Evolution of the Biphenyl Dioxygenase BphA from Burkholderia xenovorans LB400 by Random Mutagenesis of Multiple Sites in Region III*

It is now established that several amino acids of region III of the biphenyl dioxygenase (BPDO) (cid:1) subunit are involved in substrate recognition and regiospecificity toward chlorobiphenyls. However, the sequence pattern of the amino acids of that segment of seven amino acids located in the C-terminal portion of the (cid:1) subunit is rather limited in BPDOs of natural occurrence. In this work, we have randomly mutated simultaneously four residues (Thr 335 -Phe 336 -Ile 338 -Ile 341 ) of region III of Burkholderia xenovorans LB400 BphA. The library was screened for variants able to oxygenate 2,2 (cid:1) -dichlorobi-phenyl (2,2 (cid:1) -CB). Replacement of Phe 336 with Met or Ile with a concomitant change of Thr 335 to Ala created new variants that transformed 2,2 (cid:1) -CB into 3,4-dihydro-3,4-dihydroxy-2,2 (cid:1) -dichlorobiphenyl, which is a dead end metabolite that was not cleaved by BphC. Replacement of Thr 335 -Phe 336 with Ala 335 -Leu 336 did not cause this type of phenotypic change. Regiospecificity toward congeners other than 2,2 (cid:1) -CB that were oxygenated more efficiently by variant Ala 335 -Met 336 than by LB400 BPDO

Based on site-directed mutagenesis and family shuffling of genes, variant BPDOs with extended PCB degrading potency were obtained (6 -9). However, several persistent congeners that are of concern for health are not degraded, even by the most performing BPDO variants described to date. Amino acid residues of the C-terminal portion of the ISP BPH ␣ subunit (BphA) were found to influence the substrate selectivity of the enzyme (6 -9). A stretch of seven amino acid residues termed region III was of particular interest (6,9). The sequence of amino acids for region III of B. xenovorans LB400 BPDO, which is the wild-type enzyme exhibiting the highest PCB degrading potency, is Thr 335 -Phe 336 -Asn 337 -Asn 338 -Ile 339 -Arg 340 -Ile 341 (4). The sequence pattern of corresponding region is Ala 335 -Ile 336 -Asn 337 -Thr 338 -Ile 339 -Arg 340 -Thr 341 in Pseudomonas pseudoalcaligenes KF707 BphA1 (10) and Gly 335 -Ile 336 -Asn 337 -Thr 338 -Ile 339 -Arg 340 -Thr 341 in Comamonas testosteroni strain B-356 BphA (11). KF707 BphA BPDO shares more than 95% homology with LB400 BphA (9). KF707 BPDO and B-356 BPDO catalyze the oxygenation of a much narrower range of PCB congeners than LB400 BPDO. Nevertheless, replacement of region III of LB400 BphA by that of KF707 or of B356 created enzymes that oxygenated a much broader range of PCB congeners (6,8,9,12). This shows that region III amino acids are not the only ones influencing the catalytic activity toward chlorobiphenyl, but it also shows that the sequence pattern of region III of LB400 BphA is not optimal for catalytic activity toward chlorobiphenyls.
In addition to their influence on the range of chlorobiphenyls used as substrate by the enzyme, amino acids of region III also appear to be implicated in regiospecificity toward 2,2Ј-CB. Thus Suenaga et al. (12) found that replacing Ile 336 of KF707 BphA with Phe as in LB400 BphA changed the regiospecificity toward 2,2Ј-dichlorobiphenyl (2,2Ј-CB) to favor production of 2,3-dihydroxy-2Ј-chlorobiphenyl.
