Introduction

Carbazole, an environmental pollutant, is a heteroaromatic compound containing nitrogen found in coal–tar creosote. Because, bioremediation has proven to be an effective technique for cleaning sites polluted by persistent compounds such as carbazole, carbazole-degrading bacteria namely Pseudomonas resinovorans strain CA10 [20, 21], Novosphingobium sp. strain KA1 [6], Janthinobacterium sp. strain J3 [11], Pseudomonas stutzeri strain OM1 [17], Sphingomonas sp. strain CB3 [22], and Nocardioides aromaticivorans strain IC177 [10] have been isolated and studied. However, these are all terrestrial bacteria, and no genetic study has been performed on carbazole-degrading marine bacteria. Previously, our group has reported on the isolation and characterization of car gene cluster from the marine isolate strain OC7 [14]. Even though this strain was isolated from marine environments, it showed closest relation to bacteria in the Lysobacter genus and no evidence being a marine bacterium. The carbazole-degrading bacterium Neptuniibacter sp. strain CAR-SF was isolated from seawater [4]. In contrary to strain OC7, growth tests using artificial seawater mineral salt plates showed no growth of strain CAR-SF when NaCl was not present. Bacteria degrade carbazole by conversion to 2-hydroxypenta-2,4-dienoate and anthranilate via dioxygenation by carbazole 1,9a-dioxygenase (CarA), aromatic ring cleavage by the meta-cleavage enzyme (CarB), and hydrolytic cleavage by the meta-cleavage compound hydrolase (CarC). Next, 2-hydroxypenta-2,4-dienoate is converted to TCA cycle intermediates by 2-hydroxypenta-2,4-dienoate hydratase (CarD), 4-hydroxy-2-oxovalerate aldolase (CarE), and acetaldehyde dehydrogenase (CarF). Anthranilate is also converted to TCA cycle intermediates via dioxygenation by anthranilate 1,2-dioxygenase (AntABC), spontaneous deamination, decarboxylation, and further degradation [16]. In this study, we isolated the first carbazole degradation genes from a marine bacterium strain CAR-SF. The genes were characterized by reverse transcription-polymerase chain reaction (RT–PCR) and biotransformant products were confirmed by GC/MS using resting cell reactions with Escherichia coli.

Materials and Methods

Gene Isolation

Total DNA of carbazole-utilizing bacterium CAR-SF was extracted from cells grown in artificial seawater mineral salts medium ONR7a [1] with carbazole as the sole carbon source at 30°C using standard protocols. Extracted total DNA was digested with EcoRI or EcoRV before ligation into plasmid vectors pUC119 or pBluescript II SK(−) linearized using similar restriction endonuclease and treated with calf intestinal alkaline phosphatase (Takara Bio Inc., Shiga, Japan). E. coli DH5alpha was transformed by recombinant plasmids constructed. Clones were screened for meta-cleavage enzyme activity by spraying with 2,3-dihydroxybiphenyl-acetone solution. Nucleotide sequences were determined as previously reported [13]. To analyze the obtained sequences, DNASIS Pro software (version 2.8; Hitachi Software Engineering Co. Ltd., Kanagawa, Japan) was employed. The BLAST program (http://www.ncbi.nlm.nih.gov/BLAST/) was used for homology searches and CLUSTAL W algorithm provided in DNASIS Pro was used to align sequences. The nucleotide sequences presented in this study have been deposited in DDBJ, EMBL, and GenBank nucleotide sequence databases under accession numbers AB365208 and AB365209.

Construction of Expression Plasmids

To express the carbazole dioxygenase genes under control of the lac promoter in E. coli, the recombinant plasmids shown in Table 1 were constructed. The forward primers were designed to add an XbaI site and a Shine–Dalgarno sequence upstream of the initial codon. The reverse primers were designed to add a SpeI site downstream of the stop codon. Genes carAa and orf6 were amplified from pBSF101 and pUSF01. Genes carAc and carAd from strain CA10, and carAcI from strain KA1 were amplified from pUCA1 [20] and pBKA102 [11]. Amplified products were cloned into pT7Blue(R) T-vector (Novagen, Madison, WI, USA) before subcloning into pBluescript II SK(−) to produce each designated plasmids.

