Changing the Substrate Reactivity of 2-Hydroxybiphenyl 3-Monooxygenase from Pseudomonas azelaica HBP1 by Directed Evolution*

The substrate reactivity of the flavoenzyme 2-hydroxybiphenyl 3-monooxygenase (EC 1.14.13.44, HbpA) was changed by directed evolution using error-prone PCR. In situ screening of mutant libraries resulted in the identification of proteins with increased activity towards 2-tert-butylphenol and guaiacol (2-methoxyphenol). One enzyme variant contained amino acid substitutions V368A/L417F, which were inserted by two rounds of mutagenesis. The double replacement improved the efficiency of substrate hydroxylation by reducing the uncoupled oxidation of NADH. With guaiacol as substrate, the two substitutions increasedVmax from 0.22 to 0.43 units mg−1protein and decreased the K′ m from 588 to 143 μm, improvingk′cat/K′ m by a factor of 8.2. With 2-tert-butylphenol as the substrate,k′cat was increased more than 5-fold. Another selected enzyme variant contained amino acid substitution I244V and had a 30% higher specific activity with 2-sec-butylphenol, guaiacol, and the “natural” substrate 2-hydroxybiphenyl. TheK′ m for guaiacol decreased with this mutant, but the K′ m for 2-hydroxybiphenyl increased. The primary structure of HbpA shares 20.1% sequence identity with phenol 2-monooxygenase from Trichosporon cutaneum. Structure homology modeling with this three-domain enzyme suggests that Ile244 of HbpA is located in the substrate binding pocket and is involved in accommodating the phenyl substituent of the phenol. In contrast, Val368 and Leu417 are not close to the active site and would not have been obvious candidates for modification by rational design.

HbpA mutants with an altered substrate reactivity should allow the synthesis of catechols that are not synthesized by wild-type HbpA. Rational protein design based on a known three-dimensional structure has been used for such purposes (15)(16)(17)(18), but random approaches have lately become more popular (19). Directed enzyme evolution (20 -23), the most often used strategy, was applied to improve substrate specificity, activity, enantioselectivity, or thermostability (21, 24 -27).
Here we report on the use of directed evolution to change the substrate reactivity of HbpA. We increased the specific activity of HbpA towards 2-hydroxybiphenyl, 2-sec-butylphenol, guaiacol (2-methoxyphenol), and 2-tert-butylphenol. Moreover, a significant increase in the efficiency of NADH utilization was achieved with one mutant monooxygenase. These results are interpreted at the structural level with the help of a threedimensional model of HbpA.

MATERIALS AND METHODS
Chemicals, Bacterial Strains, and Plasmids-E. coli JM101 and plasmid pAA1 were used for cloning and gene expression. Plasmid pAA1 is a pUC18 (28) derivative harboring the hbpA gene as a SalI/NsiI fragment (29) cloned into the SalI/PstI sites of the pUC18 polylinker.
Commercially available chemicals were purchased from Fluka AG (Buchs, Switzerland). Catalase from beef liver was obtained from Roche Molecular Biochemicals. Taq DNA polymerase, restriction enzymes, and T4 DNA ligase were purchased from Roche Molecular Biochemicals. 2,3-Dihydroxybiphenyl and 3-sec-butylcatechol were prepared by whole-cell biotransformations, using a recombinant E. coli JM101 containing the hbpA gene (14).
Cloning Procedures-Competent E. coli JM101 cells were prepared using a modified CaCl 2 -based method (31). A 5-ml culture of E. coli JM101 was grown overnight in LB. The cells were diluted 1:100 into fresh LB and grown until they reached an A 600 of 0.5 Ϯ 0.1. After centrifugation (5500 ϫ g, 4°C, 8 min), the supernatant was discarded, and the pellet was resuspended in 0.2 volumes of 10 mM sodium acetate (pH 5.8), 50 mM MnCl 2 , and 5 mM NaCl. The suspension was put on ice for 30 min and recentrifuged as before. The pellet was dissolved in 0.1 volumes of 10 mM sodium acetate (pH 5.8), 70 mM CaCl 2 , 5 mM MnCl 2 , and 5% glycerol and stored in aliquots of 100 l at Ϫ70°C until use.
