The Molecular Basis for Inhibition of BphD, a C-C Bond Hydrolase Involved in Polychlorinated Biphenyls Degradation

The microbial degradation of polychlorinated biphenyls (PCBs) by the biphenyl catabolic (Bph) pathway is limited in part by the pathway's fourth enzyme, BphD. BphD catalyzes an unusual carbon-carbon bond hydrolysis of 2-hydroxy-6-oxo-6-phenylhexa-2,4-dienoic acid (HOPDA), in which the substrate is subject to histidine-mediated enol-keto tautomerization prior to hydrolysis. Chlorinated HOPDAs such as 3-Cl HOPDA inhibit BphD. Here we report that BphD preferentially hydrolyzed a series of 3-substituted HOPDAs in the order H > F > Cl > Me, suggesting that catalysis is affected by steric, not electronic, determinants. Transient state kinetic studies performed using wild-type BphD and the hydrolysis-defective S112A variant indicated that large 3-substituents inhibited His-265-catalyzed tautomerization by 5 orders of magnitude. Structural analyses of S112A·3-Cl HOPDA and S112A·3,10-diF HOPDA complexes revealed a non-productive binding mode in which the plane defined by the carbon atoms of the dienoate moiety of HOPDA is nearly orthogonal to that of the proposed keto tautomer observed in the S112A·HOPDA complex. Moreover, in the 3-Cl HOPDA complex, the 2-hydroxo group is moved by 3.6 Å from its position near the catalytic His-265 to hydrogen bond with Arg-190 and access of His-265 is blocked by the 3-Cl substituent. Nonproductive binding may be stabilized by interactions involving the 3-substituent with non-polar side chains. Solvent molecules have poor access to C6 in the S112A·3-Cl HOPDA structure, more consistent with hydrolysis occurring via an acyl-enzyme than a gem-diol intermediate. These results provide insight into engineering BphD for PCB degradation.

Polychlorinated biphenyls (PCBs) 4 were manufactured extensively in the 20th century for industrial and commercial applications, including use in electrical transformers, hydraulics, and plasticizers. 5 Although banned in the United States since 1977, environmentally persistent PCBs have been linked to cancer (2), childhood neurodevelopmental deficits arising from prenatal exposure (3), and a host of other effects attributed to endocrine disruption (4). Indeed, concerns about high concentrations of PCBs and other contaminants in some freshwater fish have prompted a recent comprehensive risk-benefit analysis of fish consumption (5).
The inhibition of Bph enzymes has recently taken on additional significance in light of discoveries in Mycobacterium tuberculosis, the etiological agent of tuberculosis. Specifically, homologues of the Bph enzymes are involved in cholesterol catabolism (15) and are critical for the survival of the pathogen in the human macrophage (16). The lack of human homologues of the cholesterol-degrading enzymes suggests that they are promising drug targets. Understanding the mechanism of BphD inhibition will facilitate the engineering of the enzyme to improve its activity toward chlorinated substrates and should inform the development of inhibitors of HsaD, the cholesterol-degrading homologue in M. tuberculosis.
BphD catalyzes the hydrolytic C-C bond cleavage of HOPDA ( Fig. 1) and has features typical of the ␣/␤-hydrolase superfamily (17,18), including the fold and conserved active site catalytic triad composed of serine, histidine, and aspartate residues. To expel the electron-rich dienoate moiety of the substrate, the enzyme first employs a His-265-mediated enol-keto tautomerization to generate a hydrolyzable keto-intermediate (E⅐S k ) (19 -24). Indeed, His-265-dependent formation of an intermediate with a red-shifted absorbance spectrum (E⅐S red ) is formed at a similar rate in both WT and S112A BphDs (19). Crystallographic data for the E⅐S red intermediate trapped in the S112A⅐HOPDA complex are most consistent with the bound HOPDA being ketonized (19), although a non-planar, distorted conformation of the enol/enolate could not be completely ruled out. Hydrolysis then proceeds via either a gem-diol intermediate (24 -28) or an acyl-enzyme intermediate (19,20). As noted above, HOPDAs that are chlorinated on the dienoate moiety are poorly transformed by BphD (8 -10). Specifically, 5-chlorination reduced the maximal rate of BphD by 3-fold, and chlorination at the 3 or 4 positions reduced the maximal rate by ϳ10 3 and ϳ10 4 -fold, respectively. Although 4-Cl HOPDA is the least efficiently transformed monochlorinated HOPDA, 3-Cl HOPDAs represent a more significant roadblock to PCB degradation in two respects. First, 3-Cl HOPDA (t1 ⁄ 2 ϳ 500 h) is more stable than 4-Cl HOPDA, which undergoes a non-enzymatic transformation to 4-OH HOPDA (t1 ⁄ 2 ϭ 2.8 h) followed by degradation to products that include acetophenone (t1 ⁄ 2 ϳ 180 h) (8). Second, 3-Cl HOPDA is a very poor substrate of BphD LB400 , effectively inhibiting the hydrolysis of HOPDA (K ic ϭ 0.57 M) more potently than either 4-Cl HOPDA (K ic ϭ 3.6 M) or 4-OH HOPDA (K ic ϭ 0.95 M). It has been unclear whether the 3-Cl substituent impairs tautomerization or hydrolysis and whether the defect is due to an electronic or steric effect.
