Introduction

Organosulfur compounds contained in fossil fuels are one of the major causes of acid rain and air pollution since the combustion of fossil fuels releases sulfur oxides into the atmosphere. During the refinery process of petroleum, sulfur is removed from petroleum by hydrodesulfurization (HDS) with chemical metal-containing catalysts under extremely high temperatures and pressures. However, heterocyclic sulfur compounds, such as dibenzothiophene (DBT) and alkylated DBTs, are recalcitrant to conventional HDS processes, and alkylated DBTs especially remain in the fraction of light gas oil. Therefore, the development of a desulfurization process for more intensive desulfurization in light of increasingly stringent environmental regulations is required. One such method, the application of a biodesulfurization process using microbial catalysts capable of desulfurizing HDS-resistant sulfur compounds, such as alkylated DBTs, has been the focus of great interest because of its reduced burden on the environment with respect to energy consumption and CO2 evolution (for review, see Monticello 2000; Ohshiro and Izumi 1999, 2002; Suzuki 1999).

Various mesophilic bacteria capable of desulfurizing DBT, a model organosulfur compound recognized as a target for more intensive desulfurization, through targeted reactions at the carbon–sulfur bonds have been isolated (for review, Ohshiro and Izumi 1999, 2002; Suzuki 1999). For example, in Rhodococcus sp. IGTS8, genes involved in sulfur-specific desulfurization of DBT were identified, and three desulfurization genes (dszA, dszB, and dszC) were confirmed to constitute a single operon (Denome et al. 1993; Li et al. 1996; Piddington et al. 1995). In this bacterium, four enzymes consisting of two monooxygenases (DszC and DszA), one desulfinase (DszB), and one flavin reductase (DszD) that couples with DszC and DszA, are responsible for DBT desulfurization (Gray et al. 1996). Furthermore, in a bacterial community residing in sulfurous-oil-containing soils, the sequence homology and classification of desulfurization genes were determined using PCR (Duarte et al. 2001).

Since most DBT-desulfurizing bacteria are mesophilic, their DBT-desulfurizing activities are high at temperatures of around 30°C and lower at temperatures above 40°C (Nakayama et al. 2002; Ohshiro et al. 1997, 1999). However, HDS-treated light gas oil would be supplied to the biodesulfurization process at a temperature of more than 40°C and thus would have to first be cooled to around 30°C before being desulfurized by the bacteria (Borgne and Quintero 2003; Mcfarland 1999). Moreover, in the refinery process, to avoid solidification of the oils, the temperature of HDS-treated light gas oil is usually maintained at around 45°C even after recovery of the residual heat of the oil by the heat exchanger (COSMO Oil 2002). Therefore, the ability to carry out thermophilic biodesulfurization at 40–50°C is desirable and could be integrated more easily into the refining process, since the energy cost for cooling the HDS-treated light gas oil is thereby omitted. Another advantage of thermophilic biodesulfurization is that contamination by mesophilic bacteria would be avoided. With this goal in mind, we previously isolated two thermophilic DBT-desulfurizing bacteria, Bacillus subtilis WU-S2B (Kirimura et al. 2001) and Mycobacterium phlei WU-F1 (Furuya et al. 2001, 2002), which can desulfurize DBT and DBT derivatives through targeted oxidation of the carbon–sulfur bonds even at 50°C. Furthermore, we confirmed that M. phlei WU-F1 efficiently desulfurized HDS-treated light gas oil over a temperature range up to 50°C (Furuya et al. 2003).

The thermophilic DBT-desulfurizing bacterium Paenibacillus sp. A11-2 was previously isolated (Konishi et al. 1997), and its genes and enzymes involved in thermophilic desulfurization were characterized (Ishii et al. 2000a,b; Konishi et al. 2000). In this bacterium, three desulfurization genes (tdsA, tdsB, and tdsC) constitute an operon similar to that of Rhodococcus sp. IGTS8, and each of the tds genes shows a considerable homology to the corresponding dsz genes (Ishii et al. 2000a). The thermal profiles of DBT desulfurization by growing and resting cells of Paenibacillus sp. A11-2 showed that DBT was desulfurized only at a temperature of around 50°C (Konishi et al. 1997). By contrast, since growing and resting cells of B. subtilis WU-S2B and M. phlei WU-F1 can desulfurize DBT over a temperature range from 30 to 52°C (Furuya et al. 2001; Kirimura et al. 2001), these strains might be more practical. Before the properties of B. subtilis WU-S2B and M. phlei WU-F1 can be applied as genetic resources for biodesulfurization, the DBT-desulfurization genes must be identified and functionally analyzed. Furthermore, in order to understand explicitly the molecular basis of bacterial desulfurization, a comparative analysis of the nucleotide sequences of desulfurization genes and the deduced amino acid sequences of the respective enzymes from different species of bacteria is required.

In this report, we describe the identification and functional analysis of the genes responsible for DBT desulfurization in B. subtilis WU-S2B. The nucleotide sequences of three open reading frames (ORFs, designated bdsA, bdsB, and bdsC for bacterial desulfurization) were found to constitute a single operon, and the individual functions of these ORFs were characterized by their heterologous expression in Escherichia coli. The effects of temperature on the desulfurizing activity of recombinant E. coli cells harboring bdsABC and that of B. subtilis WU-S2B were compared. The results showed that the gene products of bdsABC were functional in the recombinant E. coli cells. In addition, we cloned and sequenced the DBT-desulfurizing genes of M. phlei WU-F1. Surprisingly, the desulfurization genes of the two bacteria were identical to each other. Therefore, we concluded that the desulfurizing abilities of B. subtilis WU-S2B and M. phlei WU-F1 over a temperature range up to 52°C are basically dependent on the function of bds gene products.

