A Recently Assembled Degradation Pathway for 2,3-Dichloronitrobenzene in Diaphorobacter sp. Strain JS3051

ABSTRACT Diaphorobacter sp. strain JS3051 utilizes 2,3-dichloronitrobenzene (23DCNB), a toxic anthropogenic compound, as the sole carbon, nitrogen, and energy source for growth, but the metabolic pathway and its origins are unknown. Here, we establish that a gene cluster (dcb), encoding a Nag-like dioxygenase, is responsible for the initial oxidation of the 23DCNB molecule. The 2,3-dichloronitrobenzene dioxygenase system (DcbAaAbAcAd) catalyzes conversion of 23DCNB to 3,4-dichlorocatechol (34DCC). Site-directed mutagenesis studies indicated that residue 204 of DcbAc is crucial for the substrate specificity of 23DCNB dioxygenase. The presence of glutamic acid at position 204 of 23DCNB dioxygenase is unique among Nag-like dioxygenases. Genetic, biochemical, and structural evidence indicate that the 23DCNB dioxygenase is more closely related to 2-nitrotoluene dioxygenase from Acidovorax sp. strain JS42 than to the 34DCNB dioxygenase from Diaphorobacter sp. strain JS3050, which was isolated from the same site as strain JS3051. A gene cluster (dcc) encoding the enzymes for 34DCC catabolism, homologous to a clc operon in Pseudomonas knackmussii strain B13, is also on the chromosome at a distance of 2.5 Mb from the dcb genes. Heterologously expressed DccA catalyzed ring cleavage of 34DCC with high affinity and catalytic efficiency. This work not only establishes the molecular mechanism for 23DCNB mineralization, but also enhances the understanding of the recent evolution of the catabolic pathways for nitroarenes.

two strains provide an opportunity to study how pathways for the closely related isomers evolved in the same microbial community exposed to both DCNB isomers.
In this study, we elucidated the 23DCNB catabolic pathway in strain JS3051 via genome sequencing, whole-cell biotransformations, recombinant expression, and biochemical analyses. Comparative genome analysis, substrate specificity, site-directed mutagenesis, and structural analysis revealed the origins of the pathway and the factors that dictate the different recent origins of the 23DCNB and 34DCNB dioxygenases.

RESULTS
Genome of Diaphorobacter sp. strain JS3051. The complete genome of strain JS3051 comprises 4.6 Mb, consisting of one circular chromosome and three circular plasmids. More details of the genomic information are summarized in Table S1 in the supplemental material. The two identical 16S rRNA gene sequences of strain JS3051 share 100% identity to those of Acidovorax sp. strain JS42, Acidovorax ebreus strain TPSY, and Diaphorobacter polyhydroxybutyrativorans strain SL-205, and 99.93% identity (1 nucleotide difference) to that of Diaphorobacter sp. strain JS3050. Comparison of the whole genomes of the above five strains revealed that strain JS3051 had the highest identity to Acidovorax sp. strain JS42 and Diaphorobacter sp. strain JS3050 (Fig. S1). The closest relationship was further identified between strain JS3051 and Acidovorax sp. strain JS42 by calculating the distances between species derived from in silico DDH (with a 0.81 DNA-DNA hybridization [DDH]) ( Table S2). The probability of DDH . 0.7 was 96%, which reaches the threshold value of same species (28).
Prediction and organization of 23DCNB catabolic genes. A previous study provided preliminary evidence that strain JS3051 degrades 23DCNB by an oxidative pathway (9), similar to the pathway of 2CNB in Pseudomonas sp. strain ZWLR2-1 (20). Therefore, a working hypothesis for an analogous pathway for 23DCNB (Fig. 1A) provided the basis to search for candidate genes in JS3051. First, a gene cluster (designated dcb, Fig. 1B) encoding a three-component dioxygenase was a strong candidate for involvement in the initial dihydroxylation of 23DCNB due to its similarity to the ring-hydroxylating dioxygenases responsible for the catabolism of naphthalene and nitroarenes (Table S3). Second, the enzyme catalyzing the ring cleavage in the 23DCNB pathway was likely to be a chlorocatechol or catechol dioxygenase. A gene cluster (designated dcc, Fig. 1B) is highly similar to the clc genes responsible for 3-and 4-chlorocatechol oxidation in Pseudomonas knackmussii B13 (29). Additionally, two other catechol 1,2-dioxygenase genes were also annotated on the chromosome of strain JS3051. The dcb and dcc clusters are not contiguous on the chromosome (Fig. 1B). Divergently transcribed LysR family regulators (dcbR and dccR, respectively) are present on both clusters. Gene annotations, locations on the chromosome, and the most closely related matches are listed in Table S4.