Herein we investigated the basis for inhibition of BphD by 3-Cl HOPDA. First, the ability of BphD to hydrolyze a series of 3-substituted HOPDAs was studied using steady-state kinetics to compare steric versus electronic effects. Second, single turnover stoppedflow kinetic analysis was used to probe the effects of 3-substitution on E⅐S red formation in both WT and hydrolytically impaired Ser112Ala variants. Finally, crystal structures of the S112A⅐3-Cl HOPDA and S112A⅐3,10-diF-HOPDA complexes were analyzed to reveal the structural correlates of the kinetic behavior.
Preparation of 4,4Ј-diF DHB-A plate of W medium (30) was streaked with frozen stock of Pandoraea pnomenusa (formerly Comamonas testosteroni) B-356 and incubated with biphenyl crystals in the lid at 30°C until colonies were visible (ϳ5 days). Several colonies were added to 3 ml of W medium containing 1 mg biphenyl and incubated at 30°C and 250 rpm until cloudy (ϳ4 days). Alternatively, this step could be shortened to ϳ2 days if a 50-l aliquot of frozen stock was used to inoculate the 3 ml of culture. One liter of W medium containing 0.5 g of biphenyl in a 2-liter flask was inoculated with 1 ml of starter culture, and the mixture was incubated as above. When the optical density at 600 nm reached 1, the culture was carefully decanted to remove biphenyl crystals and then centrifuged for 10 min at 7000 ϫ g. The pellet was washed two times with potassium phosphate buffer (I ϭ 0.1 M, pH 7.5) to remove residual biphenyl and resuspended in 500 ml of buffer supplemented with 100 mg of 4,4Ј-diF-biphenyl and 7 mg of 3-chlorocatechol. The latter was included to inactivate DHBD (31) and thereby prevent enzymatic transformation of the produced 4,4Ј-diF-DHB. The culture was incubated as above, and 50-l aliquots were analyzed by high-performance liquid chromatography at 30-min intervals using a Prodigy ODS Prep column, 2.1 ϫ 250 mm (Phenomenex, Torrance, CA), operating at a flow rate of 1.5 ml/min. The mobile phase initially consisted of a 30:70 ratio of solvent A (0.5% aqueous H 3 PO 4 ) to solvent B (methanol) for the first 5 min of the run, then a gradient was used to achieve 100% B at 10 min. The retention time of 3-chlorocatechol was 2.6 min, 4,4Ј-diF-DHB eluted at 5.4 min, and 4,4Ј-diF-biphenyl eluted at 13 min. When the maximum concentration of 4,4Ј-diF-DHB in the culture was reached as judged by high-performance liquid chromatography (after ϳ3 h), the culture was filtered to remove undissolved starting material, and then extracted three times with ϳ200 ml of ethyl acetate. The pooled fractions were dried over anhydrous MgSO 4 and rotary evaporated to dryness. The crude extract was dissolved in an appro-
Steady-state Kinetic Measurements-Initial rates of BphDcatalyzed hydrolysis at varying substrate concentrations were obtained by monitoring the substrate absorbance maximum versus time in potassium phosphate buffer (I ϭ 0.1 M, pH 7.5) using a Varian Cary 5000 spectrophotometer equipped with a thermostatted cuvette holder (Varian Canada, Mississauga, Ontario, Canada) maintained at 25.0 Ϯ 0.5°C, controlled by Cary WinUV software version 2.00. Reactions were carried out in a 1-ml volume and were initiated by the addition of 5 l of an appropriately diluted enzyme solution. The 3-Me HOPDA was generated in situ by adding 80 g of polyhistidine-tagged DHBD to the cuvette containing 4-Me DHB. After obtaining the rate of background decay (0.5-1 min), BphD was added and the activity of the enzyme was determined by correcting for the background. The Michaelis-Menten equation was fit to the data using LEONORA (32). The molar absorptivity of 3,10-diF HOPDA (⑀ 438 ϭ 37.9 Ϯ 2.1 mM Ϫ1 cm Ϫ1 ) was calculated from its absorption spectrum after quantification by measuring the amount of dioxygen consumed in the ring-opening reaction of 4,4Ј-diF DHB catalyzed by DHBD. Dioxygen consumption was measured using a Clark-type polarographic oxygen electrode (Yellow Springs Instruments model 5301, Yellow Springs, OH) as previously described for other HOPDAs (33). The molar absorptivity of 3-Me HOPDA (⑀ 430 ϭ 18.7 Ϯ 0.4 mM Ϫ1 cm Ϫ1 ) was determined by recording the absorption spectrum of a known quantity of 4-Me DHB immediately after cleavage by DHBD.