Materials and methods

Bacterial strains and culture conditions

B. subtilis WU-S2B (FERM P-17041, Kirimura et al. 2001) and M. phlei WU-F1 (FERM P-17717, Furuya et al. 2001) were used as sources of desulfurization genes. For isolation of total DNA, B. subtilis WU-S2B was cultivated as described previously (Kirimura et al. 2001) and M. phlei WU-F1 as described previously (Furuya et al. 2001) with the modifications of Poupin et al. (1999). E. coli XL1-Blue MRA(P2) and lambda DASH II phage vector (Stratagene, CA) were used for cloning of the desulfurization genes, and E. coli JM109 and cloning vector pGEM-T (Promega, WI) for cloning and sequencing of PCR products. E. coli JM109 and cloning vector pUC19 were used for subcloning of genes from the phage vector, and E. coli JM109 and pKK223-3 (Amersham Biosciences, NJ) for expression of the genes. E. coli strains were grown in either Luria–Bertani (LB) medium or NZY medium (Sambrook and Russell 2001). Recombinants were selected on LB agar plates or in liquid media containing 50 μg ampicillin/ml.

DNA manipulation

Total DNA of B. subtilis WU-S2B and M. phlei WU-F1 was prepared with a QIAGEN Genomic-tip (QIAGEN, CA). Phage DNA was purified with a QIAGEN Lambda Midi Kit, and plasmid DNA was purified with a QIAGEN Plasmid Mini Kit. DNA fragments were purified with a GFX PCR DNA and Gel Band Purification Kit (Amersham Biosciences). Digestion with restriction endonucleases, ligation, and deletion experiments were carried out using standard procedures (Sambrook and Russell 2001) under the conditions recommended by the manufacturers. Automated DNA sequencing was performed with an ABI PRISM 310 Genetic Analyzer (Applied Biosystems Japan, Tokyo, Japan). The sequence was determined by complete sequencing of both strands, with multiple sequencing of some regions. Sequence assembly and analysis were done using GENETYX-MAC ver10.1 (SDC, Tokyo, Japan). When necessary, PCR-derived fragments were sequenced to confirm that no point mutation had occurred. The GenBank and SwissProt databases were searched for similarities of nucleic acid and amino acid sequences using the FASTA program of the DNA Data Bank of Japan (DDBJ) and the BLAST program of the National Center for Biotechnology Information (NCBI). The analysis of amino acid homology by the phylogenetic tree was carried out using the ClustalW program of the DDBJ.

Construction of a total DNA library

Total DNAs of B. subtilis WU-S2B and M. phlei WU-F1 were partially digested with Sau3AI and treated with bacterial alkaline phosphatase. The DNA fragments were ligated to BamHI-digested lambda DASH II arms (Stratagene). The recombinant phage DNA was packaged into phage heads in vitro with the Gigapack III Gold packaging kit (Stratagene). The packaged phages were propagated in E. coli XL1-Blue MRA(P2) cells.

Oligonucleotide primers, probes, and hybridization

The nucleic acid sequences of the PCR primers were designed using conserved amino acid sequences of DBT-desulfurizing enzymes from Rhodococcus sp. IGTS8 and Paenibacillus sp. A11-2. PCR was done with the GeneAmp PCR system 9700 (Applied Biosystems Japan) as follows. The PCR reaction mixture (50 μl ), containing 10 mM Tris–HCl (pH 7.0), 50 mM KCl, 1.5 mM MgCl2, dNTP (each species, 200 μM), PCR primers (1 μM each), DNA template, and Taq DNA Polymerase (1.25U, Nippon Roche, Tokyo, Japan), was reacted according to the following PCR program: 5 min at 95°C; 30 cycles of 1 min at 95°C, 1.5 min at 52°C, and 3 min at 72°C; followed by a final 10 min at 72°C. The PCR product was cloned in E. coli and sequenced.

A DNA probe was prepared by randomly labeling the PCR product with digoxigenin-11-dUTP using the DIG DNA Labeling Kit (Nippon Roche) and then used for screening clones harboring desulfurization genes. The lambda DASH II library of B. subtilis WU-S2B total DNA was plated on a lawn of E. coli XL1-Blue MRA(P2) cells grown on NZY agar plates and transferred to Hybond-N+ nylon membranes (Amersham Biosciences). The membranes were probed at 68°C in 5×SSC, 1% (w/v) blocking reagent (Nippon Roche), 0.1% (w/v) N-lauroylsarcosine and 0.02% (w/v) SDS, and washed twice at 25°C in 2×SSC and 0.1% (w/v) SDS, and at 68°C in 0.1×SSC and 0.1% (w/v) SDS.