A Reiske-iron dioxygenase catalyzes the dihydroxylation of 23DCNB to 34DCC. The genes encoding the predicted 23DCNB dioxygenase (dcbAaAbAcAd) are related to the Rieske non-heme iron oxygenases, comprising an oxidoreductase (DcbAa), an ironsulfur ferredoxin protein (DcbAb), and a terminal oxygenase (a-subunit DcbAc and b-subunit DcbAd). To determine whether the putative dcbAaAbAcAd-encoded dioxygenase is responsible for the initial dihydroxylation reaction in 23DCNB degradation, dcbAaAbAcAd from strain JS3051 was cloned into pETDuet-1 and expressed in Escherichia coli strain BL21(DE3) cells. When a whole-cell biotransformation assay was performed with 23DCNB as the substrate, a single product was detected by high-performance liquid chromatography (HPLC) and identified as 3,4-dichlorocatechol (34DCC) based on comparison of the retention time and UV absorption spectrum with those of authentic 34DCC (Fig. S2). The identity was further confirmed by gas chromatography-mass spectrometry (GC-MS) analysis ( Fig. 2A). During the biotransformation, 23DCNB was converted stoichiometrically to 34DCC and NO 2 2 (Fig. 2B). The results indicated that DcbAaAbAcAd is a 23DCNB dioxygenase capable of oxidizing 23DCNB to 34DCC with concomitant nitrite release. No other nitroarene dioxygenase candidates were found in the genome of JS3051.
The amino acid at position 204 changes the substrate specificity of 23DCNB dioxygenase (23DCNBDO) and 2NT dioxygenase (2NTDO). DcbAc, the large subunit determining the substrate specificity, shows highest identity (97%) to its counterpart in 2NT dioxygenase from strain JS42. The enzyme 2NT dioxygenase (2NTDO) was reported  Organizations of the dcb gene cluster encoding the Nag-like dioxygenase (a) and the dcc gene cluster encoding the chlorocatechol catabolic enzymes (b). The identification of dcb genes was based on the 2-chloronitrobenzene (2CNB) dioxygenase from Pseudomonas stutzeri ZWLR2-1 and the dcc genes were based on the 3-and 4-chlorocatechol catabolic genes (clc genes) from Pseudomonas knackmussii B13. The flanking mobile elements are shown as different shades of gray.
to transform chloronitrobenzenes (24), but its ability to transform dichloronitrobenzenes was not reported. All 14 amino acid differences between 23DCNBDO and 2NTDO are located in the catalytic domain near the C-terminus (Fig. 3). In order to investigate the impact of these substitutions on the substrate specificity, relative activities of 23DCNBDO and 2NTDO toward different nitroarenes were analyzed. Although nitrobenzyl alcohols are also sometimes side products from methyl-substituted nitroarenes (30), we focus here on the ring-dihydroxylated products. The enzyme 2NTDO is active not only with 2NT, but also with NB, 2CNB, and 23DCNB (Fig. 4A). In contrast, 23DCNBDO exhibits a preference for meta-substituted substrates, such as 23DCNB, 3CNB, and 3NT (Fig. 4B). Both 2NTDO and 23DCNBDO had minimal activity toward para-substituted nitroaromatic substrates, including 4NT and 4CNB ( Fig. 4A and B).
To determine the contributions of the 14 differing amino acids to substrate specificities of 23DCNBDO and 2NTDO, mutants were made by replacing each of the amino acids present in one protein with the corresponding amino acid of its counterpart. Activity assays indicated that the residue at position 204 plays a key role in determining the specificity of 23DCNBDO and 2NTDO ( Fig. 4C and D), whereas the other 13 residues had minimal effect (Table S5). The single replacement of Ile to Glu at position 204 of 2NTDO reduced the activities by 19-, 46-, and 17-fold with 2NT, 2CNB, and NB, respectively. In contrast, the activity with 23DCNB (9-fold higher than that of 2NT) became the primary activity (Fig. 4C). Likewise, the E204I mutation of 23DCNBDO caused a shift in preference from meta-to ortho-substituted substrates, such as 2CNB, 2NT, and 23DCNB (Fig. 4D). Unexpectedly, 23DCNB was still a primary substrate of the 23DCNBDO E204I variant, indicating that some of the other 13 amino acids affect the activity toward 23DCNB.
