Control of catalysis in flavin-dependent monooxygenases

Archives of Biochemistry and Biophysics 493 (2010) 26–36

Contents lists available at ScienceDirect

Archives of Biochemistry and Biophysics journal homepage: www.elsevier.com/locate/yabbi

Review

Control of catalysis in flavin-dependent monooxygenases Bruce A. Palfey *, Claudia A. McDonald Department of Biological Chemistry, University of Michigan Medical School, 1150 W. Medical Center Dr., Ann Arbor, MI 48109-5606, USA

a r t i c l e

i n f o

Article history: Received 19 September 2009 and in revised form 17 November 2009 Available online 26 November 2009 Keywords: Flavin Monooxygenase Hydroxylase Hydroperoxide Peroxide Oxygen

a b s t r a c t Flavoprotein monooxygenases reduce flavins, speed their reaction with oxygen, and stabilize a C4a-oxygen adduct long enough to use this reactive species to transfer an oxygen atom to a substrate. The flavin– oxygen adduct can be the C4a-peroxide anion, in which case it reacts as a nucleophile. The protonated adduct – the C4a-hydroperoxide – reacts as an electrophile. The elimination of H2O2 competes with substrate oxygenation. This side-reaction is suppressed, preventing the waste of NAD(P)H and the production of toxic H2O2. Several strategies have been uncovered that prevent the deleterious side-reaction while still allowing substrate hydroxylation. Ó 2009 Elsevier Inc. All rights reserved.

Introduction Reduced flavoenzymes react with oxygen. For many flavoenzymes, this is a deleterious side-reaction and is often suppressed by the protein, but oxidases and monooxygenases have evolved to use O2 as a physiological substrate. Monooxygenases (a.k.a. hydroxylases) oxygenate a substrate by taking advantage of the high reactivity of O2 reduced to the peroxide level. Oxygenation of a substrate cleaves the oxygen–oxygen bond and produces water. Monooxygenases activate O2 by synthesizing a covalent flavin–oxygen adduct, the C4a-(hydro)peroxide. The electrophilic or nucleophilic reactions of the peroxyflavin intermediate parallel those of organic peroxides used by synthetic chemists. A number of mechanistically diverse reactions comprise the catalytic cycles of monooxygenases; several strategies for promoting and controlling these reactions are now evident. Conformational changes appear to be generally important in coordinating critical reactions and stabilizing labile intermediates. This brief review describes the inherent chemical constraints on flavin–oxygen chemistry, some enzymatic strategies for dealing with these constraints, and examples of enzymes that use this chemistry for various biological tasks. Flavin–oxygen reactions Flavin hydroquinones (2-electron reduced flavins) react in solution relatively slowly with O2 despite the extremely favorable free energy change (Fig. 1). For instance, the bimolecular rate constant * Corresponding author. Fax: +1 734 764 3509. E-mail address: [email protected] (B.A. Palfey). 0003-9861/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.abb.2009.11.028

for the reaction of reduced tetraacetyl riboflavin is 250 M1 s1, about seven orders of magnitude lower than the diffusion-limited collision frequency [1]. This is because total electron spin must be conserved in chemical reactions. The ground state of O2 is unusual because it is a triplet – it has two unpaired electrons. Most organic molecules are singlets – all their electrons are paired. Conservation of spin in the reaction of oxygen with a singlet requires two unpaired electrons in the products, thus forming superoxide and an organic radical. Most organic radicals are very unstable, and superoxide is not exceedingly stable either, so their formation is unfavorable, preventing rapid reactions. However, flavin radicals – the semiquinones – have some stability in solution, allowing the slow reaction of O2, although not at rates fast enough to support biochemistry without catalysis. The stabilities of intermediates such as flavin and oxygen radicals can be altered enormously by the protein environment. Consequently, most flavin monooxygenases react with oxygen 2–4 orders of magnitude faster than free flavins. The superoxide–semiquinone pair, caged in aqueous solvent upon formation, reacts by radical coupling before the intermediates can diffuse apart [2]. An oxygen–carbon bond forms between superoxide and the isoalloxazine at C4a, a site of high spin-density in the neutral semiquinone [3]. The resulting C4a-peroxide anion is protonated in water, forming the C4a-hydroperoxide. Hydrogen peroxide is rapidly eliminated in a buffer-catalyzed reaction by the deprotonation of N5 and the protonation of the oxygen of the leaving group to form oxidized flavin. In fact, the hydroperoxide has not been seen in the pathway because the rate-determining step at attainable oxygen concentrations is the initial bimolecular reaction. The hydroperoxide, generated in water in pulse-radiolysis experiments, eliminates H2O2 with a rate constant of 260 s1,

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

27

Fig. 1. The reaction of reduced flavin with O2. The initial step in the reaction of reduced flavins with O2 is a single-electron transfer to form a solvent-caged semiquinone– superoxide pair. Radical recombination gives the flavin-C4a-peroxide anion, followed by protonation to the hydroperoxide, and elimination of H2O2.

much larger than the pseudo-first order rate constant of 0.5 s1 that could be obtained with high oxygen concentrations [4]. Extensive model chemistry has shown that hydroperoxyflavins react like other organic hydroperoxides [2]. Alkylating N5 of flavin models prevents the rapid elimination of hydrogen peroxide, allowing their reactivity with substrates to be studied. The isoalloxazine moiety attached to the proximal peroxy–oxygen is electron-withdrawing, polarizing the O–O bond and activating it for heterolytic cleavage. Thus nucleophiles such as amines or sulfides attack the distal oxygen of the hydroperoxide and displace the flavin oxide leaving group. Brønsted analysis suggests that a partial charge of about 0.4 develops on the proximal oxygen in the transition state. Tasks and challenges for monooxygenases Effectively harnessing the reactive potential of the peroxyflavin requires the promotion of a number of reactions, and, just as importantly, the inhibition of side-reactions. Peroxyflavins are synthesized by first reducing oxidized flavin with a pyridine nucleotide, followed by the reaction with O2. Once synthesized, the peroxyflavin oxygenates the substrate; often this substrate must be activated. After oxygen transfer, the flavin hydroxide eliminates water to form oxidized flavin. The wildly different transition states in the catalytic cycle – hydride transfer, O2 activation, substrate oxygenation, and water elimination – require different stabilizing interactions from the protein. Hydroxylases spread the chemistry across the protein using coordinated conformational changes. While monooxygenases promote these diverse reactions, they also prevent the elimination of H2O2 from the peroxy adducts by blocking the access of solvent to N5 and sometimes speed the competing oxygen transfer reaction. Thus monooxygenases balance many chemical requirements, making them the most complex of the ‘‘simple’’ flavoenzymes. The biological benefit of successful monooxygenation is (usually) a hydroxylated compound used in metabolism or signal transduction. Failed attempts not only deprive organisms of a valuable metabolite, but also waste reduced pyridine nucleotides and generate toxic hydrogen peroxide. Thus monooxygenase evolution must be guided by selective pressure to develop control mechanisms assuring that committing NAD(P)H results in successful hydroxylation. Assessing the appropriateness or presence of the substrate is difficult because several reaction steps intervene