Ligation mixtures containing pUC18 (cut BamHI/SphI), amplified hbpA gene (cut BamHI/SphI), 0.5 units of T4 DNA ligase (Roche Molecular Biochemicals), and 10ϫ ligation buffer were incubated overnight at 4°C. The ligation mixture was directly used for transformation. For this, an aliquot of competent E. coli JM101 was thawed on ice. The cells were mixed with 1 l of ligation mixture and placed on ice for 30 min. The heat pulse was performed at 42°C for 90 s and followed by incubation on ice for 1 min. After the addition of 1 ml of LB, the cells were incubated at 37°C for 1 h. The complete transformation mixture was transferred onto selective LB plates and incubated overnight at 30°C.
Screening-The screening procedure for the desired modified HbpA was based on the instability of the reaction products. At neutral pH, catechols autoxidize to quinones and semiquinones, which readily form reddish or brownish, undefined, high molecular weight compounds (32). After transformation, E. coli JM101 transformants were transferred directly onto LB plates containing 150 g ml Ϫ1 ampicillin, 0.04‰ (w/v) 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal), 200 M isopropyl-1-thio-␤-D-galactopyranoside, and a 0.1-0.5 mM concentration of the 2-substituted phenol to be screened for. Incubation at 30°C resulted in three colony types: blue colonies, which did not contain the amplified hbpA gene; white colonies, which contained an hbpA gene that encoded for an inactive enzyme towards the aromatic test substrate, or an inactive lacZ gene due to frameshifted ligation; and reddish brown colonies, which contained an enzyme with activity towards the added 2-substituted phenol. The time-dependent intensity of color formation in the latter colonies could be used to distinguish between different activities. For substrates that were only poorly transformed to the corresponding catechol, the color formation could be intensified by the addition of 1.5 mM ferric chloride and 50 g ml Ϫ1 p-toluidine (33). If positive clones were detected, the plasmids containing the amplified hbpA genes were isolated and retransformed into E. coli JM101.
Preparation of Cell Extracts-Cells from a 5-ml LB culture were spun down at 5000 ϫ g for 15 min and resuspended in 800 l of 50 mM phosphate buffer (pH 7.2). This suspension was transferred to a 1.5-ml Eppendorf tube containing 1.2-g glass beads (diameter, 0.1-0.2 mm), and the cells were disrupted in a Retsch mill (Retsch GmbH, Hann, Germany) for 10 min at 90% power. The cell extracts were separated from the glass beads by centrifugation (15,000 ϫ g, 15 min) and supplemented with FAD to a final concentration of 50 M. Cell extracts could be stored at Ϫ20°C for 2-3 weeks without significant loss of HbpA activity.
Enzyme Purification-The purification of HbpA and its mutants was based on a simplified version of a procedure described earlier (1). 6 g (cell wet weight) of frozen E. coli JM101, harboring a pUC18 derivative encoding either the wild-type or the amplified hbpA gene, were suspended in 25 ml of phosphate buffer (10 mM, pH 7.5). Cell extract was prepared by twice passing the suspension through a French pressure cell (20 K; Sim Aminco) at 70 bars, followed by ultracentrifugation at 4°C (Beckmann L8 -60 M, 40,000 ϫ g, 30 min).
Fractions containing HbpA activity were pooled, supplemented with 0.9 M ammonium sulfate, and loaded onto a hydrophobic interaction chromatography column (1 ϫ 8 cm; butyl-Sepharose 4 Fast Flow; Amersham Biosciences, Inc.) equilibrated with 0.75 M ammonium sulfate in 100 mM sodium phosphate buffer (pH 7.0). Elution was carried out with a linear gradient from 0.75 to 0 M ammonium sulfate in 100 mM sodium phosphate buffer (pH 7.0). Fractions containing HbpA were pooled and concentrated in an Ultrafree-15 centrifugal filter device (Biomax-50K; Millipore Corp., Bedford, MA).
