
146
flavin molecule is probably entirely related to the hybridization state of the N(5) atom
(Moonen et al., 1984).
147
N(3) N(5)
Figure 37. One-dimensional 15N-NMR data obtained using the DEPT pulse sequence without broad-band decoupling in the fully reduced state.
148
15N Chemical Shifts (ppm)
a a - a Atom FMNH2 FMNH TARFH2 BM-3
N(1) 128.0 182.6 116.7 186.4
N(3) 149.7 149.3 145.8 148.4
N(5) 58.0 57.7 60.4 47.1
N(10) 87.3 97.2 72.2 98.1
a From Vervoort et al (Vervoort et al., 1986)
Table 14. 15N Chemical Shifts for Free and Bound FMN in the Reduced State at pH 7.0, 300°K
149
Coupling constants (Hz)
N(3)H N(5)H
BM-3 ox 90.2 _
BM-3 hq 90.6 70.3
The coupling constants were determined with an accuracy of + 0.8 Hz.
Table 15. Coupling constants for N(3)H and N(5)H in the oxidized and fully reduced states of BM-3.
150
From one-dimensional 15N-NMR experiments alone we cannot obtain quantitative estimates of the hydrogen bonding strengths of the nitrogen atoms that are involved in a hydrogen bond, specifically, N(3), and N(5) in the reduced state. The temperature coefficient of proton chemical shifts from NMR data has been used as an indicator of hydrogen-bonding strength. For protons involved in hydrogen bonding, an upfield shift in resonance with increasing temperature has been rationalized as due to the weakening of hydrogen bonds, leading to an alteration in the distribution between the hydrogen-bonded and nonhydrogen-bonded species (Chang & Swenson, 1999). Thus, the two-dimensional
1H-15N HSQC spectra of oxidized and fully reduced BM-3 were obtained at pH 7.0 at different temperatures. The presence of the HSQC signal for the protons at all temperatures tested implies that these protons are exchanging slowly with solvent due to their relative inaccessibility and strong hydrogen bonding interactions with the protein.
The low values obtained for the temperature coefficients of –2.96, -1.65 and –
0.647ppb/°K for N(3)H oxidized, N(3)H reduced and N(5)H reduced, respectively as seen in Figure 38 are consistent with strong hydrogen bonding.
151
15N-1H HSQC
12
10
8
6
H chemical shift shift chemical H 4 1
2
0 260 270 280 290 300 310 320
o Real Temp ( K)
Figure 38. Temperature dependencies of the proton chemical shift for the N(3)H oxidized (closed circles), N(3)H reduced (open circles) and N(5)H reduced (closed squares) of the bound 15N-labeled FMN for BM-3.
152
DISCUSSION
The FMN-containing reductase domain of P450BM-3 is a structurally independent unit
that is highly homologous to the bacterial flavodoxins. Strong experimental evidence in
support of this was provided when the domain was expressed separately in E.coli and
also in combination with the FAD/NADPH containing reductase domain as well as the
heme domain. This characteristic of the FMN-binding domain enables us to study its
properties while eliminating complications caused by the presence of the other domains.
Furthermore, the crystal structure of the putative electron-transfer complex between the
heme domain of P450BM-3 and the FMN-binding domain was solved recently and that
allows for a more detailed analysis.
A unique characteristic of BM-3 redox chemistry is its inability to thermodynamically stabilize the one electron reduced anionic semiquinone form of the flavin. Although the flavin semiquinone is formed transiently, it is converted to the hydroquinone during the titration (10min/point). Thus, reduction by the second electron is thermodynamically more favorable than the first one electron reduction, which corresponds to a positive difference between the second and first one-electron midpoint potentials, Esq/hq and Eox/sq of 65mV. Because of the inherent problems in obtaining these individual midpoint potential values, the difference between the two couples is not an accurate estimate. Differences of 20mV, 29 mV and 39mV have been observed (Daff et al., 1997). Regardless, it is the fully reduced state that is thermodynamically favored over the semiquinone state which is rare among flavoproteins that act to stabilize the semiquinone. The changes in the midpoint potential values with pH for the individual
153
couples clearly indicate that protonation is associated with the transfer of the second
electron, hence the semiquinone that is transiently formed is the anionic species. This was
also indicated by a transient increase in absorbance at 380nm. The pH dependency for the
two-electron reduction (Eox/hq) provide conclusive evidence for the formation of the
anionic form of the hydroquinone. This is further corroborated by the chemical shift of
the N(1) atom in the fully reduced state which is similar to the value of anionic flavin in a
polar solvent (Table 14).
The hydrogen-bonding interaction to N(5) is crucial to flavoproteins in general, being a dominant factor in the stabilization of the blue neutral semiquinone as well as the hydroquinone state. For BM-3, the hydrogen bond donated to N(5) of the flavin may play a role in the lack of semiquinone stabilization. Donation of a hydrogen bond by the protein may disfavor reduction by increasing the cost of protonating N(5). While this may explain the tendency to form the anionic semiquinone in this protein, why is the hydroquinone stabilized? What happens to the hydrogen bond at N(5) upon reduction? To address these questions, the NMR experiments were done. The most striking feature of the 15N NMR spectra of the oxidized BM-3 is the extreme high-field shift of the N(5)
resonance which is indicative of a strong hydrogen bond between the N(5) atom and the
protein. In the reduced state, the N(5) atom resonates at higher field than N(5) of free
flavin – 47.1ppm versus 57.7ppm respectively. It follows that the N(5) atom is more sp3 hybridized. This hybridization state is derived also from the 15N(5)-1H coupling constant and indicates that reduced flavin does not need to acquire a structure bent along the N(5)-
N(10) axis of the molecule, as usually believed, but rather that the two nitrogen atoms
154
can take configurations independent of each other (Beinert et al., 1985; Moonen et al.,
1984). It is likely that the geometry of the binding site plays a role in inducing this hybridization state of N(5) in the reduced state. It is believed that the isoalloxazine ring becomes significantly nonplanar and the π system weakens in a highly bent flavin, and therefore the presence of an aromatic stacking interaction would tend to inhibit bending of the flavin ring along the N(5)-N(10) axis (Haynes et al., 2002). In BM-3, the coplanar
Trp574 that is stacked against the isoalloxazine ring could play this very role, resulting in only the N(5) atom being distorted out of the plane of the molecule.
155
A H-N N(10) C(9a) C(5a) N(5)
B
H-N
C
H-N
Figure 39. Diagrammatic depiction of the N(10)-C(9a)-C(5a)-N(5) edge of the flavin ring detailing the events occurring upon reduction. The hydrogen bond to a backbone amide is shown in the oxidized state with all the flavin atoms in one plane (A), the clash with the protonated N(5) if no conformational changes were to take place in the fully reduced state (B) and the puckering at N(5) that relieves the clash in the reduced state (C). In this state, only the N(5) atom is lifted out of the plane.The NMR data does not yield information about the configuration of the H atom at N(5) (axial or equatorial) or the conformation of the pyrazine ring (bent up or down).
156
BM-3 is not unique in this behavior. OYE has also been noted to possess a similar fully sp3 hybridized N(5) atom in the fully reduced state (Beinert et al., 1985). What
could be the reason for this? As previously mentioned, a very strong hydrogen bond to
the backbone amide proton of Asn537 is present to N(5) in the OX state. Asn537 is
located on a loop region made up of residues 536Y-N-G-H539and contacts the N(5) edge of the flavin ring. This loop is quite characteristic and unlike those seen in other flavoproteins. The hydroxyl group of Tyr536 provides a hydrogen bond to the phosphate moiety of the FMN cofactor. The presence of two hydrogen bonds within the turn region help to rigidify the loop in addition to the two proline residues that are present in tandem at the end of the turn. Formation of the red anionic SQ obviates the need for the hydrogen bond at N(5), however, it can not be avoided in the HQ state as the pKa for the dianionic
HQ is very high and is therefore unstable in solution. Once protonated, a conformational change becomes necessary in order to avoid a steric clash between the proton on N(5) and the amide proton of Asn537. Rather than undergo a conformational change within the protein, as is seen in some flavodoxins, BM-3 undergoes a conformational change within the flavin ring. This effectively eliminates the steric clash between the two protons without the need for structural changes in the FMN binding loop that would be associated with a greater energetic cost due to the rigidity of the loop (Figure 39). Furthermore, sp3 hybridization of N(5) prevents electron delocalization into the isoalloxazine ring which increases the electron density at N(5) and results in the large upfield shift for that atom.
The enhanced basicity of the nitrogen atom also increases its proton affinity and therefore, a lack of a hydrogen bond partner is tolerated. The enhanced rigidity of this
157
loop region due to the presence of two proline residues in tandem is probably an
important point of consideration. Protein conformational changes in this region would be
severely restricted and therefore, flavin conformational changes are preferred.
This protein is clearly an enigma with the interaction at N(5) puzzling. Donating a hydrogen bond rather than accepting one at this position is significant and likely to play a key role in understanding the redox properties of this protein. Why is the hydroquinone form favored over the neutral semiquinone species, both of which are protonated at N(5)?
Furthermore, formation of the FMN hydroquinone in the holoprotein leads to a dead enzyme that is unreactive in electron transfer to the heme (Sevrioukova et al., 1996a). It
is hard to imagine why this species would be thermodynamically favored. Why is the
anionic semiquinone that is not protonated at N(5) not stabilized? Old yellow enzyme
makes a similar hydrogen bond to N(5) with the backbone amide proton of Thr37 and yet
stabilizes the anionic semiquinone state at the level of 15 – 20% under equilibrium
conditions (Stewart & Massey, 1985). What role does the sp3 hybridized nature of the
N(5) atom play in the reduced states? A more detailed evaluation of this interaction and identification of possible mechanisms of redox tuning in this protein are needed. Some insight into these questions may be gained from the studies detailed in the next chapter.