Construction of the expression plasmids for desulfurization genes

In order to adjust the restriction site to the cloning site of pKK223-3 for construction of expression plasmids, a fragment from −41 to 386 bp in the direction of the start codon of bdsA was amplified by PCR using the primers 5′-GCC AAG CTT TGG TCT CCG GTA ACT GAT CCC-3′ and 5′-GAC CAC ATT CCA CGA GAT CCG GC-3′ to generate a HindIII site and BssSI site, respectively. The amplified 0.4 kb HindIII–BssSI fragment and a 3.4-kb BssSI-MluI fragment of pU1Nh were cloned into the HindIII site of pKK223-3 to construct plasmid pKBDS, since MluI has the same cohesive ends as HindIII. The deletion plasmids pKBDSA, pKBDSB and pKBDSC were constructed from pKBDS with restriction endonuclease digestion and self-ligation.

Chemicals

DBT, 2-hydroxybiphenyl (2-HBP), and 2,8-dimethylDBT were purchased from Tokyo Kasei (Tokyo, Japan). DBT sulfone (DBTO2) and benzo[b]naphtho[2,1-d]thiophene (3,4-benzoDBT) were purchased from Aldrich (Milwaukee). 4,6-DimethylDBT, 2′-hydroxybiphenyl-2-sulfinate (HBPSi) and 2-ethylnaphtho[2,1-b]thiophene (2-ethylNTH) were kindly supplied by the laboratory of Japan Cooperation Center, Petroleum (Shizuoka, Japan). All other reagents were of analytical grade and commercially available.

Thermal profile of desulfurization by resting cells of the recombinant E. coli

Recombinant E. coli JM109 cells were cultivated overnight at 30°C in LB-ampicillin medium and then diluted 100-fold with fresh LB-ampicillin medium followed by shaking at 37°C for 16 h. Cells were harvested at 4°C by centrifugation at 8,000 g for 5 min, washed twice and resuspended in 0.1 M potassium phosphate buffer (pH 7.0) to give an optical density at 660 nm (OD660) of 20. Fifty microliters of substrate solution, DBT and its metabolic intermediates (5.4 mM, in n-tridecane or ethanol), were added to 1 ml of the cell suspension in a screw-cap test tube. For measurement of desulfurizing activity in an organic/aqueous two-phase reaction system, n-tridecane was used as the organic phase since it dissolved DBT and showed no harmful effect on E. coli. The reaction was done at 50°C or at the temperatures indicated in the figures with inverted shaking in a rotating incubator at 50 rpm for 1 h. After the reaction, the bacterial suspension was acidified by adding 20 μl of 6 M HCl and extracted with 0.5 ml ethylacetate. The extract was filtered through a 0.20-μm polytetrafluoroethylene membrane filter (Advantec Toyo, Tokyo, Japan) and then analyzed by high-performance liquid chromatography (HPLC, type LC-10A; Shimadzu, Kyoto, Japan) using a Puresil C18 column (Waters, MA), as described by Kirimura et al. (2001).

Substrate specificity of desulfurization by growing cells of recombinant E. coli

The recombinant E. coli JM109 cells harboring plasmids were incubated overnight at 30°C in LB-ampicillin medium. Aliquots (50 μl) of these cultures were inoculated into 5 ml of fresh, modified M9-IPTG-ampicillin medium using MgCl2 instead of MgSO4, and substrate solution (50 μl) containing DBT or its derivatives (5.4 mM, in n-tridecane) was added. After cultivation with reciprocal shaking at 37°C for 72 h, the culture broth was acidified by adding 100 μl of 6 M HCl and extracted with 2 ml ethylacetate. The resulting extract was then analyzed by gas chromatography (GC, type GC-2010; Shimadzu) equipped with a flame ionization detector, a flame photometric detector and a 30-m type DB-5 capillary column (J&W Scientific, CA). The flow rate of the nitrogen carrier gas was 15 ml/min, the column temperature was set at 210°C (240°C for 3,4-BenzoDBT), and the injector and detector temperatures were maintained at 260°C. B. subtilis WU-S2B was cultivated with reciprocal shaking at 45°C for 72 h using A-2 medium (Kirimura et al. 2001) instead of LB or modified M9 medium.

Accession number

The nucleotide sequence data for the desulfurization genes of B. subtilis WU-S2B were submitted to the DDBJ/EMBL/GenBank nucleotide sequence databases under the accession number AB076745.