Stopped-flow Spectrophotometry-Single turnover reactions (BphD ϭ 8 M; substrate ϭ 4 M) in the phosphate buffer described above were monitored using an SX.18MV stopped-flow reaction analyzer (Applied Photophysics Ltd., Leatherhead, UK) equipped with a photodiode array detector. The drive syringe chamber and optical cell were maintained at 25°C by a recirculating water system. Multiple wavelength data from the time courses of single shots were acquired using the Xscan software (Applied Photophysics Ltd.), then saved as CSV files in the RISC Pro-K software and exported to Excel where replicate measurements from at least four shots were averaged. Selected single wavelength datasets were then imported into the SX18MV software (Applied Photophysics Ltd.) where multiple exponential equations were fit to the data to obtain reciprocal relaxation times and amplitudes. Good fits were characterized by random variation in the residuals.
Crystallization and Preparation of Complexes-Crystals of the substrate-free S112A variant of BphD were grown at 20°C in 1.9 M sodium malonate, pH 7.0, by sitting drop vapor diffusion, as previously reported (19). Complexes were prepared by incubating crystals for 30 min in 60 l of reservoir solution supplemented with ϳ10 mM 3-Cl-HOPDA or ϳ65 mM 3,10difluoro-HOPDA. Crystals were prepared for flash freezing by serial transfer into solutions containing higher concentrations of sodium malonate (3.4 M and 3.7 M, pH 7.0) augmented with ϳ5 mM of 3-Cl HOPDA or trace amounts of 3,10-difluoro-HOPDA. The incubation time was 3-6 s per step. Crystals were frozen by immersion in liquid N 2 .
Diffraction Experiments and Structure Analysis-All diffraction data were acquired by the use of the SERCAT facilities at the Advanced Photon Source, Argonne National Laboratory. The x-ray wavelength was 1.0 Å, and crystals were maintained at ϳ100 K during data collection. Diffraction images were recorded by a MarMosaic 300 charge-coupled device detector (Mar USA, Inc., Evanston, IL). For each crystal, ϳ100 frames were collected with a 1°rotation per frame; exposure times were 1-10 s per degree. All images were processed using DENZO, and intensities were merged and scaled using SCALEPACK; both programs were from the HKL2000 program suite (34).
The initial model for both complexes included only the protein atoms from the crystal structure of the S112A⅐malonate complex (PDB code 2PU6) (19). Rigid body refinement performed in REFMAC (35) from the CCP4 package (36) was followed by iterative cycles of restrained atomic parameter refinement via REFMAC and manual density fitting using the molecular graphics program O (37). PRODRG (38) was used to construct structures of 3-Cl HOPDA, 3,10-difluoro-HODPA, and malonate for density fitting and establishment of refinement restraints. The bond lengths and bond angles of the ligands were restrained to values expected for the enol isomer. Torsion angles in the non-aromatic portion of the substrates were unrestrained. The stereochemical properties of the models and the hydrogen bonding were analyzed by using the programs PROCHECK (39) and REDUCE (40).

RESULTS
Steady-state Kinetics-To assess the basis of inhibition of BphD by 3-Cl HOPDA, the enzyme-catalyzed hydrolysis of 3,10-diF HOPDA and 3-Me HOPDA was studied using steadystate kinetics. The 10-fluoro substituent is not expected to greatly affect catalysis because HOPDAs with small, electronwithdrawing substituents (e.g. -Cl and -CF 3 ) at this position are hydrolyzed by BphD with k cat values within 25% of that for HOPDA (8,28). Significantly, reduction of the volume of the electronegative 3-substituent by 40% upon chlorine-to-fluorine substitution resulted in a 150-fold increase in k cat ( Table 1), indicating that larger 3-substituents interfere with catalysis. This conclusion was supported by the observation that the substrate with the largest 3-substituent, 3-Me HOPDA, had the lowest apparent k cat (Table 1). Although a K m value was reported for 3-Cl HOPDA (8), K m could not be determined for 3-Me HOPDA due to a combination of lower activity, lower molar absorptivity, and higher background decay rates, which together prevented reliable initial rate measurements at low substrate concentrations. Finally, we also determined the stability of HOPDAs in the enzyme reaction buffer (Table 1), but we discerned no obvi-ous correlation between either the electronegativity of the substituents and the nonenzymatic decomposition, or between the nonenzymatic and enzymatic reactions. In conclusion, k cat was strongly anticorrelated with the volume of the substituent at C3 and uncorrelated with electronegativity over the series of HOPDAs studied, suggesting that turnover is largely dictated by steric, not electronic, factors.