158
CHAPTER 6
THE FMN-BINDING DOMAIN OF P450BM-3: INVESTIGATION INTO THE
POSSIBLE MECHANISMS OF REDOX TUNING
INTRODUCTION
The cloning and characterization of the FMN-binding domain of P450BM-3 laid the groundwork for additional experimental studies. It is evident from data on several flavoproteins that conformational changes, either in the protein or in the flavin, play a functional role in maintaining the redox properties of the cofactor (Drennan et al., 1999;
Hoover et al., 1999; Kasim & Swenson, 2000; O'Farrell et al., 1998). Surprisingly, these conformational changes are localized at or near the N(5) region of the flavin. This locus is unique in BM-3 when compared to the flavodoxins in that a main chain amide
(Asn537) serves as a strong hydrogen bond donor to N(5) of the flavin in the oxidized state. This observation is supported by both the crystal structure of this protein in complex with the heme domain (Sevrioukova et al., 1999), as well as from one- and two- dimensional NMR experiments presented here. The adoption of a sp3 hybridized conformation of the N(5) atom upon reduction is a unique property that was first identified in old yellow enzyme (OYE) (Beinert et al., 1985). 159
Old yellow enzyme was the first flavoprotein identified (Warburg & Christian,
1933) and is characterized by its ability to bind phenolic ligands with a perturbation of the flavin spectra and the formation of a striking long wavelength (500 – 800nm) absorbance band (Abramovitz & Massey, 1976). There are several points of similarity between old yellow enzyme and BM-3. 1) In the reduced states, the N(5) atom is puckered out of the plane for both proteins and adopts a sp3 hybridized conformation. 2)
Strong hydrogen bonding interactions with the N(5) in the oxidized state are present. This occurs with the amide of Thr37 and Asn537 in OYE and BM-3, respectively. This may influence the puckering of the N(5) atom in the reduced state. 3) Both proteins have a tendency to form the anionic semiquinone, with the maximal amount of semiquinone formed being 15 – 20% under equilibrium conditions for old yellow enzyme, and none for BM-3, 4). In each case, Eox/sq is more negative than Esq/hq. Finally, 5) the hydroquinone spectra of these two proteins also resemble one another. Therefore, considerable information can be gained by comparison to old yellow enzyme.
One of the interesting differences between the two proteins is the inability of BM-
3 to stabilize the anionic semiquinone to even a lesser degree than old yellow enzyme.
The difference between Eox/sq and Esq/hq is similar between the two proteins:- old yellow enzyme; Eox/sq = -245mV, Esq/hq = -215mV and BM-3; Eox/sq = -206 mV and
Esq/hq = -177mV (Daff et al., 1997; Stewart & Massey, 1985). The factor(s) responsible for the lack of thermodynamic stabilization of the semiquinone species in BM-3 need to be identified as that holds the key to understanding the functional properties of this protein. There is no argument that this species does form, evidence for this has been
160
obtained by equilibrium and transient state spectrophotometric studies (Sevrioukova et
al., 1996a) and by electron paramagnetic resonance (EPR) studies (Murataliev et al.,
1997). Furthermore, it was found that the thermodynamically stable hydroquinone species when formed resulted in the formation of an inactive enzyme that was incapable of reducing the heme (Sevrioukova et al., 1996a).
A plausible approach might be to alter the two single-electron potentials differentially, aiming to separate the two couples such that Eox/sq is now less negative than Esq/hq. This should result in the thermodynamic stabilization of the semiquinone species. It is known that the hydrogen bonding state of the N(5) atom influences the redox potential of the flavin, as well as the chemical reactivity of the C(4a) site (Ghisla &
Massey, 1989). In the oxidized state, the presence of a hydrogen bond to N(5) is expected to stabilize that state, thus decreasing the intrinsic redox potential of the flavin and making its reduction less favorable. Elimination of this hydrogen bond should destabilize the oxidized state, affecting primarily Eox/sq. Similarly, instead of eliminating the interaction entirely, the hydrogen bond strength could be affected by altering the turn stability preferences as was seen in the flavodoxin from C. beijerinckii (Kasim &
Swenson, 2000). Clearly, this interaction is critical and therefore, its perturbation is likely to generate results that are greatly different from wild-type.
The inner flavin-binding loop of BM-3 adopts a type I′ turn conformation and is composed of residues 536Tyr-Asn-Gly-His, the conformation of which is suited to provide
a strong hydrogen bond with the backbone amide of Asn537. Turn stability preferences
indicate that the positioning of the glycine residue is critical in order to maintain this
161
conformation that requires the residue to adopt positive φ angles (Yan et al., 1995; Yang et al., 1996). To test this requirement for glycine, Gly538 was mutated to an alanine. This was done in combination with mutations at Asn537 to Gly/Ala in order to directly compare the turn stability with previously calculated values for dipeptides having similar combinations of residues. While directly addressing the role of turn stability in this flavoprotein, these mutations would also potentially alter the turn conformation and in turn alter the hydrogen bond strength to N(5). As the hydrogen bond donor to N(5) is a backbone amide, it is challenging to disrupt this interaction utilizing routine mutagenesis techniques while still maintaining the integrity of the protein. Introduction of a proline residue at that position would effectively eliminate the N(5) hydrogen bond. However, the structural constraints introduced by a proline residue, that has limited conformational freedom, would have to be taken into account. Acknowledging the limitations and difficulties with this approach, it was still tried as the data from this mutant would yield preliminary information on the importance of the N(5) interaction. Furthermore, the resulting –537Pro-Gly- sequence is incapable of adopting the type I′ turn conformation which would add to the conclusions from the other Gly/Ala mutations.
A more direct approach of introducing a positive charge near the N(5) atom was also followed with an aim to stabilize the reduced state(s) via favorable electrostatic interactions. Site-directed mutagenesis of Tyr98 in the D.vulgaris flavodoxin to basic residues, histidine and arginine, resulted in substantial stabilization of the anionic hydroquinone state (Swenson & Krey, 1994). Little effect on the neutral semiquinone species was seen that is stabilized by this flavodoxin. Therefore, while Eox/sq was not
162
substantially altered, an 180mV increase in the midpoint potential for Esq/hq was
observed, corresponding to an apparent stabilization of 4.0 kcal/mol over wild-type.
Thus, the cationic nature of the side chains stabilized the flavin hydroquinone anion
through favorable electrostatic interactions. It could therefore be argued that similar
electrostatic stabilizing interactions could affect the stability of the flavin semiquinone
anion in BM-3 and possibly overcome the factors responsible for destabilizing this redox
state.
Thus, this study was designed to address the role of the N(5) and electrostatic interactions in this flavoprotein. Several mutant proteins were generated, each differentially perturbing the N(5) hydrogen bond. The results demonstrate the clear importance of this N(5) hydrogen bond and also point toward the inner FMN-binding loop in playing a structural role in the stability of this protein. The data also indicate a lack of electrostatic control in the regulation of the redox-potentials in this flavoprotein.
MATERIALS AND METHODS
Preparation of apoprotein. Apoprotein was prepared by dialysis against a 2M potassium bromide solution in 0.1M sodium phosphate, pH 7.0 in the presence of 0.3mM EDTA
(Mayhew, 1971). After all the flavin was released, the protein was dialyzed extensively against 50mM sodium phosphate buffer containing 0.3mM EDTA.
All other information regarding materials and methods can be found in Chapter 2.
163
RESULTS
Construction of recombinant vectors encoding the mutant FMN-binding domains of
P450BM-3. The oligonucleotides for performing PCR amplification of the wild-type gene have already been described in the preceding chapter. For each mutant, an oligonucleotide was synthesized that introduced the appropriate mutation in the FMN- binding loop adjacent to the N(5) edge of the flavin. PCR amplification of the mutant gene was done in a two-step process as detailed in Chapter 2. Protein purification was as for wild-type, however it was noticed that some of these mutants bound the cofactor considerably weaker and this resulted in the loss of FMN during the purification process.
164
0.3
0.2 Absorbance 0.1
0.0 300 400 500 600
Wavelength (nm)
Figure 40. Representative UV-visible absorbance spectra of the N537A mutant in the oxidized and fully reduced states. Data were collected during a reductive titration with sodium dithionite under anaerobic conditions in 50mM sodium phosphate buffer, pH 7.0 at 25°C. The data shown is that for the N537A mutant which is qualitatively similar to that of N537G.
165
-160
-180 h E
-200
-220 -1 0 1
log[OX/HQ]
Figure 41. Representative determinations of Eox/hq for the N537A (closed circles) and the N537G (open circles) mutants. The data are plotted as the linear version of the Nernst equation using the system potential (Eh) as determined by the concentration of the oxidized and reduced forms of the redox indicator dye in equilibrium with the proteins during the titration.
166
160
120
80
40 Fluorescence Intensity
0 0 20406080100120 µ Apoprotein ( L)
Figure 42. Representative binding titrations for the determination of the dissociation constant of the FMN cofactor in the oxidized state for the N537A (closed circles) and the N537G (open circles) mutants. Data were corrected for dilution. The solid lines represent the best nonlinear regression fit of the data to a binding isotherm for a 1:1 complex, from which the dissociation constant was derived.
167
Mutations designed to alter turn stability. Four different mutants were made to introduce the various combinations of glycine and alanine at the two central positions of the turn.
These fall into two classes, on the basis of turn stability, one having glycine at the third position of the turn and the other having alanine at the third position of the turn. The residue at the second position is not involved in dictating turn stability.
N537A (-537Ala-Gly-) and N537G (-537Gly-Gly-): Both these mutant holoproteins
exhibited an A278/A468 ratio of 5.1 + 0.1, a value similar to wild-type and characteristic of a 1:1 cofactor-protein complex. The spectral characteristics of the two redox states (OX and HQ) were recorded during a reductive titration with sodium dithionite under anaerobic conditions. The visible absorbance spectra for both proteins in the oxidized state were similar to wild-type, exhibiting the characteristic 20nm red shift in the first transition and the presence of a shoulder between 550nm and 700nm. As for wild-type, reduction followed a single two-electron process with no evidence for the accumulation of the semiquinone species. This conclusion is strongly supported by the presence of a single isosbestic point at 348nm throughout the titration (Figure 40). Midpoint potential values of –196mV and –202mV were obtained for N537A and N537G respectively
(Figure 41). The similarity of the value for the N537A mutant with wild-type is not fortuitous. It has been proven that the influence of turns on midpoint potentials is dictated by the presence or absence of side chains at the two central positions and not by the physico-chemical nature of the particular side chain (Kasim & Swenson, 2000). The dissociation constants for the FMN complex were measured by monitoring the quenching in the flavin fluorescence upon binding (Figure 42). Both mutants displayed similar Kd
168
values of 30 + 5nM that is an average of at least 4 independent titrations. This represents a 5-fold increase compared to wild-type.