E204 is a unique residue among Nag-like dioxygenases. Alignment of the amino acid sequences of the a subunits of Nag-like dioxygenases showed that the residues at position 204 (or the corresponding residues) appear to be variable in nitroarene dioxygenases while conserved in naphthalene dioxygenases (Fig. 4E). All of the residues are nonpolar amino acids except for the glutamic acid of 23DCNBDO (Fig. 4E).
To gain insight into how E204 affects the substrate specificity of 23DCNBDO, a homology model of the 23DCNBDO a subunit was constructed based on the crystal structure of NBDO (95% amino acid sequence identity) (31). The 23DCNBDO protein has a similar hydrophobic pocket to that of other Nag-like dioxygenases, with residues including Phe200, Leu293, Leu305, and Phe222 (Fig. 4F). The isoleucine at position 204 of 2NTDO also contributes to the hydrophobic environment in the active site. Substituting the Ile204 with a glutamic acid changes the hydrophobic environment around the C 3 atom of 2NT and, consequently, affects its correct positioning. On the other hand, the glutamic acid seems to interact with the C 3 chlorine atom through a halogen bond (Fig. 4F) (32), , and 23DCNBDO E204I mutant (D) shown by relative activities monitored with whole-cell nitrite assays in E. coli. For 2NTDO and 23DCNBDO E204I mutants, the relative activities were compared with 2NT (3.1 mmol mg 21 min 21 and 1.1 mmol mg 21 min 21 , respectively). For 23DCNB and 2NTDO I204E mutant, the relative activities were compared with 23DCNB (2.6 mmol mg 21 min 21 and 1.5 mmol mg 21 min 21 , respectively). The red and brown colors represent the 2NTDO-and 23DCNBDOderived enzymes, respectively. The diagonal stripe shows the 2NTDO-like activity and the horizontal stripe shows the 23DCNBDO-like activity. The homology model of 23DCNBDO with 23DCNB in the active site was generated to gain insight into the relationship of key residues and substrate specificity. (E) A cropped multiple sequence alignment of representative sequences of a subunit of Nag-like dioxygenases. (F) The (Continued on next page) Li et al. which could be the cause of the higher activity for 3CNB and 23DCNB than for 3NT ( Fig. 4B and C).
Structural comparison of nitroarene dioxygenases. The lack of activity of both 2NT and 23DCNB dioxygenases toward 34DCNB ( Fig. 4A and B) is consistent with the observation that the system in JS3051 is not closely related to that of JS3050 (Fig. 5), which was isolated from the same habitat based on its ability to degrade 34DCNB. Substrate specificity assays revealed that 24DNT dioxygenase accepts 34DCNB, but not 23DCNB, as a substrate for ring hydroxylation (data not shown), which is consistent with the phylogenetic analysis of Nag-like dioxygenases (Fig. 5). The 23DCNB and 34DCNB dioxygenases share more similar substrate preferences with 2NT dioxygenase and 24DNT dioxygenase than with each other.
Identification of products revealed that the dihydroxylation occurred at the analogous positions of 2NT and 23DCNB (both have a C 2 substituted group) and also the analogous positions of 24DNT and 34DCNB (both have C 3 and C 4 substituted groups) (Fig. 6A). Molecular docking indicated that 2NT dioxygenase accommodates both 2NT and 23DCNB in similar orientation in the substrate-binding pocket (Fig. 6B). Notably, the pocket shows strong steric hindrance to the C 4 substituted groups. This is supported by the fact that 2NT dioxygenase has a 6-fold higher activity for 2CNB than for 24DCNB. Similarly, both 2NT and 23DCNB dioxygenases have minimal activity with 4NT and 4CNB ( Fig. 4A and B). In the same way, 24DNT and 34DCNB also have similar orientation in the substrate-binding pocket of 24DNT dioxygenase, which accounts for the regiospecific dihydroxylation (Fig. 6C). Residue 258 of the 24DNT dioxygenase a subunit is a valine, and it does not form a hydrogen bond with the nitro group as does the Asn258 in 2NT dioxygenase. In contrast, the Val258 and Trp256 are very close to the C 2 of the substrates (;3.6 Å), and thus increase the steric hindrance to the C 2 substituted group (Fig. 6C).
