The C-terminal Domain of 4-Hydroxyphenylacetate 3-Hydroxylase from Acinetobacter baumannii Is an Autoinhibitory Domain*

Background: The flavin reduction rate of C1 reductase is enhanced ∼20-fold upon binding to HPA (effector). Results: The truncated C1 variant lacking the C-terminal domain is reduced as fast as the wild-type enzyme in the presence of HPA. Conclusion: The C-terminal domain is an autoinhibitory domain. Significance: These findings demonstrate a novel principle that may be used in a wide variety of oxygenases. p-Hydroxyphenylacetate (HPA) 3-hydroxylase from Acinetobacter baumannii consists of a reductase component (C1) and an oxygenase component (C2). C1 catalyzes the reduction of FMN by NADH to provide FMNH− as a substrate for C2. The rate of reduction of flavin is enhanced ∼20-fold by binding HPA. The N-terminal domain of C1 is homologous to other flavin reductases, whereas the C-terminal domain (residues 192–315) is similar to MarR, a repressor protein involved in bacterial antibiotic resistance. In this study, three forms of truncated C1 variants and single site mutation variants of residues Arg-21, Phe-216, Arg-217, Ile-246, and Arg-247 were constructed to investigate the role of the C-terminal domain in regulating C1. In the absence of HPA, the C1 variant in which residues 179–315 were removed (t178C1) was reduced by NADH and released FMNH− at the same rates as wild-type enzyme carries out these functions in the presence of HPA. In contrast, variants with residues 231–315 removed behaved similarly to the wild-type enzyme. Thus, residues 179–230 are involved in repressing the production of FMNH− in the absence of HPA. These results are consistent with the C-terminal domain in the wild-type enzyme being an autoinhibitory domain that upon binding the effector HPA undergoes conformational changes to allow faster flavin reduction and release. Most of the single site variants investigated had catalytic properties similar to those of the wild-type enzyme except for the F216A variant, which had a rate of reduction that was not stimulated by HPA. F216A could be involved with HPA binding or in the required conformational change for stimulation of flavin reduction by HPA.

Coordination of the transfer of reduced flavin or of electron transfer within or between flavin-dependent enzymes is important for minimizing both wasteful consumption of reducing agents such as NAD(P)H and production of reactive oxygen species such as hydrogen peroxide or superoxide. The control of interdomain electron transfer has been extensively studied for nitric-oxide synthase (NOS) (1) and cytochrome P450 reductase (2). Evidence suggests that these enzymes coordinate and control electron transfer processes via protein conformational changes (3,4) and domain movements (5). NOS regulates redox activities via an autoinhibitory domain that slows its redox reactions when calmodulin and Ca 2ϩ are not present (6,7). Among the two-component flavin-dependent monooxygenases in which reduced flavin transfer between reductase and oxygenase components is required as part of their catalyses there are three known mechanisms for transfer of reduced flavin. Bacterial luciferase and alkane-sulfonate monooxygenase (8 -11) use protein-protein interactions to mediate reduced flavin transfer, whereas simple diffusion can account for the reduced transfer in p-hydroxyphenylacetate hydroxylase (HPAH) 3 from Acinetobacter baumannii (12) and the oxygenase involved in actinorhodin biosynthesis (ActVA-B) (13). Styrene monooxygenase (14) and HPAH from Pseudomonas aeruginosa (15) use both modes of transferring reduced flavin. However, little is known about how these transfers are regulated in these flavin-dependent enzymes.
