Active Site Engineering of the Epoxide Hydrolase from Agrobacterium radiobacter AD1 to Enhance Aerobic Mineralization of cis -1,2-Dichloroethylene in Cells Expressing an Evolved Toluene ortho- Monooxygenase

Chlorinated ethenes are the most prevalent ground-water pollutants, and the toxic epoxides generated during their aerobic biodegradation limit the extent of transformation. Hydrolysis of the toxic epoxide by epoxide hydrolases represents the major biological detoxification strategy; however, chlorinated epoxyethanes are not accepted by known bacterial epoxide hydrolases. Here, the epoxide hydrolase from Agrobacterium radiobacter AD1 (EchA), which enables growth on epichlorohydrin, was tuned to accept cis-1,2-dichloroepoxyethane as a substrate by accumulating beneficial mutations from three rounds of saturation mutagenesis at three selected active site residues, Phe-108, Ile-219, and Cys-248 (no beneficial mutations were found at position Ile-111). The EchA F108L/I219L/C248I variant coexpressed with a DNA-shuffled toluene ortho-monooxygenase, which initiates attack on the chlorinated ethene, enhanced the degradation of cis-dichloroethylene (cis-DCE) an infinite extent compared with wild-type EchA at low concentrations (6.8 microm) and up to 10-fold at high concentrations (540 microm). EchA variants with single mutations (F108L, I219F, or C248I) enhanced cis-DCE mineralization 2.5-fold (540 microm), and EchA variants with double mutations, I219L/C248I and F108L/C248I, increased cis-DCE mineralization 4- and 7-fold, respectively (540 microm). For complete degradation of cis-DCE to chloride ions, the apparent Vmax/Km for the Escherichia coli strain expressing recombinant the EchA F108L/I219L/C248I variant was increased over 5-fold as a result of the evolution of EchA. The EchA F108L/I219L/C248I variant also had enhanced activity for 1,2-epoxyhexane (2-fold) and the natural substrate epichlorohydrin (6-fold).

TOM also oxidizes TCE primarily to Cl − and CO 2 in vivo (22,23) and aerobically degrades various other chlorinated ethenes (20,24,25). TOM-Green originated from the first DNA shuffling of a non-heme monooxygenase (TOM) and has enhanced activity for both TCE degradation and naphthalene oxidation due to a single amino acid substitution, V106A, in TomA3 (26). In contrast to GSTs, which require glutathione as the cofactor for their enzymatic activity (27), EHs do not require a cofactor (1). Unfortunately, there are no EHs of microbial origin known to have activity toward chlorinated epoxyethanes. Nevertheless, a number of microorganisms contain EHs with various substrate ranges (28)(29)(30)(31)(32), and various directed evolution and rational protein engineering techniques may be used to alter enzymatic activity (33,34). Hence, it was investigated here whether an epoxide hydrolase could be tuned to accept chlorinated epoxyethanes as a substrate. template and sites F108 and I219 were randomized individually. In the third round, pBS(Kan)EH F108L/C248I (containing amino acid substitutions F108L and C248I in EchA) was used as the template and site I219 was subjected to saturation mutagenesis. Two degenerate PCR fragments were produced for each site with 463 bp and 749 bp for site F108, 457 bp and 754 bp for site I111, 800 bp and 414 bp for site I219, and 853 bp and 327 for site C248. After purifying from agarose gels, the two fragments for each site were combined at a 1:1 ratio as templates to obtain the full-length PCR product with the EH Front and EH Rear primers. The resulting randomized PCR product (1167 bp) was cloned into pBS(Kan)EH after double digestion with KpnI and SacI, replacing the corresponding fragment in the original plasmid. IPTG (0.5 mM) was added along with 5 mM sodium succinate (as a substrate to produce NADH). After 2 hrs of incubation at 37°C and 250 rpm, the whole-cell reaction was quenched by heating the vials in boiling water for 90 sec and centrifuging (16,000 × g, 4 min) to collect the supernatant. Chloride irons