Analytical Methods-As is also the case for other flavin-containing oxygenases, NADH oxidation is partially uncoupled from substrate hydroxylation in HbpA (1). Therefore, specific activities were determined both for NADH oxidation and substrate consumption/product formation. NADH oxidation was followed spectrophotometrically at 340 nm or polarographically by monitoring oxygen consumption with an oxygen electrode (7). The assay contained 0.2-1 mM substrate, 0.3 mM NADH, 20 mM air-saturated phosphate buffer (pH 7.5), and 10 -20 l of cell extract or purified protein in a total volume of 1 ml. To determine substrate utilization and product formation, the enzymatic reaction was stopped by the addition of perchloric acid, and the resulting precipitate was removed by centrifugation (15,000 ϫ g, 10 min). The samples were diluted 1:1 with MeOH, 0.1% phosphoric acid and analyzed with a Hypersil ODS column (5 m, 4.5 ϫ 125 mm) using a Hewlett Packard HP 1050 Ti high pressure liquid chromatograph coupled to a diode array detector (HP DAD 1040M). The elution was carried out under isocratic conditions with MeOH/H 2 O (0.1% phosphoric acid) as mobile phase.
Steady-state kinetic parameters were calculated by weighted nonlinear regression analysis (Enzfitter; Elsevier-Biosoft, UK). Uncoupling of product formation from oxygen consumption was resolved by performing HbpA activity assays in the absence and in the presence of catalase (34). The catalase recycles 50% of the oxygen, which is used for hydrogen peroxide formation. By determining initial reaction rates in the absence and presence of catalase, the substrate related uncoupling of product formation from oxygen consumption could be calculated.
HbpA Modeling-The sequence of HbpA was aligned to the recently corrected sequence of phenol 2-monooxygenase from Trichosporon cutaneum (35) with ClustalX (36) using the PAM 350 matrix. Model building of HbpA was performed with MODELLER (37) using the CVFF force field (38). The closed form of the phenol 2-monooxygenase structure (Protein Data Bank entry 1foh) was used as a template. The model was verified after several rounds of energy minimization. The stereochemical quality of the homology model was verified by PROCHECK (39), and the protein folding was assessed with PROSAII (40), which evaluates the compatibility of each individual residue with its environment. The FAD and the substrate 2-hydroxybiphenyl were placed in identical positions and orientations as the FAD and phenol in the template structure.
Nomenclature-Subscript letters indicate the substrate on which the mutant was screened (G, guaiacol; T, 2-tert-butylphenol); numbers indicate the round of error-prone PCR. HbpA* represents HbpA variants in general.

RESULTS
Directed Evolution of HbpA-2-Hydroxybiphenyl 3-monooxygenase (HbpA) was subjected to in vitro manganese mutagenesis with error-prone PCR (41) and subsequent in situ screening for enzymes with an altered substrate reactivity (HbpA*). The base substitution rate in the error-prone PCR was tuned to an exchange rate of 1-3 per hbpA gene, to produce an average of one amino acid substitution per HbpA* (24). The Mn 2ϩ concentration was adapted to template composition, template length, dNTP concentration, and polymerase type. We tested different Mn 2ϩ concentrations in a range of 0.1-1 mM. At a concentration of 0.5 mM Mn 2ϩ , 60-70% of the amplified hbpA genes encoded for active HbpA. Sequencing of randomly picked active or inactive clones showed that on average there were 1.2 amino acid substitutions per HbpA*. With respect to the base substitutions, transitions exceeded transversions by a factor of 2.
The mutant library was plated on substrate-containing medium, where active HbpA produces aromatic polymers. The improvement of enzyme activity is generally associated with a decrease of the K m towards the substrate (24). We used aromatic substrate concentrations that were lower than the K m of the parent enzyme. The color of the polymer formed depended on the screening substrate used; the polymer formed from 3-methoxycatechol was brownish, whereas the 3-tert-butylcatechol polymer was reddish. Fig. 2 shows E. coli JM101 growing on solidified LB medium containing 0.5 mM 2-tert-butylphenol and expressing either wild-type hbpA or hbpA T2 .
After the first round of mutagenesis, active clones were screened for increased activity on guaiacol and 2-tert-butylphenol. From each experiment (500 clones), we took 6 -8 clones, which showed increased color formation, and determined NADH oxidation in crude cell extracts as a function of the test substrate and the physiological substrate 2-hydroxybiphenyl. In cases where the activity towards the test substrate or the ratio between the activities for the test substrate and 2-hydroxybiphenyl was higher than for the parent enzyme, product formation was analyzed by reverse phase HPLC. 1 The levels of HbpA and HbpA* were checked with SDS-PAGE to correct for different expression levels.