p-Hydroxyphenylacetate 3-hydroxylase from A. baumannii catalyzes the ortho-hydroxylation of p-hydroxyphenylacetate (HPA) to generate 3,4-dihydroxyphenylacetate. HPAH belongs to a large family of two-component flavin-dependent monooxygenases. Catalytic and structural studies of the enzymes from several bacterial species, including Pseudomonas putida (16,17), Pseudomonas aeruginosa (15), Escherichia coli (18), Klebsiella pneumonia (19), Sulfolobus tokodaii (20), Thermus thermophilus (21,22), and A. baumannii (12,(23)(24)(25)(26)(27)(28)(29)(30)(31), have been carried out. HPAH consists of a reductase component (C 1 ) that catalyzes reduction of FMN to generate FMNH Ϫ for its oxygenase (C 2 ), which catalyzes the oxygenation of HPA (12,23). C 2 has been shown to hydroxylate HPA via a C4a-hydroperoxyflavin typical of flavin-dependent hydroxylases (29 -31). C 1 can catalyze the reduction of FMN by NADH, and under aerobic conditions, the resulting FMNH Ϫ will react with oxygen to generate H 2 O 2 (constituting NADH oxidase activity) (25). In the presence of HPA, the substrate for C 2 , the rate of reduction of FMN is greatly stimulated, implying that HPA acts as an effector of the C 1 enzymatic activity. Such allosteric control of the reductase component by a substrate of the partner oxygenase has been reported only for C 1 and for nitrilotriacetate and EDTA monooxygenases (32,33) but not for HPAH from other organisms or reductase components of other twocomponent flavin-dependent enzymes (34). It has been speculated that the C-terminal half of C 1 , which does not contain any sequences for NADH or flavin-binding folds, is a regulatory domain that when bound to HPA permits rapid reduction of FMN (25).
The N-terminal domain(s) of C 1 (residues 1-178; 18 kDa) is homologous with several smaller flavin reductases with known structures (HPAH systems from S. tokodaii (20) and from T. thermophilus (21) and the reductase component of phenol 2-hydroxylase from Bacillus thermoglucosidasius A7 (PheA2) (35)), and its structure is predicted to be similar. The remaining C-terminal residues (192-315) have ϳ20% identity to MarR proteins (36). MarR is a repressor protein that binds to its DNA target to control expression of MarA. MarA is a protein that regulates expression of the AcrAB-TolC efflux pump that excretes xenobiotics from bacteria (37). Upon binding to an aromatic effector such as salicylate (and presumably antibiotics), MarR dissociates from DNA to derepress the expression of MarA and the AcrAB-TolC efflux pump, thereby promoting excretion of antibiotics or aromatic xenobiotics (37,38).
In this study, three truncated C 1 variants as well as single site variants of residues Arg-21, Phe-216, Arg-217, Ile-246, and Arg-247 were constructed to investigate the role of the C-terminal domain in regulating HPA-stimulated C 1 activity. The results show that residues 179 -230 and particularly residue Phe-216 are important in the HPA-stimulated NADH oxidase activity. The data indicate that the C-terminal domain is an autoinhibitory domain that upon binding HPA undergoes conformational changes to allow more rapid flavin reduction and release of FMNH Ϫ .
Modeled Structure of C 1 -A three-dimensional structure of C 1 was modeled using the Swiss-PdbViewer (or DeepView) program via SWISS-MODEL, a web site that analyzes and predicts three-dimensional structures of target amino acid sequences (40 -42). The C 1 structure obtained from this site was used with the PyMOL program (version 0.99) to calculate vacuum electrostatics and identify amino acids that could be important for interactions between the N-and C-terminal domains of wild-type C 1 and would be suitable for application of site-directed mutagenesis.
Construction of Truncated C 1 Variants-Genes of truncated variants, t178C 1 , t230C 1 , and t259C 1 , were amplified by PCR using a GeneAmp PCR system (Applied Biosystems, model 2004). The PCR contained 4 g of C 1 -pET11a plasmid as a DNA template, 0.5 M primers (supplemental Tables S1 and S2), a 200 M concentration of each dNTP, and 2.0 units of Pfu DNA polymerase (Fermentas) in 1ϫ Pfu buffer and sterile water in a total volume of 50 l. The final PCR products were digested by NdeI and BamHI and ligated into the corresponding sites of pET11a using T 4 ligase. The ligated plasmids were analyzed for their sequences by Macrogen (Korea) and kept at Ϫ80°C until used.