-537Gly-Ala- and –537Ala-Ala-: Both these mutant proteins were purified as the apo form
indicative of their altered ability to bind the flavin cofactor. Therefore, first the
dissociation constants were measured by monitoring the spectral changes in the flavin
absorbance upon binding (Figure 43). Although both mutant proteins were capable of
binding FMN, as evidenced by the distinct difference spectra, this ability was
significantly impaired. Only 20 –30% of the apoprotein was active in binding and
dissociation constants of 430nM and 970nM were obtained for the -537Gly-Ala- and –
537Ala-Ala- mutants; respectively. This represents a 72-fold and 162-fold increase relative
to wild-type.
These mutants emphasize the requirement for a glycine residue at the third
position in order to maintain the type I′ turn conformation that seems to be critical in
binding the flavin cofactor.
169
Figure 43. Determination of the dissociation constant for the -537Gly-Ala- (A) and – 537Ala-Ala- (B) mutant. In each case, an FMN solution (~5 µM) in 50 mM sodium phosphate buffer pH 7.0 at 25°C was titrated with substoichiometric amounts of freshly prepared apoprotein. After correction for dilution, the changes in extinction coefficients at 442nm and 494nm (insets) associated with complex formation were plotted as a function of the added apoprotein. The solid line shows the best fit to the data to a single- site binding isotherm.
170
0.05 A
0.04
0.08
494-442nm 0.03
0.04 0.02
Absorbance 0.00 Absorbance ∆ ∆ ∆ ∆ 0.01
-0.04 350 400 450 500 550 600 Wavelength 0.00 0 50 100 150 200
B 0.05
0.04 498-444nm 0.03 0.08 ance b
0.02 0.04 sor Absorbance Ab ∆ ∆ ∆ ∆ 0.01 0.00
400 500 600 Wavelength 0.00 0 50 100 150 200 Apoprotein (µL)
Figure 43.
171
15N Chemical Shifts (ppm)
Atom FMNa TARFa BM-3 -537Ala-Ala-
N(1) 190.8 199.9 189.0 188.8
N(3) 160.5 159.8 160.5 158.6
N(5) 334.7 344.3 321.5 332.8
N(10) 164.6 150.2 162.6 162.3
a From Vervoort et al (Vervoort et al., 1986).
Table 16. 15N Chemical Shifts for Free and Bound FMN in the Oxidized State at pH 7.0, 300°K for the -537Ala-Ala- mutant.
172
As the flavin-binding site appeared to be disrupted in the –537Gly-Ala- and –
537Ala-Ala-, a one-dimensional NMR spectroscopy experiment was performed initially
for the –537Ala-Ala- mutant. This would help directly identify the interactions that are perturbed in this mutant. The chemical shift values obtained for the four nitrogen atoms were decidedly different (Table 16), particularly for N(5). The chemical shift for N(5) in the –537Ala-Ala- mutant is significantly down field from that observed in the wild type
BM-3 and closer to that of free flavin in solution. This could indicate that the strong N(5)
hydrogen bond observed in wild-type is severely impaired in this mutant.The chemical
shift value for the N(3) and N(10) atoms are much less sensitive to changes in hydrogen
bonding, so their similarities to the wild type and to free flavin is not very surprising.
However, the upfield shift for N(3) relative to wild type (~2 ppm) could indicate a
weakened hydrogen bond at this position. In the crystal structure, the backbone carbonyl
group of Thr577 provides the hydrogen bond donor to N(3) (Sevrioukova et al., 1999).
Alteration in the orientation of the FMN in the binding site would likely disrupt this
interaction. Trp574, which is situated on the same loop as Thr577, is responsible for the
strong charge-transfer band seen for this domain. This interaction has also been affected
as evidenced by the decrease in intensity of this band resulting in the loss of color in this
mutant. These results seem to suggest that the disruption of the loop through the
introduction of the methyl side chain at the third position in this Type I′ turn has
significantly disrupted the strong hydrogen bond at N(5) which in turn affect the other
interactions made with the FMN. The similarity in the chemical shift value for the N(1)
atom with wild type suggests that the interactions with this edge of the flavin have not
173
been altered. It should be noted that beacuase of the weaker binding of FMN to the –
537Ala-Ala- mutant and the similarity of the chemical shift values to unbound flavin, one
might be concerned that the 15N-FMN did not remain bound throughout the entire
experiment. However, the Kd for the FMN complex of this mutant is approximately
1µM. Because the concentration of the holoprotein was significantly higher than this in
the NMR experiment (~ 1 mM), it seems reasonable to conclude that the 15N-FMN should remain bound throughout unless the protein denatures during the extended experiment (overnight at room temperature). Several attempts were made to establish that the 15N-FMN remained bound at the end of the experiment. The visible absorbance and fluorescence spectrum of the NMR sample was measured in addition to attempts of running the sample through a gel filtration column in order to separate free flavin from bound. The data did seem consistent with a bound FMN, although, unfortunately, this conclusion was not completely unequivocal.
Mutation eliminating the N(5) hydrogen bond – N537P: The hydrogen bond donated to
N(5) in BM-3 is through the backbone amide of Asn537. Elimination of this interaction was effectively achieved by replacing the asparagine with a proline residue, which lacks the main chain amide proton. This mutant was also purified as the apo form. The ability of the apoprotein to bind flavin was severely impaired. Interestingly, the –537Pro-Gly- sequence is incapable of adopting a type I′ turn conformation due to the restricted φ angles for proline. This suggests that the N(5) hydrogen bonding interaction that is provided by the type I′ turn conformation is essential in maintaining the integrity of the flavin-binding site.
174
Mutations introducing a positive charge near the flavin. In order to introduce the positive
charge near the flavin, the substitution would have to be either on the inner FMN-binding
loop (-537YNGH-) or at the si face tryptophan residue. Previous studies indicated that mutation of the W574 to non-polar residues abolished FMN-binding, therefore this site was not considered to introduce the substitution (Klein & Fulco, 1993). The mutants designed to alter turn stabilities indicated that while mutations at the glycine residue were not tolerated, mutations at Asn537 had no major effect on the redox potentials.
Furthermore, this residue provides the hydrogen bond donor to N(5) through its amide proton and it was preferred to leave this interaction intact. That left the surface exposed side chain of Tyr536 that makes re face contacts with the flavin ring. It was shown that substitution of Tyr to a polar residue did not entirely abolish FMN binding and therefore this residue was chosen as the site at which to introduce the positive charge . It is worth mentioning that the hydroxyl group of the tyrosine also participates in a hydrogen bond with the phosphate moiety of the FMN cofactor.
175
0.2
0.1 Absorbance
0.0 300 400 500 600 700
Wavelength (nm)
Figure 44. UV-visible absorbance spectra of the Y536H mutant in the oxidized and fully reduced states. Data were collected during a reductive titration with sodium dithionite under anaerobic conditions in 50mM sodium phosphate buffer, pH 7.0 at 25°C.
176
Figure 45. Representative determination of Eox/hq for the Y536H mutant in 50mM sodium phosphate, pH 7.0 at 25°C. (A) Spectral changes occurring upon reduction. (B) Plot of the data as the linear version of the Nernst equation using the system potential (Eh) determined by the concentration of the oxidized and reduced forms of the redox indicator dye, anthraquinone-2,6-disulfonate, that is in equilibrium with the protein.
177
0.3 A
0.2
Absorbance 0.1
0.0 300 400 500 600 700 Wavelength (nm) B -160
-180 h E
-200
-220 -1.0 -0.5 0.0 0.5 1.0
log[OX/HQ]
Figure 45.
178
Y536H: The holoprotein exhibited spectral characteristics similar to wild-type, displaying
the characteristic charge-transfer band. Reduction by sodium dithionite under anaerobic
conditions followed a single two-electron process with no evidence for the accumulation
of the semiquinone species (Figure 44). However, it was observed that the protein was
relatively unstable, having a tendency to release flavin over time as monitored by the
difference in spectra after reduction and reoxidation, as well as by fluorescence. The
redox potential for Eox/hq was measured using anthraquinone-2,6-disulfonate and a
value of –188mV was obtained, which is close to that of wild-type (Figure 45). The
dissociation constant measured by monitoring the spectral changes of FMN upon binding
was 350nM, which is an average of 5 independent experiments (Figure 46). The difference spectra exhibited troughs and peaks similar to wild-type (see inset Figure 46).
179
0.06
0.04 0.04 502-440nm 0.03
0.02
0.01
0.02 0.00 Absorbance δ δ Absorbance δ δ -0.01
-0.02
-0.03 400 500 600 Wavelength 0.00 0 20406080
Apoprotein (µL)
Figure 46. Determination of the dissociation constant for the complex between oxidized FMN and the mutant Y536H. An FMN solution (~5 µM) in 50 mM sodium phosphate buffer pH 7.0 at 25°C was titrated with substoichiometric amounts of freshly prepared apoprotein. After correction for dilution, the changes in extinction coefficients at 440nm and 502nm (inset) associated with complex formation were plotted as a function of the added apoprotein. The solid line shows the best fit to the data to a single-site binding isotherm.
180
0.3
0.2 Absorbance 0.1
0.0 300 400 500 600 700
Wavelength
Figure 47. UV-visible absorbance spectra of the Y536H mutant in the oxidized and fully reduced states. Data were collected during a reductive titration with sodium dithionite under anaerobic conditions in 50mM sodium phosphate buffer, pH 6.1 at 25°C.