The ortho-substituted PCB congeners such as 2,2Ј-, 2,6-, and 2,6,2Ј,6Ј-CB are among the most resistant to microbial attack. LB400 BPDO is among the few BPDO of natural occurrence that can oxygenate 2,2Ј-CB efficiently (9). In previous work we found that evolved BPDOs selected for their ability to oxygenate 2,2Ј-CB at a higher rate than the parental LB400 BPDO were able to catalyze a broader range of PCB congeners than LB400 BPDO (6). Although the amino acid residues of region III BphA influence greatly the range of PCB congeners oxygenated by the enzyme, based on the alignment of sequences of BphAs that are found in data bases, the diversity of sequence patterns of this stretch of amino acids is rather limited (6). In this work, we have used an approach that combines a PCRbased method of random mutagenesis at targeted multi-sites plus DNA shuffling to create a library of BPDO variants that were randomly mutated simultaneously on four residues Thr 335 -Phe 336 -Asn 338 -Ile 341 of region of LB400 BphA. The purpose of the work was to evaluate the extent of variation that can occur in region III of BphA of LB400 with no loss of activity toward 2,2Ј-CB and the influence of new sequence patterns on catalytic activity toward this congener and other chlorobiphenyls. One of the most potent variant of the library (variant p4) exhibited a higher turnover rate of oxygenation of 2,2Ј-CB, but its regiospecificity toward this congener was altered. The rates of metabolism and the metabolite profiles of this variant toward a range of chlorobiphenyl congeners were compared with those obtained when LB400 BPDO catalyzed the reaction. The data suggest that despite the fact that residues 335 and 336 exert a strong influence on the regiospecificity toward 2,2Ј-CB and on the turnover rate toward several congeners, the pattern of chlorine substitution on the PCB substrate imposes its orientation toward the catalytic active center.
Preparation of DNA and PCRs to Evolve BPDO and Screening Protocol-Seven degenerated primers, each of which contained degenerated nucleosides at positions to be mutated, were used with appropriate antisense primers to amplify fragments of LB400-bphAE that overlapped over the stretch of DNA that encode for region III (Fig. 2). Primers 1 and 2 were used in conjunction with the reverse primer 3 to amplify the 1065-bp fragments A and B that corresponded to the right end portion of bphA plus bphE. Primer 1 and 2 were degenerated at bp positions 1021, 1022, and 1023 to introduce a mutation to replace Ile 341 of the protein by any other amino acid. In addition, primer 1 was degenerated at position 1008, and primer 2 was degenerated at bp position 1007. The base pairs at positions 1007 and 1008 belong to the codon encoding amino acid 336 (Phe) of the protein. Fragment A and B extended from the beginning of the DNA stretch that encoded for region III of BphA (sequence Thr 335 -Phe 336 -Asn 337 -Asn 338 -Ile 339 -Arg 340 -Ile 341 ) to the end of the gene. The five other primers, primers 4, 5, 6, 7, and 8, containing degenerated nucleotides were used in conjunction with primer 9 to amplify a 1023-bp fragment that corresponded to the left end portion of bphA. These five sets of primers amplified DNA fragments C, D, E, F, and G that extended from the beginning of bphA to bp position 1023, which corresponded to the end of region III of BphA (Fig.  2). Therefore fragments C, D, E, F, and G overlapped with fragment A and B inside the DNA stretch that encoded region III. Primer 4 was degenerated at bp positions 1006, 1007, and 1008 to mutate Phe 336 of the protein; primer 5 was degenerated at bp positions 1003, 1004, and 1005 to mutate Thr 335 of the protein; primer 6 was degenerated at bp positions 1012, 1013, and 1014 to mutate Asn 338 of the protein; primer 7 was degenerated at bp positions 1003, 1005, 1006, and 1008 to mutate Thr 335 and Phe 336 together; and primer 8 was degenerated at bp positions 1006, 1008, 1011, and 1014 to mutate Phe 336 and Asn 338 together.
The seven fragments that overlapped inside region III were shuffled to generate a library of 1238-bp fragments (Fig. 3). The DNA fragments were treated with DNase I to prepare the 50 -100-bp fragments for DNA shuffling (18). Taq DNA polymerase was used for the primerless PCR step. The conditions were set as recommended by Roche Applied Science except that the program was as follows for 60 cycles: 94°C for 45 s, 42°C for 45 s, and 72°C for 1.5 min. The primerless PCR product was PCR-amplified using the following antisense oligonucleotides 10 (5Ј-ACGGCTGGGCCTACGACATC-3Ј) and 11 (5Ј-CTGCGCTTCGCGG-TAGTAGAACTGC-3Ј). Taq DNA polymerase was used, setting conditions as recommended by Roche Applied Science. The resulting library of 1238-bp fragments was digested with MluI and AvrII, and the resulting 864-bp MluI/AvrII fragments were ligated to pQE31[LB400-bphAE] previously deleted from its MluI/AvrII fragment. The library of plasmids was transformed into E. coli DH11S pYH31[LB400-bphF-GBC] cells that expressed the entire set of enzymes required to convert chlorobiphenyls into corresponding choro-HOPDAs.