Eight clones were initially selected following in situ screening on guaiacol. In five cell extracts, the recombinant protein level was increased, but the specific HbpA* activity remained constant. Two cell extracts contained HbpA* with a lower in vitro activity than the wild type enzyme. Sequencing revealed that both enzymes contained one amino acid substitution, which probably decreased the enzyme stability in vitro but not in vivo. One mutant monooxygenase was found to have an increased specific activity towards guaiacol and was named HbpA G1 .
Six clones were initially selected following in situ screening on 2-tert-butylphenol. Only one clone harbored an hbpA* gene with a base substitution that led to an amino acid exchange and a higher activity towards this substrate. This gene (hbpA T1 ) was used for another round of error-prone PCR and in situ screening. From two selected clones, we obtained one mutant monooxygenase with a higher activity towards 2-tert-butylphenol. This variant was named HbpA T2 .
Purification and Characterization of Mutant Enzymes-The hbpA T1 and the hbpA G1 genes each contained one and the hbpA T2 gene contained two base substitutions that led to an amino acid change (Table I). In addition, hbpA T1 and hbpA T2 each carried one silent mutation, and hbpA G1 carried 2 base substitutions that did not result in an amino acid change.
Wild-type and mutant HbpA were purified from recombinant E. coli JM101 harboring a pUC18 derivative carrying the corresponding hbpA gene. Using this system, HbpA and HbpA* could be overexpressed to about 20% of total cell protein. The enzymes were purified as tetramers to homogeneity with yields around 30%. Purity was confirmed by SDS-PAGE (Fig. 3), which showed only HbpA or HbpA* monomers.
The mutant enzymes followed Michaelis-Menten kinetics with all substrates tested as does wild-type HbpA. For KЈ m determinations, HbpA and HbpA* activities were measured by NADH oxidation at different substrate concentrations. All mutants showed an increased KЈ m towards the natural substrate 2-hydroxybiphenyl, whereas KЈ m remained unchanged for 2-sec-butylphenol. The KЈ m towards guaiacol was significantly decreased for all mutants (Table II).
Because HbpA and HbpA* showed uncoupling of NADH oxidation from substrate hydroxylation for all substrates tested, apparent turnover rates were determined by measuring product formation and/or substrate consumption with reverse phase HPLC (Table III). Using this method, HbpA G1 showed a 30% increased specific activity towards 2-hydroxybiphenyl, 2-sec-butylphenol, and guaiacol. Compared with HbpA, HbpA T2 showed half the activity towards 2-hydroxybiphenyl and a 12% lower activity towards 2-sec-butylphenol. At the same time, it revealed a 5-fold increase in activity towards 2-tert-butylphenol, the substrate on which the enzyme was screened, and twice the activity of HbpA towards guaiacol and salicylaldehyde.
Uncoupling of NADH Oxidation and Substrate Hydroxylation-The NADH oxidase activity of the mutant enzymes was determined spectrophotometrically in the absence of the aromatic substrate. At saturating concentrations of the coenzyme  In the presence of the aromatic substrate, the rate of NADH oxidation by HbpA or HbpA* generally exceeds the rate of substrate consumption, due to uncoupling of NADH oxidation from substrate hydroxylation (Table IV), thus lowering the hydroxylation efficiencies of these enzymes. Interestingly, whereas the hydroxylation efficiencies of HbpA and HbpA G1 were similar for each of the substrates tested, it was considerably higher for HbpA T1 and HbpA T2 . The uncoupling of substrate hydroxylation from NADH oxidation can have different origins (1,3). One possibility is that substrate is bound to the enzyme but not hydroxylated due to inefficient oxygen transfer from the flavin (C4a)-hydroperoxide. Alternatively, product remains bound to the enzyme and can not be hydroxylated but can induce the elimination of hydrogen peroxide (Fig. 4). To distinguish between substrate-and product-related uncoupling for HbpA and HbpA G1 , oxygen consumption was monitored in activity assays in the absence and in the presence of catalase. Whereas for 2-sec-butylphenol and guaiacol, most uncoupling could be ascribed to the substrates, for 2-hydroxybiphenyl more than half of the uncoupling could be attributed to the product 2,3-dihydroxybiphenyl. This was confirmed by incubating the enzymes with the reaction product 2,3-dihydroxybiphenyl, which resulted in an NADH oxidase activity of 1.9 units mg Ϫ1 protein for HbpA and 2.0 units mg Ϫ1 protein for HbpA G1 . With both enzyme variants, 2,3-dihydroxybiphenyl acted as a true nonsubstrate effector, since no product formation could be de-tected. In contrast, 3-sec-butylcatechol and 3-methoxycatechol hardly stimulated NADH oxidation in wild-type HbpA or any of the mutants.
