The Genes Coding for the Conversion of Carbazole to Catechol Are Flanked by IS6100 Elements in Sphingomonas sp. Strain XLDN2-5

Background Carbazole is a recalcitrant compound with a dioxin-like structure and possesses mutagenic and toxic activities. Bacteria respond to a xenobiotic by recruiting exogenous genes to establish a pathway to degrade the xenobiotic, which is necessary for their adaptation and survival. Usually, this process is mediated by mobile genetic elements such as plasmids, transposons, and insertion sequences. Findings The genes encoding the enzymes responsible for the degradation of carbazole to catechol via anthranilate were cloned, sequenced, and characterized from a carbazole-degrading Sphingomonas sp. strain XLDN2-5. The car gene cluster (carRAaBaBbCAc) and fdr gene were accompanied on both sides by two copies of IS6100 elements, and organized as IS6100::ISSsp1-ORF1-carRAaBaBbCAc-ORF8-IS6100-fdr-IS6100. Carbazole was converted by carbazole 1,9a-dioxygenase (CARDO, CarAaAcFdr), meta-cleavage enzyme (CarBaBb), and hydrolase (CarC) to anthranilate and 2-hydroxypenta-2,4-dienoate. The fdr gene encoded a novel ferredoxin reductase whose absence resulted in lower transformation activity of carbazole by CarAa and CarAc. The ant gene cluster (antRAcAdAbAa) which was involved in the conversion of anthranilate to catechol was also sandwiched between two IS6100 elements as IS6100-antRAcAdAbAa-IS6100. Anthranilate 1,2-dioxygenase (ANTDO) was composed of a reductase (AntAa), a ferredoxin (AntAb), and a two-subunit terminal oxygenase (AntAcAd). Reverse transcription-PCR results suggested that carAaBaBbCAc gene cluster, fdr, and antRAcAdAbAa gene cluster were induced when strain XLDN2-5 was exposed to carbazole. Expression of both CARDO and ANTDO in Escherichia coli required the presence of the natural reductases for full enzymatic activity. Conclusions/Significance We predict that IS6100 might play an important role in the establishment of carbazole-degrading pathway, which endows the host to adapt to novel compounds in the environment. The organization of the car and ant genes in strain XLDN2-5 was unique, which showed strong evolutionary trail of gene recruitment mediated by IS6100 and presented a remarkable example of rearrangements and pathway establishments.


Introduction
Carbazole is an N-heterocyclic compound that is known to possess mutagenic and toxic activities [1]. To date, a number of bacterial strains capable of degrading carbazole have been isolated and characterized. Phylogenetically, almost all of these strains belong to pseudomonads and sphingomonads. The degradation of carbazole starts with angular dioxygenation in all studied strains so far [2,3,4,5,6,7,8]. In this pathway, carbazole is initially attacked at the 1 and 9a positions by carbazole 1,9adioxygenase (CARDO), resulting in the formation of a highly unstable hemiaminal, which gives rise to 29-aminobiphenyl-2,3diol after spontaneous cleavage and rearomatization. An extradiol dioxygenase attacks the hydroxylated ring of 29-aminobi-phenyl-2,3-diol at the meta position to generate 2-hydroxy-6-(29aminophenyl)-6-oxo-2,4-hexadienoic acid, which is hydrolyzed to produce anthranilic acid and 2-hydroxypenta-2,4-dienoic acid. The resulting metabolite, anthranilate, is converted to catechol in a single step by anthranilate 1,2-dioxygenase (ANTDO) (Figure 1) [9,10]. In this upper carbazole degradation pathway, CARDO is considered as the key enzyme. In Pseudomonas resinovorans CA10, the most intensively studied pseudomonad, CARDO consists of a terminal oxygenase, a ferredoxin, and a ferredoxin reductase, encoded by carAa, carAc, and carAd, respectively. In pseudomonads, the upper pathway genes, carAaBaBbCAcAd, are in the car cluster, which is transcribed as a single transcriptional unit [6]. Interestingly, unlike the well-organized operons in pseudomonads, the catabolic genes in sphingomonads are often dispersed or not coordinately regulated. For example, plasmid pCAR3 contains multiple gene sets, which are involved in the carbazole degradation pathway in a carbazole-degrader Novosphingobium sp. KA1 (previous Sphingomonas sp. KA1) [10]. These unusual organizations of the degradative genes were often observed in other sphingomonads [11,12].
Insertion sequences (ISs) are a subject of increasing interests for biodegradation because of the variety of their structures, modes of action, and the biodegradation abilities they confer bacteria [13]. ISs are small and mobile genetic elements that are ubiquitously distributed within bacterial genomes, and play an important role in evolution by facilitating horizontal gene transfers between bacterial populations, which contribute significantly to the diversity of bacteria by enhancing the organisms' adaptive and evolutionary capacities. IS6100 is an important IS that flanks a range of catabolic operons, for example, the operons for metabolism of various aromatic substrates [14,15]. In this work, we report that two loci coding for the enzymes that convert carbazole to catechol were found to be flanked by IS6100 elements. Evidence was given for the involvement of these genes in the degradation of carbazole in Sphingomonas sp. XLDN2-5.