between the reduction of the flavin and the transfer of oxygen from the hydroperoxide. Two strategies have evolved to prevent wasteful NAD(P)H oxidase activity – ‘‘bold’’ and ‘‘cautious’’. ‘‘Cautious’’ monooxygenases require a hydroxylatable substrate to be bound in order to allow rapid flavin reduction. Thus hydroperoxide formation from the reduced enzyme is triggered by the presence of the substrate to be hydroxylated. Enzymes employing this strategy stabilize the hydroperoxide from wasteful elimination only moderately, relying on hydroxylation to out-compete the side-reaction. ‘‘Bold’’ monooxygenases allow rapid flavin reduction and subsequent hydroperoxide formation regardless of the presence of a hydroxylatable substrate, but effectively protect the flavin hydroperoxide from elimination. These enzymes stall the catalytic cycle until a competent substrate is encountered, again preventing futile NAD(P)H oxidase activity. The reactions and control mechanisms outlined above are implemented in different ways by different enzymes. The large number of permutations of catalytic control, reactivity of the hydroperoxide, and protein structures precludes an extensive discussion of every type of hydroxylase. What follows are examples selected either because they illustrate general hydroxylase paradigms, or they represent interesting biological applications of monooxygenase chemistry. Aromatic hydroxylases – ‘‘Cautious’’ hydroxylases p-Hydroxybenzoate hydroxylase p-Hydroxybenzoate hydroxylase (PHBH)1 is the most extensively studied flavoprotein monooxygenase. Its mechanism is understood in unparalleled detail and illustrates the chemistry shared by all aromatic hydroxylases, as well as some features that are uniquely adapted to PHBH. PHBH is a homodimer of 45 kDa subunits and uses a non-covalently bound FAD as a prosthetic group. It catalyzes the hydroxylation of p-hydroxybenzoate (pOHB) to 3,4-dihydroxybenzoate at the expense of NADPH and O2 [5–7]. PHBH is one of the many microbial hydroxylases involved in lignin catabolism. The catalytic cycle of PHBH consists of two half-reactions (Fig. 2). In the reductive half-reaction, the oxidized FAD prosthetic 1 Abbreviations used: PHBH, p-hydroxybenzoate hydroxylase; pOHB, p-hydroxybenzoate; KMO, kynurenine 3-monooxygenase; MICAL, Molecule Interacting with CasL; EGCG, epigallocatechin-3-gallate; CH, calponin homology; CRMP, collapsing response mediator protein; FMO, flavin-dependent monooxygenase.

28

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

group of the enzyme–pOHB complex is reduced by NADPH, and NADP dissociates. The oxidative half-reaction is comprised of the reaction of O2 with the reduced enzyme–pOHB complex, oxygenation of the substrate, and regeneration of the oxidized flavin. A steady-state kinetic analysis showed that pOHB and NADPH bind to the oxidized enzyme in a random order, and an irreversible step occurs before the reaction of O2, giving a bi-uni uni-bi ping-pong kinetic mechanism [8]. The complexities of the chemistry and regulation of PHBH and other hydroxylases are largely invisible to steady-state kinetics. Stopped-flow investigations on the half-reactions, which take advantage of changes in the flavin spectrum, have illuminated the intricacies of catalysis in detail that is far beyond the reach of steady-state kinetics.

Reductive half-reaction The reductive half-reaction has been studied in stopped-flow experiments by mixing anaerobic solutions of the oxidized PHBH–pOHB complex with anaerobic solutions of NADPH; the changes to the flavin absorbance spectrum allow events to be followed in detail. An oxidized flavin–NADPH charge-transfer complex forms rapidly [9], followed by the transfer of the proR hydride of NADPH to N5 of FAD [10]. The reduced enzyme–pOHB–NADP complex produced by the hydride transfer has a charge-transfer absorbance, which disappears after NADP dissociates.

The reductive half-reaction not only fulfills its chemical function of providing the reduced FAD needed to synthesize the flavin hydroperoxide in the oxidative half-reaction, it is also the key point for regulating the catalytic cycle and preventing harmful NADPH oxidase activity. The reaction rate between NADPH and the enzyme is stimulated dramatically – 105-fold – when pOHB is bound to the enzyme even though pOHB plays no chemical role in the reaction [8]. This effect is entirely kinetic; the reduction potential of the enzyme, 163 mV, does not change when pOHB binds. Furthermore, the kinetics of pOHB binding are markedly altered upon flavin reduction [11]. When oxidized, the enzyme binds pOHB too rapidly to be observed in stopped-flow experiments, placing a lower limit of 106 M1 s1 on the rate constant for association [12]. In contrast, the association and dissociation are very slow with reduced enzyme, with rate constants of 160 M1 s1 and 3.4  103 s1, respectively. Again, this is a kinetic effect – the affinity of pOHB is essentially unchanged for the oxidized and reduced enzymes. These two kinetic devices – the linkage of rapid reduction to pOHB binding, and the squelching of pOHB dissociation – ensure that PHBH is not reduced wastefully. pOHB must be present and after reduction, pOHB is kinetically trapped at the active site, available for hydroxylation in the oxidative half-reaction. The mechanism controlling the stimulation by pOHB of reduction is at least partly understood. Crystal structures of PHBH show

Fig. 2. The catalytic cycle of PHBH. The enzyme commits to catalysis in the reductive half-reaction, which is rapid only when pOHB is bound. pOHB reacts in the oxidativehalf-reaction by an electrophilic aromatic substitution. Note that the initial non-aromatic hydroxylation product is not observed, presumably for kinetic reasons. The elimination of H2O2 from the hydroperoxide is undetectable in the wild-type enzyme but can be significant with poor substrates or mutant enzymes.

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

that the isoalloxazine can adopt two conformations (Fig. 3). In one conformation it is mostly buried within a crevice in the protein placing C4a close to pOHB, where it must be if it is to form the hydroperoxide and hydroxylate the substrate. The N5 position of the flavin in this conformation – the ‘‘in’’ conformation – is inaccessible to solvent and NADPH. Flavin reduction is made possible when the isoalloxazine swings like a pendulum 30° out of it protected crevice to the more exposed ‘‘out’’ conformation [13–16]. The movement of the flavin from ‘‘in’’ to ‘‘out’’ is initiated by the binding of NADPH, but also requires pOHB. The presence of pOHB is sensed by an internal proton-transfer network [17,18]. The phenolic oxygen of pOHB is within hydrogen-bonding distance to the side-chain of Tyr 201, which hydrogen-bonds to Tyr 385, two internal water molecules, and finally His 72, which communicates with the surface approximately 12 Å from pOHB [19]. This hydrogenbond network and the positive electrostatic potential (partly due to a helix dipole) near pOHB and the isoalloxazine of the ‘‘in’’ conformation lower the pKa of the phenolic proton of pOHB by 2 units [11,20,21]. Deprotonation is thought to be coupled to isoalloxazine movement (Fig. 4). Generating the phenolate of pOHB causes electrostatic repulsion with the peptide carbonyl of Pro 293, and this torque is transmitted through the protein backbone, moving Asn 300, a residue hydrogen-bonded to the C2-carbonyl of FAD, freeing the flavin for movement to the ‘‘out’’ position [22]. Reduction of the flavin generates the anionic hydroquinone, which is thought to be drawn back into the active site by the positive electrostatic potential [23]. Coincident with this is the reprotonation of the phenolate of pOHB by movement of protons along the

Fig. 3. Flavin conformations and the proton-transfer network of PHBH. The ‘‘in’’ and ‘‘out’’ conformations, taken from 1pbe and 1dod, are indicated. The hydrogen-bond network is also indicated. It acts as a sensor for pOHB in the reductive half-reaction, allowing the flavin to move to the ‘‘out’’ conformation, and deprotonates pOHB to enhance its nucleophilicity in the oxidative half-reaction.