DccA catalyzes the ring cleavage of 34DCC. The downstream genes involved in the catabolism of 34DCC were further investigated. BLAST analysis revealed three can-  Bacterial Catabolism of 2,3-Dichloronitrobenzene ® didate genes (I3K84_08270, I3K84_12300, and I3K84_15960) encoding putative (chloro) catechol dioxygenases that might catalyze 34DCC ring cleavage. Among them, only the I3K84_08270 (designated dccA) encoded an enzyme able to catalyze conversion of 34DCC into 2,3-dichloromuconate (Fig. 7A). Cell extract from E. coli carrying DccA exhibited a specific activity of 0.37 U/mg for catechol and 64% relative activity toward 34DCC. Cell extracts from strain JS3051 exhibited a specific activity of 0.08 U/mg for  34DCC and its relative activity against 34DCC (76%) was similar to that from E. coli cells carrying DccA ( Table 1). The results are consistent with the hypothesis that DccA catalyzes the 34DCC ring cleavage in strain JS3051. The kinetic parameters of purified H 6 -DccA (Fig.  S3) indicate that it has a higher affinity for 34DCC (K m , 0.48mM) and 4CC (K m , 0.73mM) among the tested substrates (Table 2). H 6 -DccA also had similar catalytic efficiency for 34DCC (k cat /K m , 44.2 min 21 mM 21 ) and 4CC (k cat /K m , 53.6 min 21 mM 21 ), suggesting that DccA has successfully adapted to the 23DCNB pathway ( Table 1). Absence of activity toward 45DCC, a 34DCC analogue and the only product from 34DCNB degradation in strain JS3050, indicates that DccA has high substrate specificity toward 34DCC.
DccB catalyzes conversion of 2,3-dichloromuconate to chlorodienelactone. DccB has 100% sequence identity to the well-studied chloromuconate cycloisomerase ClcB B13 that catalyzes lactonization of 2,4-dichloromuconate, 2-chloromuconate, and 3-chloromuconate (33). However, its activity toward 2,3-dichloromuconate, the ring-cleavage product of 34DCC, was unknown. Heterologously coexpressed DccA and DccB were used in an in vitro sequential catalytic assay to detect the cycloisomerization of 2,3-dichloromuconate with 34DCC as the initial substrate. The reaction catalyzed by DccA was completed rapidly and, due to the relatively slow reaction rate of DccB, the 2,3-dichloromuconate (l max = 268 nm) accumulated in the reaction mixture (Fig. 7B), followed by a slower spectral shift from 268 nm to 293 nm (Fig. 7C), consistent with formation of 5-chlorodienelactone (33). The isobestic point at 275 nm indicated direct conversion. The spectral change was not detected with the crude extracts containing DccA only, indicating that DccB is a functional chloromuconate cycloisomerase acting on 2,3-dichloromuconate as substrate.

DISCUSSION
This study revealed that the catabolism of 23DCNB by strain JS3051 is initiated by a Rieske-type 23DCNB dioxygenase that adds both atoms of molecular oxygen to the benzene ring with the release of nitrite and formation of 34DCC. The 34DCC product is subsequently degraded via a modified ortho-cleavage pathway (Fig. 1A). The evolution of the 23DCNB pathway is most likely from an ancestral Nag-like naphthalene degradation pathway and a chlorocatechol pathway, both with modified enzyme specificity. Analyses of the 23DCNB dioxygenase using biochemical and structural approaches  Bacterial Catabolism of 2,3-Dichloronitrobenzene ® also revealed key factors determining substrate specificity and the recent divergence of nitroarene dioxygenases. The 23DCNB dioxygenase genes (dcb) share a recent common ancestor with the genes encoding 2-nitrotoluene dioxygenase. The conclusion that the dcb genes share a recent common ancestor with the genes encoding 2-nitrotoluene dioxygenase is well supported by the surprisingly high identity and the same organization between dcb genes of strain JS3051 and the ntd genes from the 2-nitrotoluene degrader strain JS42 (Table S3) (Fig. 8A). Additional evidence dictated that orf2 within the dcb operon has the same start codon and high similarity with the N terminus of salicylate hydroxylase large subunits (NagG) from strains containing nag-like genes, such as strain U2. The truncated nagG remnant is a strong indication that dcb originated from a nag-like naphthalene dioxygenase gene cluster (14). The identical sequence between orf2 and its counterpart in the ntd operon (34) (Fig. 8A) indicates the close evolutionary relationship of strains JS3051 and JS42.