Results
Screening of the genomic library, DNA sequencing, and genome walking The car probe, labeled with DIG, was used for screening the genomic library, and a positive clone, designated as pBY13 (Figure 2A), was sequenced and analyzed to contain a DNA insert of 6.8 kb. BLAST search results revealed that there were five intact open reading frames (ORFs), carR, carAa, carBa, carBb, and carC, which were found to be 99% identical to the corresponding genes of Novosphingobium sp. strain KA1 [10]. A closer look at the left region suggested that an IS6100, exhibiting 100% identity to that of Mycobacterium fortuitum [16], was interrupted by a novel insertion element ISSsp1. Therefore, the interrupted IS6100 was designated IS6100::ISSsp1. The putative transposase for ISSsp1 was transcribed in the same direction of the car cluster, but in the opposite direction as the interrupted IS6100. ISSsp1 belongs to the IS256 family of prokaryotic ISs, a subgroup of the mutator family of transposases (Pfam00872). ISSsp1 which had imperfect inverted repeats (32-bp in length, one mismatch) was flanked by two copies of 8-bp direct repeat ( Figure 2B). The product in dashed square is unstable and has not been detected. Enzyme names: carbazole 1,9a-dioxygenase (carAaAcfdr); metacleavage enzyme (carBaBb); hydrolase (carC); anthranilate 1,2-dioxygenase (antAcAdAbAa). Compound I: carbazole; compound II: 29-aminobiphenyl-2,3-diol; compound III: 2-hydroxy-6-oxo-6-(29-aminobiphenyl)-hexa-2,4-dienoic acid; compound IV: 2-hydroxypenta-2,4-dienoate; compound V: anthranilic acid; VI: catechol. doi:10.1371/journal.pone.0010018.g001 At the 39-end of pBY13, an incomplete ORF (238 bp) showed 100% identity to carAcI of KA1. In order to clone the full gene and its flanking sequences, a 1-kb DNA was amplified by genome walking. After ligation of the fragment to pMD18-T followed by transformation to E. coli DH5a, a positive clone, designated p6100-a, was obtained. DNA sequencing indicated that p6100-a contained the left part of carAc, an ORF8, and interestingly a partial IS6100 sequence. The presence of the partial IS6100 sequence led to the hypothesis that there was a complete IS6100 and might form a transposon, and therefore triggered further study. The presence of an intact IS6100 copy downstream of ORF8 was successfully validated by PCR using carC-sp3 and IS6100-F2 and sequencing. This transposon-like context (IS6100::ISSsp1-ORF1-carRAaBaBbCAc-ORF8-IS6100) was designated TnCar, tentatively.