29

proton-transfer network [18]. The observation that the phenolate of tetrafluoro-p-hydroxybenzoate also stimulates reduction is explained by this model [24]. This complex model is based on a large number of experiments; only some highlights can be mentioned here. The ‘‘in’’ and ‘‘out’’ flavin conformations observed in crystal structures correlate with spectral changes caused by ligand binding, providing a useful tool with millisecond time-resolution for tracking the flavin position [14]. Movement from ‘‘in’’ to ‘‘out’’ is barely observable in the first few milliseconds after mixing the wild-type enzyme with NADPH because the subsequent reduction reaction follows rapidly. The two processes are resolved by the 10-fold isotope effect on the hydride transfer when deuterated NADPH is used [17]. Spectral changes corresponding to the movement of the flavin from ‘‘in’’ to ‘‘out’’ were also observed in a number of altered PHBH systems, including site-directed mutants and enzyme substituted with 6azido FAD. Irradiation of 6-azido FAD produces N2 and a reactive nitrene. When the flavin is ‘‘in’’, the nitrene inserts into Pro 293, covalently fixing the isoalloxazine. Enzyme with the isoalloxazine fixed in the ‘‘in’’ conformation takes days to react with NADPH, emphasizing how important the movement to the ‘‘out’’ position is for reduction [25]. A number of experiments demonstrate the coupling of proton movement to the conformational changes of the isoalloxazine. The rate constant for flavin reduction increases with pH and is controlled by the pKa of the phenolic oxygen of pOHB [17]. Reduction reactions performed in the absence of buffer but in the presence of a pH-indicator dye showed that at pH 6.1 (below the pKa of bound pOHB), the PHBH–pOHB complex ejects a proton upon NADPH binding [18]. Flavin reduction is accompanied by proton uptake, whose rate is determined by hydride transfer, as demonstrated by kinetic isotope effects. Disrupting the proton-transfer network by mutagenesis prevents the phenolic proton of pOHB from equilibrating rapidly with bulk solvent. The His72Asn mutant enzyme was particularly instructive. The population of enzyme with the phenolate of pOHB bound reacts slightly more rapidly than wild-type enzyme with a rate constant that is independent of pH [17]. Isotope effects showed that hydride transfer determined the rate of flavin reduction [18]. In contrast, the population of enzyme that had the phenolic form of pOHB bound reacted much slower with NADPH. The substrate isotope effect was abolished. The rate constant had a significant solvent isotope effect, indicating that proton movement – presumably coupled to attaining the ‘‘out’’ conformation – determined the rate of flavin reduction. The importance of generating an anion on the aromatic ligand is illustrated by the behavior of p-aminobenzoate, an isosteric analog of pOHB. While p-aminobenzoate binds to PHBH with the same affinity as pOHB, it does not stimulate flavin reduction by NADPH, nor can it be deprotonated to the amide anion. Interestingly, the rate constant of flavin reduction of the Pro293Ser mutant

Fig. 4. The coupling of pOHB ionization and flavin conformation in the reductive half-reaction. Deprotonation of the phenolic oxygen of pOHB generates an anion which triggers flavin movement from the ‘‘in’’ to the ‘‘out’’ conformation, allowing hydride transfer. The anionic reduced flavin is drawn back to the ‘‘in’’ conformation by electrostatics, accompanied by reprotonation of pOHB and dissociation of NADP.

30

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

enzyme is independent of pH, presumably because the loss of rigidity of the protein backbone does not couple the phenolatebackbone carbonyl repulsion into a torque transmitted to Asn 300 [22]. It is not yet known how NADPH binding triggers flavin movement from the ‘‘in’’ to the ‘‘out’’ conformation. Structures of pyridine nucleotide have been elusive. A structure (1k0j.pdb) was obtained by soaking crystals of the Arg220Gln mutant enzyme in the absence of pOHB with a very high concentration of NADPH [26]. The resulting complex likely had reduced FAD and NADPH, but this was not verified. The pyridine nucleotide was bound in an elongated conformation. The AMP moiety was well-defined by electron density, but the nicotinamide-end of the molecule was not, indicating high mobility. Importantly, the isoalloxazine and the dihydronicotinamide moieties are not in a position to react. A plausible model of a reactive complex was built by constraining the adenosine moiety to its well-defined pocket, replacing the FAD of the structure with the flavin in the ‘‘out’’ conformation obtained in a wild-type structure, and curling the pyridine nucleotide so that the proR hydride of the dihydronicotinamide was over the re-face of the isoalloxazine and 4 Å from N5. This sandwich complex is unusual but not unprecedented [27]. Several protein residues that interact with the pyridine nucleotide in this model, such as Tyr 38 and His 162, had been implicated by previous studies using mutagenesis and chemical modification, but the previously proposed NADPH complexes differed markedly from the one based on the Arg220GlnNADPH structure [28,29]. The structure of PHBH was the first of a flavin hydroxylase to be determined. Since then, several other structures of aromatic hydroxylases have been determined and movement of the flavin between ‘‘in’’ and ‘‘out’’ conformations is proving to be a common feature. Interestingly, while flavin reduction in all aromatic hydroxylases is stimulated by the aromatic substrate (though usually not as much as 105-fold), the proton-transfer network of PHBH is unique – other hydroxylases do not have an analogous network and therefore must use other regulatory mechanisms to control the position of the flavin. These mechanisms remain to be uncovered. In PHBH, it has been suggested that the elaborate mechanism for sensing pOHB prevents the consumption of p-aminobenzoate, which is a precursor to folates [17]. Besides its involvement in reduction, flavin movement is also thought to be a component of the process for binding or releasing aromatic ligands. The pOHB binding site is inaccessible to solvent and a conformational change of some type is needed [14,26]. In support of this, photochemically linking the flavin in the ‘‘in’’ conformation slows pOHB binding markedly [25]. The change in the kinetics of ligand binding and release between oxidized and reduced enzyme might also be attributed to flavin movement; presumably it is more difficult to move the anionic reduced isoalloxazine out of the positive electrostatic field centered on the ‘‘in’’ conformation. Oxidative half-reaction After the reductive half-reaction, the reduced flavin is in the ‘‘in’’ conformation and N5 is inaccessible to solvent. pOHB is kinetically trapped in a site close to C4a, and its phenolic oxygen is part of the proton-transfer network. With these pieces in place, O2 reacts rapidly with the reduced flavin with a rate constant of 3.6  105 M1 s1 [12]. Although computational studies occasionally suggest that the protein guides O2 to a pocket near the flavin in monooxygenases (and oxidases) [30], there is no experimental evidence for oxygen binding. The putative semiquinone–superoxide intermediate has not been observed. Molecular modeling suggests that O2 reacts to form superoxide in a small pocket over the C4a–C10a of the isoalloxazine that contains water in many crystal structures [31]. This pocket is in the midst of a positive electrostatic potential, which would