Site-directed Mutagenesis-The genes encoding the variants R247A, I246W, R217A, F216A, R201A, and R21A were constructed according to the PCR protocol and primers described in supplemental Table S2. DpnI was added to the resulting PCR products to remove the template plasmids. Plasmids encoding for C 1 variants were propagated in E. coli and purified according to the QIAprep Spin Miniprep kit protocols (Qiagen, Germany). The sequences of the purified plasmids were analyzed by Macrogen, 1st BASE Pte. Ltd. (Singapore), or the DNA Sequence Core (University of Michigan). The variant C 1 plasmids were kept at Ϫ80°C until used.
Protein Expression/Purification-Most of the C 1 variants were expressed and purified according to the protocol reported for the wild-type C 1 (24). For variants t230C 1 , t259C 1 , and R201A, only the crude extracts were prepared and tested for their NADH oxidation activities. After purification, the purities of the t178C 1 , R247A, I246W, R217A, F216A, and R21A variants were estimated by SDS-PAGE.
NADH Oxidase Activity-NADH oxidase activity of C 1 was measured spectrophotometrically at 340 nm. A typical assay reaction contained enzyme (ϳ0.02 M; no extra FMN) and NADH (96 M) in 50 mM sodium phosphate buffer, pH 7.0 at 25°C. Reactions were carried out in the absence and presence of HPA to test whether variants exhibited HPA-stimulated NADH oxidation activities.
Rapid Reaction Experiments-Reduction of C 1 variants by NADH, oxidation of reduced t178C 1 by oxygen, and the transfers of reduced flavin from t178C 1 to C 2 or to cytochrome c were investigated using a stopped-flow spectrophotometer (TgK Scientific). All experiments were carried out in 50 mM sodium phosphate, pH 7.0 at 4°C unless noted otherwise. Experiments and preparations for anaerobic solutions were similar to the protocols in previous reports (12,25,26). Rates of reduced flavin transfer from t178C 1 to C 2 or to cytochrome c were measured by mixing a solution of reduced t178C 1 with a solution containing C 2 and equilibrated with various concentrations of oxygen or mixed with a solution containing cytochrome c and equilibrated with oxygen-free N 2 or argon.
Data Analysis-Observed rate constants (k obs ) from kinetic traces were calculated by fitting data to exponential functions in Program A. 4 Rate constants were determined from plots of k obs and NADH concentrations using Marquardt-Levenberg nonlinear fitting algorithms included in KaleidaGraph (Synergy Software) according to Equation 1.

RESULTS
Modeled Structure of Wild-type C 1 -The SWISS-MODEL web site (40 -42) was used to predict the three-dimensional structure of C 1 from its sequence. The N-terminal half of C 1 (residues 14 -168) is 38% identical to the reductase component of PheA2 (Protein Data Bank code 1RZ0) (35,24), and the C-terminal half of C 1 (residues 187-314) is 19% identical to MarR, a transcription regulator involved in bacterial antibiotic resistance (Protein Data Bank code 3BPV) (36). The modeled structure of C 1 is shown in Fig. 1, and the results of the alignment of the C 1 sequence with known structures are shown in supplemental Fig. S1. The modeled structure of C 1 suggests that reduction of FMN occurs in the N-terminal half, whereas the C-terminal half likely acts as a regulatory domain that controls the rate of catalysis. Several truncated variants (t178C 1 , t230C 1 , and t259C 1 ) were constructed and overexpressed to evaluate which parts of the C-terminal half are necessary for controlling the redox activity of C 1 (Fig. 2).
PyMOL was used to analyze the modeled C 1 structure for surface electrostatics (Fig. 1B) and to identify amino acids that might be important for controlling the C 1 reaction. This analysis identified Arg-21 and Arg-217 as important for the interactions between the N-and C-terminal domains that likely participate in the regulatory function by HPA. MarR structures (Protein Data Bank codes 3BPX and 1JGS) (36,38) show that Arg-16 and Arg-86, which are homologous to Arg-21 and Arg-217 in C 1 , bind to the carboxylate of its salicylate regulatory ligand perhaps analogously to C 1 binding HPA. Therefore, several variants around Arg-21 and Arg-217 positions (R247A, I246W, R217A, F216A, R201A, and R21A) were constructed to test their functional roles.