Results

Cloning of the DBT-desulfurization genes from B. subtilis WU-S2B

DBT-desulfurization genes constituting operons, such as dszABC and tdsABC, were previously cloned from the mesophilic bacterium Rhodococcus sp. IGTS8 (Denome et al. 1993; Piddington et al. 1995) and the thermophilic bacterium Paenibacillus sp. A11-2 (Ishii et al. 2000a), respectively. Since the conserved regions in the deduced amino acid sequences of these genes were searched and identified as AEARNFG, between 429 and 447 nucleotides from the initiation codon of dszA, and GFDRFWR, between 99 and 116 nucleotides from the termination codon of dszC, they were chosen as appropriate regions for PCR primers. Based on these amino acid and nucleotide sequences, degenerated primers (5′-GCI GAR GCI MGI AAY TTY GG-3′ as the sense primer and 5′-CGT IGC GCC AIA AGC GGT C-3′ as the anti-sense primer; I, inosine) were synthesized and used for PCR with total DNA prepared from B. subtilis WU-S2B as the template. A 3.2-kb amplified DNA fragment was cloned into the pGEM-T vector, and the nucleotide sequence of this PCR product was determined. Since the amino acid sequence deduced from the nucleotide sequence showed approximately 60% homology to both dsz and tds, this PCR product was presumed to be a part of the DBT-desulfurization genes of B. subtilis WU-S2B. Primers for amplification of the internal region of this PCR product (5′-GCA TGA CAT CCG ATA CG-3′ as the sense primer and 5′-TAG TTT GGG TGG GTT CC-3′ as the anti-sense primer) were subsequently synthesized and used for another PCR with total DNA of B. subtilis WU-S2B. The PCR product amplified with these primers was labeled and used as probe to screen a total DNA library of B. subtilis WU-S2B in the lambda DASH II vector by plaque hybridization. Among approximately 8,000 plaques screened by plaque hybridization, three positive plaques were obtained. As shown in Fig. 1, restriction and Southern hybridization analyses suggested that all insert DNAs from three positive lambda clones contained a part of the same region of the total DNA of B. subtilis WU-S2B and that a 7.1-kb NheI fragment of positive clone no. 1 covered the entire region (bdsABC) encoding all the DBT-desulfurizing enzymes, as described later. The 7.1-kb NheI fragment was then cloned into XbaI site of pUC19 to give pU1Nh plasmid.

Fig. 1
figure 1

Physical maps of three positive lambda clones and the desulfurization genes from Bacillus subtilis WU-S2B. The construction of the deletion plasmids is described in “Materials and methods”. The reaction catalyzed by recombinant Escherichia coli cells harboring each deletion plasmid is indicated on the right side of plasmid map. The restriction sites from the vector are shown with an asterisk; sites removed by blunting and ligation are shown in parenthesie. B Bpu1102I, C CpoI, E EcoRI, H HindIII, K KpnI, M MluI, N NotI, Nh NheI, P PstI, Ph PshAI, S SacII. Plac and Ptac (open triangles) show the lac and tac promoter directions of pUC18 and pKK223-3, respectively

Full size image

Characterization of the nucleotide sequence of the bds genes

The nucleotide sequence hybridizable with the probe and the sequence adjacent to the hybridizable region of pU1Nh were determined. Three ORFs were identified to be aligned in the same direction. These ORFs show significant homology to the dsz operon of Rhodococcus sp. IGTS8 and the tds operon of Paenibacillus sp. A11-2. Each ORF homologous to dszA, dszB, and dszC of IGTS8 was therefore designated as bdsA, bdsB, and bdsC, respectively, for bacterial desulfurization. The overall nucleotide sequence of the bds operon, bdsABC, shows 61.0% homology to dszABC of Rhodococcus sp. IGTS8 and 57.8% to tdsABC of Paenibacillus sp. A11-2. In this operon structure of bdsABC, the end of bdsA has a 4-bp overlap with the translational initiation site of bdsB, similar to the dsz genes of Rhodococcus sp. IGTS8 and the tds genes of Paenibacillus sp. A11-2. Furthermore, bdsB also has a 4-bp overlap with bdsC, which is in contrast to the dsz and tds genes.

The first ORF, bdsA, encodes a protein of 453 amino acids, and the corresponding gene product, BdsA, shows significant homology with several proteins, as determined by BLAST and FASTA programs. BdsA has 78.8 and 64.4% homologies to DszA of Rhodococcus sp. IGTS8 and TdsA of Paenibacillus sp. A11-2, respectively. Low homologies were found between BdsA and several FMNH2-dependent monooxygenases, except for DszA and TdsA. For example, BdsA shows 41.5, 41.2, and 32.2% homologies to SnaA, pristinamycin IIA synthase subunit A of Strepromyces pristinaespiralis (Blanc et al. 1995), NtaA, nitrilotriacetate monooxygenase of Chelatobacter heintzii (Knobel et al. 1996), and EmoA, EDTA monooxygenase of EDTA-degrading bacterium BNC1 (Bohuslavek et al. 2001), respectively. Alignment of the amino acid sequences of BdsA and its equivalents, DszA and TdsA, from two other desulfurizing bacteria (Fig. 2) showed that BdsA has high homology in its N-terminal region, approximately 250 amino acids, and its C-terminal region, approximately 50 amino acids, to the equivalent proteins. When compared to the FMNH2-dependent monooxygenase family, BdsA also shows high homology in the regions conserved within desulfurizing enzymes. However, BdsA shows no similarity in its N-terminal region of approximately 100 amino acids to members of this family, such as NtaA, SnaA, and EmoA (data not shown). In this family, as shown in Fig. 3a, DBT-desulfurizing enzymes could be assigned to one cluster on the phylogenetic tree, but, as shown in Fig. 3b, bdsA is independently located from three clusters of the partial dszA genes classified by Duarte et al. (2001).