promote the formation of superoxide [19]. It is also the binding site identified crystallographically for bromide [14]; monovalent anions such as halides and azide inhibit the oxidative half-reactions of aromatic hydroxylases [12,32]. Superoxide in the anion-binding pocket over the semiquinone would be ideally situated for C-O bond formation, making the flavin peroxide [20]. The distal oxygen, situated near C10a, would be protonated by solvent, which has access to the edge of the anion-binding pocket that is opposite N5. Rotation of the hydroperoxide towards pOHB (180°) positions the distal oxygen for reaction. The mechanism of oxygen-transfer to the aromatic substrate was the subject of intense debate for decades and drove much of the early interest in the mechanism of PHBH. Flavoproteins only hydroxylate aromatic substrates that have electron-donating substituents, suggesting that the reaction is an electrophilic aromatic substitution with the hydroperoxide acting as the electrophile, producing a C4a-hydroxyflavin [33]. However, there was doubt that the hydroperoxide was a powerful enough electrophile, prompting alternative proposals, including formation of a ringopened carbonyl oxide derived from the hydroperoxide as the actual electrophile, and a homolytic reaction in which a hydroxyl radical-equivalent was transferred to form a flavinoxy radical-oxygenated substrate radical intermediate pair [12,34–36]. Studies on the reaction using alternate substrates, pH-dependencies, and artificial flavins, by stopped-flow and pulse-radiolysis, drove this debate. Ultimately, all the evidence is consistent with a simple electrophilic aromatic substitution. A linear free energy relationship using PHBH reconstituted with artificial flavins substituted at the 8-position showed that the rate constant of hydroxylation increased as the predicted pKa of the hydroxyflavin decreased [35]. The slope of the Brønsted plot suggested a charge of 0.4 in the transition state. Aromatic substrates act as nucleophiles in flavoprotein hydroxylation reactions. In accord with classic electrophilic aromatic substitutions from organic chemistry, the site of enzymatic hydroxylation is always ortho – or para – to an electron-donating ring substituent. The immediate hydroxylation product in an electrophilic substitution is non-aromatic. Oxygen transfer to the 3-position of pOHB creates an sp3-hybridized carbon, and the activating –OH group becomes a carbonyl. Unless the phenolic proton of pOHB can be removed, hydroxylation to form the very unstable protonated carbonyl will not occur. However, solvent does not have access to pOHB – the active site is buried to prevent access to N5 of the flavin, which would enable the elimination of H2O2 from the hydroperoxide. The proton-transfer network (Tyr 201, Tyr 385, two water molecules, and His 72; Fig. 3), described above for its role in regulating the reductive half-reaction, solves the chemical problem of deprotonating pOHB in the oxidative halfreaction [11,37]. Thus the Tyr201Phe mutant enzyme forms the hydroperoxide like the wild-type enzyme but is unable to hydroxylate pOHB effectively and mostly eliminates H2O2. This also indicates that the protein protects the hydroperoxide to a limited extent – rapid hydroxylation of the substrate is necessary to avoid waste. The immediate non-aromatic hydroxylation product tautomerizes to the aromatic product – carbon-3 is deprotonated and the carbonyl oxygen is protonated as it becomes an –OH. The non-aromatic intermediate has not been observed during pOHB hydroxylation, presumably because the rate constant for tautomerization is much larger than the rate constant for hydroxylation. However, PHBH also hydroxylates 2,4dihydroxybenzoate and p-aminobenzoate, although not efficiently [12]. An extra intermediate with a high extinction coefficient is observed in both cases. It is now believed that these spectra are the sum of the conjugated non-aromatic hydroxylation [35] product and the flavin C4a-hydroxide produced by

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

O–O bond scission and that the aromatization is slow, allowing the detection of the non-aromatic intermediate. Substrate hydroxylation produces the C4a-hydroxyflavin, which must eliminate water in order to regenerate oxidized isoalloxazine. Such an elimination would be analogous to the elimination of H2O2 from the hydroperoxide. As mentioned above, the protein of PHBH shields N5 from the solvent, preventing the elimination of C4a adducts. It is likely that the solution to this problem is the movement of the flavin to the solvent-exposed ‘‘out’’ conformation where N5 can be deprotonated and the OH-leaving group protonated. This has not been demonstrated directly, but explains why high levels of pOHB (or other aromatic substrates) inhibit PHBH [38]. Exceedingly high levels of pOHB trap the enzyme as the hydroxyflavin intermediate, which is only possible if 3,4-dihydroxybenzoate vacates the aromatic binding site while the hydroxyflavin is intact. Since aromatic ligand release and binding appear to require flavin movement, exposure of the hydroxyflavin to solvent in the ‘‘out’’ conformation seems likely. Substrate binding pushes the equilibrium to the stable complex with the flavin in the ‘‘in’’ conformation. Inhibition by the aromatic substrate is common among the aromatic hydroxylases, suggesting that flavin movement is usually a component of ligand-exchange and the discharge of the hydroxyflavin. Activation of an aromatic by an – NH2 – kynurenine 3monooxygenase Kynurenine, an aromatic amine, is hydroxylated to 3-hydroxykynurenine by kynurenine 3-monooxygenase (KMO) in a pathway that degrades tryptophan. In mammals, 3-hydroxykynurenine is a precursor to NAD and is intimately involved in neurochemistry. Inhibitors of KMO have potential uses for treating strokes and Huntington’s disease. Mammalian KMOs are membrane-bound and have been too difficult to purify from tissue or to express in bacteria to allow enzymological studies. However, a soluble homolog was found in Pseudomonas fluorescens, providing a tractable model system to study important medicinal chemistry [39]. KMO uses NADPH as its reducing substrate. The reductive halfreaction follows the outline established for PHBH. Although an oxidized enzyme–NADPH charge-transfer complex is not observed before the hydride transfer, after the reaction a reduced enzyme–NADP complex is prominent. Bound kynurenine stimulates the rate of reduction by 2500-fold although the reduction potential of the enzyme (188 mV) does not change in the presence of ligands. It is very likely that flavin movement is an important part of the mechanism of stimulation of reduction by ligands, although structures of the enzyme are not yet available. Deprotonating the amino group of kynurenine to the amide anion would be too unfavorable to allow the ionization of the ligand to trigger the events leading to reduction. Interestingly, benzoylalanine (2-amino-4-oxo-4-phenylbutanoate) and nitrobenzoylalanine (2-amino-4-(3-nitrophenyl)-4-oxobutanoate), two non-hydroxylatable ligands, also stimulate the reaction of KMO with NADPH, though not to the extent that the substrate does [39]. This suggests that other features besides an activating substituent on the aromatic ring are sensed for reduction. The reaction of the reduced enzyme with O2 (k = 1.5  105 M1 s1) forms a flavin hydroperoxide. The hydroxyflavin is not observed – the hydroperoxide appears to convert directly to oxidized flavin, even though kynurenine is hydroxylated – indicating that the rate constant for oxygen transfer (5.5 s1) is much lower than that of water elimination (Fig. 5). The final reaction of the oxidative half-reaction is the release of the product from the oxidized enzyme (1.9 s1) and is the rate-determining step in catalysis. It is accompanied by large spectral changes which suggest flavin movement. Interestingly, when the non-reactive ligand benzoylalanine is bound to the enzyme during the reaction with O2, the hydroperoxide forms