Enzyme Preparations of Truncated C 1 Variants-Three variants of C 1 truncated at the C-terminal domain (t178C 1 , t230C 1 , and t259C 1 ; Fig. 2) were constructed and overexpressed in E. coli ("Experimental Procedures" and supplemental data). The effect of HPA in stimulating NADH oxidase activity was ascertained by assaying activities of the truncated variants in the absence and presence of HPA ("Experimental Procedures") (23,25). The t230C 1 and t259C 1 variants exhibited markedly increased rates of NADH oxidation in the presence of HPA similar to wild-type C 1 (data not shown), implying that these truncated variants can bind HPA normally. Therefore, we con-

The C-terminal Domain of A. baumannii HPAH Is Autoinhibitory
clude that residues 230 -315 are not critical for regulating C 1 activities. In contrast, the t178C 1 variant catalyzed NADH oxidation at the same rate both in the absence and presence of HPA, implying that the C-terminal residues 179 -229 are important for HPA binding and are stimulating C 1 activity. Single site mutations of positively charged residues in this region were constructed to more precisely probe their roles in regulating C 1 activity (results shown below).
The t178C 1 truncated variant was purified according to the protocol used for the wild-type enzyme (24). The subunit molecular mass of the purified t178C 1 was estimated to be ϳ20 kDa from SDS-PAGE analysis (supplemental Fig. S2A), and the native molecular mass determined via gel filtration analysis is ϳ45 kDa (supplemental Fig. S2B). These results indicate that like wild-type C 1 and PheA2 (35) the truncated enzyme is also a homodimer (23). Table 1 compares biochemical and biophysical characteristics of t178C 1 and wild-type C 1 . In general, the properties of t178C 1 are similar to those of wild-type C 1 (12,25,26). Most of the properties of C 1 regarding catalytic and redox activities are retained in the N-terminal domain of the t178C 1 truncated variant, but there is no regulation by HPA. These data are therefore consistent with the C-terminal domain being a regulatory domain that controls HPA-stimulated activity.
Reductive Half-reaction of t178C 1 -The kinetics of reduction of the t178C 1 -bound FMN by NADH were investigated using stopped-flow spectrophotometry. Fig. 3 shows the kinetic traces recorded at 458 and 690 nm of the reduction by NADH of t178C 1 -bound FMN in the absence and presence of HPA. HPA had no measurable effect on the kinetics. Moreover, these traces are nearly indistinguishable from those for the reduction of wild-type C 1 in the presence of 0.2 mM HPA (after mixing). For reference, the kinetic trace of wild-type C 1 in the absence of HPA is overlaid (Fig. 3).
Traces monitored by absorbance changes at 458 and 690 nm were biphasic, and the data were analyzed according to the model used for analysis of wild-type C 1 in the presence of HPA (see Scheme 2 in Ref. 25). The first phase of the reaction (dead time, ϳ0.003 s) resulted in formation of a charge transfer complex (C 1 -FMN-HPA⅐NADH) as noted by the absorbance increase at 690 nm. The second phase (0.003-0.01 s), demarked by a large decrease in absorbance at 458 nm and a small increase in absorbance at 690 nm, is due to reduction of FMN to form a second charge transfer species (C 1 -FMNH 2 -HPA⅐NAD ϩ ). Apparent rate constants describing this phase were dependent on NADH concentration, and the value of k 3 (flavin reduction) was analyzed according to Equation 1. The third phase (0.01-0.05 s) resulted in the decay of the second charge transfer species, and this phase was independent of NADH concentration.
The kinetic parameters of t178C 1 are summarized in Table 2. The rate constants for reduction of FMN by t178C 1 (276 Ϯ 6 s Ϫ1 ) were not affected by the presence of HPA. This result implies that in wild-type enzyme the C-terminal part of C 1 (residues 179 -315) inhibits reduction of FMN in the N-terminal domain, and HPA binding to the C-terminal domain alleviates this inhibition. Without the C-terminal domain, t178C 1 is constitutively active with or without HPA present.