Structure Homology Modeling-To assess the effects of the amino acid replacements in the mutant enzymes, a sequence alignment between HbpA (586 residues) and phenol 2-monooxygenase from T. cutaneum (PHHY; 664 residues), the most closely related enzyme with known three-dimensional structure (42), was performed. Fig. 5 shows the alignment with the three conserved sequence motifs with a putative dual function in FAD/NAD(P)H binding (2). Most sequence homology between HbpA and PHHY was found in the N-terminal part of the proteins, which consists of the FAD-binding and substratebinding domains and constitutes the enzyme active site. A sequence identity of 24.4% was calculated when only these parts of HbpA and PHHY were taken into account. There was less homology in the C-terminal part of the proteins, reducing the overall sequence identity to 20.1%. The only known function of the C-terminal domain of PHHY is its participation in subunit association (42).
Structure homology modeling with PHHY confirmed that HbpA consists of three domains. The FAD-binding and substrate-binding domains of both enzymes are structurally conserved, but the structure of the C-terminal domain of HbpA is more uncertain. Like p-hydroxybenzoate hydroxylase, which   contains no extra domain (43,44), bacterial HbpA contains considerably fewer surface loops than eukaryotic PHHY. Fig. 6 shows a model of the HbpA subunit, as obtained by using the "closed" subunit of PHHY (Protein Data Bank entry 1foh) as the template file. For PHHY it was reported that the two subunits in the homodimer do not have an identical conformation and that the largest difference involves a loop (residues 170 -210, PHHY numbering) which can act as a lid that opens and closes the active site (42). In HbpA, this active site loop (residues 142-164) is much shorter, but the model suggests that it is still able to cover the active site.
Further examination of the three-dimensional model of HbpA revealed that the amino acids Ile 244 , Val 368 , and Leu 417 that were changed in the variants are spread throughout the structure. Ile 244 is located in the substrate binding pocket and is rather close to the flavin ring. Interestingly, Ile 244 corresponds with Tyr 289 of PHHY. This tyrosine is believed to play an important role in catalysis by positioning the aromatic substrate for attack at the ortho position (42). Val 368 and Leu 417 are both located near the protein surface of the HbpA subunit, and Leu 417 is positioned far away from the substrate binding site. Val 368 corresponds with Ile 414 in PHHY and is part of a conserved helix, whereas Leu 417 corresponds with Val 480 in PHHY and is located in the beginning of the C-terminal domain. Although Ile 414 and Val 480 are far away from the dimer interface of PHHY, the possibility cannot be excluded that in HbpA, Val 368 and Leu 417 play a role in tetramer formation or tetramer stabilization. DISCUSSION Most members of the family of flavoprotein hydroxylases are involved in the degradation of aromatic compounds by soil microorganisms (45). However, the application of these redox enzymes is not restricted to the metabolism of pollutants in our environment. Due to their high regioselectivity, flavoprotein hydroxylases also have considerable potential in the synthesis of new fine chemicals.
Directed Evolution of HbpA-Assuming an electrophilic aromatic substitution reaction mechanism (3), we chose 2-tertbutylphenol and guaiacol (2-methoxyphenol) as model substrates for directed evolution of HbpA. These substrates differ significantly from the natural substrate: (i) the bulky side chain of 2-tert-butylphenol requires more room in the enzyme active site, and (ii) the methoxy group of guaiacol is more polar and withdraws more charge from the aromatic system (ϩM, ϪI compared with ϩM, ϩI) (46). Furthermore, so far no activity for the ortho-hydroxylation of guaiacol and 2-tert-butylphenol has been described. Microbial degradation of guaiacol proceeds only via demethylation (47)(48)(49), whereas chlorinated guaiacols can also be degraded via para-hydroxylation (50,51). Thus, HbpA variants with an increased catalytic activity towards guaiacol and 2-tert-butylphenol will allow the biotechnological production of the corresponding catechols and may yield information about the structure-function relationship of HbpA.