181
The pKa of a histidine residue is in the range of 6.0 – 6.8 (Nozaki & Tanford,
1967; Tanokura, 1983). Therefore, it is likely not to be completely ionized at pH 7.0. The anaerobic reductive dithionite experiments were performed at lower pH in order to observe what effect having a fully charged residue at that position would have. However, the protein was relatively unstable at lower pH, resulting in aggregate formation with the release of flavin. This was monitored by absorbance as well as fluorescence spectroscopy. As glycerol is known to have a stabilizing effect on proteins, the experiment was repeated at different concentrations of glycerol, ranging from 5 – 15% and it was observed that at 15% glycerol, the protein was relatively stable. Under these conditions, a reductive dithionite experiment was performed and that also followed a single two-electron process (Figure 47). No evidence for the formation of the
semiquinone species was seen indicative of perhaps electrostatics not playing a role in the
modulation of the midpoint potentials in this protein.
Y536R: Unlike the Y536H mutant, this mutant protein lost FMN by the end of the
purification process. Reconstitution of the protein with flavin yielded an estimate of the
dissociation constant in the micromolar range (Figure 48). Preliminary reductive
dithionite titrations indicated that this mutant, like the Y536H mutant, does not stabilize
the anionic semiquinone species, adding to the conclusion that electrostatics do not seem
to be a major determinant of the reduction potential.
182
0.035
0.030
0.025
0.04 0.020 500-440nm
0.02 0.015
0.00 0.010 Absorbance Absorbance δ δ δ δ 0.005 -0.02 350 400 450 500 550 600 Wavelength 0.000 0 50 100 150 200 µ Apoprotein ( L)
Figure 48. Determination of the dissociation constant for the Y536R mutant. An FMN solution (~5 µM) in 50 mM sodium phosphate buffer pH 7.0 at 25°C was titrated with substoichiometric amounts of freshly prepared apoprotein. After correction for dilution, the changes in extinction coefficients at 440nm and 500nm (inset) associated with complex formation were plotted as a function of the added apoprotein. The solid line shows the best fit to the data to a single-site binding isotherm.
183
DISCUSSION
The mechanism of action of NADPH-P450 reductase (CPR) is of considerable interest as
it serves as an electron carrier to several P450-dependent microsomal monooxygenation
reactions via its FAD and FMN prosthetic groups. Unlike CPR, the flavoprotein
P450BM-3, which is functionally analogous to the microsomal NADPH-dependent P450
oxidoreductase, is soluble and serves as a good model for structure-function relationships
in this class of proteins. One of the key differences between these two proteins is the
electron-donating species to the P450, which is the hydroquinone for CPR and the
anionic semiquinone for P450BM-3. Interestingly, P450BM-3 does not
thermodynamically stabilize the anionic semiquinone species despite it being the
kinetically relevant intermediate. This study was devoted to identifying the factors
responsible for the destabilization of the semiquinone.
The crystal structure of the FMN-binding domain of P450BM-3 in complex with the heme was recently solved at 2.03 Å resolution (Sevrioukova et al., 1999). Although
electrophoresis analysis of the dissolved crystals indicated that the linker between the
heme and the FMN-binding domains was proteolyzed, the precise positioning of the
methyl groups of the FMN toward the heme indicated that this complex was specific and
facilitated by long-range electrostatic interactions. The flavin binding site is surrounded
mainly by neutral and hydrophobic residues, contrasting with that of both flavodoxins as
well as CPR. A further look at this unique binding site of P450BM-3 revealed many
distinguishing characteristics, some of which were the target of this study.
184
Data for the Gly/Ala substitutions at the central position of the turn correlate well with the predicted turn stability. The inner flavin binding loop is comprised of residues -
537YNGH- and forms a sharp turn near the N(5) edge of the flavin. The torsion angles
(φ,ψ) adopted by the central residues of the turn resemble a type I′ turn conformation –
angles for N537 and G538 being (48.4,37.8) and (91.6, 1.6) respectively, differing
slightly from the idealized values of (60,30) and (90,0). These values fall in the left-
handed helical region of the Ramachandran plot. In the left-handed helical region, ~60%
of the residues are populated by glycine, and the others are non-glycine residues (Takano
et al., 2001). It has been postulated that the high proportion of Gly residues is because the
non-Gly residues have unfavorable energies from the local steric interaction of backbone
atoms with the side-chain Cβ atom. This means that substitution of a Gly with a non-Gly
residue would destabilize the protein as is also evident from the turn stabilities calculated
for this particular conformation (Yan et al., 1995; Yang et al., 1996). Therefore, the type
I′ turn conformation is most stable when it has a glycine at the third position as it can
adopt the required positive φ conformation. This is supported by the data in this study
where the mutants N537A (-537Ala-Gly-) and N537G (-537Gly-Gly-), both having a
glycine at the third position, resembled the wild-type protein in their ability to bind FMN
with high affinity. On the other hand, both -537Gly-Ala- and –537Ala-Ala- that have a non-
Gly (Ala) residue at that position, resulted in proteins that were severely impaired in
FMN-binding. These data support the hypothesis that turn conformation and stability play an important role in maintaining the properties of this flavoprotein, as was evidenced for the flavodoxins.
185
Importance of the hydrogen bond to N(5). Unique to this flavoprotein, is the
strong hydrogen bonding interaction to the N(5) atom of the bound flavin cofactor in the
oxidized state. The hydrogen bond donor is the amide proton of Asn537 that resides on
the inner FMN-binding loop. When the hydrogen bond donor group is a side chain of a
residue, it is possible to either alter its strength or remove the interaction entirely by
appropriate substitutions (Bradley & Swenson, 1999; Bradley & Swenson, 2001). It is
almost impossible to do that in the case of a backbone interaction while still utilizing the
natural amino acids. The amide linkage could be substituted for an ester linkage by
protein splicing, although such techniques are outside the scope of this project. Instead a
proline was substituted for Asn537 which effectively eliminated the hydrogen bond to
N(5) and yet utilized a natural amino acid. Arguably, the conformational restrictions
imposed by a proline residue would need to be taken into account; molecular modeling
suggested that this residue could be accommodated at this position without steric
interference with the flavin ring. Furthermore, it is located on a surface exposed loop that
should allow for compensatory structural changes. However, the N537P mutant while
being soluble like wild-type, was severely impaired in its ability to bind flavin. If you
take into account the turn stabilities mentioned earlier, it is perhaps not surprising.
Proline, due to its restricted φ angle of –60 + 25°, cannot adopt a type I′ turn conformation. While it is still unclear as to the exact effect of this hydrogen bond in this flavoprotein, there remains no doubt as to the absolute requisite for the type I′ turn conformation in maintaining the stability and FMN binding characteristics.
186
Effect of electrostatics. The environment surrounding the flavin is predicted to have a role in stabilizing specifically the reduced forms of the flavin. Reduction increases the electron density in the region of the ‘enediamine’ subfunction (i.e. positions N(1),
C(4a) and N(5)) of the isoalloxazine ring, and thus the nature of the protein environment is key to modulating the reduction potential. Understandably then, introduction of a positive charge near this locus would help stabilize the anionic semiquinone as well as the anionic hydroquinone. With this aim in mind, the Y536H and Y536R mutants were generated. While the histidine mutant would have an ionizable side chain, the effective charge of which could be regulated with pH (Chang & Swenson, 1997), the arginine mutant would retain the positive charge in the pH range testable. Both these mutations perturbed flavin binding significantly and it could be likely due to the loss of the hydrogen bond to the phosphate moiety of the FMN. There was no evidence for the substantial stabilization of the anionic semiquinone species leading to the conclusion that electrostatics are not an important means of stabilizing and/or destabilizing the reduced states.
There exists contradictory data on electrostatic control of the isoalloxazine environment. In the model of the electron-transferring flavoprotein (ETF) from
Methylophilus methylotrophus (sp W3A1), an arginine (Arg237) is located close to the
FAD isoalloxazine ring, with its guanidinium group positioned over the si face of the dimethylbenzene region of the flavin. The guanidinium group is thus located to help stabilize the increased electron density on reduction of the flavin to the anionic semiquinone. Redox properties of the mutant ETF in which Arg237 is replaced by Ala
187
indicate that the arginine residue provides ~200mV stabilization to the anionic
semiquinone (Talfournier et al., 2001). On the other hand, in morphinone reductase, the
side chain of Arg238 is located close to the N(1)/C(2) carbonyl region of the
isoalloxazine ring. This positioning of a positively charged side chain at this locus is
common to many flavoproteins. By potentiometry of wild type and several mutant forms
(Arg238X), it was shown that the presence of this positively charged side chain is not a
requirement for stabilization of reduced flavin (Craig et al., 2001). Whereas for the
flavodoxin from D.vulgaris, introduction of positively charged histidine and arginine residues in the flavin-binding site afforded nearly 4.0 kcal/mol stabilization to the anionic hydroquinone state (Swenson & Krey, 1994). Whether electrostatics plays a role in redox regulation may be specific to the particular protein with several other factors also needing to be considered. In the FMN-binding domain of P450BM-3, there appears to be no evidence for the added stabilization afforded to the anionic semiquinone state by the presence of a positively charged residue.
At this time, it is not clear what structural aspects of the protein help modulate the redox properties of the cofactor. Lack of electrostatic control in redox regulation indicates the presence of another factor that precludes the formation of the anionic semiquinone species. The absolute requirement for the hydrogen bonding interaction to
N(5) and the maintenance of the type I′ turn conformation for binding of the flavin is evident. It remains to be established what effect these have on the redox properties of the
FMN. Is it likely that the N(5) interaction dominates over all other favorable interactions, such as electrostatics, and is primarily responsible for destabilizing the semiquinone? The
188
overriding importance of the main chain interaction with N(5)H, in the C. beijerinckii
flavodoxin, has already been established in the thermodynamic stabilization of the neutral
semiquinone state (Kasim & Swenson, 2001). While the importance of this interaction
still remains in BM-3, its role could be reversed, possibly because the semiquinone
species involved is anionic instead of the blue neutral. Complete understanding of the
conformational changes occurring at N(5) upon reduction should help answer these
questions.