Transformants were inoculated on nylon membranes on the surface Mutagenesis at Targeted Multi-sites of bphA of LB agar plates (17). After 18 h of incubation at 37°C, the membranes were transferred to fresh LB plates containing 1 mM isopropyl ␤-Dthiogalactopyranoside (IPTG), the plates were incubated for 3 h at 37°C, and then crystals of 4-CB or of 2,2Ј-CB were placed into the lid of the Petri dishes. 4-CB is a chlorobiphenyl congener that is normally oxidized at rate similar or higher than biphenyl by most BPDOs of natural occurrence, including LB400 BPDO (8). It was therefore used to determine the proportion of the clones of the library expressing an enzyme for which the mutations that occurred in region III did not alter their capacity to oxygenate this easily oxidizable chlorobiphenyl. The plates were incubated at 37°C, and colonies exposed to 2,2Ј-CB were periodically inspected visually over a period of 48 h to search for those that developed a stronger yellow color than a control expressing LB400 BphA.
In some experiments, the metabolites were analyzed from catalytic oxygenation using reconstituted BPDO prepared from His-tagged purified enzymes components that were obtained following protocols pub-lished previously (2). Catalytic activities were determined from measurement of substrate depletion recorded by GC-MS analysis (15).

Screening for BPDO Variants Able to Oxygenate 2,2Ј-CB-To
randomly mutate targeted residues of region III simultaneously, we used the approach described under "Experimental Procedures" and summarized in Fig. 3. A library of E. coli transformants carrying pQE31[bphAE] plus pYH31[bphFGBC] was generated. Transformants were grown on nitrocellulose membranes and exposed to the vapors of a chlorobiphenyl. The screening assay was based on the production of yellow chloro-HOPDA by these transformants. In a first set of experiments, we found that 20% of the transformants were able to oxygenate the easily oxidizable 4-CB; 50% of them were unable to transform 2,2Ј-CB into the yellow chloro-HOPDA. Therefore approximately half of the variants that remained active toward 4-CB contained (a) mutation(s) that caused a loss of activity toward 2,2Ј-CB.
70,000 colonies of the library were exposed to 2,2Ј-CB vapors. Among the ϳ7,000 yellow-colored colonies, only seven exhibited a clearly more intense color than control colonies expressing parental LB400 BPDO. Based on the assumption that variant exhibiting a higher turnover rate of reaction toward 2,2Ј-CB would produce larger amount of chloro-HOPDA from 2,2Ј-CB, we retained these colonies for further examination.
We also noticed that a fraction of the 70,000 transformants examined exhibited a pinkish gray color rather than the expected yellow color when exposed to 2,2Ј-CB. The latter were presumed to generate a metabolite that BphC was unable to cleave. Therefore, these variants were expected to have changed their regiospecificity toward 2,2Ј-CB because it has been established that catalytic oxygenation of this congener by LB400 BPDO produces as major metabolite 2,3-dihydroxy-2Јchlorobiphenyl (20,21) that can be cleaved by BphC. These pinkish gray variants were also characterized.
All of the variants retained were examined for their capacity to degrade a mixture of 18 PCB congeners. Based on previous reports (20,(22)(23)(24), with the exception of 2,5,2Ј,5Ј-CB and 2,3,2Ј,3Ј-CB, all chlorobiphenyls of the mix are metabolized at a low rate by purified preparations of LB400 BPDO or by E. coli clones expressing this enzyme. For this reason, under our experimental conditions the depletion levels of these congeners by IPTG-induced E. coli cells expressing LB400 BPDO was lower than 20% after 18 h of incubation. These levels of depletion were too low to report statistically significant variations and were therefore not reported precisely (Table I). The mix of 18 congeners did not contain 2,2Ј-CB because it was not discriminated from 2,6-CB by our GC-ECD chromatography setting, but its degradation was demonstrated in other experiments (see below).