Fig. 2
figure 2

Multiple alignments of the amino acid sequence of desulfurizing enzymes from various desulfurizing bacteria. Amino acid residues identical in all the members are indicated by white letters in black boxes. Amino acid residues identical in more than half of the members are indicated by shaded boxes. Numbers indicate the positions of the residues in the complete amino acid sequence of each enzyme. a Comparison of BdsA, DBTO2 monooxygenase from B. subtilis WU-S2B, with its equivalents from other strains. b Comparison of BdsB, HBPSi desulfinase from B. subtilis WU-S2B, with its equivalents from other strains. c Comparison of BdsC, DBT monooxygenase from B. subtilis WU-S2B, with its equivalents from other strains. DszA, DszB, and DszC are desulfurization enzymes from Rhodococcus sp. IGTS8. TdsA, TdsB, and TdsC are desulfurization enzymes from Paenibacillus sp. A11-2

Full size image
Fig. 3
figure 3

a Unrooted phylogenetic tree of FMNH2-dependent monooxygenase family relating to BdsA and b classification of gene equivalents to dszA from several desulfurizing bacteria. Scale bars 0.1 substitution per site. a The positions of DBTO2 monooxygenases are boxed. BdsA DBTO2 monooxygenase of B. subtilis WU-S2B, DszA DBTO2 monooxygenase of Rhodococcus sp. IGTS8 (L37363), TdsA DBTO2 monooxygenase of Paenibacillus sp. A11-2 (AB033997), SnaA pristinamycin IIA synthase subunit A of Strepromyces pristinaespiralis (U21215), NtaA nitrilotriacetate monooxygenase of Chelatobacter heintzii (U399411), EmoA EDTA monooxygenase of EDTA-degrading bacterium BNC1 (AF176664). b The group of the Rhodococci dsz genes is boxed. bdsA Gene encoding DBTO2 monooxygenase of B. subtilis WU-S2B, tdsA gene encoding DBTO2 monooxygenase of Paenibacillus sp. A11-2 (AB033997), IGTS8 gene encoding DBTO2 monooxygenase of Rhodococcus sp. IGTS8 (L37363), A1/1 uncultured Rhodococcus sp. A1/1 DBT-desulfurization enzyme-like protein (dszA) gene (AF322033), A1/2 uncultured Rhodococcus sp. A1/2 DBT-desulfurization enzyme-like protein (dszA) gene (AF322034), A2/1 uncultured Rhodococcus sp. A2/1 DBT-desulfurization enzyme (dszA) gene (AF322036), A2/2 uncultured Rhodococcus sp. A2/2 DBT-desulfurization enzyme-like protein (dszA) gene (AF322035), A2/3 uncultured Rhodococcus sp. A2/3 DBT-desulfurization enzyme-like protein (dszA) gene (AF322037), A69 Rhodococcus sp. A69 DBT-desulfurization enzyme (dszA) gene (AF322041), JDVE-1 uncultured Rhodococcus sp. JDVE-1 DBT-desulfurization enzyme-like protein (dszA) gene (AF322038), JDVE-2 uncultured Rhodococcus sp. JDVE-2 DBT-desulfurization enzyme (dszA) gene (AF322039), JDVE-3 uncultured Rhodococcus sp. JDVE-3 DBT-desulfurization enzyme (dszA) gene (AF322040), M41 Rhodococcus sp. M41 DBT-desulfurization enzyme-like protein (dszA) gene (AF322042)

Full size image

The second ORF, bdsB, encodes a protein of 356 amino acids, and the deduced amino acid sequence shows 68.1 and 53.1% homology to those of DszB of Rhodococcus sp. IGTS8 and TdsB of Paenibacillus sp. A11-2, respectively. As shown in Fig. 2, while the N-terminal region of BdsB shows high homology to those of the equivalents, the amino acid sequence of this region has no homology to other enzymes, except to DszB and TdsB. BdsB also shows partial homology (approximately 20–30%) to SsuA (van der Ploeg et al. 1999), AsfC (Vermeij et al. 1999), and AtsR (Kahnert and Kertesz 2000), which are substrate-binding proteins responsible for sulfonate or sulfate ester utilization and are members of sulfate starvation-induced proteins.

The last ORF, bdsC, encodes a protein of 415 amino acids, and the deduced amino acid sequence of BdsC shows 72.9 and 51.2% homology to DszC of Rhodococcus sp. IGTS8 and TdsC of Paenibacillus sp. A11-2, respectively. BdsC also shows partially homology to acyl-CoA dehydrogenases of several microorganisms (P45857, P52042, and Q06319).

Heterologous expression of the DBT-desulfurization genes (bdsABC) in E. coli

Recombinant E. coli cells harboring pU1Nh, containing the entire region of bdsABC, degraded DBT to produce 2-HBP (data not shown), indicating that bdsABC, encoding the enzymes responsible for the ability of DBT-desulfurization, could be expressed in E. coli cells. Subsequently, the contribution of each ORF (bdsA, B and C) to DBT-desulfurization was characterized. As shown in Fig. 1, the recombinant E. coli cells harboring pKBDSA carrying only bdsA but lacking bdsB and bdsC could convert DBTO2 into HBPSi, but no degradation of DBT and HBPSi was observed. The recombinant E. coli cells harboring pKBDSB carrying only bdsB desulfurized HBPSi into 2-HBP, and the recombinant E. coli cells harboring pKBDSC carrying only bdsC oxidized DBT into DBTO2. These results clearly indicated that the gene products of bdsA, bdsB, and bdsC are responsible for the desulfurization of DBT to 2-HBP via DBTO2 and HBPSi.