Finally, it is important to keep in mind that these studies were performed with an isolated domain, a 20kDa fragment of an 119kDa holoprotein. It is possible that interactions with the other domains help tailor the redox potentials in this domain. Studies on nucleotide binding in P450BM-3 show that the role of the bound nucleotide
(NADPH/NADP+) is to maintain the redox potentials of the flavin cofactors so that rapid catalysis can occur and is likely the reason why equilibrium redox potential measurements differ from the conditions operating during catalysis (Murataliev &
Feyereisen, 2000). This would suggest that while these mutations did not seem to alter the equilibrium potentials significantly, they might be affecting the rates of catalysis. It is premature to speculate what these effects may be, data for that must await additional stopped-flow experiments where the electron-transferring rates to a donor such as cytochrome c can be measured.
189
CHAPTER 7
GENERAL CONCLUSIONS AND FUTURE DIRECTIONS
The structural simplicity and unique redox properties of the flavodoxin, when compared to other flavoproteins, make this a good model system in which to study the protein-flavin interactions responsible for establishing some of the biochemical properties of the cofactor. One such interaction that has been extensively studied is the hydrogen bonding interaction to N(5) of the flavin (Chang & Swenson, 1999; Hoover et al., 1999;
Ludwig et al., 1997; O'Farrell et al., 1998). This is the dominant interaction that is primarily responsible for the stabilization of the blue neutral semiquinone that is stabilized significantly in flavodoxins. To a lesser extent, it also stabilizes the fully reduced state of the cofactor. In this way, the oxidation-reduction potential of the functionally important sq/hq couple is poised at very low values, in some cases approaching –500mV (pH 7 vs SHE). This contributes to their role as low-potential electron-transfer proteins in a number of essential biological processes such as nitrogen fixation and photosynthesis (Ludwig & Luschinsky, 1992; Mayhew & Tollin, 1992).
In the C. beijerinckii flavodoxin, the preferential stabilization of the blue neutral semiquinone has been demonstrated to be mainly due to interactions with a reverse turn in the 50’s loop. The hydrogen bond to N(5) in the reduced states occurs as a result of a 190
change in the configuration of the unusual central Gly57-Asp58 peptide bond in this turn.
In the oxidized state, the peptide bond occurs in a cis conformation with the carbonyl
group of Gly57 oriented away from the flavin (cis-O-down). Upon reduction, the carbonyl group gets “flipped” over to point toward the flavin ring and adopts a trans conformation (trans-O-up). This structural rearrangement brings the carbonyl group in close proximity to the flavin ring such that it can serve as a hydrogen bond acceptor to
N(5)H of the reduced flavin. Unique to this flavodoxin is that this structural rearrangement resembles a conversion of a type II β-turn in the oxidized state to a type II′
β-turn in the reduced states. Substitutions at the two central positions of the turn (Gly57-
Asp58) that dictate the stability of the turn conformation, clearly demonstrated that turn energetics is directly coupled to the modulation of the oxidation-reduction potentials in this flavodoxin.
Furthermore, elimination of all other side chain interactions within this reverse turn while still allowing for the conformational change to occur, resulted in a functional protein, one able to bind FMN and thermodynamically stabilize the blue neutral semiquinone species of the flavin. The loss of favorable and/or unfavorable interactions with the side chains of Met56 and Asp58 were found to be additive, with this additive effect breaking down when Glu59 was considered. The reason for non-additivity involving Glu59 is not clear, but may involve either an indirect electrostatic interaction or structural perturbations such that the residues no longer behave independently. This just goes to demonstrate that in some instances single mutations will not suffice and that interpretations of single substitutions must be done with caution. Overall, these studies
191
emphasize the overriding importance of the main chain interaction with N(5)H of the
FMN and the associated conformational change that occurs in this turn upon reduction,
especially in the thermodynamic stabilization of the blue neutral form of the FMN
semiquinone in this flavodoxin.
Since the flavodoxin-like structural motif has been identified for the FMN- binding domains in a number of other more complex flavoproteins, the existence of similar stabilizing interactions as seen in the flavodoxins was an interesting avenue to investigate. The FMN-binding domain of P450BM-3 was a good system in which to do so because despite the structural similarity, this domain is unique in many ways. The lack of thermodynamic stabilization of a semiquinone species was just one aspect pursued with an aim to investigating the N(5) interaction in this flavoprotein. Given the lack of structural data on this system in the reduced state, conclusions on structural events occurring upon reduction are not easy to make. However, the NMR studies indicate a novel conformational change occurring within the flavin ring. Specifically, the N(5) atom alone is pushed out of plane when it adopts the sp3 hybridized conformation. This was proven not only by the chemical shift value of N(5) in the reduced state but also by its coupling constant of 70.3 Hz. This establishes that conformational changes do occur within the flavin ring, specifically at the N(5) atom, indicating the importance of this interaction in this flavoprotein. The sequence specificity of the turn that is in contact with the flavin, homologous to the 50’s turn in the C. beijerinckii flavodoxin, was found to be an important determinant for the overall stability of the protein. The maintenance of the type I′ turn conformation was an absolute prerequisite for flavin binding. It seems that the
192
turn conformation of the inner flavin-binding loop is critical in maintaining the stability
and redox properties of flavoproteins in general. Efforts to thermodynamically stabilize
the anionic semiquinone form of the flavin in BM-3 by stabilizing electrostatic
interactions was unsuccessful, possibly due to the lack of electrostatic control of the
isoalloxazine environment in this flavoprotein. The possibility of the N(5) interaction and
the associated conformational change in playing a role in destabilizing the semiquinone
state still remains to be answered. The answer may lie in the differing electronic
properties between the anionic semiquinone and the hydroquinone anion. In the former,
the electron density is thought to be delocalized on the N(5)-C(4a)-C(4)=O edge whereas
in the latter its on the N(1)-C(2)=O locus.
Future directions. The inner FMN-binding loop of BM-3 is an obvious target for future mutagenesis. Unlike other loop regions, this loop ends in two proline residues that are believed to help rigidify its conformation. It would be interesting to investigate what functional role having two proline residues in tandem play by introducing alanine substitutions, either single or double in this flavoprotein. It would be more interesting to target the second proline as one proline is found at that position in CPR. Furthermore, in that context, it would be interesting to introduce the mutations mentioned in the preceding chapter as it may be likely that these mutations may be better accommodated when the loop is more flexible. Another avenue that can be explored is the introduction of an additional amino acid in the loop region to mimic the loop structure found in CPR. It has been noted in alignments between CPR and BM-3 that due to the unique conformation of the BM-3 loop where it turns sharply at Asn537, it seems to be one
193
residue shorter than the other FMN-binding domains. Is it possible that introduction of an
additional residue would make the turn adopt a conformation resembling that of CPR? If
so, would it then stabilize a semiquinone species?
Further kinetic and stopped-flow studies would also help enlighten the role of some the mutants already generated. In BM-3, the anionic semiquinone of the FMN is kinetically stabilized, existing only transiently in order to reduce the heme iron when substrate is bound. While none of the current mutants affected the thermodynamic stability of the semiquinone, the rate of its formation may be altered during catalytic turnover. In addition, it is important to remember that the isolated FMN-binding domain is less stable than when other protein domains surround it and future work may be more productive if utilizing the holoprotein instead of the isolated domain.
Use of flavin analogs might also prove useful, especially 5-deazaflavin that would preclude hydrogen bond formation at the 5 position of the ring. Those studies may help shed some light on the role of the hydrogen bond to N(5) of the FMN, although interpretations of such data are not as direct as mutagenesis studies due to the alterations in the electronic distribution of the isoalloxazine ring. Structural consequences of such replacements have already been acknowledged. For instance, the introduction of an additional proton at C(5) in 5-deazaflavin would likely necessitate some structural readjustments in the flavin binding site.
Additional experiments that would complement the current data could include obtaining the resonance Raman spectra of this flavoprotein in the oxidized and fully reduced state. The assignment of the resonance Raman bands of the isoalloxazine ring on
194
the basis of isotopic data and normal mode vibrational analysis has already been done
(Bowman & Spiro, 1981; Kitagawa et al., 1979; Tegoni et al., 1997). Comparison of the frequency differences for the flavin bands could indicate a slight change in isoalloxazine conformation and changes in hydrogen bond strength, specially at N(5) that would strengthen the NMR data. In addition, BM-3 has a long wavelength band in the absorption spectra, the so-called charge-transfer band. Excitation within this region would help provide more structural detail of the flavin-binding site.
Only by extending these studies to other flavoenzyme systems will the general significance of the N(5) interaction and the associated conformational change become apparent. These studies will further strengthen the impact of our observations made with the C. beijerinckii flavodoxin.
195
LIST OF REFERENCES
Abramovitz, A. S. & Massey, V. (1976). Interaction of phenols with old yellow enzyme. Physical evidence for charge-transfer complexes. J Biol Chem 251(17), 5327-36.
Alderton, W. K., Cooper, C. E. & Knowles, R. G. (2001). Nitric oxide synthases: structure, function and inhibition. Biochem J 357(Pt 3), 593-615.
Anderson, R. F. (1983). Energetics of the one-electron reduction steps of riboflavin, FMN and FAD to their fully reduced forms. Biochim Biophys Acta 722(1), 158- 62.
Beinert, W. D., Ruterjans, H. & Muller, F. (1985). Nuclear magnetic resonance studies of the old yellow enzyme. 1. 15N NMR of the enzyme recombined with 15N-labeled flavin mononucleotides. Eur J Biochem 152(3), 573-9.
Binsch, G., Lambert, J. B., Roberts, B. W. & Roberts, J. D. (1964). Nitrogen-15 Magnetic Resonance Spectroscopy. II. Coupling Constants. J. Am. Chem. Soc 86, 5564- 5570.