Table 1 Bacterial strains and plasmids used in this study
Full size table

RNA Extraction and RT–PCR

Total RNA samples were extracted on strains grown in ONR7a containing 0.1% (w/v) carbazole or 0.1% (w/v) anthranilate for 24 h. Samples from strain grown with 0.1% (w/v) succinate was used as negative control. For RT–PCR, primers were designed to amplify a 300-bp fragment of 16S rRNA, carAa, and antA. RT–PCR was run as follows: 50°C for 30 min; 94°C for 2 min; 30 cycles of 94°C for 2 min, 55°C for 15 s and 72°C for 30 s; and 72°C for 5 min. Total RNA extractions and RT–PCR preparations were conducted as reported previously [14]. The primer sets used were CarAa-L (5′-gctcaagctccttggtgaac-3′) and CARAa-R (5′-ggaggggtatctccatcacc-3′) specific for carAa gene, AntA-L (5′-ataagaccgacggcaaactg-3′) and AntA-R (5′-aaccatcaagaccgttttcg-3′) specific for antA gene, 16S-L (5′-tcgggaactctgagacaggt-3′) and 16S-R (5′-ttcatggagtcgagttgcag-3′) specific for 16S rDNA. Intensities of gene transcripts were measured and compared using Quantity One software (Bio-Rad Laboratories, Hercules, CA).

Resting Cell Reaction

Resting cell reaction experiments were prepared as reported previously [14]. 0.01% (w/v) carbazole was used and reaction was continued for 24 h at 37°C. Reaction contents were extracted with the same volume of ethyl acetate, dried over anhydrous Na2SO4 and the solvent was removed under reduced pressure. After trimethylsilylation with N-methyl-N-trimethylsilyltrifluoroacetamide (Nacalai Tesque, Kyoto, Japan) at 70°C for 20 min, each sample was injected into a GC/MS system (Thermo Electron, Kanagawa, Japan). Plasmids pUCARA [21] and pBluescript II SK(−) were used as positive and negative controls, respectively. The carbazole conversion ratio of each sample was determined by the formula given in Urata et al. [25]. All experiments were carried out in triplicate.