Oxidative Half-reaction of t178C 1 -The oxidation of reduced t178C 1 by oxygen was investigated using stopped-flow experiments and compared with the results for wild-type C 1 . Studies with wild-type C 1 showed that superoxide dismutase simplified the reaction from highly complex to biphasic kinetics. The data were interpreted according to a model in which the reduced C 1 contains mixed populations with one species oxidizing faster than the other (25). The presence of HPA pro-  motes the wild-type enzyme to shift toward the faster reacting species that releases FMNH Ϫ at a greater rate. Kinetic traces for the oxidation of the flavin of t178C 1 monitored at 458 nm in the absence and presence of HPA were single exponential (Fig. 4). Therefore, the reduced t178C 1 reacted with O 2 in a simple bimolecular reaction with the rate constant of 247 Ϯ 2 M Ϫ1 s Ϫ1 (Table 2 and Scheme 1), which is 50% that of wild-type enzyme (25). The presence of HPA had no effect on the oxidation of the t178C 1 variant (Fig. 4). The experiment in Fig. 4 did not contain superoxide dismutase because superoxide dismutase introduced complicated multiphasic kinetics on t178C 1 oxidation (supplemental Fig. S3). Therefore, oxidations of reduced wild-type C 1 and of t178C 1 occur via different mechanisms in which superoxide might be accumulated in the oxidation of the wild type but not significantly in the t178C 1 variant. Nevertheless, these results for t178C 1 , like those for flavin reduction (Fig. 3), show no effects of HPA.
Flavin Transfer from t178C 1 to C 2 -Using the stopped-flow spectrophotometer, a solution of reduced t178C 1 was mixed with solutions of C 2 containing various concentrations of O 2 .
Kinetics of the reaction monitored at 380 and 458 nm in the absence and presence of HPA (Fig. 5) were used to estimate the rate constant for transfer of reduced flavin (FMNH Ϫ ) from C 1 to C 2 . The rate constant for free FMNH Ϫ binding to C 2 is very large, estimated to be Ͼ10 7 M Ϫ1 s Ϫ1 . The resulting C 2 ⅐FMNH Ϫ complex reacts with oxygen to form a C4a-hydroperoxy-FMN intermediate with a second-order rate constant of 1.1 ϫ 10 6 M Ϫ1 s Ϫ1 , which yields an observed rate constant of 143 s Ϫ1 at 0.13 mM oxygen concentration (12,28). Therefore, the kinetics of the absorbance increase at 380 nm (formation of C4a-hydroperoxy-FMN) can be used to estimate the rate constant for reduced flavin transfer from t178C 1 to C 2 . The observed rate constant for formation of the C4a-hydroperoxyflavin in Fig. 5 is less than 143 s Ϫ1 (the rate constant for C 2 -FMNH Ϫ reacting with O 2 ) because this rate is limited by the transfer of FMNH Ϫ from C 1 to C 2 ; the observed value (42 Ϯ 4 s Ϫ1 ), which was the same with or without HPA (Fig. 5), therefore reflects the rate constant for FMNH Ϫ transfer in this variant. These results are quite different from those for wild-type C 1 (12) (Fig. 5 and Table 2). In the absence of HPA, when wild-type C 1 -FMNH Ϫ is mixed with an aerobic C 2 solution, the C4a-hydroperoxyflavin is formed with a rate constant of only 0.35 s Ϫ1 due to the slow release of FMNH Ϫ from C 1 . The presence of HPA causes more rapid release of FMNH Ϫ from C 1 , permitting formation of C4ahydroperoxyflavin with a rate constant of 80 s Ϫ1 (12). Although HPA has no effect on the release of FMNH Ϫ from C 1 , the rate constant of FMNH Ϫ transfer in the variant (42 Ϯ 2 s Ϫ1 ) is slightly less than that of the wild-type C 1 in the presence of HPA (80 s Ϫ1 ).