Characterization of Yellow-colored Variants-Seven yellow transformants were confirmed to metabolize 2,2Ј-CB at a higher rate than LB400 BPDO. Suspensions of IPTG-induced cells of these E. coli clones metabolized 2,2Ј-CB at rates ranging from 3.0 to 3.5 nmol/h compared with 1.5 nmol/h for E. coli cells expressing LB400 BPDO. Sequence analysis of these clones indicated that most of them were single or double mutants for which the amino acid sequence patterns of region III were identical or very similar to the patterns of variants ob-tained by Mondello et al. (9). Three variants (y7, y8, and y9) exhibited the sequence pattern Ala 335 -Phe 336 -Asn 338 -Ile 341 , two variants (y1 and y5) exhibited the sequence pattern Gly 335 -Phe 336 -Asn 338 -Ile 341 , and two variants (y10 and y11) exhibited the sequence pattern Ala 335 -Leu 336 -Asn 338 -Ile 341 . With respect to the range of PCB substrates degraded, the data confirmed those of Mondello et al. (9) where the replacement of Thr 335 by a small hydrophobic amino acid (Ala or Gly) did not extend the range of PCB congeners used as substrate (not shown). Replacement of Thr 335 -Phe 336 simultaneously by Ala-Leu created variant y10, which was able to oxygenate more efficiently doubly para-substituted congeners such as 4,4Ј-CB and 2,2Ј,4,4Ј-CB, confirming previous report (9). In addition, the y10 variant was able to oxygenate efficiently the doubly meta-substituted congeners such as 3,3Ј-CB that LB400 BPDO degraded poorly ( Table I).
Characterization of the Pinkish Gray-colored Variants-Approximately 50 of the 70,000 clones examined for their ability to convert 2,2Ј-CB into chloro-HOPDA exhibited a pinkish gray coloration. Seven were sequenced. Variant p1 was a triple mutant in which Thr 335 -Phe 336 -Ile 341 was replaced by Ala 335 -Leu 336 -Val 341 . Variant p8 was a double mutant where Thr 335 -Phe 336 was replaced by Ala 335 -Ile 336 . It is noteworthy that of the seven variants, five (p2, p3, p4, p5, and p7) were double mutants that shared an identical amino acid sequence pattern where Thr 335 was replaced by Ala and Phe 336 was replaced by Met.
Induced E. coli cell suspension expressing p1 BPDO metabolized 2,2Ј-CB at a rate of 1.5 nmol/h, which was similar to that of cells expressing LB400 BPDO. The rate of depletion of 2,2Ј-CB by cells expressing p8 BPDO was, however, much lower (0.38 nmol/h). Assays measuring the rate of depletion of 2,2Ј-CB by His-tagged purified preparations of p4 variant (Ala 335 -Met 336 ) revealed that its specific activity toward 2,2Ј-CB was in the range of 70 -90 nmol 2,2Ј-CB oxidized per min per mg enzyme compared with values of 35-55 nmol/ min/mg enzyme for LB400 BPDO. The specific activity of purified preparations of p4 BPDO was significantly higher than that of LB400 when biphenyl was the substrate. Values of 350 -400 nmol biphenyl converted per min per mg p4 BPDO were obtained, compared with 120 -150 nmol/min/mg LB400 BPDO. Therefore, the turnover rate of oxidation of biphenyl had increased significantly for this enzyme. It was also the case for several other PCB congeners that were oxygenated more efficiently by E. coli cells expressing p4 BPDO than those expressing LB400 BPDO (Table I). On the other hand, cells expressing p8 BPDO oxygenated 3,3Ј-CB more efficiently than those expressing LB400 BPDO, but overall, the range of chlorobiphenyls degraded was much smaller than for variant p1 and p4.