Blyth, A. W. (1879). J. Chem. Soc. 35, 530-539.
Bourn, A. J. R. & Randall, E. W. (1964). Mol. Phys. 8, 567-571.
Bowman, W. D. & Spiro, T. G. (1981). Normal mode analysis of lumiflavin and interpretation of resonance Raman spectra of flavoproteins. Biochemistry 20(11), 3313-8.
Bradley, L. H. & Swenson, R. P. (1999). Role of glutamate-59 hydrogen bonded to N(3)H of the flavin mononucleotide cofactor in the modulation of the redox potentials of the Clostridium beijerinckii flavodoxin. Glutamate-59 is not responsible for the pH dependency but contributes to the stabilization of the flavin semiquinone. Biochemistry 38(38), 12377-86.
Bradley, L. H. & Swenson, R. P. (2001). Role of hydrogen bonding interactions to N(3)H of the flavin mononucleotide cofactor in the modulation of the redox potentials of the Clostridium beijerinckii flavodoxin. Biochemistry 40(30), 8686-95. 196
Breinlinger, E. C. & Rotello, V. M. (1997). Model Systems for Flavoenzyme Activity. Modulation of Flavin Redox Potentials through π-Stacking Interactions. J Am Chem Soc 119, 1165-6.
Briggs, W. R., Christie, J. M., Knieb, E. & Salomon, M. (1999). . In Flavins and Flavoproteins 1999 (Ghisla, S., Kroneck, P., Macheroux, P. and Sund, H., ed.), pp. 299-308. Rudolf Weber Agency for Scientific Publications, Berlin.
Burnett, R. M., Darling, G. D., Kendall, D. S., LeQuesne, M. E., Mayhew, S. G., Smith, W. W. & Ludwig, M. L. (1974). The structure of the oxidized form of clostridial flavodoxin at 1.9-A resolution. J Biol Chem 249(14), 4383-92.
Byron, C. M., Stankovich, M. T., Husain, M. & Davidson, V. L. (1989). Unusual redox properties of electron-transfer flavoprotein from Methylophilus methylotrophus. Biochemistry 28(21), 8582-7.
Chang, F. C. & Swenson, R. P. (1997). Regulation of oxidation-reduction potentials through redox-linked ionization in the Y98H mutant of the Desulfovibrio vulgaris [Hildenborough] flavodoxin: direct proton nuclear magnetic resonance spectroscopic evidence for the redox-dependent shift in the pKa of Histidine-98. Biochemistry 36(29), 9013-21.
Chang, F. C. & Swenson, R. P. (1999). The midpoint potentials for the oxidized- semiquinone couple for gly57 mutants of the clostridium beijerinckii flavodoxin correlate with changes in the hydrogen-bonding interaction with the proton on N(5) of the reduced flavin mononucleotide cofactor As measured by NMR chemical shift temperature dependencies [In Process Citation]. Biochemistry 38(22), 7168-76.
Clark, W. M. (1972). Oxidation-reduction potentials of organic systems, Robert E. Krieger Publishing Company, Huntington, N.Y.
Craig, D. H., Barna, T., Moody, P. C., Bruce, N. C., Chapman, S. K., Munro, A. W. & Scrutton, N. S. (2001). Effects of environment on flavin reactivity in morphinone reductase: analysis of enzymes displaying differential charge near the N-1 atom and C-2 carbonyl region of the active-site flavin. Biochemical Journal 359(Pt 2), 315-23.
Cuello, A. O., McIntosh, C. M. & Rotello, V. M. (2000). Model Systems for Flavoenzyme Activity. The Role of N(3)-H Hydrogen Bonding in Flavin Redox Processes. J Am Chem Soc 122, 3517-21.
197
Daff, S. N., Chapman, S. K., Turner, K. L., Holt, R. A., Govindaraj, S., Poulos, T. L. & Munro, A. W. (1997). Redox control of the catalytic cycle of flavocytochrome P- 450 BM3. Biochemistry 36(45), 13816-23.
Davidson, V. L., Husain, M. & Neher, J. W. (1986). Electron transfer flavoprotein from Methylophilus methylotrophus: properties, comparison with other electron transfer flavoproteins, and regulation of expression by carbon source. J Bacteriol 166(3), 812-7.
Doddrell, D. M., Pegg, D. T. & Bendall, M. R. (1982). Distortionless Enhancement of NMR Signals by Polarization Transfer. Journal of Magnetic Resonance 48, 323- 327.
Draper, R. D. & Ingraham, L. L. (1968). A potentiometric study of the flavin semiquinone equilibrium. Arch Biochem Biophys 125(3), 802-8.
Drennan, C. L., Pattridge, K. A., Weber, C. H., Metzger, A. L., Hoover, D. M. & Ludwig, M. L. (1999). Refined structures of oxidized flavodoxin from Anacystis nidulans. J Mol Biol 294(3), 711-24.
Druhan, L. J. & Swenson, R. P. (1998). Role of methionine 56 in the control of the oxidation-reduction potentials of the Clostridium beijerinckii flavodoxin: effects of substitutions by aliphatic amino acids and evidence for a role of sulfur-flavin interactions. Biochemistry 37(27), 9668-78.
Dubourdieu, M., le Gall, J. & Favaudon, V. (1975). Physicochemical properties of flavodoxin from Desulfovibrio vulgaris. Biochim Biophys Acta 376(3), 519-32.
Dudley, K. H., Ehrenberg, A., Hemmerich, P. & Müller, F. (1964). Helv. Chim. Acta 47, 1354-1382.
Edmondson, D. E. & Newton-Vinson, P. (2001). The covalent FAD of monoamine oxidase: structural and functional role and mechanism of the flavinylation reaction. Antioxid Redox Signal 3(5), 789-806.
Ehrenberg, A., Muller, F. & Hemmerich, P. (1967). Basicity, visible spectra, and electron spin resonance of flavosemiquinone anions. Eur J Biochem 2(3), 286-93.
Eren, M. (1990). Ph.D. Thesis. Ohio State University, Columbus, Ohio.
Eren, M. & Swenson, R. P. (1989). Chemical synthesis and expression of a synthetic gene for the flavodoxin from Clostridium MP. Journal of Biological Chemistry 264(25), 14874-9.
198
Feng, Y. & Swenson, R. P. (1997). Evaluation of the role of specific acidic amino acid residues in electron transfer between the flavodoxin and cytochrome c3 from Desulfovibrio vulgaris. Biochemistry 36(44), 13617-28.
Flashner, M. I. & Massey, V. (1974). Purification and properties of L-lysine monooxygenase from Pseudomonas fluorescens. J Biol Chem 249(8), 2579-86.
Fox, K. M. & Karplus, P. A. (1994). Old yellow enzyme at 2 A resolution: overall structure, ligand binding, and comparison with related flavoproteins. Structure 2(11), 1089-105.
Fukuyama, K., Matsubara, H. & Rogers, L. J. (1992). Crystal structure of oxidized flavodoxin from a red alga Chondrus crispus refined at 1.8 A resolution. Description of the flavin mononucleotide binding site. Journal of Molecular Biology 225(3), 775-89.
Ghisla, S. & Massey, V. (1989). Mechanisms of flavoprotein-catalyzed reactions. Eur J Biochem 181(1), 1-17.
Ghisla, S., Massey, V., Lhoste, J. M. & Mayhew, S. G. (1974). Fluorescence and optical characteristics of reduced flavines and flavoproteins. Biochemistry 13(3), 589-97.
Goodman, A. J., Breinlinger, E. C., McIntosh, C. M., Grimaldi, L. N. & Rotello, V. M. (2001). Model Systems for Flavoenzyme Activity. Control of Flavin Recognition via Specific Electrostatic Interactions. Org Lett 3(10), 1531-4.
Govindaraj, S., Poulos, T. L. & Tollin, G. (1997). The domain architecture of cytochrome P450BM-3. Journal of Biological Chemistry 272(12), 7922-6.
Griffin, K. J., Degala, G. D., Eisenreich, W., Muller, F., Bacher, A. & Frerman, F. E. (1998). 31P-NMR spectroscopy of human and Paracoccus denitrificans electron transfer flavoproteins, and 13C- and 15N-NMR spectroscopy of human electron transfer flavoprotein in the oxidised and reduced states. European Journal of Biochemistry 255(1), 125-32.
Gruez, A., Pignol, D., Zeghouf, M., Coves, J., Fontecave, M., Ferrer, J. L. & Fontecilla- Camps, J. C. (2000). Four crystal structures of the 60 kDa flavoprotein monomer of the sulfite reductase indicate a disordered flavodoxin-like module. J Mol Biol 299(1), 199-212.
Gunasekaran, K., Gomathi, L., Ramakrishnan, C., Chandrasekhar, J. & Balaram, P. (1998). Conformational interconversions in peptide beta-turns: analysis of turns in proteins and computational estimates of barriers. J Mol Biol 284(5), 1505-16.
199
Haines, D. C., Sevrioukova, I. F. & Peterson, J. A. (2000). The FMN-binding domain of cytochrome P450BM-3: resolution, reconstitution, and flavin analogue substitution. Biochemistry 39(31), 9419-29.
Hall, D. A., Vander Kooi, C. W., Stasik, C. N., Stevens, S. Y., Zuiderweg, E. R. & Matthews, R. G. (2001). Mapping the interactions between flavodoxin and its physiological partners flavodoxin reductase and cobalamin-dependent methionine synthase. Proc Natl Acad Sci U S A 98(17), 9521-6.
Hall, L. H., Orchard, B. J. & Tripathy, S. K. (1987). The Structure and Properties of Flavins: Molecular Orbital Study Based on Totally Optimized Geometries. II. Molecular Orbital Structure and Electron Distribution. International Journal of Quantum Chemistry 31, 217-242.
Haynes, C. A., Koder, R. L., Miller, A. F. & Rodgers, D. W. (2002). Structures of nitroreductase in three states: effects of inhibitor binding and reduction. Journal of Biological Chemistry 277(13), 11513-20.