Reduced Flavin Transfer from t178C 1 to Cytochrome c-The experiments in Fig. 6 are analogous to those in Fig. 5 but use an independent means of detecting the release of FMNH Ϫ from C 1 . A solution of t178C 1 -FMNH Ϫ was mixed with a solution of oxidized cytochrome c to determine the rate constant for release of FMNH Ϫ from C 1 . Cytochrome c is a small hemoprotein that under the conditions for Fig. 6 is reduced by FMNH Ϫ very rapidly (Ͼ500 s Ϫ1 ) to generate ferrocytochrome c (12). The oxidized and reduced cytochrome c species have distinct spectral properties, permitting facile monitoring of its reduction at 550 nm (Fig. 6). Therefore, the observed rate constant for reduction of cytochrome c (measured by monitoring after mixing reduced C 1 and cytochrome c together) is related to the rate constant for release of FMNH Ϫ from C 1 . Kinetic traces recorded at 550 nm showed that both in the absence and presence of HPA (Fig. 6) cytochrome c was reduced by t178C 1 with a rate constant of 75 Ϯ 1 s Ϫ1 . For the wild-type enzyme, the observed rate constants for cytochrome c reduction are 0.35 s Ϫ1  in the absence of HPA and 80 s Ϫ1 in the presence of HPA (ϳ230-fold difference; Table 2). These results again confirmed that the autoinhibition and the effect of HPA on stimulation of FMNH Ϫ release in C 1 were absent in the t178C 1 variant. Enzyme Preparation of Single Site Variants of C 1 -All of the above results suggest that the C-terminal domain of C 1 is a regulatory domain for controlling the production of FMNH Ϫ . The results in Figs. 3 and 4 suggest that for the wild-type C 1 the C-terminal domain acts to inhibit both the reduction of the flavin and the release of FMNH Ϫ to the solution. In the presence of HPA, we suggest that domain movement occurs to relieve the inhibition, resulting in ϳ20-fold faster flavin reduction and ϳ230-fold faster release of FMNH Ϫ . Because the crystal structures of C 1 with and without HPA bound are currently not available, putative binding sites of HPA on the C-terminal domain of C 1 were postulated from its homology with the x-ray structure of the MarR⅐salicylate complex (Protein Data Bank codes 3BPX and 1JGS). The Arg-86 and Arg-16 residues of MarR undergo a conformational change upon binding salicylate to optimize its ability to bind its target DNA (36,38). On the basis of its homology to MarR and the results obtained with the truncated C 1 variants (Figs. 2 and 3), we propose that the HPA binding site is located near residues 193-201 and 220 -264 of C 1 . In addition, we deduced that a negatively charged moiety of HPA would interact with positively charged residues in that region. Therefore, residues Arg-21 and Arg-217, which are located between the N-and C-terminal domains, were identified as putative residues that are important for controlling the NADH oxidation activity of C 1 . Single site variants of these residues and of other residues located in the same region (Arg-21, Arg-201, Phe-216, Arg-217, Ile-246, and Arg-247) were constructed, changing these residues to Ala or Trp (for Ile-246) (Figs. 1 and 2 and supplemental Table S1). Most of these variants (R247A, I246W, R21A, R217A, and F216A) were expressed and purified using the same protocol as used for the wild-type enzyme, and their flavin reduction kinetics were measured in the absence and presence of HPA (see below). With R201A, only the crude extract was tested for the NADH oxidation activity in the absence and presence of HPA. R201A is as active as the wild-type C 1 , and its NADH oxidase activity can be enhanced by the presence of HPA just as with the wild-type enzyme (supplemental Fig. S4). Therefore, the involvement of this residue with the HPA binding was ruled out.
Kinetics of Flavin Reduction of R247A, I246W, R21A, R217A, and F216A in the Absence and Presence of HPA-Solutions of oxidized R247A, I246W, R217A, R21A, and F216A variants were mixed anaerobically with various NADH concentrations in the absence and presence of HPA using the stopped-flow spectrophotometer, and the reactions were monitored by the absorbance changes at 458 and 690 nm. Figs. 7 and 8 show time courses for the reduction of these variants by 6.4 mM NADH in the absence and presence of HPA. The data were analyzed as described for Fig. 3. For most variants, the presence of HPA increased the rate constant for the reduction of enzyme-bound   flavin about 17-24-fold, which is similar to the effect of HPA observed with the wild-type C 1 (ϳ20-fold; Table 3). Therefore, these residues are not directly involved in the HPA binding. The only exception was observed for the F216A variant in which there was no stimulation of flavin reduction by HPA. This result implies that Phe-216 is likely involved with the HPA binding and/or the stimulation of flavin reduction by HPA.