Transposable elements are often responsible for transfer of catabolic genes during adaptive evolution in response to the introduction of xenobiotics (35)(36)(37). The dcb cluster is surrounded by a single copy of an IS21-like insertion sequence, together with a Tn3 transposase gene upstream and an IS4 family insertion sequence downstream, that are identical to those in the ntd cluster from the 2NT degrader Acidovorax sp. strain JS42 (Fig. 8A). Additionally, a TnpA transposase, an IS91, an ISCsp2, and two copies of IS1071 transposons are flanked by the upstream Tn3 transposase in JS3051 (Fig. S4). All are absent from the ntd cluster but located on other sites of the chromosome or the plasmid pAOVO01 of strain JS42. The above evidence, together with the close phylogenetic relationship between strains JS3051 and JS42 (Table S2), suggests strongly that the dcb cluster originated from a within-species lineage related to strain JS42.
Structure-activity relationships among nitroarene dioxygenases. The striking similarity among the large subunits provides strong support for the previous observation that all the genes encoding Nag-like nitroarene dioxygenases (Fig. 5) share a recent common ancestor with the Nag-like naphthalene dioxygenases (18). The divergence of the sequences, substrate specificities, and structural characteristics of nitroarene dioxygenases can be summarized in the following features. The 24DNT and 34DCNB dioxygenases are in a clade separated from the other known nitroarene dioxygenases (Fig. 5). The clade II nitroarene dioxygenases accept 2/3-substituted nitroarenes, whereas the clade I nitroarene dioxygenases prefer 3,4-substituted substrates (Fig. 4A and B) (21,31,(38)(39)(40). The position of the substituents seems to have more influence than the type of the groups on the substrate preferences of nitroarene dioxygenases. The catalytic domains of nitroarene dioxygenases share many common features, but notable divergence is introduced by residue 258. In class II nitroarene dioxygenases, Asn258 plays an important role in positioning the substrates by interacting with the nitro groups through hydrogen bonding (Fig. 4F) (31). In contrast, the class I nitroarene dioxygenases possess a nonpolar residue (Val or Ile) at position 258, which is consistent with Nag-like naphthalene dioxygenases (Fig. 4E). Other conserved residues in the catalytic domain, including I207, A223, A254, Q259, Q293, G309, F314, I329, and D384, are also only present in class I dioxygenases.
Multiple structural features can affect accommodation of new substrates by Reiske dioxygenases. For instance, the channel of the active site of some Reiske dioxygenases can be a bottleneck that controls substrate access (41,42). Similarly, evolution of 2NT dioxygenase into a productive 4NT dioxygenase resulted from artificial laboratory evolution involving direct selection of spontaneous mutants (15). The increased 4NT activity was attributed to three missense mutations outside the substrate-binding site of the catalytic domain. In contrast, the residue 204, which determines the substrate specificities of 2NTDO and 23DCNBDO observed in this study, is located in the substratebinding site (Fig. 4F). Understanding the mechanisms of substrate selectivity of nitroarene dioxygenases will be beneficial for the potential applications in creating variants with novel activities.
A chlorocatechol cluster involved in the lower pathway of 23DCNB catabolism. The evolution of new substrate preferences that allowed the initial attack and elimination of the nitro groups of the various compounds was necessary but not sufficient for assembly of productive catabolic pathways. In the downstream pathway of 23DCNB catabolism, the striking similarity between the entire dcc and clc clusters, the presence of the almost identical flanking fragments between these two clusters, and the large fragment containing the dcc cluster flanked by two direct repeats of IS66-family insertion sequences and a Tn3-like element (Fig. 8B) strongly indicate that the dcc genes were recruited through recent lateral transfer. The facts that the positions of ntd and cat in JS42 are similar to those of tcb and cat in the genome of JS3051 and the dcc genes are absent from the genome of JS42 support the argument that the dcc genes of JS3051 were recruited by horizontal gene transfer.