31

and eliminates H2O2, and the final spectral change that normally accompanies ligand-exchange also occurs, despite the fact that the bound and free ligands are identical. This shows that the oxidized enzyme directly produced by the elimination of the C4a-adduct is different – probably in conformation – than the resting oxidized enzyme. The examination of the mechanism of KMO accomplished so far demonstrates the impact that detailed kinetic and spectral studies can have on biomedical problems. Analogs of kynurenine such as benzoylalanine and m-nitrobenzoylalanine inhibit the production of 3-hydroxykynurenine, a desirable therapeutic effect in some diseases, and therefore, have been considered as drug leads. However, as mentioned above, these compounds are also effectors of the reduction of KMO but are not hydroxylated. Instead, these compounds lead to the production of H2O2 – hardly a desirable therapeutic effect [39]. MICAL In order for the nervous system to develop correctly, axons are guided by a number of signaling proteins. Cues from semaphorins, such as Sema-1a, which binds to PlexA, a plexon guidance receptor, cause the collapse of neuronal growth cones and actin depolymerization by an unknown mechanism, halting axon growth. Molecule Interacting with CasL (MICAL) is a newly discovered aromatic hydroxylase activated by plexins and necessary for the guidance of neuronal processes [40]. MICALs are large cytosolic proteins (118 kDa) composed of more than 1000 amino acids. Discovered in 2002 by Suzuki et al. [41], MICAL-1 was identified as a molecule that interacts with the CasL SH3 domain by Far Western screening in hematopoietic cells. MICAL is evolutionarily conserved. A single gene has been identified in Drosophila while in vertebrates (humans and rodents), three genes have been identified (MICAL-1, MICAL-2 and MICAL-3) [40–42]. Various tissues express MICAL such as the lung, thymus, spleen, testis and brain [40,41]. However, most studies have centered on MICAL expression in the developing and/or mature central nervous systems of Drosophila or rat. In studies involving rodents, MICAL-1 and MICAL-3 are broadly expressed in the embryonic and developing nervous system while MICAL-2 increases in expression as the rat matures into adulthood. Interestingly, MICAL-2 is not present in several brain structures such as the hypothalamus and the striatum [43]. MICALs have also been found to interact with several proteins such as CasL, vimentin, rab1 and plexinAs [40– 42,44]. Kolodkin and colleagues demonstrated that semaphorins require the hydroxylase domain to develop a functional nervous system by mutating the three glycine residues of the GXGXXG motif of the FAD-binding domain to tryptophan residues. Drosophila with this mutation had axon motor guidance defects [40]. A main catechin contained in green tea, (-)-epigallocatechin-3-gallate (EGCG), has been suggested to be a general inhibitor of hydroxylases based on experiments performed using PHBH (noncompetitive) [45] and squalene epoxidase (competitive) [46]. EGCG also inhibits in vitro the semaphorin-mediated growth-cone collapse in rat dorsal root ganglia [40]. MICALs contain five domains, the largest being an N-terminal flavoprotein monooxygenase domain of 500 amino acids, which is highly conserved and binds the FAD prosthetic group (Fig. 6). The structure of the hydroxylase domain of Mus musculus MICAL1 has been determined [47,48]. It is highly homologous to PHBH and other aromatic hydroxylases, with motifs present such as the consensus GXGXXG motif (the Rossmann fold), the conserved GD sequence of hydroxylases important for binding the ribose moiety of FAD [40], as well as the conserved DG motif, involved in binding pyrophosphate moiety of FAD. Sedimentation velocity experiments

32

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

Fig. 5. The hydroxylation of kynurenine. Oxygen transfer from the flavin hydroperoxide of KMO first forms the hydroxyflavin-non-aromatic intermediate, which decays rapidly to the oxidized flavin-product complex.

Fig. 6. Domain organization of MICAL. The N-terminal monooxygenase domain is very similar to PHBH. The other four domains are probably involved in signal-transduction.

demonstrated that the flavoprotein domain is a monomer [48], which is different from most aromatic hydroxylases, which are most often dimers or tetramers. The isoalloxazine of MICAL adopts ‘‘in’’ and ‘‘out’’ conformations similar to PHBH, depending on the oxidation state of the flavin. When MICAL was reduced, the flavin adopted the ‘‘in’’ conformation and when oxidized it adopted the ‘‘out’’ conformation [47]. The other domains of MICAL comprise the C-terminus of the protein. These domains are the calponin homology (CH) domain, followed by a LIM domain, a proline-rich region (PPKPP) for Src homology 3 (SH3) domain, and a coiled-coil region. These types of domains are generally important for signal transduction, protein–protein interactions and cytoskeletal organization. MICAL interacts with the CasL SH3 domain through the proline-rich PPKPP sequence on the C-terminus [41]. There are two regions on MICALs, the first right after the CH domain and the second after the LIM domain, which, based on sequence analysis, cause the variance in size of the three MICAL isoforms [40]. Catalysis by MICAL has barely been studied. The hydroxylase domain has high NADPH oxidase activity which is inhibited by EGCG [48], and lower NADH oxidase activity. NADPH oxidase activity is very unusual for an aromatic hydroxylase suggesting that the separate domain lacks the usual regulatory mechanism that inhibits reduction in the absence of the aromatic substrate. The aromatic substrate, if it exists, has not been identified. It has been suggested that MICAL actually is an NADPH oxidase, and the H2O2 produced causes actin depolymerization. Alternatively, a signaling protein (CRMP – collapsing response mediator protein) has been suggested to be a recruiting protein for the substrate of MICAL. CRMP was found to alter the NADPH oxidase activity of MICAL using con-

structs that were constitutively active. Constructs consisting of the hydroxylase-CH-LIM domains in the presence of CRMP produced less hydrogen peroxide. The co-expression of CRMP with full-length MICAL did not inhibit the production of hydrogen peroxide [49]. The high oxidase activity of the hydroxylase domain suggests that one of the four missing domains of full-length MICAL regulates reduction. Non-aromatic hydroxylases – ‘‘Bold’’ monooxygenases Cyclohexanone/Baeyer–Villiger monooxygenases There are many monooxygenases that use flavin peroxides as nucleophiles in Baeyer–Villiger oxidations of ketones to esters. A few related microbial enzymes have been studied mechanistically and structurally, illustrating a strategy for avoiding wasteful NADPH oxidase activity that is very different from that employed by aromatic hydroxylases. The catalytic cycle consists of a reductive half-reaction and an oxidative half-reaction (Fig. 7). In the reductive half-reaction, NADPH binds to the enzyme and reduces the FAD prosthetic group with the proR hydride and NADP remains tightly bound to the enzyme [50]. The presence of a ketone makes no difference in the reaction rate – flavin reduction is not controlled by its substrate. The reduced enzyme–NADP complex reacts with oxygen quite rapidly (>106 M1 s1) to form the flavin C4a-peroxide. In the absence of a ketone, this adduct is remarkably stable. For instance, at pH 7.2, 4 °C, the intermediate formed on cyclohexanone monooxygenase from Acinetobacter decays to oxidized enzyme and H2O2 with a half-life of 5 min. In contrast, the flavin hydroperoxides of

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

33

Fig. 7. The catalytic cycle of a generic Baeyer–Villiger monooxygenase. Pyridine nucleotide does not dissociate after flavin reduction. Its presence stabilizes the flavin hydroperoxide, which oxidizes ketones.

aromatic hydroxylases without substrate generally decay in milliseconds. The stability of the C4a-adduct of Baeyer–Villiger monooxygenases depends on bound NADP. In its absence, oxidized enzyme forms within a second [50–52]. O2 reacts with the reduced flavin of the NADP complex to make deprotonated flavin peroxide. This is the nucleophilic form of the flavin–oxygen intermediate, required for a Baeyer–Villiger reaction (Fig. 8). Protonation of the peroxide to the hydroperoxide was observed at pH 7.2 with a large shift in the absorbance maximum from 366 to 383 nm and a rate constant of 3.3 s1 [51]. This is quite slow compared to proton-transfer reactions of small molecules in aqueous solution. Double-mixing pH-jump stopped-flow experiments were used to observe the conversion of the hydroperoxide to the peroxide by first aging the intermediate formed by reaction with O2 at pH 7.2 in the first mix and then mixing the hydroperoxide with buffer at pH 9.0; deprotonation, observed as a shift in the absorbance maximum from 383 to 366 nm, occurred with a rate constant of 4.7 s1. A value of 8.4 was found for the pKa of the hydroperoxide. The peroxide, not the hydroperoxide, is the nucleophile that attacks the carbonyl of the substrate. This was demonstrated in double-mixing stopped-flow experiments at 4 °C. The flavin peroxide, formed rapidly at pH 7.2, was allowed to age for various times