The ratio of 2,3-dihydroxy-2Ј-chlorobiphenyl to cis-3,4-dihydro-3,4-dihydroxy-2,2Ј-chlorobiphenyl was determined for all of the pinkish gray and yellow variants obtained in this study. The metabolite profile of all yellow variants was similar to that of LB400 BPDO (Table II). However, it had changed for all of the pinkish gray variants that generated a larger amount of cis-3,4-dihydro-3,4-dihydroxy-2,2Ј-dichlorobiphenyl than LB400 BPDO. No other metabolites than 2,3-dihydroxy-2Јchlorobiphenyl, cis-3,4-dihydro-3,4-dihydroxy-2,2Ј-chlorobiphenyl, and 3,4-dihydroxy-2,2Ј-dichlorobiphenyl were produced from 2,2Ј-CB by any of these variants. Table III and Fig. 4 show the metabolite profile obtained from selected chlorobiphenyls when LB400 and p4 BPDOs were used to catalyze the oxygenation. The metabolites were identified by GC-MS analysis in comparison with the available data reported in the literature. The table also shows the relative rate of oxygenation of these congeners by p4 BPDO compared with that of LB400 BPDO. LB400 BPDO oxygenates 3,3Ј-CB very poorly (20,23). There are two possible sites of attack of 3,3Ј-CB (carbons 5 and 6 or carbons 4 and 5). In previous report (24), based on the analysis of metabolites generated from 3,3Ј-CB by coupled reactions involving the four enzymes of the biphenyl catabolic pathway, 5,6-dihydro-5,6-dihydroxy-3,3Ј-dichlorobiphenyl was identified as the major metabolite, and 4,5-dihydro-4,5-dihydroxy-3,3Јdichlorobiphenyl was identified as the minor one. This was confirmed using C. testosteroni B-356 BPDO. A purified preparation of B-356 BPDO produced a single metabolite from 3,3Ј-CB whose GC retention time and mass spectral features were identical to the major metabolite produced from LB400 BPDO (not shown), and this metabolite is converted to 3-chlorobenzoate by C. testosteroni B-356 cells (25). Variant p4 produced the same two metabolites as LB400 BPDO in much larger amounts. When 1 nmol of enzyme was used to catalyze the reaction, the area under the GC-MS peaks of metabolites generated by p4 BPDO after 2 min of incubation was 1000 times higher than corresponding peaks of metabolites produced by 1 nmol of LB400 BPDO during the same time.
The co-planar congener 3,4,3Ј,5Ј-CB and the mono-ortho coplanar 2,4,3Ј,4Ј-CB are not oxygenated by purified preparations of LB400 BPDO or E. coli cells expressing the enzyme. A single metabolite identified as a trichlorinated dihydroxybiphenyl was detected when IPTG-induced E. coli cells expressing p4 BPDO were used to catalyze the oxygenation of 3,4,3Ј,5Ј-CB. It is, however, noteworthy that this metabolite was only detected when the substrate was added at a concentration lower than 10 M in the culture medium. Bedard et al. (26) have previously reported the fact that the more recalcitrant PCB congeners are degraded more efficiently when present at very low concentration. There is no clear explanation for this observation; however, substrate inhibition of BPDO reaction caused by uncoupling of NADH oxidation to substrate oxidation has been demonstrated for several chlorobiphenyls (27). Nevertheless, clearly, variant p4 has acquired the capacity to oxygenate this co-planar congener. Although the exact position of attack could not be determined, it is noteworthy that the oxygenation does not occur on the two chlorine-free ortho-meta carbons 5 and 6. Otherwise, the metabolite would have been tetrachlorinated.
IPTG-induced E. coli cells expressing p4 BPDOs produced a single metabolite from 2,4,3Ј,4Ј-CB in a small amount (representing a conversion yield of ϳ1-5% of the substrate). The mass spectral features of its butylboronate derivative were those of a tetrachlorinated dihydro-dihydroxybiphenyl (Table  III). However, we were unable to determine whether the oxygenation occurred on the 2,4-or 3,4-chlorinated ring.