Hazzard, J. T., Govindaraj, S., Poulos, T. L. & Tollin, G. (1997). Electron transfer between the FMN and heme domains of cytochrome P450BM-3. Effects of substrate and CO. Journal of Biological Chemistry 272(12), 7922-6.
Hoover, D. M., Drennan, C. L., Metzger, A. L., Osborne, C., Weber, C. H., Pattridge, K. A. & Ludwig, M. L. (1999). Comparisons of wild-type and mutant flavodoxins from Anacystis nidulans. Structural determinants of the redox potentials. J Mol Biol 294(3), 725-43.
Jabs, A., Weiss, M. S. & Hilgenfeld, R. (1999). Non-proline cis peptide bonds in proteins. J Mol Biol 286(1), 291-304.
Jenkins, C. M., Genzor, C. G., Fillat, M. F., Waterman, M. R. & Gomez-Moreno, C. (1997). Negatively charged anabaena flavodoxin residues (Asp144 and Glu145) are important for reconstitution of cytochrome P450 17alpha-hydroxylase activity. J Biol Chem 272(36), 22509-13.
Jorns, M. S., Wang, B. & Jordan, S. P. (1987). DNA repair catalyzed by Escherichia coli DNA photolyase containing only reduced flavin: elimination of the enzyme's second chromophore by reduction with sodium borohydride. Biochemistry 26(21), 6810-6.
200
Karrer, P., Schöpp, K. & Benz, F. (1935). Helv. Chim. Acta 18, 426-429.
Kasim, M. & Swenson, R. P. (2000). Conformational energetics of a reverse turn in the Clostridium beijerinckii flavodoxin is directly coupled to the modulation of its oxidation-reduction potentials. Biochemistry 39(50), 15322-32.
Kasim, M. & Swenson, R. P. (2001). Alanine-scanning of the 50's loop in the Clostridium beijerinckii flavodoxin: evaluation of additivity and the importance of interactions provided by the main chain in the modulation of the oxidation- reduction potentials. Biochemistry 40(45), 13548-55.
Kitagawa, T., Nishina, Y., Kyogoku, Y., Yamano, T., Ohishi, N., Takai-Suzuki, A. & Yagi, K. (1979). Resonance Raman spectra of carbon-13- and nitrogen-15-labeled riboflavin bound to egg-white flavoprotein. Biochemistry 18(9), 1804-8.
Klein, M. L. & Fulco, A. J. (1993). Critical residues involved in FMN binding and catalytic activity in cytochrome P450BM-3. J Biol Chem 268(10), 7553-61.
Kunkel, T. A. (1985). Rapid and efficient site-specific mutagenesis without phenotypic selection. Proc Natl Acad Sci U S A 82(2), 488-92.
Laudenbach, D. E., Straus, N. A., Pattridge, K. A. & Ludwig, M. L. (1988). Sequence and Structure of Anacystis nidulans Flavodoxin : Comparisons with Flavodoxins from other Species. In Flavins and flavoproteins, 1987 : proceedings of the Ninth International Symposium on Flavins and Flavoproteins, Atlanta, Georgia, USA, June 7-12, 1987 (Edmondson, D. E. & McCormick, D. B., eds.), pp. 249-260. W. de Gruyter, Berlin.
Lostao, A., El Harrous, M., Daoudi, F., Romero, A., Parody-Morreale, A. & Sancho, J. (2000). Dissecting the energetics of the apoflavodoxin-FMN complex. J Biol Chem 275(13), 9518-26.
Lostao, A., Gomez-Moreno, C., Mayhew, S. G. & Sancho, J. (1997). Differential stabilization of the three FMN redox forms by tyrosine 94 and tryptophan 57 in flavodoxin from Anabaena and its influence on the redox potentials. Biochemistry 36(47), 14334-44.
Ludwig, M. L. & Luschinsky, C. L. (1992). Structure and Redox Properties of Clostridial Flavodoxin. In Chemistry and Biochemistry of Flavoenzymes (Müller, F., ed.), Vol. III, pp. 427-466. CRC Press, Boca Raton, FL.
201
Ludwig, M. L., Pattridge, K. A., Metzger, A. L., Dixon, M. M., Eren, M., Feng, Y. & Swenson, R. P. (1997). Control of oxidation-reduction potentials in flavodoxin from Clostridium beijerinckii: the role of conformation changes. Biochemistry 36(6), 1259-80.
Luschinsky, C. L., Dunham, W. R., Osborne, C., Pattridge, K. A. & Ludwig, M. L. (1991). Structural Analysis of Fully Reduced A. nidulans Flavodoxin. In Flavins and flavoproteins 1990 : proceedings of the tenth international symposium, Como, Italy, July 15-20, 1990 (Curti, B., Ronchi, S. & Zanetti, G., eds.), pp. 409- 413. W. de Gruyter, Berlin New York ;.
Massey, V. & Hemmerich, P. (1980). Active-site probes of flavoproteins. Biochem Soc Trans 8(3), 246-57.
Massey, V. & 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(10), 3181-9.
Matthew, J. B., Weber, P. C., Salemme, F. R. & Richards, F. M. (1983). Electrostatic orientation during electron transfer between flavodoxin and cytochrome c. Nature 301(5896), 169-71.
Mayhew, S. G. (1971). Studies on flavin binding in flavodoxins. Biochim Biophys Acta 235(2), 289-302.
Mayhew, S. G. (1999). The effects of pH and semiquinone formation on the oxidation- reduction potentials of flavin mononucleotide. A reappraisal. Eur J Biochem 265(2), 698-702.
Mayhew, S. G. & Tollin, G. (1992). General Properties of Flavodoxin. In Chemistry and Biochemistry of Flavoenzymes (Müller, F., ed.), Vol. III, pp. 389-426. CRC Press, Boca Raton, FL.
McCarthy, A. A., Walsh, M. A., Verma, C. S., O'Connell, D. P., Reinhold, M., Yalloway, G. N., D'Arcy, D., Higgins, T. M., Voordouw, G. & Mayhew, S. G. (2002). Crystallographic investigation of the role of aspartate 95 in the modulation of the redox potentials of Desulfovibrio vulgaris flavodoxin. Biochemistry 41(36), 10950-62.
Mildvan, A. S., Weber, D. J. & Kuliopulos, A. (1992). Quantitative interpretations of double mutations of enzymes [published erratum appers in Arch Biochem Biophys 1992 Aug 15;297(1):188]. Arch Biochem Biophys 294(2), 327-40.
202
Miles, C. S., Ost, T. W., Noble, M. A., Munro, A. W. & Chapman, S. K. (2000). Protein engineering of cytochromes P-450. Biochim Biophys Acta 1543(2), 383-407.
Möhle, K., Gubmann, M. & Hofmann, H. (1997). Structural and Energetic Relations between β Turns. Journal of Computational Chemistry 18, 1415-1430.
Moonen, C. T., Vervoort, J. & Muller, F. (1984). Carbon-13 nuclear magnetic resonance study on the dynamics of the conformation of reduced flavin. Biochemistry 23(21), 4868-72.
Müller, F. (1991). Chemistry and biochemistry of flavoenzymes, 1, CRC Press, Boca Raton.
Müller, F. & Massey, V. (1969). Flavin-sulfite complexes and their structures. J Biol Chem 244(15), 4007-16.
Murataliev, M. B. & Feyereisen, R. (2000). Functional interactions in cytochrome P450BM3. Evidence that NADP(H) binding controls redox potentials of the flavin cofactors. Biochemistry 39(41), 12699-707.
Murataliev, M. B., Klein, M., Fulco, A. & Feyereisen, R. (1997). Functional interactions in cytochrome P450BM3: flavin semiquinone intermediates, role of NADP(H), and mechanism of electron transfer by the flavoprotein domain. Biochemistry 36(27), 8401-12.
Murty, C. V. & Adiga, P. R. (1982). Pregnancy suppression by active immunization against gestation-specific riboflavin carrier protein. Science 216(4542), 191-3.
Nakajima, N., Higo, J., Kidera, A. & Nakamura, H. (2000). Free energy landscapes of peptides by enhanced conformational sampling. J Mol Biol 296(1), 197-216.
Nicholson, L. K., Yamazaki, T., Torchia, D. A., Grzesiek, S., Bax, A., Stahl, S. J., Kaufman, J. D., Wingfield, P. T., Lam, P. Y., Jadhav, P. K. & et al. (1995). Flexibility and function in HIV-1 protease. Nat Struct Biol 2(4), 274-80.
Nozaki, Y. & Tanford, C. (1967). Examination of Titration Behavior. Methods Enzymol. 11, 715-734.
O'Farrell, P. A., Walsh, M. A., McCarthy, A. A., Higgins, T. M., Voordouw, G. & Mayhew, S. G. (1998). Modulation of the redox potentials of FMN in Desulfovibrio vulgaris flavodoxin: thermodynamic properties and crystal structures of glycine- 61 mutants. Biochemistry 37(23), 8405-16.
203
Ohmae, E., Iriyama, K., Ichihara, S. & Gekko, K. (1998). Nonadditive effects of double mutations at the flexible loops, glycine-67 and glycine-121, of Escherichia coli dihydrofolate reductase on its stability and function. J Biochem (Tokyo) 123(1), 33-41.
Pace, C. N., Vajdos, F., Fee, L., Grimsley, G. & Gray, T. (1995). How to measure and predict the molar absorption coefficient of a protein. Protein Sci 4(11), 2411-23.
Pal, D. & Chakrabarti, P. (1999). Cis Peptide Bonds in Proteins: Residues Involved, their Conformations, Interactions and Locations. J Mol Biol 294(1), 271-288.
Reibenspies, J. H., Guo, F. & Rizzo, C. J. (2000). X-ray crystal structures of conformationally biased flavin models. Org Lett 2(7), 903-6.
Reynolds, R. A., Watt, W. & Watenpaugh, K. D. (2001). Structures and comparison of the Y98H (2.0 A) and Y98W (1.5 A) mutants of flavodoxin (Desulfovibrio vulgaris). Acta Crystallogr D Biol Crystallogr 57(Pt 4), 527-535.