The key factor for a successful directed evolution experiment is an effective screening or selection procedure (24,52). With HbpA, the instability of the formed 3-substituted catechols offered a good basis for the development of a suitable in situ screening procedure. The time-dependent color formation could be efficiently used for qualitative estimation of enzyme activity directly after construction of the mutant library. The high reliability of our in situ screening procedure was illustrated by the fact that only 2-8 clones per round of mutagenesis had to be selected to obtain a successfully modified enzyme. Furthermore, the in situ screening was not restricted to the directed evolution of HbpA towards 2-tert-butylphenol. A second mutant library was screened on guaiacol, and enzymes with an increased activity towards this substrate could also be selected. This demonstrates that in situ screening provides an easy and rapid method for the detection of specific enzyme features if the corresponding assay can be applied on solid media.
Enzyme Kinetics-The mutant proteins, which were selected for higher activity towards 2-tert-butylphenol and guaiacol, were characterized with respect to their catalytic properties and substrate specificity. Mutations V368A/L417F changed the substrate reactivity of HbpA for the hydroxylation of different 2-substituted phenols, whereas mutation I244V reduced the substrate spectrum but increased the turnover rate. Interestingly, HbpA G1 (I244V) showed an increased activity with guaiacol but not towards salicylaldehyde, whereas HbpA T2 (V368A/ L417F) showed a doubled activity towards both phenols. This suggests that the substrate side chain causes mostly steric rather than inductive effects. This conclusion is supported by the results obtained with 2-sec-butylphenol. This substrate has a flexible side chain, which can move freely in several directions. This results in an approximately equal activity and KЈ m values for the wild-type enzyme and all mutants. Most enzymes in nature do not function under V max conditions but catalyze reactions at [S]/K m ratios between 0.01 and 0.1 (53). The catalytic efficiency under these conditions is described by the ratio k cat /K m . This ratio was determined for HbpA and its variants. KЈ m values for all mutant enzymes were decreased compared with wild-type HbpA towards guaiacol but increased for the physiological substrate 2-hydroxybiphenyl. This was even the case for HbpA G1 , which has an increased activity towards 2-hydroxybiphenyl. This suggests that evolutionary advantage can be more easily achieved with a low K m rather than a high activity, probably because environmental substrate concentrations are low.
A low K m for the substrate of interest is also important for the biotechnological production of 3-substituted catechols with HbpA variants. The substituted phenols as well as the formed catechols are highly bactericidal and inactivate the whole-cell biocatalyst. However, this problem can be solved by processes with integrated in situ product recovery or two-liquid-phase bioconversions (13,54,55). These processes are based on the principle that both substrate and product are present at low concentrations in the aqueous phase of the bioreactor. Therefore, a desired enzyme feature for these processes is a high activity at the lowest possible substrate concentration (i.e. a low K m ).
Location of Mutations in the HbpA Model-Lacking information on the three-dimensional structure of HbpA, an interpretation of the structural effects of the obtained amino acid substitutions is speculative. However, structure homology modeling with PHHY allows some statements on the effects of the modifications in the HbpA variants. Interestingly, Ile 244 of HbpA corresponds with Tyr 289 of PHHY (42) and Tyr 222 of phydroxybenzoate hydroxylase (56,57). In p-hydroxybenzoate hy- FIG. 5. Alignment of 2-hydroxybiphenyl 3-monooxygenase (HbpA) from P. azelaica HBP1 and phenol 2-monooxygenase (PHHY) from T. cutaneum containing the conserved sequence motifs of flavoprotein aromatic hydroxylases. Letters and symbols between the two sequences represent identical (letters) and similar (ϩ) residues. The consensus profiles according to Eppink et al. (2) shown below the alignment include strictly conserved residues in boldface letters. Uppercase letters are amino acid residues. Lowercase letters are as follows: hydrophobic residues (h); small residues (s); charged residues (c); all residues (x). -, gap. The shaded boxes mark the location of the amino acid substitutions in the HbpA mutants. droxylase, Tyr 222 interacts with the carboxylic moiety of the substrate and is critically involved in closing the active site, allowing efficient substrate hydroxylation (for a review, see Ref. 58). In PHHY, Tyr 289 has been shown to play an important role in orienting the substrate for attack at the ortho position by forming a hydrogen bond with the hydroxyl moiety of the phenol. Moreover, Tyr 289 favors the flavin out rather than the flavin in position through formation of a hydrogen bond between the N-3 of the isoalloxazine ring and the phenolic hydroxyl group (35). In HbpA, Ile 244 clearly cannot fulfill a similar function. However, its position close to the side chain of the phenolic substrate suggests a direct influence on the shape of the substrate binding pocket (Fig. 7). From the above considerations and the catalytic properties of the I244V variant, we suggest that substitution of Ile 244 by Val in HbpA has a small but significant effect on substrate binding due to the reduced size of the side chain.