before being mixed with cyclohexanone in buffer at pH 9.0. When the flavin–oxygen adduct was aged for only a short time, mixing with ketone resulted in most of the flavin oxidizing rapidly (110 s1) as the peroxide was consumed in the Baeyer–Villiger reaction; neither the Criege intermediate nor the hydroxyflavin were observed, presumably because the rate constants for their reaction were much greater then that of nucleophilic attack. In contrast, aging the initial peroxide at pH 7.2 allows it to become protonated; subsequent mixing with cyclohexanone in pH 9.0 buffer results in the formation of oxidized flavin with a rate constant of 5 s1, indicating that deprotonation of the hydroperoxide is determining the rate of the reaction. At intermediate age-times, the enzyme has a mixture of the flavin peroxide and hydroperoxide, and the reaction of each species is observed. After flavin oxidation, spectral changes ascribed to the dissociation of e-caprolactam and NADP were observed. Separate experiments showed that NADP binds to oxidized enzyme in a two-step process. The reverse rate constant is about the same as that measured for proton transfers to the oxygen adducts, suggesting that the same conformational change controls access to the active site [51]. Structures of related Baeyer–Villiger monooxygenases – phenylacetone monooxygenase from Thermobifida fusca and cyclohexanone monooxygenase from Rhodococcus – reveal the origin of the

Fig. 8. Substrate oxidation by cyclohexanone monooxygenase. The flavin peroxide anion formed by the reaction of O2 with the reduced enzyme can be protonated slowly to the hydroperoxide (pKa of 8.4), or it can react with cyclohexanone to form the Criege intermediate, which collapses to e-caprolactam and hydroxyflavin, followed by the elimination of water. Note that NADP must be bound during the reaction to protect the C4a-hydroperoxide but is not shown for the sake of clarity.

34

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

kinetic behavior of these enzymes [53,54]. Baeyer–Villiger monooxygenases are not closely related to aromatic hydroxylases, but more closely resemble disulfide oxidoreductases. The active site lies in a crevice between two domains – an NADPH binding domain inserted into an FAD-binding domain – joined by a signature-sequence for Baeyer–Villiger monooxygenases [55]. The structures of free oxidized enzyme (phenylacetone monooxygenase) [53] and NADP complexes (cyclohexanone monooxygenase) show that conformational changes of residues at the active site and between domains control the catalytic cycle [54]. After hydride transfer from NADPH, domain rotation moves the nicotinamide ring to a position that blocks access to N5 of FAD, thus protecting the hydroperoxide from elimination after the reaction of O2. FMO Mammalian microsomes contain a flavin-dependent monooxygenase (FMO) that hydroxylates a very broad range of substrates. FMOs in mammals are key detoxification enzymes responsible for the clearance of many xenobiotics and drugs [56]. FMO-homologs have also been found in plants, yeast, and bacteria. In plants, FMOs are involved in the biosynthesis of auxins and in defense against pathogens [57]. Microbial FMOs have been suggested to catalyze disulfide bond formation during protein folding and in natural product biosynthesis [58]. The pig liver FMO has been studied most extensively, and the broader family seems to behave similarly. The molecular weight of a subunit is 59 kDa and the enzyme purifies as a tetramer or octamer with bound lipid. The kinetic mechanism resembles that of the Baeyer–Villiger monooxygenases (Fig. 9). The catalytic cycle is composed of reductive and oxidative half-reactions. Two populations of enzyme were detected in the reductive half-reaction in stopped-flow experiments [59]. NADPH is the preferred substrate, with rate constants for flavin reduction of 2.5 and 0.9 s1. NADH also reduces the enzyme with the same maximal rate constants but binds 20-fold weaker. The presence of a substrate does not change the rate of reduction. There is a 6-fold primary deuterium isotope effect on both phases of flavin reduction by NADPH labeled at the 4-proR position. NADP remains tightly bound to the enzyme after flavin reduction as a charge-transfer complex with weak long-wavelength absorbance. The reduced enzyme–NADP complex reacts with O2 with a bimolecular rate constant of 2000 M1 s1 at 15 °C – a very low value for a monooxygenase [60]. The hydroperoxide formed by the reaction is very stable, taking 2 h at 4 °C to decay to oxidized enzyme. This long-lived hydroperoxide reacts with many different substrates, forming a C4a-hydroxyflavin whose elimination of water is the rate-determining step in catalysis. Most substrates are soft nucleophiles – amines, hydroxylamines, thiols, thioethers, disulfides, sulfoxides, and even iodide [56] – but boronates [61]

and aldehydes [62] – electrophiles – can also be hydroxylated. The variety of substrates that are accepted is vast; a recent review listed steady-state kinetic parameters for about two hundred compounds [63]. Obviously, the structural requirements for a compound to be a substrate are not stringent; hydrophobic compounds are favored over charged compounds. This is consistent with the physiological role of FMO as a scavenger of xenobiotics; hydrophobic foreign compounds can penetrate the non-polar cell membranes. Interestingly, primary amines are not substrates, while tertiary amines are better substrates than secondary amines. This trend is opposite to that seen with hydroxylations in solution by a flavin hydroperoxide model compound. Structures of two soluble microbial FMOs (from Schizosaccharomyces pombe [64] and Methylophaga sp. [65] strain SK1) explain how NADP stabilizes the flavin hydroperoxide (Fig. 10). The structures resemble the Baeyer–Villiger monooxygenases. They have an FAD-binding domain and an NADP(H) binding domain. NADP binds in a crevice between the domains and blocks access of the solvent to the flavin. The amide nitrogen of the nicotinamide ring of NADP is within hydrogen-bonding distance of N5 and the carbonyl oxygen at the 4-position. C4 of NADP is too far from N5 for hydride transfer. For the Methylophaga enzyme, this was interpreted as indicating that hydride transfer to the flavin occurred in another conformation, attained by a shift between domains, and the complex whose structure was determined would protect the hydroperoxide by blocking access to N5 [65]. A model of the flavin hydroperoxide in the structure suggests that the 20 -OH of the ribose of NADP could form a hydrogen-bond to the hydroperoxy moiety, stabilizing it further. Although the structures of the S. pombe enzyme were badly misinterpreted mechanistically, they agree completely with the analysis of the Methylophaga enzyme [64]. Without NADP, N5 of the flavin hydroperoxide would be accessible to solvent, allowing rapid elimination of H2O2. In fact, when reduced enzyme without NADP is mixed with oxygenated buffer, a direct conversion from reduced to oxidized enzyme, without a hydroperoxide intermediate, is observed [60].

Two-component systems The most extensively studied monooxygenases, like those described above, accomplish all their chemistry with one polypeptide and use a single flavin as a prosthetic group. However, not all flavin hydroxylases are as concise as the single-component enzymes – two-component systems, consisting of two separate enzymes, are now becoming widely known. In these systems, the flavin is a substrate rather than a prosthetic group. It is reduced by a reductase and then hydroxylates the substrate after transferring to a separate monooxygenase. The reductase itself may be a flavoprotein whose prosthetic group is reduced by pyridine nucleotide and oxidized by

Fig. 9. The catalytic cycle of FMO. Like the Baeyer–Villiger monooxygenases, NADP remains bound to FMO after the reductive half-reaction (not shown in the oxidative halfreaction for the sake of clarity). Unlike the Baeyer–Villiger monooxygenases, FMOs usually hydroxylate nucleophilic substrates.