In addition to recruitment, the subsequent evolution of genes encoding enzymes with modified specificities for the downstream pathways would have been essential. Although ClcA B13 and DccA differed by only six amino acids (Table S6), ClcA B13 exhibited a strong preference for 35DCC (43) rather than 34DCC (Table 1). This situation is similar to that of CbnA (preferring 35DCC) and TcbC (preferring 34DCC), which differ in 12 amino acids (44). Sequence analysis indicates that both DccA and TcbC share the same Val48, Ala52, and Met73 residues, and both CbnA and ClcA B13 share the same Leu48, Val52, and Ile73 residues (Table S6). These three residues seem to be responsible for the preferential activity toward 34DCC and 35DCC (44).
Different recent origins of genes encoding catabolic pathways for DCNB isomers in strains JS3050 and JS3051. Although the two closely related strains isolated from the same location utilize Nag-like nitroarene dioxygenases to catalyze the initial dioxygenation reaction of DCNBs, several lines of evidence indicate their different recent ancestries. First, the 34DCNB dioxygenase is closer to 24DNT dioxygenase than to 23DCNB dioxygenase in sequence, structure, and substrate specificity. Second, an IS30like insertion sequence flanked by the dcnAd of JS3050 is identical to that of strain DNT (45), and totally different from the mobile elements of JS3051 in organization and sequence (Fig. 8A). Finally, the genes involved in the chlorocatechol pathway were discontinuously distributed on the chromosome and a plasmid of strain JS3050 and showed relatively low identity with the contiguous dcc genes in strain JS3051 (Fig. 8B). Analyses of substrate compatibility in the active sites (Fig. 6), combined with biochemical characterization, provided insight into the molecular mechanisms underlying the different origins of DCNBs dioxygenases. It is clear that the active sites of the different precursor dioxygenases evolved separately for their respective DCNB substrates. The lower pathways seem to have been selected by the regiospecific differences in metabolites of DCNBs. Theoretically, the dioxygenation of 34DCNB could generate both 34DCC and 45DCC. In fact, 34DCNB dioxygenase from JS3050 specifically transformed 34DCNB to 45DCC (21). Such regiospecificity would be consistent with the presence of a 45DCC pathway in JS3050, whereas a 34DCC pathway is required for productive degradation of 2,3DCNB by JS3051.
Genome sequencing and analysis. Sequencing of genomic DNA of strain JS3051 was performed by Shanghai OE Biotech Co., Ltd. (Shanghai, China) using the Pacific Bioscience (PacBio) RS technology (47). The complete genome sequence of strain JS3051 was assembled using Falcon (48) and Circulator (49). The genome was annotated by the Prokaryotic Dynamic Programming Genefinding Algorithm (Prodigal V2.6.3) (50) and RAST annotation service (51).
Site-directed mutagenesis. Site-directed mutagenesis of dcbAc and ntdAc was performed by PCR. Briefly, plasmids pETDuet-DCB and pETDuet-NTD were used as templates for mutagenesis. The templates were amplified by PFU DNA polymerase (Vazyme Biotech Co., Ltd) following the manufacturer's protocol with the mutagenic oligonucleotides listed in Table 4. The products were transformed into E. coli DH5a and screened on LB agar with ampicillin.
Whole-cell biotransformation assays. To determine the function of dcbAaAbAcAd, E. coli strain BL21(DE3) (pETDuet-DCB) was grown in LB medium to an optical density at 600 nm (OD 600 ) of 0.6 and gene expression was induced at 30°C for 5 h after addition of IPTG (isopropyl-b-D-thiogalactopyranoside) (0.3 mM). The cells were harvested, washed twice with phosphate-buffered saline (PBS) and suspended in MSB containing 0.1 mM 23DCNB. E. coli strain BL21(DE3) containing the pETDuet-1 vector was used as negative control. Cell suspensions were incubated with shaking (220 rpm, 30°C) and sampled at appropriate intervals for the subsequent analyses. Concentrations of 23DCNB and 34DCC were quantified by HPLC. Nitrite was detected by the Griess method as described previously (52).