Rui et al., UConn
concentrations in 500 µL of supernatant were measured spectrophotometrically at 460 nm as indicated above. Cells contacted with the same amount of DMF were used as the negative control, and at least three independent experiments were analyzed.
To determine the kinetics of cis-DCE mineralization, the OD of the cell culture was 1.2, and the initial cis-DCE concentrations were 6.8 to 540 µM (using different stock solutions of 6.25 mM, 25 mM, 125 mM, and 500 mM in DMF at 0.2-0.4 vol%). The supernatant chloride ion concentrations generated from mineralizing cis-DCE for each concentration were measured at 9 min for 6.8 µM and 13. Fractions with purified EchA (with the highest EH activity and the highest purity as visualized on SDS-PAGE) were pooled and dialyzed against TEMAG buffer overnight. The final product was stored at -20˚C with glycerol (10% v/v) for future use; variant F108L/I219L/C248I and wild-type enzyme were purified from 10% to 80% and 90%, respectively. Activity of column fractions was determined using a polypropylene 96-well plate format with styrene oxide (5 mM) as the substrate rather than the reported epichlorohydrin (35). Column fractions (10 µL) added to 136 µL TE buffer were incubated with 5 mM styrene oxide at 37˚C for 15 minutes followed by the sequential addition of 100 mM 4- Homology structural modeling. The three dimensional coordinates of the EchA variants were generated with SWISS-MODEL Server (50-52) using a structure model of wild-type EchA as the template (36), and visualized with Swiss-PdbViewer (50)(51)(52). The use of the structure model instead of the original X-ray structure of EchA as the structural template for homology modeling was due to the fact that the X-ray structure was obtained from an inactive enzyme possibly with false crystal packing forces, which resulted in one of the catalytic triad residues, Asp246, to be positioned outside of the active site (36); here we used the EchA structure model with the loop containing Asp246 rebuilt in the more likely active conformation of EchA (36) both because it represents a common picture of active site of α/β hydrolase-fold enzymes (36) and was further confirmed by mutagenesis studies (37).