Temperature effects on DBT desulfurization by resting cells of recombinant E. coli

The effects of temperature on DBT desulfurization were investigated using resting cells of recombinant E. coli harboring pKBDS. Since the amounts of DBT degraded and 2-HBP produced were especially increased between 16 and 20 h of cultivation of recombinant E. coli cells, the desulfurizing activities of the resting cells after 16, 20, and 24 h of cultivation were measured. The activity was highest in cells cultivated for 16 h (data not shown). The resting cells of recombinant E. coli showed 2-HBP-producing activity over a temperature range of 30–52°C, most efficiently at 40°C (0.08 nmol/min/mg dry cell weight) (Fig. 4). In addition, resting cells degraded DBT most efficiently at 50°C (0.16 nmol/min/mg dry cell weight). These thermal profiles of desulfurizing activity in recombinant E. coli cells agreed with those in the original strain B. subtilis WU-S2B (Kirimura et al. 2001). Resting cells of recombinant E. coli showed DBT-degrading and 2-HBP-producing activities approximately two- and three-fold, respectively, higher than those of B. subtilis WU-S2B at 40°C, and showed DBT-degrading activity approximately five times as high as B. subtilis WU-S2B at 50°C (Kirimura et al. 2001). For the resting cell reactions, the relationship between the amount of DBT degraded and that of 2-HBP produced does not seem to be stoichiometric, especially at temperatures above 40°C, as observed for B. subtilis WU-S2B (Kirimura et al. 2001).

Fig. 4
figure 4

Effects of temperature on DBT desulfurization by resting cells of recombinant E. coli harboring pKBDS. Resting cell reaction was carried out using n-tridecane solution containing DBT as substrate. DBT-degrading activities by recombinant E. coli (open circles) and B. subtilis WU-S2B (closed circles) are shown. 2-HBP-producing activities by recombinant E. coli (open triangles) and B. subtilis WU-S2B (closed triangles) are also indicated

Full size image

The thermal profile of BdsA was studied with resting recombinant E. coli cells harboring pKBDSA using DBTO2 as substrate, and those of BdsB and BdsC were studied with resting recombinant E. coli cells harboring pKBDS using HBPSi and DBT, respectively. Although the effects of temperature on each DBT-desulfurization step were measured in resting cells of recombinant E. coli JM109 harboring pKBDSA, pKBDSB, and pKBDSC, cells harboring pKBDSB and pKBDSC showed low activities of BdsB and BdsC, respectively. Therefore, these two activities were instead measured using recombinant E. coli cells harboring pKBDS. BdsA was measured in recombinant E. coli cells harboring pKBDSA since the activity was as high as in cells harboring pKBDS. As shown in Fig. 5, the optimal temperatures of BdsC, BdsA, and BdsB were about 50, 45, and 37°C, respectively. BdsB activity was highest at temperatures below 40°C. The relative activities of BdsC and BdsA were higher than that of BdsB at temperatures above 45°C, and the activity of BdsC was the highest of the three enzymes at temperatures above 50°C. Furthermore, DBTO2 was not detected as an intermediate in the reaction of resting cells over a temperature range of 30–52°C (data not shown). HBPSi was not quantitatively extracted with ethylacetate due to its water solubility, and was detected in low amounts in the reaction mixture at 50°C. However, the amount of HBPSi could be calculated with the equation HBPSi=initial DBT−residual DBT−DBTO2−2-HBP, and was 0.050 nmol in the reaction mixture at 50°C. Based on the amount of DBT degraded, 0.055 nmol in the reaction mixture, at 50°C the rate of HBPSi was calculated to be 91%. These results suggest that BdsC and/or BdsA are the rate-determining enzymes in DBT desulfurization at temperatures below 40°C, and that BdsB is the rate-determining enzyme at temperatures above 45°C. These tendencies agreed with the relationship between DBT-degrading activity and 2-HBP-producing activity (Fig. 4).

Fig. 5
figure 5

Effects of temperature on each step of DBT desulfurization by resting cells of recombinant E. coli. The reaction was carried out using an ethanol solution of DBT or one of its metabolic intermediates as substrate. BdsB activity was determined by the amount of 2-HBP-production, since HBPSi as a substrate could not be quantitatively extracted with ethylacetate due to its water solubility. DBT-degrading activity (circles), DBTO2-degrading activity (squares), and 2-HBP-producing activity (triangles) are shown

Full size image

Substrate specificity of desulfurization by the recombinant E. coli cells

Alkylated DBT derivatives detected in light gas oil are more recalcitrant to conventional hydrodesulfurization than DBT (Suzuki 1999), while B. subtilis WU-S2B can desulfurize various DBT derivatives (Kirimura et al. 2001). As shown in Table 1, 2,8-dimethylDBT, 4,6-dimethylDBT, and 3,4-benzoDBT could be degraded by growing cells of recombinant E. coli harboring pKBDS in modified M9 medium, and each desulfurized metabolite corresponding to 2-HBP derived from DBT was detected by GC analysis (data not shown). When cells were cultivated using LB medium instead of modified M9 medium, the recombinant E. coli cells grew well in LB medium supplemented with each substrate. However, the degradation rates for 4,6-dimethylDBT and 3,4-benzoDBT were very low (data not shown). By contrast, recombinant E. coli cells were unable to grow in modified M9 medium without supplementation of DBT derivatives as the sole source of sulfur (data not shown). The low degradation rates of 4,6-dimethylDBT and 3,4-benzoDBT by the recombinant E. coli cells might be related to a difference in the uptake and/or utilization of DBT derivatives.