Results and Discussion

Isolation and Identification of car Gene Cluster

Three different plasmids were isolated and identified from clones showing meta-cleavage activity. A plasmid containing 4.2-kb fragment of ScaI-digested total DNA ligated into EcoRV-digested pBluescript II SK(−) was designated pBSF101. Two other plasmids containing a 6.8- and a 6.7-kb fragment of EcoRI-digested total DNA ligated into EcoRI-digested pUC119 were designated pUSF01 and pUSF09. Plasmids pBSF101 and pUSF01 contained genes of the upper pathway and pUSF09 contained the genes responsible for the lower pathway degradation (Fig. 1). Nine complete open reading frames (ORFs) were identified from pBSF101 and pUSF01, among them five involved in carbazole degradation (Table 2). The deduced amino acid sequence of ORF2 had 77% identity to the terminal oxygenase component of carbazole 1,9a-dioxygenase (CarAa) from Janthinobacterium sp. strain J3. Strain J3 is known to possess carbazole degrading genes (car genes) with high similarity to P. resinovorans strain CA10 [11]. A Rieske-type [2Fe–2S] cluster and mononuclear nonheme iron-binding site [15, 18] motifs were also conserved. ORF3 and ORF4 were deduced as the small and large subunits of 2′-aminobiphenyl-2,3-diol 1,2-dioxygenase and showed 48 and 59% similarity with CarBa and CarBb of strain CA10. ORF5 gene product showed 67% identity with the meta-cleavage compound hydrolase (CarC) from strain J3. ORF6 showed a low similarity to Rieske [2Fe–2S] domain protein such as the ferredoxin component from Burkholderia thailandensis strain E264 (16% identity; [12]) and from strain J3 (12% identity). Notably, the 5′ terminal region of ORF6 showed 33% identity with CarAcJ3 and CarAcCA10. However, no conserved motif of a Rieske [2Fe–2S] iron–sulfur protein was found. The ORF2 until ORF5 (carAaBaBbC) formed the car gene cluster which is similar to car gene clusters of strain CA10 and J3 (carAaAaBaBbCAcORF7Ad). Even though the ferredoxin component was not included in the isolated fragments, a weak signal of a CarAcCA10-like ferredoxin was detected from the total DNA extract using Southern hybridization (result not shown). This result suggested that a possible functional ferredoxin for CarAaCAR-SF existed but separated from the isolated car gene cluster. Plasmid pUSF09 contained six complete ORFs and one partial ORF (Fig. 1 and Table 3). The partial ORF10 was deduced to be a putative AraC/XylS-type transcriptional regulator and orientated similarly to the same gene found in Pseudomonas fluorescens strain MB214 [19]. The deduced amino acid sequences of ORF11 and ORF12 were similar to the large and small subunits of anthranilate 1,2-dioxygenase and showed 72 and 61% identity with AntA and AntB of strain CA10 [25]. A Rieske-type [2Fe–2S] cluster-binding site motif and mononuclear nonheme iron-binding site motif were also conserved. ORF13 gene product containing a plant-type [2Fe–2S] cluster-binding motif and FAD/NAD-binding domains showed 53% identity to anthranilate dioxygenase reductase (AntC) of Acinetobacter sp. strain ADP1 [2]. ORF15 gene product showed similarity to a meta-cleavage enzyme catechol 2,3-dioxygenase from various bacteria such as the Acinetobacter sp. strain YAA (78% identity; [3]), P. stutzeri strain OM1 (77% identity; [17]), and P. putida strain G7 (75% identity; [5]). This enzyme showed strong cleaving activity toward 2,3-dihydroxybiphenyl which led to the isolation of this gene cluster. Such strong activity indicated the possibility of this enzyme being able to assist the ring cleavage in the upper pathway of carbazole degradation. The genes of this cluster suggested that strain CAR-SF converts anthranilate to catechol by anthranilate 1,2-dioxygenase (AntABC), similar to strain CA10. However, strain CA10 continues the downstream conversion using an ortho-cleavage enzyme, whereas strain CAR-SF presumably utilizes a meta-cleavage enzyme. Catechol catabolic genes such as the xyl genes of the TOL plasmid pWW0 and the nah genes of plasmid NAH7 [7, 8] also convert catechol to a TCA cycle intermediate through meta-cleavage. Similarly, the carbazole-degrading P. stutzeri strain OM1 metabolizes catechol using the same pathway [17]. Nevertheless, detection of metabolism products is needed to confirm the degradation pathway for strain CAR-SF.

Fig. 1
figure 1

Organization of the genes responsible for the upper and lower carbazole degradation pathways of the strain CAR-SF. Restriction sites shown: B, BamHI; E, EcoRI; P, PstI; S, ScaI. Pentagons indicate the direction and region of gene transcription

Full size image
Table 2 Homology of deduced ORFs in the 4.2-kb ScaI and 6.8-kb EcoRI fragments of the strain CAR-SF
Full size table
Table 3 Homology of deduced ORFs in the 6.7-kb EcoRI fragment of strain CAR-SF
Full size table