Richardson, J. S. (1981). The anatomy and taxonomy of protein structure. Adv Protein Chem 34, 167-339.
Roberts, D. L., Frerman, F. E. & Kim, J. J. (1996). Three-dimensional structure of human electron transfer flavoprotein to 2.1-A resolution. Proceedings of the National Academy of Sciences of the United States of America 93(25), 14355-60.
Rose, G. D., Gierasch, L. M. & Smith, J. A. (1985). Turns in peptides and proteins. Adv Protein Chem 37, 1-109.
Ruettinger, R. T., Wen, L. P. & Fulco, A. J. (1989). Coding nucleotide, 5' regulatory, and deduced amino acid sequences of P- 450BM-3, a single peptide cytochrome P- 450:NADPH-P-450 reductase from Bacillus megaterium. J Biol Chem 264(19), 10987-95.
Salazar, D., Zhang, L., deGala, G. D. & Frerman, F. E. (1997). Expression and characterization of two pathogenic mutations in human electron transfer flavoprotein. Journal of Biological Chemistry 272(42), 26425-33.
Sevrioukova, I., Shaffer, C., Ballou, D. P. & Peterson, J. A. (1996a). Equilibrium and transient state spectrophotometric studies of the mechanism of reduction of the flavoprotein domain of P450BM-3. Biochemistry 35(22), 7058-68.
204
Sevrioukova, I., Truan, G. & Peterson, J. A. (1996b). The flavoprotein domain of P450BM-3: expression, purification, and properties of the flavin adenine dinucleotide- and flavin mononucleotide-binding subdomains. Biochemistry 35(23), 7528-35.
Sevrioukova, I. F., Li, H., Zhang, H., Peterson, J. A. & Poulos, T. L. (1999). Structure of a cytochrome P450-redox partner electron-transfer complex. Proc Natl Acad Sci U S A 96(5), 1863-8.
Sevrioukova, I. F. & Peterson, J. A. (1995). NADPH-P-450 reductase: structural and functional comparisons of the eukaryotic and prokaryotic isoforms. Biochimie 77(7-8), 562-72.
Smith, W. W., Burnett, R. M., Darling, G. D. & Ludwig, M. L. (1977). Structure of the semiquinone form of flavodoxin from Clostridum MP. Extension of 1.8 A resolution and some comparisons with the oxidized state. J Mol Biol 117(1), 195- 225.
Stewart, R. C. & Massey, V. (1985). Potentiometric studies of native and flavin- substituted Old Yellow Enzyme. J Biol Chem 260(25), 13639-47.
Susin, S. A., Lorenzo, H. K., Zamzami, N., Marzo, I., Snow, B. E., Brothers, G. M., Mangion, J., Jacotot, E., Costantini, P., Loeffler, M., Larochette, N., Goodlett, D. R., Aebersold, R., Siderovski, D. P., Penninger, J. M. & Kroemer, G. (1999). Molecular characterization of mitochondrial apoptosis-inducing factor. Nature 397(6718), 441-6.
Swenson, R. P. & Krey, G. D. (1994). Site-directed mutagenesis of tyrosine-98 in the flavodoxin from Desulfovibrio vulgaris (Hildenborough): regulation of oxidation- reduction properties of the bound FMN cofactor by aromatic, solvent, and electrostatic interactions. Biochemistry 33(28), 8505-14.
Takano, K., Yamagata, Y. & Yutani, K. (2001). Role of non-glycine residues in left- handed helical conformation for the conformational stability of human lysozyme. Proteins 44(3), 233-43.
Talfournier, F., Munro, A. W., Basran, J., Sutcliffe, M. J., Daff, S., Chapman, S. K. & Scrutton, N. S. (2001). alpha Arg-237 in Methylophilus methylotrophus (sp. W3A1) electron-transferring flavoprotein affords approximately 200-millivolt stabilization of the FAD anionic semiquinone and a kinetic block on full reduction to the dihydroquinone. Journal of Biological Chemistry 276(23), 20190-6.
205
Tanokura, M. (1983). 1H-NMR study on the tautomerism of the imidazole ring of histidine residues. I. Microscopic pK values and molar ratios of tautomers in histidine-containing peptides. Biochim Biophys Acta 742(3), 576-85.
Tegoni, M., Gervais, M. & Desbois, A. (1997). Resonance Raman study on the oxidized and anionic semiquinone forms of flavocytochrome b2 and L-lactate monooxygenase. Influence of the structure and environment of the isoalloxazine ring on the flavin function. Biochemistry 36(29), 8932-46.
Ullmann, G. M., Hauswald, M., Jensen, A. & Knapp, E. W. (2000). Structural alignment of ferredoxin and flavodoxin based on electrostatic potentials: implications for their interactions with photosystem I and ferredoxin-NADP reductase. Proteins 38(3), 301-9. van Schagen, C. G. & Muller, F. (1981). A 13C nuclear-magnetic-resonance study on free flavins and Megasphaera elsdenii and Azotobacter vinelandii flavodoxin. 13C-enriched flavins as probes for the study of flavoprotein active sites. Eur J Biochem 120(1), 33-9.
Veine, D. M., Arscott, L. D. & Williams, C. H., Jr. (1998). Redox potentials for yeast, Escherichia coli and human glutathione reductase relative to the NAD+/NADH redox couple: enzyme forms active in catalysis. Biochemistry 37(44), 15575-82.
Vervoort, J., Muller, F., Mayhew, S. G., van den Berg, W. A., Moonen, C. T. & Bacher, A. (1986). A comparative carbon-13, nitrogen-15, and phosphorus-31 nuclear magnetic resonance study on the flavodoxins from Clostridium MP, Megasphaera elsdenii, and Azotobacter vinelandii. Biochemistry 25(22), 6789-99.
Wang, M., Roberts, D. L., Paschke, R., Shea, T. M., Masters, B. S. & Kim, J. J. (1997). Three-dimensional structure of NADPH-cytochrome P450 reductase: prototype for FMN- and FAD-containing enzymes. Proceedings of the National Academy of Sciences of the United States of America 94(16), 8411-6.
Warburg, O. & Christian, W. (1933). Biochem. Z. 266, 377-411.
Wassink, J. H. & Mayhew, S. G. (1975). Fluorescence titration with apoflavodoxin: a sensitive assay for riboflavin 5'-phosphate and flavin adenine dinucleotide in mixtures. Anal Biochem 68(2), 609-16.
Watenpaugh, K. D., Sieker, L. C. & Jensen, L. H. (1976). A Crystallographic Structural Study of the Oxidation States of Desulfovibrio Vulgaris Flavodoxin. In Flavins and flavoproteins : proceedings (Singer, T. P., ed.), pp. 405-410. Elsevier Scientific, Amsterdam ; New York.
206
Watt, W., Tulinsky, A., Swenson, R. P. & Watenpaugh, K. D. (1991). Comparison of the crystal structures of a flavodoxin in its three oxidation states at cryogenic temperatures. J Mol Biol 218(1), 195-208.
Weber, P. C. & Tollin, G. (1985). Electrostatic interactions during electron transfer reactions between c- type cytochromes and flavodoxin. J Biol Chem 260(9), 5568-73.
Wells, J. A. (1990). Additivity of mutational effects in proteins. Biochemistry 29(37), 8509-17.
Whitby, L. G. (1953). A New Method for Preparing Flavin-adenine Dinucleotide. Biochemical Journal 54, 437-442.
White, H. B., 3rd & Merrill, A. H., Jr. (1988). Riboflavin-binding proteins. Annu Rev Nutr 8, 279-99.
Witanowski, M., Stefaniak, L. & Webb, G. A. (1981). Annu. Rep. NMR Spectrosc. 11B, 1-493.
Yalloway, G. N., Mayhew, S. G., Malthouse, J. P., Gallagher, M. E. & Curley, G. P. (1999). pH-dependent spectroscopic changes associated with the hydroquinone of FMN in flavodoxins. Biochemistry 38(12), 3753-62.
Yan, Y., Erickson, B. E. & Tropsha, A. (1995). Free Energies for Folding and Refolding of Four Types of β Turns: Simulation of the Role of D/L Chirality. J. Am. Chem. Soc. 117, 7592-7599.
Yang, A. S., Hitz, B. & Honig, B. (1996). Free energy determinants of secondary structure formation: III. beta- turns and their role in protein folding. J Mol Biol 259(4), 873-82.
Zanetti, G. & Aliverti, A. (1991). . In Chemistry and Biochemistry of Flavoenzymes (Müller, F., ed.), Vol. II, pp. 305-351. CRC Press, Boca Raton, FL.
Zhang, X. J., Baase, W. A. & Matthews, B. W. (1991). Toward a simplification of the protein folding problem: a stabilizing polyalanine alpha-helix engineered in T4 lysozyme. Biochemistry 30(8), 2012-7.
Zhao, Q., Modi, S., Smith, G., Paine, M., McDonagh, P. D., Wolf, C. R., Tew, D., Lian, L. Y., Roberts, G. C. & Driessen, H. P. (1999). Crystal structure of the FMN- binding domain of human cytochrome P450 reductase at 1.93 A resolution. Protein Science 8(2), 298-306.
207
Zheng, Y.-J. & Ornstein, R. L. (1996). A theoretical study of the structures of flavin in different oxidation and protonation states. J. Am. Chem. Soc. 118, 9402-9408.
Zhou, Z. & Swenson, R. P. (1995). Electrostatic effects of surface acidic amino acid residues on the oxidation-reduction potentials of the flavodoxin from Desulfovibrio vulgaris (Hildenborough). Biochemistry 34(10), 3183-92.
Zhou, Z. & Swenson, R. P. (1996). The cumulative electrostatic effect of aromatic stacking interactions and the negative electrostatic environment of the flavin mononucleotide binding site is a major determinant of the reduction potential for the flavodoxin from Desulfovibrio vulgaris [Hildenborough]. Biochemistry 35(50), 15980-8.
208