Further attempts were carried out to evaluate the properties of the F216A variant. Gel filtration chromatography data indicated that F216A is a dimeric enzyme similar to the WT and the truncated variant t178C 1 . CD spectra of wild type and the F216A variant were similar and were unaffected by addition of HPA. These results imply that the presence of HPA does not cause any significant change in secondary structures of the proteins. Using the method used previously (27), the K d value for the binding of HPA to F216AC 1 was measured as 18 Ϯ 5 M, which is higher than the K d for the HPA binding to WT (1.4 Ϯ 0.2 M).

DISCUSSION
The results reported herein imply that the C-terminal domain of C 1 is an autoinhibitory domain that suppresses the redox activity of C 1 in the absence of its effector, HPA. Upon binding HPA, the repression of C 1 activity is alleviated presumably by a protein conformational change(s) that moves the C-terminal domain away from the active site. The regulatory mechanism illustrated for C 1 is unusual among the two-component flavoprotein monooxygenases. Most of the flavin reductase components in these two-component enzyme systems are about half the size of C 1 (ϳ16 -18 kDa); all of the reductases with known three-dimensional structures consist of protein folds that are mainly involved in catalyzing flavin reduction by NADH with no regulatory features. Only a few proteins have been identified that have extra C-terminal domains that are not directly involved in their redox reactions, namely the reductase of HPAH from P. putida 5 and the reductase of the nitrilotriacetate monooxygenase system from Chelatobacter strain ATCC 29600 (cB) (32). Recently, the structure of the reductase of nitrilotriacetate monooxygenase (NTA_MoB; Protein Data Bank code 3NFW) from Mycobacterium thermoresistibile has been reported (43). However, NTA_MoB (189 residues) is smaller than cB from Chelatobacter sp. (335 residues).
Results from Tables 2 and 3 and Figs. 3, 5, and 6 indicate that the truncated t178C 1 is catalytically as active and except for its size and its lack of interactions with HPA has biochemical and biophysical properties similar to those of wild-type C 1 . The t178C 1 variant catalyzed flavin reduction with rate constants of 276 Ϯ 6 and 285 Ϯ 4 s Ϫ1 in the absence and presence of HPA, respectively, and it can transfer FMNH Ϫ to C 2 or to cytochrome c with rate constants of 42 Ϯ 2 and 75 Ϯ 1 s Ϫ1 , respectively. The presence of HPA had no influence on these kinetic constants, again implying that the C-terminal domain that had been removed is responsible for binding HPA and for any resultant conformational changes. We attribute the slightly greater rate of reduction of cytochrome c than of transfer of FMNH Ϫ to C 2 to the cytochrome c being able to accept electrons from FMNH Ϫ before it completely leaves C 1 , whereas formation of the C4a-FMN hydroperoxide clearly requires complete dissociation from C 1 and association with C 2 before it reacts with O 2 . 5 C. Durgan and D. P. Ballou, unpublished data.  Table 3. The inset shows kinetic traces of F216A mutant at 458 and 690 nm.  Table 3. The inset shows kinetic traces of F216A mutant at 458 and 690 nm.  (25).

The C-terminal Domain of A. baumannii HPAH Is Autoinhibitory
The C-terminal domain of C 1 has homology to MarR proteins whose structures are characterized by a helix-winged protein fold for binding to DNA targets (38). The binding of salicylate and other aromatic compounds to MarR pushes apart the two DNA binding lobes, thereby abolishing the ability of MarR to bind to DNA (36). Because binding of HPA can alleviate the autoinhibitory effect of the somewhat homologous C-terminal domain of C 1 , we suggest that conformational changes similar to those of MarR may also occur in C 1 upon binding HPA. We suspect that the binding of HPA would disrupt interactions between the C-and N-terminal domains of C 1 so that the enzyme-bound flavin becomes more accessible to NADH and to solvent (Figs. 1 and 9). The observation that other aromatic compounds in addition to HPA can stimulate C 1 activity (23) suggests that this regulatory mechanism of C 1 is a general means for HPAH to increase the rate of reduced flavin production when a potential substrate for the oxygenase component is present.