RESULTS
Plasmid construction. To create clone libraries via electroporation and reliably screen them for enhanced EH activity, a stable plasmid that expresses EchA constitutively, pBS(Kan)EH (Fig. 2), was constructed that utilizes a constitutive lac promoter and kanamycin resistance gene. Use of kanamycin circumvents segregational instability and avoids feeder colonies that are associated with ampicillin resistance vectors.
The resulting epoxide hydrolase expressed in E. coli TG1 had activity towards its natural substrate epichlorohydrin (10 mM) based on the preliminary EH assay using 4-nitrobenzylpyridine (data not shown). TOM-Green was expressed from pMMB206-TOM-Green (Fig. 2), a wide-host-range, low-copynumber vector that is compatible with pBS(Kan)EH. individually on the four EchA sites F108, I111, I219, and C248, which we chose based on their close vicinity to the catalytic triad residues (D107, D246, and H275; Fig. 3) (36). Two of these residues, F108 and C248, were hypothesized previously to influence substrate binding in this or a related enzyme, although mutagenesis was not performed at these sites (2,36). By cloning DNA fragments from saturation mutagenesis back into the corresponding position of pBS(Kan)EH, all possible amino acids were introduced at the three sites respectively. A library containing ~2,000 colonies for each site was obtained and about 300 of those colonies were screened in 96-well plates for cis-DCE degradation since 292 independent clones from saturation mutagenesis at one site need to be screened for a 99% probability that each possible codon has been tested (49).
Whole cells expressing the EchA variants and TOM-Green with enhanced cis-DCE mineralization, as indicated by increased Cl − released, were found from three of the mutagenesis libraries, and the beneficial amino acid substitutions were F108L, I219F, and C248I, indicating each of these positions is important for adapting EchA to the substrate cis-DCE epoxide ( Table 2). No beneficial amino acid substitution was found at position I111. Although the three variants enhanced cis-DCE mineralization to a similar extent when co-expressed with TOM-Green (Table 2, 2.4 to 2.7-fold), the C248I mutation was slightly superior so it was used as a new template for a second round of saturation mutagenesis to combine the beneficial Rui et al., UConn mutations at positions F108 and I219; saturation mutagenesis was used to introduce the new residues at these positions rather than site-directed mutagenesis since it was not clear that how the three positions would interact.
Around 300 colonies from each of the two resulting libraries were again screened for improved cis-DCE mineralization activity using 96-well microtiter plates. The beneficial mutations that resulted in further improvements in cis-DCE mineralization from the two libraries were F108L/C248I (7.1-fold) and I219L/C248I (4.2-fold). As EchA F108L/C248I enhanced cis-DCE mineralization more than EchA I219L/C248I, it was used as the new template for a third round of saturation mutagenesis at position I219.
The same-sized library was screened, and four positive variants were found, all containing I219L. Thus, the best EchA variant for enhancing cis-DCE mineralization was created by three rounds of saturation mutagenesis with amino acid substitutions F108L, I219L, and C248I. The mutation I219F that was discovered in the first round of saturation mutagenesis as beneficial was lost in the further mutagenesis experiments, which indicates I219F might not be compatible with other mutations at C248 and/or F108.
The whole process shows that beneficial mutations can be quickly accumulated by multiple rounds of saturation mutagenesis and screening relatively small libraries. Because the mineralization of cis-DCE is the concerted reaction by both TOM-Green and EchA, whole cells were used. Naphthol synthesis assays were used to monitor TOM-Green activity in the cis-DCE degradation experiments of EchA mutants F108L/C248I and F108L/I219L/C248I to ensure the difference in cis-DCE mineralization rate was not caused by differences in TOM-Green activity. It was assumed that EchA should have no effect on naphthol formation either because no naphthalene epoxide was formed during the TOM-Green transformation or because naphthalene epoxide (if formed) was not within the

Rui et al., UConn
substrate range of EchA. TOM-Green activity was relatively constant with each EchA isoform (Table 3) at approximately 1 nmol/min•mg protein at 0.24 mM naphthalene.
In addition, the EchA expression levels of all the mutants listed in Table 2 was characterized using SDS-PAGE (41). The TOM-Green α (size 54.4 kDa) and β (size 37.7 kDa) subunits were clearly seen as well as EchA (size 34 kDa), and the expression level was the same for all the EchA mutants as well as for TOM-Green (data not shown). Hence, the enhancements in cis-DCE activity were not due to changes in protein expression.  Table 2. In comparing the enhancement of cis-DCE mineralization by EchA variant to the wild-type, the part of cis-DCE mineralized of by TOM-Green alone (in TG1/pMMB206-TOM-Green/pBS(Kan)) was subtracted as background signal as no EchA was involved. Table 2 that there was only slight increase in cis-DCE mineralization rate by wild-type EchA compared to the EchA − strain, indicating that cis-DCE epoxide is a poor substrate of wild-type EchA. Although the single mutation variants at the three separate sites (F108, I219, and C248) did not result in a large change in cis-DCE mineralization enhancement, combination of beneficial mutations did lead to a step-by-step improvement and finally brought about 10-fold enhancement in cis-DCE mineralization rate with the variant containing the triple mutations F108L/I219L/C248I ( Table 2). As the cell systems are isogenic, there was equivalent EchA protein expression level, and there was similar TOM-Green activity, these results indicate that the EchA mutants, especially F108L/I219L/C248I, were tailored to accept cis-DCE epoxide within their substrate range and to participate in the biological degradation of cis-DCE epoxide generated as the primary intermediate by TOM-Green.