Table 1 Degradation of DBT derivatives by growing cells of recombinant E. coli harboring pKBDS
Full size table

Cloning of the gene corresponding to B. subtilis bdsABC from M. phlei WU-F1

Using a strategy similar to that used for cloning of bdsABC from B. subtilis WU-S2B, the corresponding gene responsible for DBT desulfurization was also cloned from the total DNA library of M. phlei WU-F1 (details not shown). The nucleotide sequences of bdsABC and the 3 kb-upstream region in M. phlei WU-F1 were found to be completely identical to the genes of B. subtilis WU-S2B, although the two strains are genetically different.

Discussion

In the present study, bdsABC, the DBT-desulfurization genes (bds genes) from B. subtilis WU-S2B and the corresponding genes from M. phlei WU-F1 were cloned based on the conserved amino acid sequences between mesophilic dsz genes and thermophilic tds genes. Based on the nucleotide sequence and heterologous expression, we confirmed that the operon structure of the bds genes is quite similar to those of DBT-desulfurization genes, such as the dsz genes and the tds genes. Since resting cells of recombinant E. coli cells could desulfurize DBT over temperatures ranging from 30 to 52°C as well as B. subtilis WU-S2B, the thermal profiles of bdsABC products were functionally reproducible in E. coli as a host. Furthermore, the recombinant E. coli cells harboring pKBDS desulfurized 2,8-dimethylDBT, 4,6-dimethylDBT, and 3,4-benzoDBT as well as the original strain. These results suggested that the thermal profile and substrate specificity of desulfurizing activity, i.e., the desulfurization ability, observed for B. subtilis WU-S2B was endowed by the gene products of bdsABC but not by other cellular properties of B. subtilis WU-S2B.

BdsA shows homologies to equivalents of the other known desulfurization genes and to several members of FMNH2-dependent monooxygenases. It should be noted that in common these enzymes require flavin reductase, similar to BdsA, and catalyze α-hydroxylation of the carbon atom next to the sulfur or nitrogen atom, although the structures of their substrates are very different (Blanc et al. 1995; Bohuslavek et al. 2001; Knobel et al. 1996; Oldfield et al. 1997). On the phylogenetic tree of the FMNH2-dependent monooxygenase family, BdsA of B. subtilis WU-S2B, DszA of Rhodococcus sp. IGTS8, and TdsA of Paenibacillus sp. A11-2 form independent cluster groups (Fig. 3a). bdsA can be distinctly classified from many Rhodococci dszA genes, which were grouped into three clusters by Duarte et al. (2001), on the phylogenetic tree based on the partial nucleotide sequences of the equivalents for dszA (Fig. 3b). Since the location on the phylogenetic tree of bdsA relative to genes equivalent to dszA (Fig. 3b) does not correspond to the location of B. subtilis WU-S2B relative to other bacteria based on 16S rRNA sequences (data not shown), the desulfurization genes such as dsz and bds are considered to have been transferred horizontally via plasmids, as discussed by Denis-Larose et al. (1997). In Fig. 3b, it is interesting to note that bdsA of B. subtilis WU-S2B is remotely located from tdsA of Paenibacillus sp. A11-2, which is taxonomically related to B. subtilis. Nevertheless, the bds genes of B. subtilis WU-S2B and the tds genes of Paenibacillus sp. A11-2 would have been independently transmitted, but not derived from identical DBT-desulfurization genes. Furthermore, the bdsABC genes of B. subtilis WU-S2B and M. phlei WU-F1, bacteria that have no taxonomic relation, were identical to each other. These findings strongly suggest that B. subtilis WU-S2B and M. phlei WU-F1 might have acquired bdsABC genes not by vertical evolution but by horizontal transmission.

The DBT-desulfurizing enzymes encoded by the tds genes of Paenibacillus sp. A11-2 showed higher activity toward benzothiophene (BT) than toward DBT. This result was supported by the observation that strain A11-2 grew more rapidly on BT than on DBT (Konishi et al. 2002). By contrast, DszC purified from R. erythropolis KA2-5-1 showed highest activity toward various alkyalted DBT derivatives and 3-methylBT among alkylated BTs. The substrate specificity of desulfurization correlated well with that of the sulfur source for growth of strain KA2-5-1 (Konishi et al. 2002). The substrate specificity of the sulfur source for growth of B. subtilis WU-S2B more closely resembled that of R. erythropolis KA2-5-1, producing Dsz enzymes, than that of Paenibacillus sp. A11-2 (Kirimura et al. 2001). Therefore, it is likely that the amino acid sequences of Bds enzymes of B. subtilis WU-S2B are similar to those of Dsz enzymes of R. erythropolis KA2-5-1 but not to those of Tds enzymes of Paenibacillus sp. A11-2. Since the sulfur components in HDS-treated light gas oils are mainly alkylated DBT derivatives and subsequently BT derivatives, the substrate specificity of Bds enzymes of B. subtilis WU-S2B is superior to that of Paenibacillus sp. A11-2 enzymes. Furthermore, Bds enzymes of resting cells of B. subtilis WU-S2B showed DBT-desulfurizing activities over a temperature range of 30–52°C (Kirimura et al. 2001), a property that is superior to those of R. erythropolis KA2-5-1 and Paenibacillus sp. A11-2 enzymes. Although the correlation between the amino acid sequences of Bds enzymes and their respective thermal profiles remains unclear, the consensus sequences between Bds enzymes and Tds enzymes might be essential for thermal stability of DBT desulfurization. This topic is currently the subject of further studies. Using site-directed mutagenesis of Rhodococcus sp. IGTS8 genes, it was found that the residues Gln-345 of DszA and Val-261 of DszC are related to recognition of C–S bond cleavage specificity and utilization of 5-methylBT, respectively (Arensdorf et al. 2002; Konishi and Maruhashi 2003). Although BdsC conserved the corresponding residue (Val-259 to Val-261 of DszC), M. phlei WU-F1 and B. subtilis WU-S2B could desulfurize 2-ethylNTH, which could not be desulfurized by R. erythropolis KA2-5-1 harboring the DszC enzyme. Therefore, it appears that the bdsABC genes are useful genetic sources for improvement of substrate specificity and wide-range thermal reactivities of desulfurization enzymes.