Although the mode of binding of 2-hydroxybiphenyl in HbpA is unknown, it is reasonable to assume that it resembles the mode of binding of phenol in PHHY. This assumption is based on the conserved mode of binding of the FAD, the similar hydrophobic nature of the substrate binding pockets, and the fact that the active site base Asp 54 of PHHY is replaced by His 48 in HbpA (Fig. 7). This histidine is likely to play an essential role in activating the 2-hydroxybiphenyl molecule prior to the regiospecific electrophilic attack by the flavin (C4a)-hydroperoxide at the 3-position of the phenolic ring.
Uncoupling of NADH Oxidation from Substrate Hydroxylation-A common feature among flavoprotein aromatic hydroxylases is the uncoupling of substrate hydroxylation from NADH oxidation with the concomitant formation of hydrogen peroxide (59). This is also observed during hydroxylation of 2-hydroxybiphenyl with HbpA (1). Mutation I244V in HbpA G1 had no significant influence on the degree of uncoupling compared with the wild-type enzyme, but the mutations V368A/L417V in HbpA T2 decreased the uncoupling with all substrates. This is a remarkable result, because both these substitutions are located far away from the substrate binding site. From the properties of the single mutant HbpA T1 , it can be concluded that the improvement of the efficiency of hydroxylation is related to the V368A substitution. This amino acid is located in the FAD binding domain, which suggests an effect on the mobility of the flavin ring. Whether this results in a stabilization and/or improved positioning of the flavin (C4a)-hydroperoxide towards the substrate remains to be investigated. The mutation L417V is located in the third domain, which in the case of PHHY is thought to be involved in subunit interactions (42). The increased hydroxylation efficiency due to substitution V368A remained, but the activity towards 2-hydroxybiphenyl decreased significantly. In combination with the increased KЈ m and higher kЈ cat for guaiacol, we conclude that the altered catalytic properties of L417V are due mainly to a changed substrate binding. Detailed structural information will be necessary to understand how the substitutions effect this change. Clearly, given their location in HbpA, substitutions V368A and L417V would not have been obvious targets for rational protein design.
For wild-type HbpA and HbpA G1 with 2-hydroxybiphenyl as the substrate, uncoupling could be ascribed for more than 50% to the reaction product 2,3-dihydroxybiphenyl. This suggests that 2,3-dihydroxybiphenyl competes with 2-hydroxybiphenyl for binding to the reduced enzyme and induces the nonproductive heterolytic cleavage of the flavin (C4a)-hydroperoxide. This interpretation is supported by results from rapid reaction kinetics studies (3). In contrast, 3-sec-butylcatechol and 3-methoxycatechol have only minor effects on the total uncoupling of wild-type HbpA and HbpA G1 , indicating that these aromatic products do not interact strongly with the reduced enzyme.
Flavoprotein aromatic hydroxylases such as p-hydroxybenzoate hydroxylase have a mechanism to decrease the rate of flavin reduction by several orders of magnitude in the absence of an aromatic substrate, thereby preventing the wasteful consumption of NAD(P)H (60 -62). Other flavin enzymes such as 4-hydroxyphenylacetate 3-hydroxylase, PHHY, and HbpA are less efficient in this respect and show some residual NAD(P)H oxidation (1,63,64). For HbpA, it has been shown by stoppedflow absorption spectroscopy that flavin reduction is the ratelimiting step in this NADH oxidation (3). An increased NADH oxidase activity of the substrate-free enzyme, as determined for all HbpA mutants, is therefore most likely related to an increase in the rate of flavin reduction.
In conclusion, we have shown that the catalytic properties and substrate reactivity of HbpA can be improved by random mutagenesis. This is the first successful modification of a flavindependent monooxygenase by molecular evolution. We expect the mutants to be useful in new biocatalytic processes and the insights obtained to be helpful in further investigations of structure-function relationships of flavin monooxygenases.