Transcriptional Analyses of car and ant Genes by RT–PCR

The transcripts of carA were detected while antA transcripts were not detected when grown with succinate. The antA transcripts detected from cells grown with carbazole were approximately 15% more when compared to the same transcripts from anthranilate-grown cells. When cells grown with carbazole or anthranilate, the intensities of carA transcripts were 380 and 460%, respectively, in comparison to the intensity of carA transcripts from cells grown with succinate (Fig. 2). These results proved the involvement of the isolated genes in carbazole degradation and the expression of both genes were induced when carbazole or anthranilate was supplied. Previously, car operon of strain CA10 was reported to be induced by the presence of carbazole or anthranilate [24]. However, car operon of strain IC177 was reported to be induced by carbazole but not anthranilate [9]. Thus, the transcriptional mechanism of car operon of strain CAR-SF may be similar to the strain CA10. The promoter region of car gene cluster was not included in the isolated gene fragments from strain CAR-SF. Thus, it is unknown whether car gene cluster of strain CAR-SF and strain CA10 share the same promoter region.

Fig. 2
figure 2

RT-PCR amplications of 16S rRNA, carAa, and antA from total RNA extracted from cells when grown with succinate, carbazole, or anthranilate as sole carbon source. Lane 1, amplication of 16S rRNA using Taq polymerase (negative control); Lane 2, amplication of 16S rRNA using RT polymerase (positive control); Lane 3, amplication using carAa specific primer set; Lane 4, amplication using antA specific primer ser

Full size image

Biotransformation of Carbazole with E. coli Cells

Based on GC/MS analyses, no product was detected other than 2′-aminobiphenyl-2,3-diol (the conversion product of carbazole through angular dioxygenation by carbazole 1,9a-dioxygenase and a spontaneous reaction). The mass spectrum of 2′-aminobiphenyl-2,3-diol detected was trimethylsililated [23] for each sample. The carbazole conversion ratios are shown in Table 4. The products of carAa, carAc, carAd, and orf6 genes from E. coli cells harboring plasmids encoding respective genes were detected using 13% SDS polyacrylamide gel (Fig. 3). Peptides with molecular masses of approximately 44, 36, 12, and 11 kDa correspond to the predicted molecular masses of carAa, carAd, orf6, and carAc gene products, respectively. E. coli JM109 harboring pBSF1 or pBSF2 converted carbazole at the rate of about 1.0%, suggesting that CarAaSF was able to weakly utilize the ferredoxin and ferredoxin reductase available in E. coli. Although a part of ORF6 showed low identity to a Rieske-type ferredoxin, it did not function as a ferredoxin. E. coli JM109 carrying pBSF3 exhibited a high conversion rate (approximately 86%), implying that CarAaSF is capable of utilizing CarAcCA10 efficiently. In addition, E. coli JM109 harboring pBSF6 converted carbazole at a rate of over 99%. This result indicates that the carbazole 1,9a-dioxygenase from strain CAR-SF is a multicomponent dioxygenase system consisting of three components, as in other well-known carbazole degrading bacteria [9, 21, 25] even though its ferredoxin and ferredoxin reductase component were not found in the isolated cluster.

Table 4 Carbazole conversion ratios obtained using E. coli cells carrying different biotransformation plasmids
Full size table
Fig. 3
figure 3

Detection of the products of carAa, carAc, carAd, and orf6. Total cellular proteins of E. coli strains were analyzed by SDS–PAGE. Lane 1, molecular mass standards of 94, 67, 43, 30, 20.1, and 14.4 kDa (top to bottom); Lane 2, JM109(pBluescript SK(−)); Lane 3, JM109(pBSF1); Lane 4, JM109(pBSF2); Lane 5, JM109(pBSF3); Lane 6, JM109(pBSF4); Lane 7, JM109(pBSF5); Lane 8, JM109(pBSF6)

Full size image

In conclusion, these genes are clustered in similar fashion with their terrestrial counterparts especially genes isolated from strain CA10 and strain J3. However, ferredoxin and ferredoxin reductase which are necessary for initial dioxygenation of carbazole were not found in the cluster and is believed to be located elsewhere. The ORF6 may be regarded as a trace of genetic movements during the formation of this cluster. This result might be a unique finding that may help reveal the evolutionary relationships among other reported car gene clusters.