IS-based PCR
The presence of IS6100 on both sides of the car gene cluster motivated further investigations using IS-based PCR. An IS-based PCR with primers IS6100-F1 and IS6100-R1 targeting the IS6100 element resulted in two distinct PCR products of about 4.3 and 2.8 kb, respectively ( Figure S1B, lane 1). The 4.3 and 2.8-kb fragments were ligated to pMD18-T to generate plasmid p6100-1 and p6100-2. DNA sequencing and BLAST search suggested that there were two partial IS6100 at each end as expected in the 4.3 and 2.8-kb fragments. The 4.3-kb fragment contains ORF1carRAa-BaBbCAcORF8 from TnCar.
BLAST search suggested that in the 2.8-kb fragment, there were two truncated ORFs and one intact ORF (Fdr) that showed 62% identity to the FdrI of strain KA1 [10]. A sequence comparison showed that the amino acid sequence of Fdr was homologous over its entire length to other members of the FAD-dependent pyridine nucleotide reductase family, containing a flavin binding domain for FAD (consensus sequence TX 6 AXGD) and two ADP binding domains (for FAD and NADH, respectively) with the consensus sequence GXGX 2 GX 3 A [17,18,19]. Two truncated ORFs encoded for a putative uncharacterized protein and a transcriptional regulator, were found upstream and downstream of fdr ( Table 1). Sequencing of PCR product revealed the existence of entire copies of IS6100 at both ends of the 2.8-kb fragment. These two IS6100 were in the same direction and formed a composite transposon (IS6100-fdr-IS6100) designated TnFdr, tentatively.
TnCar and TnFdr were each flanked by two copies of IS6100 in tail-to-head configuration. Although tail-to-head was the most abundant case for two copies of IS6100, other configurations were also reported [14]. Early attempts to amplify head-to-head and tail-to-tail configurations proved to be unsuccessful. Considering that the amplification of head-to-head and tail-to-tail configurations may be suppressed by intramolecular hybridization between two IS6100 copies during the primer annealing phase of PCR, TaKaRa LA Taq and GC Buffer I were used in the following experiments. One specific fragment (5.2 kb in size, Figure S1C, lane 3) was amplified using IS6100-R1, while no specific fragment was amplified under the same PCR conditions using IS6100-F1 ( Figure S1C, lane 2). The 5.2-kb fragment was gel-purified and cloned to pMD18-T to generate p6100-3, and the nucleotide sequence of the 5.2 kb insert was determined. In the sequenced region, five intact ORFs were found to be almost identical (only one bp mismatch) to andRAcAdAbAa genes of KA1 whose products were expected for catalyzing the conversion of anthranilate to catechol. The five ORFs were designated antR, antAc, antAd, antAb, and antAa ( Figure 2C). Strain XLDN2-5 ANTDO was a threecomponent dioxygenase and composed of a two-subunit oxygenase (antAcAd), a Rieske-type ferredoxin (antAb), and a ferredoxin reductase (antAa). antR, encoding a putative transcriptional activator, was in the same direction of antAcAdAbAa, which was different from that of Burkholderia cepacia DBO1 whose antR is in the opposite direction to its structure genes [20]. In the sequenced region, there were two partial IS6100 at each end as expected. The existence of intact IS6100 elements at both ends was confirmed by PCR and PCR product sequencing. These two IS6100 elements were in the head-to-head configuration and formed a composite transposon designated TnAnt, tentatively.

The positional relation of TnCar, TnFdr and TnAnt
There were two copies of IS6100 elements at both ends of the three transposon-like entities except that the upstream flanking IS6100 on TnCar was disrupted by a novel insertion sequence ISSsp1. It was likely that one IS6100 element was shared by two transposon-like units. In order to analyze the positional relation of TnCar, TnFdr, and TnAnt, five primers were designed outside of the IS6100 elements ( Figure S2A). If there was a shared IS6100 element, a 0.9-kb DNA fragment containing IS6100 could be amplified by PCR using different primer pairs. As shown in Figure  S2B, the expected 0.9-kb fragment could only be amplified using primers TnCar-F1 and TnFdr-R1. These results suggested that TnFdr shared an IS6100 element with TnCar, but not with TnAnt. The conclusion was confirmed by second-round PCR ( Figure S2C and S2D). Thus, we renamed TnCar and TnFdr to the car locus, and TnAnt to the ant locus.