Stopped-flow Kinetics-To better elucidate the influence of 3-substituents on catalysis, stopped-flow experiments were conducted to resolve individual catalytic steps. In a previous single turnover ([E] Ͼ [S]) stopped-flow experiment using WT BphD and HOPDA, we observed rapid (ϳ500 s Ϫ1 ) formation of E⅐S red ( max ϭ 492 nm) from the free HOPDA enolate ( max ϭ 434 nm) (Fig. 2E) (20). A similar E⅐S red intermediate ( max ϭ 506 nm) was rapidly formed (ϳ500 s Ϫ1 ) and trapped in the hydrolytically impaired S112A variant (Fig. 2F), and crystallographically was observed in a non-planar conformation (19). In contrast, E⅐S red did not accumulate when 3-Cl HOPDA was used as a substrate ( Fig. 2A). Under single turnover conditions ( occurred at the absorbance maximum of the 3-Cl HOPDA enolate (432 nm) ( Table 2). This decay corresponded to a ϳ20% loss in absorbance and occurred together with a slight blue shift to 427 nm. This may correspond to formation of an enzyme-bound planar 3-Cl HOPDA enolate, E⅐S e . This species very slowly decayed (1/ 2 ϭ 0.0077 s Ϫ1 ) at a rate that matches the previously measured k cat value (Table 1). Interestingly, this rate is similar to that of tautomerization of HOPDA in solution as measured by deuterium exchange experiments (20,22). The absence of E⅐S red accumulation suggests that 3-Cl HOPDA tautomerization is slower than hydrolysis. In an attempt to observe E⅐S red accumulation for 3-Cl HOPDA, the S112A variant was used. This variant catalyzes E⅐S red formation at the same rate as WT BphD but possesses severely impaired hydrolytic activity (19). For 3-Cl HOPDA, E⅐S red did not detectably accumulate in this variant (Fig. 2B). Similar to the behavior of WT, S112A caused an initial decrease (1/ 1 ϭ 26 s Ϫ1 ) corresponding to ϳ15% loss in 3-Cl HOPDA absorbance (Table 2), albeit without a correlated blue shift. This was followed by a very small (Ͻ1% of total 3-Cl HOPDA absorbance) and relatively slow (1/ 2 ϭ 0.55 s Ϫ1 ) increase in absorbance at 432 nm. Subsequent decay was not detected on the time scale examined (Ͻ1 min). In summary, an E⅐S red spe-  and HOPDA (bottom). A, wild type plus 3-Cl HOPDA; B, S112A plus 3-Cl HOPDA; C, wild type plus 3,10-diF HOPDA; D, S112A plus 3,10-diF HOPDA; E, wild type plus HOPDA, from Ref. 20; and F, S112A plus HOPDA, from Ref. 19. Arrows indicate the direction of absorbance changes with time. Separate relaxations are indicated by numbered arrows, and correspond to the phases described in Table 2.

TABLE 1 Steady-state parameters of BphD hydrolysis of 3-substituted HOPDAs
Errors are less than 15%. cies of 3-Cl HOPDA was not detected in either wild type or S112A, suggesting that an E⅐S e complex is the predominant enzyme-bound species and that catalysis is limited by formation of E⅐S red . The improved turnover of 3,10-diF HOPDA compared with 3-Cl HOPDA (Table 1) implies that 3,10-diF HOPDA may be more readily ketonized by the enzyme. The reaction of BphD with 3,10-diF HOPDA yielded a detectable E⅐S red ( max ϳ 468 nm, Fig. 2C). The low intensity of this spectrum suggests that only a small amount of E⅐S red accumulates, and thus tautomerization may be only partially rate-limiting. Decay of the signal at the absorbance maximum of 3,10-diF HOPDA (438 nm) can be described by three phases ( Table 2): (i) a loss in absorbance (1/ 1 ϭ 51 s Ϫ1 ) analogous to the initial decay in the 3-Cl HOPDA reaction, whereby a slightly blue-shifted feature ( max ϭ 427 nm) was generated together with a ϳ30% loss in absorbance; (ii) a small (20% of total absorbance decrease) second decay (1/ 2 ϭ 7.6 s Ϫ1 ) that generated E⅐S red ( max ϳ 468 nm), and (iii) a final phase, 1/ 3 ϭ 1.3 s Ϫ1 .

3-X HOPDA
Although E⅐S red ( max ϳ 517 nm) accumulated to a greater extent in S112A than in wild type, stoichiometric formation of this species was not observed using 3,10-diF HOPDA (Fig. 2D). E⅐S red is formed quickly (1/ 1 ϭ 90 s Ϫ1 ), but consumes only ϳ35% of the substrate based on absorbance. This is followed by a minor relaxation of Ͻ1% total absorbance, which occurs as an increase at 438 nm (1/ 2 ϭ 1.1 s Ϫ1 ) and a corresponding decrease at 517 nm. The incomplete E⅐S red formation suggests that the 3-F substituent destabilizes E⅐S red relative to the E⅐S e complex and/or free substrate. Curiously, 3,10-diF HOPDA E⅐S red formation is faster in S112A than wild type, suggesting that the Ser-112 hydroxyl function may impede tautomerization by further stabilizing the E⅐S e complex relative to E⅐S red .