Kinetics of cis-DCE mineralization by the best EchA variant.
EchA F108L/I219L/C248I co-expressed with TOM-Green was further characterized for enhancement in cis-DCE mineralization rate at different substrate concentrations, and the saturation constants, apparent V max and apparent K m for the coexpression system, were obtained (  Although we expected an enhancement in the cis-DCE degradation rate as well (initial disappearance rate), the parallel experiments monitoring cis-DCE degradation via GC did not show a significant difference in the initial degradation rates between the strains with wild-type EchA and the F108L/I219L/C248I variant (data not shown). For example, at an initial liquid cis-DCE concentration of 135 µM, about 55% cis-DCE was consistently depleted within 38 min for both strains. However, for the F108L/I219L/C248I variant, the degraded cis-DCE was almost completely mineralized as indicated by the Cl − production, while only 36% of the degraded cis-DCE was mineralized with wild-type EchA. As the two strains are isogenic with only three amino substitutions, the enhanced Cl − formation arises from the additional conversion route of cis-DCE epoxide by the evolved EchA (Fig. 1).

Enhanced 1,2-epoxyhexane and epichlorohydrin hydrolysis.
To obtain direct evidence that the EchA isoforms were functionally expressed in the system, EH activity towards an epoxide was examined.
Though cis-DCE epoxide would be the best substrate for this study, it is commercially unavailable and difficult to synthesize and utilize (16), so 1,2-epoxyhexane, a good substrate of wild-type EchA (35), was chosen as the alternative substrate to determine EH activity of wild-type EchA, EchAF108L/C248I, and EchA F108L/I219L/C248I. The same whole-cell system used for the cis-DCE mineralization experiments, TG1/pMMB206-TOM-Green/pBS(Kan)EH, was used for determining EH activity. For whole cells, there was a 2.1-fold increase in the 1,2-epoxyhexane activity by EchA F108L/I219L/C248I compared to the wild-type enzyme (Table 3). To corroborate these results, purified EchA was tested, and the k cat for 1,2- A six-fold improvement in epichlorohydrin hydrolysis was also obtained using purified enzymes (94 ± 8 µmol/min•mg for the F108L/I219L/C248I variant vs. 16 ± 2 µmol/min•mg for the wild-type enzyme).
Hence, EchA was optimized for more than just cis-DCE epoxide by the three mutations.

DISCUSSION
It is clearly shown in this paper that by active site engineering at carefully selected residues (EchA F108, I219, and C248) and by accumulating beneficial mutations via saturation mutagenesis, EchA was engineered to accept cis-DCE epoxide as a substrate. This is significant since the aerobic biodegradation of chlorinated ethenes requires the detoxification of the reactive epoxides formed as the primary intermediates after oxygenase attack. To our knowledge, this is the first report of protein engineering of epoxide hydrolases at these or analogous sites for any application. Aspergillus niger epoxide hydrolase (AnEH, PDB accession code 1QO7) (2). The structural model of a human microsomal epoxide hydrolase based on AnEH was also considered (2). These enzymes contain the canonical α/β hydrolase fold with conserved catalytic triad (2), indicating their common phylogenic origin; however, there are many structural differences due to their low sequence homology (20-30% amino acid identity in the core region) which yields an extremely-versatile substrate range (2). Based on the hypervariability at key structural residues that may contribute to the shape and substrate binding properties of the active site cavity (2,36,38,39), and in turn might affect the substrate specificity, we altered the active site residues Phe108 and Cys248 of EchA for acceptance of a new chlorinated substrate.