In the heterologous expression analysis, recombinant E. coli cells harboring bdsC oxidized DBT to DBTO2, those harboring bdsA converted DBTO2 into HBPSi, and those harboring bdsB desulfurized HBPSi to 2-HBP. This pathway for DBT desulfurization is in accordance with the sulfur-specific pathway of DBT desulfurization reported previously (Kirimura et al. 2001) based on the analyses of metabolites. Thus, the bds genes could be individually expressed with the tac promoter of the vector and each gene product could function fully and independently. The monooxygenases of Dsz enzymes of Rhodococci and Tds enzymes of Paenibacillus sp. A11-2 require flavin reductase as a coupling enzyme for desulfurizing activity (Konishi et al. 2000; Ohshiro et al. 1994). The recombinant E. coli cells harboring the bdsABC genes could degrade DBT to 2-HBP without introduction of flavin reductase genes, indicating that BdsA and BdsC expressed in E. coli couple with E. coli oxidoreductases, such as HpaC (Galán et al. 2000a), which were reported to be able to enhance the DBT-desulfurizing activity of DszA and DszC of Rhodococcus sp. IGTS8 (Galán et al. 2000b).

Although the DBT-degrading activity of recombinant E. coli cells was highest at 50°C, the DBT-desulfurizing activity (2-HBP-producing activity) was highest at 40°C, as shown in Fig. 4. As shown in Fig. 5, this is probably due to the heat liability of BdsB, which catalyzed the reaction followed by the conversion DBTO2 to HBPSi. Thus, BdsB might be the rate-determining enzyme at temperatures above 45°C. However, since HBPSi is easily dissolved in water, DBT desulfurization from the oil phase can be achieved. Moreover, it was reported that HBPSi as a byproduct of desulfurization was converted to valuable surfactants and utilized further (Mcfarland 1999). Therefore, the accumulation of HBPSi would not become an issue in the modification of process design.

The genes responsible for DBT desulfurization in M. phlei WU-F1 were also cloned and and found to be completely identical to bdsABC from B. subtilis WU-S2B. Moreover, the nucleotide sequence for the 3-kb upstream region, containing the putative promoter sequence, of bdsABC was identical for the genes of B. subtilis WU-S2B and M. phlei WU-F1, suggesting that the same promoter is functioned in the two strains (details not shown). Since the DBT desulfurization profiles of M. phlei WU-F1, especially the thermal properties, were very similar to those of B. subtilis WU-S2B (Kirimura et al. 2001; Furuya et al. 2001), the fact that B. subtilis WU-S2B and M. phlei WU-F1 have the same DBT-desulfurization genes is perhaps unusual but not irrational. Although the DBT-desulfurizing activity of resting cells of B. subtilis WU-S2B was lower than that of resting cells of M. phlei WU-F1, this difference might be due to other cellular factors, such as flavin reductases coupling with BdsA and BdsC. This possibility is the subject of current studies.

The desulfurizing activity of the recombinant E. coli cells used in this study was 0.08 nmol/min/mg dry cell weight, and was similar to the level reported for other E. coli recombinant strains harboring DBT-desulfurization genes (Ishii et al. 2000a). For example, the desulfurizing activities of recombinant E. coli cells without coexpression of flavin reductase was 0.025 nmol/min/mg dry cell weight (Ishii et al. 2000a). The target level of desulfurizing activity needed for practical industrial applications has been evaluated and was determined to be 20–50 nmol/min/mg dry cell weight (Monticello 2000; Maruhashi 2001). Therefore, at this stage, the desulfurizing activity achieved in this study is not sufficient for use in practical biodesulfurization. However, having determined the properties of thermally stable BdsABC enzymes, improvement of desulfurizing activity by genetic engineering, using the bdsABC genes and a desirable host strain such as M. phlei WU-F1, should be possible .

In conclusion, we identified and functionally analyzed the genes responsible for DBT desulfurization in B. subtilis WU-S2B and M. phlei WU-F1. It was confirmed that the desulfurizing activity of B. subtilis WU-S2B and M. phlei WU-F1 at temperatures of 30–52°C is basically dependent on the function of products of the bds genes, which are both unique and distinct from other DBT-desulfurization genes, such as dsz and tds. Genetic improvement of desulfurizing bacteria will allow their practical application to novel biodesulfurization processes.