Description of car and ant loci
The structures of car locus and ant locus are depicted in Figure 2 and the ORFs are given in Table 1. Both car cluster and fdr gene were sandwiched between two copies of IS6100. Interestingly, two identical copies of the IS6100 element also flanked the ant gene cluster, possibly making a composite transposon. All five sequenced copies of IS6100 from strain XLDN2-5 were identical over the entire 880 bp. Direct repeats (TGCGCAGG) were found directly upstream and downstream of ISSsp1, whereas no direct repeat was found outside of the IS6100 box. Furthermore, the ISSsp1 consisted of inverted repeats of 32 bp, of which only one base pair was not identical, and a 1224-bp ORF (tnpA ISSsp1 ), encoding a 407 aa putative transposase that showed similarities to transposases of the IS256 family. In order to illustrate that the genes on these two loci were really working in the degradation of carbazole in strain XLDN2-5, transcriptional and functional analyses were performed.

Transcriptional analyses of car and ant genes
The expression of the genes presenting on the two loci was studied by reverse transcription (RT)-PCR experiments. The primer sets for the carAaBaBbCAc, fdr, and antAcAdAbAa genes could amplify DNA fragments with the expected sizes ( Figure 3). No fragments could be amplified using RNA from glucose-grown XLDN2-5 cells as a template (data not shown). These results revealed that carAaBaBbCAc, fdr, and antAcAdAbAa genes were expressed in carbazole grown XLDN2-5 cells, suggesting that the gene products should be involved in the transformation of carbazole to catechol. These results also indicated that carAaBaBb-CAc and antAcAdAb gene clusters were operonic. In order to confirm that Fdr was active in the CARDO system, functional analyses were performed.

Functional analyses of putative CARDO and ANTDO
Biotransformation experiments were carried out using E. coli cells expressing putative CARDO components (CarAa, CarAc and Fdr) and ANTDO components (AntAc, AntAd, and AntAaAb) ( Figure 4). 29-Aminobiphenyl-2,3-diol and hydroxycarbazole could be detected for incubations of cell harboring pUcarAaAcfdr ( Figure S3). About 80% carbazole was converted to 29-aminobiphenyl-2,3-diol ( Figure 4A). These results clearly indicated that CarAa could catalyze angular dioxygenation and lateral dioxygenation of carbazole using CarAc and Fdr as an electron-transfer system. E. coli DH5a (pUcarAaAc) encoding only the oxygenase and ferredoxin components transformed no more than 0.1% of the carbazole, which suggested that CarAc had the ability to accept electrons from unidentified reductase in E. coli, with a much lower efficiency than the natural reductase. Furthermore, CarAc showed similarity with the cytochrome P 450 -type reductase component (RedA2) of dioxin dioxygenase from Sphingomonas wittichii RW1 [17] and putidaredoxin-type ferredoxins (CamA) from Pseudomonas putida [19]. CARDO systems of strain XLDN2-5 can be classified in the class IIA Rieske non-heme iron oxygenase system [10]. The IPTG-induced E. coli cells carrying pUantAcdba or pUantAcdb readily converted anthranilate to catechol. E. coli DH5a (pUAntAcdba) completely transformed 1 mM anthranilate to catechol in 24 h ( Figure 4B), while E. coli DH5a (pUAntAcdb) encoding only ferredoxin and oxygenase transformed only 5% of the anthranilate to catechol (data not shown). These results suggested that the expression of CARDO and ANTDO in E. coli requires the presence of the ancillary reductases for full enzymatic activity.