Protein expression and purification. The dccA gene was amplified with primers DccA-F and DccA-R (Table 4) from genomic DNA of strain JS3051 and ligated into expression vector pET-28a(1) between the NdeI and BamHI restriction sites. Recombinant DccA containing an N-terminal 6ÂHis tag was expressed and purified as described previously (53). The purified DccA was used to determine the kinetic  (Table 4) and ligated to a pET28a(1) vector between NdeI and BamHI sites. Expression conditions and preparation of cell extract containing DccAB were as described above for DccA. Enzyme assays and kinetic measurements. To elucidate the reaction catalyzed by DccA, cell lysates containing DccA were centrifuged at 15,000 Â g for 60 min to remove the debris. The supernatant was collected and used for enzyme assays. Cell extracts containing ClcA B13 and DccAB were prepared the same as for DccA. E. coli BL21(DE3) cells harboring the pET-28a(1) vector were used as a negative control. Extracts were prepared from 23DCNB-grown cells of strain JS3051 by the same method.
The reaction mixture contained crude enzyme (3 to 10 mg protein) in 50 mM Tris-HCl buffer (pH 8.0) and the reaction was initiated by the addition of (chloro)catechol substrates (50 mM). All assays were performed with a Lambda 25 spectrophotometer (PerkinElmer/Cetus, Norwalk, CT). The activities toward catechol or substituted catechols were determined by the increase in absorption at A 260nm due to the accumulation of muconate or corresponding chloromuconates. One unit of enzyme activity (U) is defined as the amount of the enzyme required for the production of one mmol of product per min at 25°C. Specific activity is expressed as units per milligram of protein. Purified DccA was used to determine the kinetic parameters as described by Potrawfke et al. (54). The extinction coefficients for chloromuconates reported by Dorn and Knackmuss (55) and Gao et al. (21) were used for determining the 1,2-chlorocatechol dioxygenase activity of DccA. The kinetic curves are shown in Fig. S4.
Activities of ring-hydroxylating dioxygenases. The activities of 23DCNBDO, 2NTDO, and their mutants toward different nitroarenes were determined based on the whole-cell biotransformation assay described above with some modifications. Specific activities were obtained by measuring the rates of nitrite accumulation at appropriate intervals (depending on the activity of each dioxygenase). The cells were collected by centrifugation, suspended in equal volumes of 0.1 M NaOH, and boiled for 10 min. Protein concentrations were determined by the Bradford method (56) with bovine serum albumin as the standard.
Homology model. Homology models of the a subunit of nitroarene dioxygenases were generated by SWISS-MODEL (57), using the NB dioxygenase a subunit (Protein Data Bank entry: 2BMO) as the template. The sequence identities between nitroarene dioxygenases and the template were more than 85%. The quality of models was estimated based on the QMQE (0.9 to 0.99) and QMEAN (20.45 to 0) scoring functions. Nitroarene substrates were docked into the active sites of dioxygenase models by AutoDock Vina (58) with default settings. The docking scores for the various poses are shown in Table S7 and the representation of all the different poses are shown in Fig. S6. The productive poses were determined based on the reference structure (31) and docking scores. The structure models were visualized by PyMOL v2.4 (http://www.pymol.org).
Analytical methods. Reverse phase high-performance liquid chromatography (HPLC) analyses were carried out with a Waters e2695 separation module equipped with a Waters 2998 photo diode array detector, using a C 18 reversed-phase column (5 mm, 4.6 Â 250 mm) at 30°C. The mobile phases were water containing 0.1% (vol/vol) acetic acid (A) and methanol (B). The elution profile was 20% of solvent B for 5 min, then linear increase to 90% B over 30 min. Gas chromatography-mass spectrometry (GC-MS) analyses were performed with a TRACE 1310 gas chromatograph (Thermo Fisher Scientific Inc., MA, USA) equipped with a capillary column HP-5MS (0.25 mm Â 30 m, Agilent technologies., CA, USA). For GC-MS analysis, biotransformation samples were extracted with diethyl ether. Then the extracts were evaporated to dryness and dissolved in anhydrous ethyl acetate.
Supplemental materials. The supplemental materials can be found at https://figshare.com/articles/ figure/20210726_SI_docx/15052137. Data availability. The whole-genome sequencing data were deposited in the NCBI database under BioProject identifier (ID) PRJNA680215.

ACKNOWLEDGMENTS
This work was funded by National Natural Science Foundation of China (NSFC) (31900075 and 31870084), China Postdoctoral Science Foundation (2019M661491), and