In summary, increasing the size of the 3-substituent appears to slow E⅐S red formation. The fluoro-substitution apparently slows tautomerization by similarly stabilizing both E⅐S red and an alternate E⅐S e binding mode. Although E⅐S red accumulation is minimal, tautomerization is not significantly slowed relative to hydrolysis, and therefore the overall rate of catalysis is not dramatically affected. In contrast, tautomerization is more severely impaired by chloro-substitution, which may preferentially stabilize E⅐S e relative to E⅐S red .
Crystal Structure of S112A⅐3-Cl HOPDA-The crystal structure of the binary complex of S112A with 3-Cl HOPDA was examined at 1.7 Å resolution. Tables 3 and 4 summarize the diffraction data and refinement statistics, respectively. With respect to the fold of the monomer and quaternary structure, the protein structure exhibits no important differences relative to the malonate (PDB entry 2PU6) and HOPDA complexes (PDB entry 2PUH) of S112A; these conclusions are also valid for the structure of the 3,10-diF HOPDA complex described below.
The binding of 3-Cl HOPDA to S112A and related electron density maps are illustrated in Fig. 3A. As shown in Fig. 3A, F o ϪF c maps, calculated before the substrate was added to the model, contained readily interpretable electron density for 3-Cl-HOPDA, which was modeled as the enol isomer, (2Z,4E)-3-Cl HOPDA, because the kinetic experiments did not detect accumulation of a species corresponding to S red in solution. Nevertheless, the nearly coplanar conformation of 3-Cl-HOPDA cannot exclude the presence of the keto isomer: the torsion angles do not uniquely predict which bonds are single and which are double. Fig. 3B shows that 3-Cl HOPDA and HOPDA bind in distinctly different conformations and slightly different overall positions. Thus, although the dienoate moiety, the 6-oxo group, and the 6-phenyl substituent occupy the same binding sites in the two complexes, the root mean square deviation is 0.86 Å for

Reciprocal relaxation times and amplitudes for single turnover reactions monitored by stopped-flow spectrophotometry at the substrate absorption maxima
The 3-Cl HOPDA reaction was monitored at 432 nm; 3,10-diF HOPDA was monitored at 438 nm. Errors are no more than 15%.    DECEMBER 14, 2007 • VOLUME 282 • NUMBER 50 the common atoms of the two ligands. Likewise, several interacting protein side chains, especially those of Asn-111, Met-171, Phe-175, Trp-266, and His-265, are in significantly different positions/conformations in the two complexes. These variations largely reflect a gross difference in the orientation of the dienoate moiety relative to the protein. In essence, the planes defined by atoms C1 through C5 in the two complexes are nearly orthogonal to each other, and the torsion angles about the C2-C3 and C1-C2 bonds differ by 155°and 90°, respectively. In consequence, the binding interactions of the 1-carboxylate and 2-hydroxo groups of 3-Cl HOPDA are remarkably different relative to those observed in the HOPDA complex. Dominant features of HOPDA binding include hydrogen bonds between both oxygen atoms of the 1-carboxylate group and the guanidinium group of conserved Arg-190, and a hydrogen bond between the 2-oxo/hydroxo oxygen and the side chain of His-265. Although the 1-carboxylate group of 3-Cl HOPDA is in the vicinity of Arg-190, it does not hydrogen bond with that residue. Rather, the 2-hydroxo group is moved by 3.6 Å from its position near His-265 in the S112A⅐HOPDA complex and hydrogen bonds with Arg-190. The Z conformation about the C2-C3 bond places the 3-Cl atom in an adjacent binding site formed by the non-polar side chains of Leu-156, Phe-175, and Phe-239 (see Fig. 3C), and at longer distances by the side chains of Pro-44 and Met-171, which is found in alternate conformations. Moreover, the chlorine atom and the side chain of Phe-175 lie between His-265 and the 2-hydroxyl group.