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

35

Fig. 10. Stereo-view of the FMO–NADP complex (2vqb.pdb). NADP (blue carbons) is bound over the re-face of the flavin (yellow carbons) and partially blocks the approach to N5.

a flavin substrate – called Class II reductases – or it may be a simple polypeptide that catalyzes the reaction of pyridine nucleotide and flavin substrates – Class I reductases [66]. There are two-component systems that use the flavin hydroperoxide as an electrophile while others use it as a nucleophile; some use FMN, while others use FAD. An extensive classification of monooxygenases has been reported [67]. The number of components appears to be irrelevant to the details of the chemistry of the catalytic cycle. However, the involvement of two separate proteins raises an issue in addition to those discussed above for single-component systems – the mechanism of the transfer of the flavin from the reductase to the monooxygenase. Two mechanisms are possible – either the reductase could release reduced flavin to solution and it could diffuse to the monooxygenase, or reduced flavin could be passed directly from the reductase to the monooxygenase in a complex of the two enzymes. It has been argued that the direct transfer of reduced flavin is needed in order to protect it from the non-enzymatic reaction with O2, which clearly must be present during monooxygenase function [66]. However, it has also been argued that the relatively slow reaction of O2 with free flavin, and the complex autocatalytic chain-reactions between oxidized, semiquinone, and hydroquinone flavins and O2 and superoxide, are still slow relative to the binding of reduced flavin to a monooxygenase [68]. Data have been published on several two-component systems consistent with either viewpoint, and it is probable that there are enzyme systems using each mechanism. Bacterial luciferase is a two-component system responsible for the bioluminescence in the sea. The monooxygenase uses reduced FMN to oxidize long-chain aldehydes to carboxylic acids in a reaction that is formally similar to a Baeyer–Villiger reaction. However, the important biological product of the reaction – light – is generated by the excited state of the flavin-C4a-hydroxide; normal Baeyer–Villiger chemistry does not produce excited states, so a more complex mechanism has been proposed [33]. Regardless, bacterial luciferase and the flavin reductases expressed in luminescent bacteria have been studied as models for flavin channeling [66,69]. The steadystate kinetic mechanism of the oxidation of NADPH by FMN catalyzed by the reductase, and the values of kinetic parameters change when the monooxygenase component is added, arguing in favor of protein–protein interactions. A complex between the reductase and monooxygenase was detected by the fluorescence anisotropy of the reductase labeled with eosin. When diluted sufficiently, the dimeric reductase dissociates to monomers, and

the monomers become available to bind to monooxygenase subunits. Another two-component system, alkanesulfonate monooxygenase, also appears to transfer reduced flavin from its reductase to its monooxygenase in a protein complex. Evidence for this comes from changes in the steady-state kinetics of the reductase activity caused by including the monooxygenase [70], and the co-elution of the monooxygenase with the histidine-tagged reductase from a metal-affinity column [71]. In contrast, a thorough transient kinetic analysis of the ActVA– ActVB system, an aromatic hydroxylase from Streptomyces coelicolor, clearly demonstrated that flavin transfer from reductase to monooxygenase by diffusion is kinetically competent – there is no need to invoke protein–protein complexes [68]. The yield of flavin hydroperoxide formed on the monooxygenase is identical when the reduced FMN–monooxygenase complex is mixed with oxygencontaining buffer and when free reduced flavin is mixed with solutions of monooxygenase and oxygen – reduced FMN is not detectably intercepted by the non-enzymatic reaction with O2. When the reductase-reduced FMN complex is used to supply the monooxygenase with reduced FMN, it is delivered at the rate for the dissociation of reduced FMN determined in the absence of monooxygenase. A curious middle-ground has been proposed for styrene monooxygenase, a two-component system catalyzing the rare flavin-dependent epoxidation of an alkene. Simulations of steady-state kinetics and the non-enzymatic oxidation of reduced FAD required both the direct transfer of reduced flavin from the reductase to the monooxygenase and the transfer via diffusion [72].

Postscript Monooxygenases hydroxylate substrates by synthesizing in situ the highly reactive and unstable flavin hydroperoxide. These enzymes have evolved to speed the reaction of O2 with reduced flavin and preserve the hydroperoxide long enough to ensure substrate hydroxylation while preventing the elimination of H2O2. Two approaches to these goals, both relying on conformational changes, are now understood – either the reaction of NAD(P)H is contingent upon the presence of the substrate to be hydroxylated, or the hydroperoxide is protected from elimination until a substrate is encountered. Similar sophisticated control of a diverse array of reactions by one or two proteins is achieved by few, if any, other enzyme systems.

36

B.A. Palfey, C.A. McDonald / Archives of Biochemistry and Biophysics 493 (2010) 26–36

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

[31] [32] [33]

[34] [35]

V. Massey, J. Biol. Chem. 269 (1994) 22459–22462. T.C. Bruice, Isr. J. Chem. 24 (1984) 54–61. Y.-J. Zheng, R. Ornstein, J. Am. Chem. Soc. 118 (1996) 9402–9408. R.F. Anderson, in: V. Massey, C.H. Williams (Eds.), Flavins and Flavoproteins, Elsevier North Holland, Inc., New York, 1982, pp. 278–283. D.P. Ballou, B. Entsch, L.J. Cole, Biochem. Biophys. Res. Commun. 338 (2005) 590–598. B. Entsch, L.J. Cole, D.P. Ballou, Arch. Biochem. Biophys. 433 (2005) 297–311. B. Entsch, W.J.H. van Berkel, FASEB J. 9 (1995) 476–483. M. Husain, V. Massey, J. Biol. Chem. 254 (1979) 6657–6666. L.G. Howell, T. Spector, V. Massey, J. Biol. Chem. 247 (1972) 4340–4350. D.J. Manstein, E.F. Pai, L.M. Schopfer, V. Massey, Biochemistry 25 (1986) 6807– 6816. B. Entsch, B.A. Palfey, D.P. Ballou, V. Massey, J. Biol. Chem. 266 (1991) 17341– 17349. B. Entsch, D.P. Ballou, V. Massey, J. Biol. Chem. 251 (1976) 2550–2563. H.A. Schreuder, P.A.J. Prick, R.K. Wieringa, G. Vriend, K.S. Wilson, W.G.J. Hol, J. Drenth, J. Mol. Biol. 208 (1989) 679–696. D.L. Gatti, B.A. Palfey, M.S. Lah, B. Entsch, V. Massey, D.P. Ballou, M.L. Ludwig, Science 266 (1994) 110–114. H.A. Schreuder, A. Mattevi, G. Obmolova, K.H. Kalk, W.G.J. Hol, F.J.T. van der Bolt, W.J.H. van Berkel, Biochemistry 33 (1994) 10161–10170. W.J.H. van Berkel, M.H.M. Eppink, H.A. Schreuder, Protein Sci. 3 (1994) 2245– 2253. B.A. Palfey, G.R. Moran, B. Entsch, D.P. Ballou, V. Massey, Biochemistry 38 (1999) 1153–1158. K.K. Frederick, B.A. Palfey, Biochemistry 44 (2005) 13304–13314. G.R. Moran, B. Entsch, B.A. Palfey, D.P. Ballou, Biochemistry 36 (1997) 7548– 7556. H. Shoun, T. Bepu, K. Arima, J. Biol. Chem. 254 (1979) 899–904. W.J.H. van Berkel, F. Müller, Eur. J. Biochem. 179 (1989) 307–314. B.A. Palfey, R. Basu, K.K. Frederick, B. Entsch, D.P. Ballou, Biochemistry 41 (2002) 8438–8446. J. Vervoort, W.J. van Berkel, F. Muller, C.T. Moonen, Eur. J. Biochem. 200 (1991) 731–738. F.J.T. van der Bolt, R.H.H. van den Heuvel, J. Vervoort, W.J.H. van Berkel, Biochemistry 36 (1997) 14192–14201. B.A. Palfey, D.P. Ballou, V. Massey, Biochemistry 36 (1997) 15713–15723. J. Wang, M. Ortiz-Maldonado, B. Entsch, V. Massey, D. Ballou, D.L. Gatti, Proc. Natl. Acad. Sci. USA 99 (2002) 608–613. R.H.H. van den Heuvel, A.H. Westphal, A.J.R. Heck, M.A. Walsh, S. Rovida, W.J.H. van Berkel, A. Mattevi, J. Biol. Chem. 279 (2004) 12860–12867. M.H.M. Eppink, H.A. Schreuder, W.J.H. van Berkel, J. Biol. Chem. 273 (1998) 21031–21039. M.H.M. Eppink, K.M. Overkamp, H.A. Schreuder, W.J.H. van Berkel, J. Mol. Biol. 292 (1999) 87–96. R. Baron, C. Riley, P. Chenprakhon, K. Thotsaporn, R.T. Winter, A. Alfieri, F. Forneris, W.J.H. van Berkel, P. Chaiyen, M.W. Fraaije, A. Mattevi, Proc. Natl. Acad. Sci. USA 106 (2009) 10603–10605. H.A. Schreuder, W.G.J. Hol, J. Drenth, Biochemistry 29 (1990) 3101–3108. P.J. Steennis, M.M. Cordes, J.G.H. Hilkens, F. Müller, FEBS Lett. 36 (1973) 177– 180. B.A. Palfey, V. Massey, Flavin-dependent enzymes, in: M. Sinnott (Ed.), Comprehensive Biological Catalysis, Radical Reactions and Oxidation/ Reduction, vol. III, Academic Press, 1998, pp. 83–154 (Chapter 29). A. Wessiak, L.M. Schopfer, V. Massey, J. Biol. Chem. 259 (1984) 12547–12556. M. Ortiz-Maldonado, D.P. Ballou, V. Massey, Biochemistry 38 (1999) 8124– 8137.