Similar to previous congeners, no metabolites were detected when purified LB400 BPDO or E. coli clones expressing this enzyme were used to catalyze the oxygenation of 2,6-CB, confirming data previously reported (6,23). Variant p4 produced three metabolites from 2,6-CB (Table III). These include the same two dichlorinated dihydro-dihydroxy metabolites that were generated by variant II-9 described previously (6) plus one monochlorinated dihydroxybiphenyl. Although the precise identity of the two dichlorinated dihydro-dihydroxy metabolites could not be determined, the fact that the dihydroxymonochlorobiphenyl metabolite represented 25% of 2,6-CB metabolites showed that variant p4 BPDO oxygenated a nonnegligible portion of 2,6-CB on vicinal ortho-meta carbons of the chlorinated ring. This was unexpected based on the observation that p4 did not prefer an ortho-meta oxygenation on the chlorinated side of the 2,2-CB ring. The data suggest that struc-  tural factors that influence the regiospecificity toward 2,2Ј-CB are not the same as those that determine the regiospecificity toward the homologous congener 2,6-CB.

DISCUSSION
In this work we have used the in vitro molecular evolution approach to examine the contribution to regiospecificity and substrate turnover of amino acid residues of BPDO ␣ subunit region III. Screening was set up to select for variants exhibiting increased turnover rate or modified regiospecificity toward 2,2Ј-CB. When compared with LB400 BPDO, several of the selected variants had increased their rates of metabolism of several PCB congeners. In vitro molecular evolution is a very powerful approach to identify structural features associated to selected phenotypes. Many variants selected on the basis of the intensity of the yellow coloration produced from 2,2Ј-CB were similar or identical to those obtained previously (9) by exchanging residues of LB400 BphA by those of KF707 BphA1. The data suggest that the number of amino acid combinations of region III leading to the increased rate of oxygenation of 2,2Ј-CB accompanied by no change of regiospecificity are rather limited. The diversity of sequence patterns of region III of variants exhibiting a change in regiospecificity toward 2,2Ј-CB was also low, but two of the three sequence patterns analyzed conferred an increase of the turnover rate of oxidation toward a range of chlorobiphenyls.
Engineering BPDOs to catalyze the oxygenation of PCBs is a formidable task. The enzyme has to learn to catalyze the oxygenation of an array of structurally distinct compounds, which is determined by the position of their chlorine substitutes. Thus some congeners can take a co-planar configuration; others, the ones that are doubly ortho-substituted on the same ring, can never be co-planar. Regiospecificity toward each chlorobiphenyl congener is directed by its chlorine substitution pattern, through interaction between the chlorine atoms and specific amino acid residues of the protein. The identity of these amino acid residues and their interactions with various chlorobiphenyl congeners is still unknown. However, several amino acids of the C-terminal portion of BphA have already been identified as being critical for enzyme turnover rate and regiospecificity toward selected chlorobiphenyl congeners. Among them, the amino acids of region III of BphA and especially Thr 335 -Phe 336 -Ile 341 of LB400 BPDO have been found to influence the substrate turnover rate toward several congeners and the regiospecificity toward 2,2Ј-CB (6,9). Changing these critical amino acids alters the orientation that 2,2Ј-CB occupies inside the catalytic pocket, and in addition it can most likely change the conformation of the biphenyl ring.
Supportive evidence provided by this work and a previous study (12) shows that residue 335 and especially residue 336 are directly implicated in enzyme turnover rates and regiospecificity. Replacing Phe 336 by Met or Ile as in p4 (Ala 335 -Met 336 ) or p8 (Ala 335 -Ile 336 ) variants resulted in a change in regiospecificity toward 2,2Ј-CB. Suenaga et al. (12) showed that a variant of KF707 BPDO in which Ile 335 (corresponding to Phe 336 of LB400 BPDO) was replaced by Phe changed its regiospecificity toward 2,2Ј-CB. Furthermore, most of the yellow variants of our collection that catalyzed a predominantly 2,3 oxygenation had retained Phe 336 , except y10 (Ala 335 -Leu 336 ). Thus the residue occupying position 336 of LB400 BPDO influences directly the regiospecificity toward 2,2Ј-CB, and Phe seems to play a significant role in this matter. Based on crystallographic analysis of naphthalene dioxygenase, Suenaga et al. (12) have developed a three-dimensional model of KF707 BphA1. Their model suggested that Ile 335 of KF707 BPDO, which corresponds to Phe 336 of LB400 BPDO, is very close to the active iron center. This residue as well as its immediate neighbors is thus likely to play an essential role in enzyme activity. The mechanism by which these residues interact with the substrate to influence access and orientation of the substrate toward the catalytic site is still unknown. However, data show that the amino acids that are found at these positions in wild-type enzymes are not those that provide the optimal structural features to metabolize an extended range of PCB congeners efficiently; Ala 335 -Met 336 variants were in this respect the most active.