Inhibition of a C-C Bond Hydrolase
Accommodation of the bulky 3-Cl substituent in an appropriate non-polar site appears to be the major cause of the differences in conformation and position of 3-Cl HOPDA relative to HOPDA as well as adjustments of the protein. If 3-Cl HOPDA assumed the conformation and position of HOPDA in the crystal structure of its complex with S112A, the 3-Cl group would be in steric conflict with the backbone nitrogens of Gly-42 and Gly-43 (distances of Ͻ2.4 Å) and the 6-oxo-substituent (2.5 Å). Similarly, when the coordinates for 3-Cl HOPDA are mapped onto the HOPDA complex, the chlorine atom is placed 2.1 Å from the C of Phe-175. Relaxation of FIGURE 3. Comparison of the binding of 3-Cl HOPDA and HOPDA to S112A. A, stereoscopic view showing: the refined model of S112A⅐3-Cl HOPDA complex; the 2F o ϪF c electron density of the refined structure (cyan, contour level ϭ 1); and the F o ϪF c electron density before 3-Cl-HOPDA was added to the model (blue, contour level ϭ 3). The carbon atoms of Ala-112, Asp-237, and His-265 are colored brown, whereas carbon atoms in other protein residues and in the ligand are gold. N, O, S, and Cl atoms are colored blue, red, yellow, and green, respectively. B, stereoscopic view of superposed models for S112A⅐3-Cl-HOPDA and S112A⅐HOPDA emphasizing the active site. All carbon atoms and covalent bonds are colored in gold for S112A⅐3-Cl-HOPDA and in gray for S112A⅐HOPDA. Other atoms are colored as in A. C, stereoscopic view of S112A⅐3-Cl-HOPDA emphasizing interactions between the 3-Cl substituent and the protein. The dotted lines mark the closest approach to side-chain atoms of Leu-156, Met-171, Phe-175, and Phe-239. The labels include contact distances in Angstroms. the latter contact may motivate the observed differences in the conformations of His-265, Trp-266, as well as other nearby residues.
At least four characteristic features of the HOPDA complex are reproduced in the 3-Cl HOPDA complex: 1) the 6-oxo group binds in the oxyanion hole formed by backbone amides of Gly-42 and Met-113; 2) C5 and C6 remain within 3.8 Å of the C␤ atom of Ala-112; 3) there is inadequate room in the active site for a water molecule to approach C6 from either face; and 4) the 6-phenyl substituent interacts with the same non-polar side chains of the NP-site, including those of Ile-153, Leu-213, Trp-216, and Val-240. With respect to point 3, the substrate binding site includes no crystallographically ordered waters and no water-sized voids. The closest ordered water is 7.5 Å away from C6 and lies in a solvent-accessible pocket beyond the phenyl substituent.
The electron density maps also suggest an additional, minor binding mode in which the phenyl substituent occupies the expected binding site, but in a different orientation, and the rest of the ligand extends away from the catalytic residues toward a possible entrance to the active site near the side chains of Phe-157, Ala-208, and Pro-212. This mode was not modeled.
Crystal Structure of S112A⅐3,10-diF HOPDA-Although the crystal structure of S112A⅐3,10-diF HOPDA was determined at higher resolution than the 3-Cl HOPDA complex (1.57 Å versus 1.7 Å; see Tables 3 and 4), the development of a definitive model for the ligand proved to be more difficult. As described below, the assessment of many electron density maps and refinement experiments led us to conclude that the C1-carboxylate and C2-hydroxyl substituents bind in multiple, poorly resolved conformations. The final model includes 3,10-diF HOPDA in a single conformation that represents the most reliable observations and is similar to the conformation of 3-Cl HOPDA.
F o Ϫ F c difference maps prior to the addition of any HOPDA atoms showed continuous density commensurate with the entire ligand only at a relatively low contour level (ϩ2, where is the standard deviation from a mean of 0 e Ϫ /Å 3 ). In addition, the density was of inconsistent quality: features near the center of the binding site and corresponding to the expected positions of C5, C6, the 6-oxo group, and a few atoms of the 6-phenyl substituent were strong and well resolved, whereas the features for the more distal atoms of the phenyl ring and the dienoate moiety were weak and poorly resolved. The better density overlaps the known binding site for malonate, a major component of the mother liquor and all stabilization solutions. Thus, a difference map was calculated with coefficients F 3,10-diF-HOPDA obs Ϫ F malonate obs and phases from the structure of the malonate complex to validate the binding of 3,10-diF HOPDA and assess its conformation. This map unequivocally defined the p-fluorophenyl substituent (i.e. 10-F). Positive density was also observed in the region occupied by C2, C3, the 3-Cl substituent, and the carboxylate in the 3-Cl HOPDA complex, but the 2-OH group was not defined and the carboxylate group was poorly resolved. Similar features were observed in difference maps based on 2.0-Å resolution data acquired from a second S112A⅐3,10-diF HOPDA crystal using a CuK ␣ rotating anode x-ray source (data not shown).