Rui et al., UConn
Phe108 is in close vicinity to the substrate (Fig. 3) as it is located next to the nucleophile Asp107, which initiates the hydrolysis reaction by attacking the substrate (36), and contributes to the formation of the structurally-conserved oxyanion hole, which is needed to stabilize the negatively-charged transition state occurring in hydrolysis (36). In addition, Phe108 has been suggested to be involved in substrate binding (36). Despite its structural and functional importance, the equivalent residues of Phe108 in the related enzymes vary considerably, with Trp125 in DhlA (39), Trp227 in human microsomal EH (mEH) (2), Trp334 in marine liver cytosolic EH (sEH) (38), Ile193 in AnEH (2), and Phe108 in EchA (36). Cys248 is one residue away from the catalytic acidic residue Asp246 (36). Its equivalent residue in AnEH, Cys350, is a constituent of the active site wall and was proposed to contribute to the geometry and character of the active site cavity (2). In addition, the side chain of Leu262, the equivalent residue in DhlA, appears to block the tunnel that connects the active site cavity with the outside solvent region (39).
Cys248 is also a hypervariant codon with the equivalent residues in other related enzymes as Cys350 in AnEH (2), Phe406 in human mEH (2), Val497 in sEH (38), and Leu262 in DhlA (39). We reasoned that mutating Cys248 may bring subtle effects on the specificity and reactivity of the enzyme.
Although there is no evidence showing that Ile219 interacts directly with substrate during the reaction nor has it been previously identified as influencing catalytic activity, we determined that it has van der Waals contact with both Phe108 and Tyr215 (within 4 Å; Fig. 3); Tyr215 was suggested to function as the proton donor in the catalytic mechanism of EchA (53) and was thought to direct initial substrate binding and positioning in the active center (2). As this Tyr residue role is conserved in other EHs (2), direct mutation at this residue could cause drastic changes in the active site properties, whereas we reasoned that mutation at Ile219, which interacts with Tyr215, could bring some subtle, beneficial effects. Change in the side chain of Ile219 was thought to bring slight changes in the position or orientation of Tyr215 as well as Phe108 and in turn could influence substrate binding.
We also tried saturation mutagenesis at position Ile111 as it is also in the vicinity of one of the catalytic residues, Asp107 (Fig. 3) and seems to be a hypervariant residue with Phe128 in DhlA (39), Phe196 in AnEH, and Leu230 in mEH (2). However we did not obtain any variant with enhanced cis-DCE mineralization when coexpressed with TOM-Green; hence, its role may be more structural than catalytic.
Concerted effects from the changes of the three residues (F108L/I219L/C248I) may optimize the size, shape, and hydrophobic character of the active site to facilitate binding and stabilization for cis-DCE epoxide and its transitional state intermediates. Interestingly, engineering EchA for the poor substrate cis-DCE epoxide also improved activity for both 1,2-epoxyhexane (Table 3) and epichlorohydrin. Hence, the substrate specificity of EchA may be extended further to epoxides of other chlorinated ethenes, such as TCE and tetrachloroethylene (PCE), by protein rational design or directed evolution. Further, in combination with metabolic pathway engineering, the chlorinated epoxyethanes may be channeled into productive metabolic pathways, potentially allowing chlorinated ethenes to be utilized as a sole carbon and energy source, since the inability of various chlorinated ethenes to support growth is not due to lack of energy during conversion (6), but because no suitable enzyme system is able to harvest the energy. (adapted from van Hylckama Vlieg and Janssen (6)). Steps 1 and 2 are the two possible spontaneous transformation pathways for cis-DCE epoxide, while step 3 (this work) and 4 (19) represent two major detoxification strategies in with cis-DCE epoxide may be biologically converted by either an epoxide hydrolase or glutathione S-transferase (IsoILR1).   EchA − : TG1/pMMB206-TOM-Green/pBS(Kan) 2 Determined via chloride ion release after 2 hr contact (cf. Table 3 with 67 min contact) 3 Total protein: 0.18 mg protein/mL·OD 4 Initial cis-DCE concentrations was 540 µM calculated based on Henry's Law with Henry's constant 0.17 (46) (1 mM were added as if all the volatile organic was in the liquid phase)  3 Total protein was 0.18 mg protein/(mL·OD) 4 Initial cis-DCE concentrations were calculated based on Henry's Law with a Henry's constant of 0.17 (46) 5 Determined via gas chromatography by monitoring 1,2-epoxyhexane degradation using whole cells (5 mM initial concentration) 6 Naphthalene was added at 5 mM although its solubility is 0.24 mM in water (54