There were major differences between variants p4 (Ala 335 -Met 336 ) and p8 (Ala 335 -Ile 336 ) with respect to their turnover rates of oxidation toward chlorobiphenyls. Variant p8 metabolized the para-substituted congeners poorly. The PCB degrading potency of variants exhibiting the Ala 335 -Ile 336 -Ile 341 sequence pattern has never been examined before. Mondello et al. (9) have examined the PCB degrading potency of a variant exhibiting an Ala 335 -Ile 336 -Thr 341 sequence pattern. This variant had acquired the capacity to oxygenate 4,4Ј-CB and other doubly para-substituted congeners. Based on these results, 341 in addition to residues 335 and 336 would be able to modulate the turnover rate of oxygenation of a range of chlorobiphenyls. Furthermore, comparison of the regiospecificity toward 2,2Ј-CB of variant y10 (Ala 335 -Leu 336 -Ile 341 ) with that of variant p1 (Ala 335 -Leu 336 -Val 341 ) further substantiate that residue 341 influences this enzyme feature.
The pinkish gray variants had changed their regiospecificity toward 2,2Ј-CB favoring oxygenation of carbons 3 and 4. At this time it is not clear why no 5,6-dihydro-5,6-dihydroxy-2,2Ј-dichlorobiphenyl was detected among the metabolites produced by 2,2Ј-dichlorobiphenyl because typically BPDO prefers ortho-meta dioxygenation of the phenyl ring. Although the catalytic activity toward a range of PCB congeners was increased considerably in p4, including doubly ortho-substituted and co-planar chlorobiphenyls, it is quite interesting that the regiospecificity toward congeners other than 2,2Ј-CB was the same for both p4 and LB400 BPDO. It is especially interesting to point out that despite the fact that the chlorinated side of the biphenyl ring was not the pre- a The ratios were determined from the relative amounts each metabolite produced in the culture media of E. coli cells expressing the indicated BPDO and are based on the area under the peak of each metabolite obtained from GC-MS analysis. monoCldiOH, 2,3-dihydroxy-2Ј-chlorobiphenyl; diCldiol, cis-3,4-dihydro-3,4-dihydroxy-2,2Ј-dichlorobiphenyl.  ferred site of attack of 2,2Ј-CB by p4 BPDO, the single metabolite obtained from 2,4,2Ј,4Ј-CB was the same as the one obtained when LB400 BPDO catalyzed the reaction, and it was identified as a trichlorinated dihydroxybiphenyl. On the other hand, 2,3,2Ј,3Ј-CB was oxygenated principally on the meta-para carbons by LB400 BPDO. The mutation that occurred in p4 did not alter the regiospecificity of the enzyme toward this substrate. It is noteworthy that the two symmetrical ortho-meta-substituted congeners 2,5,2Ј,5Ј-CB and 2,3,2Ј,3Ј-CB are both oxygenated principally on meta-para carbons by both LB400 and p4 BPDO. Altogether, data suggest that despite the fact that residues 335 and 336 exert a strong influence on the regiospecificity toward 2,2Ј-CB and on the turnover rate toward several congeners, the pattern of chlorine substitution on the PCB substrate imposes its orientation toward the catalytic active center. We will need to wait for the crystal structure of the enzyme to understand how the residues of region III interact with individual congeners to influence their rate of oxygenation and, for some of them, their site of oxygenation. However, by analogy with cytochrome P450 cam (28), it is likely that hydrophobic interactions or hydrogen bonding between the chlorine substitutes on the biphenyl skeleton and specific amino acid residues of the protein influence the orientation of the biphenyl ring inside the catalytic pocket. The identification of the amino acid residues involved in these interactions are mandatory to design strategies to broaden the range of PCB congeners that the enzyme can oxygenate efficiently.