From these observations, 3,10-diF-HOPDA was modeled and refined with the dienoate moiety in the same 2Z,4E conformation as 3-Cl HOPDA. The resulting F o Ϫ F c maps suggested an alternative location for the carboxylate group consistent with a 2E,4E conformer. To test this observation, maps were calculated with coefficients (F 3,10-diF-HOPDA obs Ϫ F 3-Cl-HOPDA obs ) and phases from the 3-Cl HOPDA complex using data from two S112A⅐3,10-diF HOPDA crystals. Both maps had negative features overlapping the carboxylate, 2-OH, and 3-Cl groups of 3-Cl HOPDA, and a positive feature confirming the 10-F substituent of diF-HOPDA (a negative peak is expected at the chlorine atom even if F occupies the same position because of the difference in atomic scattering power). There was no comparable difference density near C5, C6, and most of the atoms of the phenyl ring, indicating these portions of the HOPDA structure are consistently located in the diF-HOPDA and 3-Cl HOPDA complexes. These maps also had minor positive and negative features indicating that malonate was bound in a significant fraction of the active sites in the S112A⅐diF HOPDA crystals, as it is bound in the previously determined structure of S112A⅐malonate. The presence of malonate-associated density inhibited confirmation of the location of the 2-OH group as its expected location in the 2E,4E conformer is within 1.3 Å of the location of one of the oxygen atoms of malonate. Maps with coefficients (F 3,10-diF-HOPDA obs Ϫ xF malonate obs ), x 1.0, were calculated to obtain the equivalent of an F obs map with the contribution of the malonate removed and to estimate the fractional occupancy of 3,10-diF HOPDA. These maps also failed to establish the location of the 2-OH group or resolve the carboxylate group.
Based on this evidence, 3,10-diF HOPDA was ultimately modeled and refined at 60% occupancy as the 2Z,4E conformer in a conformation similar to that of 3-Cl HOPDA. We also tested all possible 2E,4E and 2Z,4E conformations and found none to be as compatible with the electron density. Taken as a whole, the electron density maps indicate the final model does not provide a full description of the active site structure and lead us to conclude that 3,10-diF HOPDA accesses multiple conformations in the crystal.

DISCUSSION
The molecular basis for the inhibition of BphD by 3-Cl HOPDA was studied by steady-state and transient-state kinetics, and x-ray crystallographic analysis of E⅐S complexes of 3-substituted HOPDAs. These studies revealed that (i) steadystate turnover is limited by steric, not electronic, features of the 3-substituent; (ii) large 3-substituents impair E⅐S red formation, and therefore tautomerization; and (iii) 3-substituted HOPDAs bind in a non-productive mode.
Steady-state kinetic studies demonstrated that steric bulk, not electronegativity, of the 3-substituent obstructs BphD catalysis. Incremental increases in the size of the 3-substituent incrementally slowed turnover: the enzyme preferentially hydrolyzed 3-substituted HOPDAs in the order H Ͼ F Ͼ Cl Ͼ Me. By contrast, there was no significant correlation between turnover and electronegativity of the 3-substituent. Stoppedflow spectrophotometry demonstrated that larger 3-substituents impede formation of the E⅐S red intermediate. For instance, DECEMBER 14, 2007 • VOLUME 282 • NUMBER 50

Inhibition of a C-C Bond Hydrolase
whereas E⅐S red is apparently completely formed from HOPDA, minor accumulation occurred from 3,10-diF HOPDA, and this intermediate was not detected when 3-Cl HOPDA was used as a substrate. Thus, large 3-substituents inhibit the formation of E⅐S red and ultimately disrupt tautomerization.
The kinetic data predict a non-productive binding mode for 3-substituted HOPDAs. For 3-Cl HOPDA, stopped-flow analysis reveals a two-step transformation: initial formation of an E⅐S e intermediate (1/ 1 ϭ 15 s Ϫ1 ), followed by its decay (1/ 2 ϭ 0.0077 s Ϫ1 ) to products, possibly via rate-determining formation of E⅐S red . For 3,10-diF HOPDA, a similar initial formation of E⅐S e (1/ 1 ϭ 51 s Ϫ1 ) was followed by two slower relaxations (1/ 2 ϭ 7.6 s Ϫ1 and 1/ 3 ϭ 1.3 s Ϫ1 ) during which a small amount of E⅐S red appeared to accumulate. The incomplete E⅐S red formation for S112A⅐3,10-diF HOPDA implies reversibility and suggests that the mechanism may be described by either three-step model in Fig. 4.