Discussion
Starting from the carAaBaBb fragment isolated from Sphingomonas sp. strain XLDN2-5, a combination of southern blot, genome walking, and IS-based PCR led to the isolation of two loci containing genes involved in the conversion of carbazole from anthranilate to catechol. The DNA fragments obtained are illustrated in Figure 2, and the proposed reaction details are given in Figure 1. All genes were embedded in transposon-like entities, implying the likely involvement of horizontal gene transfer in the evolution of carbazole degradation pathway. Transcriptional and functional analyses suggested that all the genes worked in the degradation of carbazole. The CARDO system catalyzed angular and lateral dioxygenation of carbazole, and ANTDO could transform anthranilate to catechol in a single step. Our research provided useful and important information for the association of IS6100 element with carbazole and anthranilate catabolic genes.
Carbazole-degrading sphingomonads have been isolated in different parts of the world. For example, Sphingomonas. sp. GTIN11 was isolated from USA [3], Novosphingobium sp. KA1 from Japan [10] and strain XLDN2-5 from China [2]. Although nearly identical car and ant genes were found in carbazoledegrading sphingomonad strains isolated from geographically dispersed locations, the organization of these genes in strain XLDN2-5 was unique. Figure 5 shows the organization of the known car genes ( Figure 5A) and ant genes ( Figure 5B) from different evolutionary origins. There were two truncated ORF1 and ORF8 encoding a putative integrase and a TonB-dependent receptor (Table 1), respectively. ORF1 and ORF8 showed 100% identity to ORF42 and ORF35 in strain KA1 [21], respectively. In strain KA1, ORF42 and ORF35 are intact ( Figure 5A), whereas the corresponding ORFs were interrupted by IS6100 in strain XLDN2-5. Furthermore, BLAST search revealed that the  terminal 27-bp region of the reported sequence (1-55 bp in AF442494) in strain GTIN11 is identical to the 27-bp left end of IS6100. This suggested that the carR gene in strain GTIN11 might be disrupted by IS6100 element. However, the entire sequence information outside the car genes in strain GTIN11 was not reported.
Interestingly, the ant cluster in strain XLDN2-5 was also bordered by copies of IS6100 elements. The organization of ant gene cluster was clearly different from that of KA1, in which catF, the upstream region of and cluster, encodes acetyl-CoA acetyltransferase, while the downstream ORF16 encodes a transposase ( Figure 5B). Although there are many ISs belonging to different family, no IS6100 element was found on pCAR3 [21]. The structure and sequence identified in this study suggested that IS6100 element might play an important role in the transfer of carbazole-degrading genes. IS6100 was originally isolated as part of the composite transposon Tn6100 from Mycobacterium fortuitum [16]. Later studies revealed its presence in a wide spectrum of host organisms in distantly related bacterial lineages, e.g. Sphingomonas paucimobilis [14], Arthrobacter sp. [15], Pseudomonas aeruginosa [22], Xanthomonas campestris [23], Salmonella enterica [24], and Corynebacterium glutamicum [25]. The existence of IS6100 in a very broad host range indicates that the IS6100 element plays a vital role in disseminating genes, including catabolic and antibiotic resistance genes, among different bacteria. The sequenced copies of IS6100 from strain XLDN2-5 were 100% identical to those of Mycobacterium fortuitum. This supports the hypothesis that IS6100 elements are distributed widely among microorganisms, whether they are Gram-negative or Gram-positive bacteria [14,15]. IS6100 belongs to the IS6 family of ISs which exclusively integrate through a cointegrative mechanism [13]. The car and ant genes might become part of strain XLDN2-5 via cointegrative captures. This type of transposition event is consistent with the truncation of the ORF1 and the ORF8 located upstream of carR and carAc, respectively ( Figure 2 and Figure 5A). Thus, the genetic structure of the car and ant genes in strain XLDN2-5 has shown strong evolutionary trails of gene recruitment and presents an extraordinary example of rearrangements and pathway evolution.
Although the terminal oxygenase and ferredoxin components of CARDO from XLDN2-5 and KA1 are identical, the reductase components have a much lower level of similarity (62%). In general, the genes encoding the components of the known Rieske oxygenases were found closely together in tightly regulated transcriptional units, as was observed with the car cluster in strain CA10 ( Figure 5A) [26]. However, it is becoming increasingly evident that the genes for catabolic pathways in sphingomonads often locate separately from each other. Recently, it is reported that multiple carbazole degradation genes dispersed on four loci (two car loci, one fdxIfdrI locus and one fdrII locus) on pCAR3 in Novosphingbium sp. strain KA1 [10]. The fdxIfdrI locus located 50 and 85 kb downstream of carI and carII gene clusters, while the fdrII gene located about 80 and 115 kb downstream of carI and carII gene clusters. Additional evidence has also been presented for the genes involved in the degradation of PAH by Sphingobium yanoikuyae B1 and Q1, and Novosphingobium aromaticivorans F199 [11,12,27], dibenzo-p-dioxin by Sphingomonas wittichii RW1 [28], pentachlorophenol by Sphingmonas chlorophenolica ATCC 39723 [29], and c-hexachlorocyclohexane by Sphingmonas paucimobilis UT26 [30]. These results suggested that loose association with reductase components might be the characteristic for this type of dioxygenase, which became more independent from the reductase component during evolution.
In the course of cloning car genes, we also discovered a novel insertion sequence ISSsp1 (see results, IS6100::ISSsp1), which insert into the upstream copy of IS6100. ISSsp1 can be classified as the mutator family that consists of transposases from prokayotes and eukaryotes. There was only one example that an IS could insert into another in Comamonas sp. strain JS46 [31]. Thus, it would be interesting to investigate the distribution of ISSsp1 in strain XLDN2-5 and its function in the establishment of the carbazole catabolic gene structure.

Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in Table 2. Sphingomonas sp. strain XLDN2-5 utilizes carbazole as the sole carbon and nitrogen source [2,32]. Strain XLDN2-5 was grown in mineral salt medium (MSM) as previously described [33], and carbazole was added as a 200 mM filter-sterilized stock solution in dimethyl sulfoxide (DMSO). Escherichia coli DH5a was used as the recipient strain in all cloning experiments. E. coli strains were grown in Luria-Bertani (LB) broth at 37uC. Ampicillin (Amp), when required, was added to a final concentration of 100 mg ml 21 .

DNA manipulation
Total DNA from pure cultures of Sphingomonas sp. strain XLDN2-5 was extracted using the WizardH Genomic DNA Purification Kit according to the recommendations of the manufacturer (Promega Corp., Madison, WI). Restriction endonucleases and T4 DNA ligase were used according to the manufacturer's instructions (TaKaRa). Isolations of DNA fragments from agarose gels were accomplished with the Qiaex II Gel Extraction Kit (Qiagen Corp., Germany). Transformations and agarose gel electrophoresis were carried out using standard methods [34].

Construction of genomic library
The genomic DNA of Sphingomonas sp. XLDN2-5 was mechanically sheared. DNA fragments of 6 to 8 kb were gel purified and used for library construction. The fragments were ligated into pUC19 digested with EcoRV, and dephosphorylated with shrimp alkaline phosphatase (Promega). The library was transformed into electrocompetent E. coli DH5a. After incubation at 37uC for 1 h, cells were spread on LB agar plates supplemented with Amp, isopropyl-1-thio-b-D-galactopyranoside (IPTG) and 5bromo-4-chloro-3-indolyl-b-D-galactopyranoside. Following overnight incubation at 37uC, plates were scored for white colonies. The white colonies were used for southern blotting using a car probe obtained by PCR.
All PCR amplifications were performed with an Eppendorf Authorized Thermal Cycler (Germany) in 50-ml reaction systems containing 5 ml of 106 buffer, 1.5 mM MgCl2, 200 mM dNTPs, 500 pmol of each primer, 10-100 ng of the template DNA, and 2.5 units of DNA polymerase. TransStart FastPfu polymerase (TransGen Biotech Co. Ltd. China) was used in PCRs whose products were used for the construction of plasmids. TaKaRa LA PCR TM Kit Ver. 2.1 was used for IS-based PCR. Taq polymerase (Generay Biotech Co. Ltd. China) was used in other PCRs. Genome walking was performed using a Genome Walking Kit (TaKaRa) according to the manufacturer's protocol. The amplified products obtained by Taq polymerase were gel purified and ligated into vector pMD18-T, followed transformation to competent E. coli DH5a. All primers used in this study are listed in Table S1.

Southern blot
Primers pcarF and pcarR were designed according to the carRAaBaBb genes of strain KA1 [21]. Products obtained using primers pcarF and pcarR were gel purified and labeled with DIG-11-dUTP, and the DNAs were transferred following the standard protocols [34]. Southern blot was performed using DIG DNA Labeling and Detection Kit (Roche) according to the manufac-

DNA sequencing and analysis
Sequencing was performed on an ABI sequencer by Shanghai Invitrogen Biotechnology Co., Ltd, China. The sequences were analyzed with Vector NTI DNA analytical software (version 8). Homology searches were performed with the BLAST programs at the National Center for Biotechnology Information website (http://www.ncbi.nlm.nih.gov/BLAST.html). The deduced amino acid sequences of ORFs were aligned using ClustalW 1.83 [35].