[36] R.F. Anderson, K.B. Patel, B. Vojnovic, J. Biol. Chem. 266 (1991) 13086–13090. [37] B.A. Palfey, B. Entsch, D.P. Ballou, V. Massey, Biochemistry 33 (1994) 1545– 1554. [38] T. Spector, V. Massey, J. Biol. Chem. 247 (1972) 4679–4687. [39] K.R. Crozier-Reabe, R.S. Phillips, G.R. Moran, Biochemistry 47 (2008) 12420– 12433. [40] J.R. Terman, T. Mao, R.J. Pasterkamp, H. Yu, A.L. Kolodkin, Cell 109 (2002) 887– 900. [41] T. Suzuki, T. Nakamoto, S. Ogawa, S. Seo, T. Matsumura, K. Tachibana, C. Morimoto, H. Hirai, J. Biol. Chem. 277 (2002) 14933–14941. [42] T. Weide, J. Teuber, M. Baeyer, A. Barnekow, Biochem. Biophys. Res. Commun. 306 (2003) 79–86. [43] R.J. Pasterkamp, H. Dai, J.R. Terman, K.J. Wahlin, B. Kim, B.S. Bregman, P.G. Popovich, A.L. Kolodkin, Mol. Cell. Neurosci. 31 (2006) 52–69. [44] J. Fischer, T. Weide, A. Barnekow, Biochem. Biophys. Res. Commun. 328 (2005) 415–423. [45] I. Abe, K. Kashiwagi, H. Noguchi, FEBS Lett. 483 (2000) 131–134. [46] I. Abe, T. Seki, K. Umehara, T. Miyase, H. Noguchi, J. Sakakibara, T. Ono, Biochem. Biophys. Res. Commun. 268 (2000) 767–771. [47] C. Siebold, N. Berrow, T.S. Walter, K. Harlos, R.J. Owens, D.I. Stuart, J.R. Terman, A.L. Kolodkin, R.J. Pasterkamp, E.Y. Jones, Proc. Natl. Acad. Sci. USA 102 (2005) 16836–16841. [48] M. Nadella, M.A. Bianchet, S.B. Gabelli, J. Barrila, L.M. Amzel, Proc. Natl. Acad. Sci. USA 102 (2005) 16830–16835. [49] E.F. Schmidt, S.-O. Shim, S.M. Strittmatter, J. Neurosci. 28 (2008) 2287–2297. [50] C.C. Ryerson, D.P. Ballou, C. Walsh, Biochemistry 21 (1982) 2644–2655. [51] D. Sheng, D.P. Ballou, V. Massey, Biochemistry 40 (2001) 11156–11167. [52] D.E. Torrex Pazmino, B-J. Baas, D.B. Janssen, M.W. Fraaije, Biochemistry 47 (2008) 4082–4093. [53] E. Malito, A. Alfieri, M.W. Fraaije, A. Marrevi, Proc. Natl. Acad. Sci. USA 101 (2004) 13157–13162. [54] I.A. Mirza, B.J. Yachnin, S. Wang, S. Grosse, H. Bergeron, A. Imura, H. Iwaki, Y. Hasegawa, P.C.K. Lau, A.M. Berghuis, J. Am. Chem. Soc. 131 (2009) 8848–8854. [55] M.W. Fraaije, N.M. Kamerbeek, W.J.H. van Berkel, D.B. Janssen, FEBS Lett. 518 (2002) 43–47. [56] L.L. Poulson, in: F. Muller (Ed.), Chemistry and Biochemistry of Flavoenzymes, vol. II, CRC Press, Boca Raton, Ann Arbor, London, 1991, pp. 87–100. [57] N.L. Schlaich, TrendsPlantSci. 12 (2007) 1360–1385. [58] H.S. Choi, J.K. Kim, E.H. Cho, Y.C. Kim, J.I. Kim, S.W. Kim, Biochem. Biophys. Res. Commun. 306 (2003) 930–936. [59] N.B. Beaty, D.P. Ballou, J. Biol. Chem. 256 (1981) 4611–4618. [60] N.B. Beaty, D.P. Ballou, J. Biol. Chem. 256 (1981) 4619–4625. [61] K.C. Jones, D.P. Ballou, J. Biol. Chem. 261 (1986) 2553–2559. [62] G.-P. Chen, L.L. Poulsen, D.M. Ziegler, Drug Metab. Disp. 23 (1995) 1390–1393. [63] S.K. Krueger, D.E. Williams, Pharmacol. Ther. 106 (2005) 357–387. [64] S. Eswaramoorthy, J.B. Bonanno, S.K. Burley, S. Swaminathan, Proc. Natl. Acad. Sci. USA 103 (2006) 9832–9837. [65] A. Alfieri, E. Malito, R. Orru, M.W. Fraaije, A. Mattevi, Proc. Natl. Acad. Sci. USA 105 (2008) 6572–6577. [66] S.-C. Tu, Antiox Redox Signal. 3 (2001) 881–897. [67] W.J.H. van Berkel, N.M. Kamerbeek, M.W. Fraaije, J. Biotech. 124 (2006) 670– 689. [68] J. Valton, C. Mathevon, M. Fontecave, V. Niviere, D.P. Ballou, J. Biol. Chem. 283 (2008) 10287–10296. [69] S.-C. Tu, Photochem. Photobiol. Sci. 7 (2008) 183–188. [70] B. Gao, H.R. Ellis, Biochem. Biophys. Res. Commun. 331 (2005) 1137–1145. [71] K. Abdurachim, H.R. Ellis, J. Bacteriol. 188 (2006) 8153–8159. [72] A. Kantz, F. Chin, N. Nallamothu, T. Nguyen, G.T. Gassner, Arch. Biochem. Biophys. 442 (2005) 102–116.