The S112A⅐3-Cl HOPDA crystal structure is consistent with the kinetic data, indicating that the 3-Cl HOPDA is bound nonproductively. More specifically, the structure reveals a substrate-binding mode (Fig. 3) wherein 3-Cl HOPDA binds in a C1-C6 coplanar conformation with the plane of the dienoate moiety orthogonal to that of the E⅐S red intermediate of the S112A⅐HOPDA complex (19). Although the crystallographic studies cannot determine the tautomeric state of the ligand in this case, the observed binding mode is consistent with the impaired E⅐S red formation of 3-Cl HOPDA observed by stopped-flow. In addition, the conformation is very similar to that observed for HOPDA in the complex with the S112A/ H265A mutant, which does not accumulate E⅐S red . The C1-carboxylate and C2-hydroxyl groups of 3-Cl HOPDA are in remarkably different positions relative to those observed for E⅐S red in the same S112A variant. Whereas both oxygen atoms of the HOPDA C1-carboxylate group hydrogen bond with the guanidinium group of Arg-190, this residue instead hydrogen bonds to the 2-oxo/hydroxyl group of 3-Cl HOPDA. As a result, the hydrogen bond between the 2-oxo/hydroxyl oxygen and the side chain of His-265 observed in the S112A complex with HOPDA is not possible for 3-Cl HOPDA. The crystal structure therefore illustrates that the preferred binding mode of 3-Cl HOPDA is incompatible with the proposed His-265-mediated proton transfer of the tautomerization reaction.
In light of the different isomeric states and binding modes of 3-Cl HOPDA and HOPDA to S112A, it is remarkable that the structures of neither complex contain a solvent molecule suitable for attack of the E⅐S C6 carbonyl in the active site, despite the extra space made available by removal of the serine hydroxyl: the closest water is more than 7 Å away. In this respect, the crystal structure of the S112A⅐3-Cl HOPDA complex is more consistent with a nucleophilic role for Ser-112.
The crystal structures also provide a possible explanation for the greater accumulation of E⅐S e for 3-Cl HOPDA relative to 3,10-diF HOPDA (Fig. 3C). The 3-Cl atom occupies a binding pocket formed by the non-polar side chains of Leu-156, Phe-175, Phe-239, and Met-171. Because these residues do not shift as a result of Ser-112 substitution, this binding pocket is probably available and also utilized in the wild-type enzyme. It is possible that the compatibility of chlorine versus fluorine with this hydrophobic pocket versus the solution may account for the greater accumulation of 3-Cl HOPDA in the E⅐S e binding mode, although this may also reflect the inability of 3-Cl HOPDA to access the E⅐S red conformation, as discussed below. Similarly, acetylcholine esterase binds huprine X such that a hydrophobic pocket is fully occupied by the inhibitor's chlorine atom, and substitution with fluorine reduces binding affinity (41,42). In the crystal, the binding of 3-Cl HOPDA is also monomorphic, whereas the carboxylate, 2-OH, and 3-F moieties of 3,10-diF HOPDA are not well ordered. This may reflect a difference in binding forces and/or a difference in the interaction between the 2-OH and 3X groups in the 2Z conformation. The proposed effects of 3-substituted HOPDAs on formation of the catalytically competent E⅐S red are summarized in Fig. 5. Unsubstituted HOPDA is rapidly converted to E⅐S red under single turnover conditions (1/ 1 ϳ 500 s Ϫ1 ), and the presence of an isosbestic point at ϳ460 nm suggests direct transformation (19,20). Large 3-substitu-  Boxes enclose structures that were crystallographically observed in the hydrolytically impaired S112A enzyme. Although the protonation state of E⅐S e is not known, the absorption spectrum is most consistent with an enzyme-bound enolate. ents may stabilize an alternate planar E⅐S e binding mode in which the 3-substituent occupies a hydrophobic pocket. In contrast, large 3-substituents are predicted to destabilize the non-planar E⅐S red conformation via a steric clash with the backbone of Gly-43 and/or an intramolecular steric conflict with other HOPDA atoms. Thus, access to E⅐S red is more severely impaired for 3-Cl HOPDA than for 3,10-diF HOPDA. Interestingly, the possibility of moderate destabilization of both E⅐S e and E⅐S red by the 3-F substituent is consistent with the S112A⅐3,10-diF HOPDA crystal structure indicating multiple binding modes for the carboxylate and 2-OH groups. Nevertheless, the association of the 6-oxo group with the oxyanion hole is maintained, which allows the scissile bond to approach the His-265 and Ser/Ala-112 residues involved in hydrolysis. Although kinetic constants were not determined for 3-Me HOPDA, the volume of the methyl substituent is only ϳ25% larger than chlorine, so it may also occupy the hydrophobic pocket.
The preference for 3-Cl HOPDA binding in the alternate planar E⅐S e mode instead of generating the productive, nonplanar E⅐S red intermediate has implications for both bioremediation of PCBs and the development of tuberculosis therapeutics. With respect to the former, the ability of the BphD homologue DxnB2 to hydrolyze 3-Cl HOPDA with a ϳ13-fold higher specificity may provide an opportunity to overcome this block in the Bph pathway (9). Nevertheless, the structural basis for the different activities of these enzymes is unclear: Gly-42 and Gly-43, which are proposed to clash with the 3-Cl substituent in the E⅐S red conformation, as well as the flanking residues are conserved between the two enzymes, as is Phe-74, the residue closest in space to Gly-43. Further studies of DxnB2 should reveal the basis for its improved turnover of 3-Cl HOPDA and may further guide protein-engineering efforts.