It can be envisaged that the C-terminal autoregulatory domain of C 1 is an important adaptation to minimize production of reactive oxygen species for two-component flavin-dependent monooxygenases. Another means for minimizing reactive oxygen species is to use protein-protein interactions as found in the bacterial luciferase system (8). Several different potential allosteric substrate-like effectors presumably can activate the autoregulatory domain of C 1 . This would permit a single reductase to be useful to and be regulated by several different molecules that would undergo oxygenation by their respective oxygenases.
Site-directed mutagenesis and kinetic investigations have identified residue Phe-216 as being important for HPA to exert its regulatory function. Results from truncated mutants have shown that residues 230 -315 are not directly involved in the autoinhibitory mechanism in C 1 because mutants truncated in this region behaved similarly to the wild-type enzyme (data not shown). Of the mutations of residues Arg-21, Phe-216, Arg-217, Ile-246, and Arg-247, only the F216A variant lacks the stimulatory effect by HPA. These C 1 variants were chosen from analogies to the MarR Protein Data Bank structures 1JGS (38) and 3BPX (36), noting how MarR binds to the carboxylate of salicylate via Arg-16 and Arg-86. The mutations listed in Table   3 were focused on the location in which the C-and N-terminal domains are postulated to interact (Fig. 1). The results in Table  3 indicate that changes in residues Arg-21, Arg-217, Ile-246, and Arg-247 did not affect the kinetic properties of the reductase, thus eliminating consideration of their involvement in controlling the autoinhibition of C 1 . However, the F216A variant, which according to the modeled structure is located in a flexible loop region near a hydrophobic zone, may be involved in promoting tighter interactions between the N-and C-terminal domains; for example, the F216A variant may not allow the N-and C-terminal domains to interact as tightly as in wild-type C 1 . This variant thereby permits rapid flavin reduction in the absence of HPA. When the x-ray structure of C 1 is available, the data reported here will become more useful for interpreting the functional roles of these residues. Recently, the crystallization and preliminary x-ray analysis of C 1 have been reported (44).
A similar controlling mechanism for flavin reduction and the subsequent electron transfer via an autoinhibitory domain were found for the reaction of endothelial and neuronal nitric-oxide synthases, which contain P450 heme (oxygenase) and diflavin (reductase) domains. Binding of calmodulin/Ca 2ϩ increases the rates of interdomain electron transfer from reduced FAD to FMN and from reduced FMN to heme iron in the oxygenase domain (1,6). Deletion of the autoinhibitory domain in NOS resulted in an enzyme that is reduced faster and has greater NO synthesis activity (7,(45)(46)(47). In addition, single site charge reversal variants of neuronal NOS, such as K842E (6) and R1229E (48), display faster flavin and heme reduction kinetics in the calmodulin-free enzyme. These substituted residues might not promote autoinhibitory interactions between domains and thus could result in enzyme variants with more accessible redox-active sites. In this respect, the control of C 1 activity by the C-terminal domain and the fact that a single site variant (F216A) can abolish the autoinhibitory feature resemble the results reported for NOS reactions.
In conclusion, this study has shown that in the absence of HPA the two-component flavin-dependent HPAH from A. baumannii utilizes the C-terminal domain of C 1 as an autoinhibitory domain to suppress both the rate of reduction of FMN and the rate of release of FMNH Ϫ from C 1 . This avoids wasteful consumption of NADH and formation of reactive oxygen species. Binding to HPA alleviates the autoinhibition by inducing conformational changes that allow C 1 to generate reduced flavin for the oxygenase domain. This catalytic feature is unusual among known two-component flavin-dependent monooxygenases but may be common in flavin reductases harboring the extra C-terminal domain.