RNA preparation and RT-PCR
After the precultivation of XLDN2-5 in 5 ml of MSM supplemented with 2 mM carbazole at 30uC, cells were harvested by centrifugation at 5,000 g for 5 min and then washed twice using MSM. The washed cells were suspended in 500 ml of MSM. Fifty microliters of the resultant cell suspension was added to 5 ml of MSM supplemented with 5 mM carbazole or with 1 g L 21 glucose and 0.5 g L 21 NH4Cl. After a 2-h incubation with reciprocal shaking (300 strokes/min) at 30uC, the cells were harvested and used for extraction of total RNA using E. Z. N. A TM Total RNA Kit I (Omega Biotech). Finally, the RNA was eluted in 50 ml of RNase-free water. A PrimeScript TM One Step RNA PCR kit (Takara) was used for RT-PCR, in which 100 ng of total RNA was used as a template. Detailed information on the RT-PCR primer sets and the conditions employed for respective gene amplifications are provided in Table S1. Control experiments without the addition of reverse transcriptase were also included.

Construction of plasmids for car and ant genes
The genes, carAa, carAc, and fdr were separately amplified by PCR using the respective primer sets shown in Table S1, which were designed to introduce appropriate restriction sites and the effective ribosome binding sites. In PCR amplification, total DNA of strain XLDN2-5 was used as a template. The amplified products were digested at the introduced restriction sites and ligated into the corresponding sites of pUC19 to produce plasmids (pUcarAa, pUcarAc, and pUfdr) for the expression of single CARDO components. For construction of pEcarAaAc, the carAa gene was amplified by carAa-F1 and carAaAcR, and carAc gene was amplified by carAaAcF and carAc-R1. carAa and carAc were then used as templates, while carAa-F1 and carAc-R1 were added to amplify carAaAc. The fragment obtained was ligated to pEASY-Blunt to generate pEcarAaAc. After their nucleotide sequences were confirmed to be identical to those designed, pEcarAaAc was double-digested using SphI and KpnI and the 1.5-kb fragment containing carAaAc was gel-purified and ligated to the corresponding sites of pUC19 and pUfdr to give pUcarAaAc and pUcarAaAcfdr.
Genes antAcAdAb and antAcAdAbAa were amplified by PCR using the respective primer sets, which were designed to introduce appropriate restriction sites and the effective ribosome binding sites. The amplified fragments were digested with HindIII and EcoRI before being ligated into pUC19 to generate pUAntAcdb and pUAntAcdba.

Biotransformation analysis
The E. coli DH5a strains harboring pUcarAaAcfdr, pUcar-AaAc, pUantAcdba or pUantAcdb were cultivated in 5 ml of LB supplemented with Amp at 37uC. Then 100 ml of the culture was transferred to 200 ml of the same medium, and IPTG was added to induce the protein at 37uC for 16 h. Then the cells were harvested by centrifugation (6,000 g, 10 min, 4uC), washed twice with MSM, and resuspended in MSM to an OD 600 of 10. Fifty microliters of carbazole (200 mM in DMSO) or anthranilate (200 mM in DMSO) was added to 10 ml of cell suspensions. After incubation on a reciprocal shaker (200 rpm) at 37uC for 20 h, the mixtures were extracted with an equal volume of ethyl acetate. After derivatization with N,O-bis-(trimethylsilyl)trifluoroacetamide (BSTFA) (Sigma) the extracts were analyzed by gas chromatography-mass spectrometry (GC-MS) as described previously [2,33].
High performance liquid chromatography was carried out to analyze the aqueous samples using an Agilent 1200 series instrument equipped with a variable-wavelength detector and a reversed-phase C18 column (4.6 mm6150 mm, Hewlett-Packard). Residual concentrations of carbazole and anthranilate and the formation of 29-animobiphenyl-2,3-diol and catechol were determined using a mobile phase of an 80:20 mixture of methanol and deionized water at a flow rate of 0.5 ml min 21 .

Author Contributions
Conceived and designed the experiments: ZG PX. Performed the experiments: ZG XW XL CT. Analyzed the data: ZG HT XH PX. Contributed reagents/materials/analysis tools: GW ZD PX. Wrote the paper: ZG PX.