Characterization of the Upper Pathway Genes for Fluorene Metabolism in Terrabacter sp. Strain DBF63

Abstract

Genes involved in the degradation of fluorene to phthalate were characterized in the fluorene degrader Terrabacter sp. strain DBF63. The initial attack on both fluorene and 9-fluorenone was catalyzed by DbfA to yield 9-fluorenol and 1,1a-dihydroxy-1-hydro-9-fluorenone, respectively. The FlnB protein exhibited activities against both 9-fluorenol and 1,1a-dihydroxy-1-hydro-9-fluorenone to produce 9-fluorenone and 2′-carboxy-2,3-dihydroxybiphenyl, respectively. FlnD is a heteromeric protein encoded by flnD1 and ORF16, being a member of the class III two-subunit extradiol dioxygenase. FlnE was identified as a serine hydrolase for the meta-cleavage products that yield phthalate.

Polycyclic aromatic hydrocarbons (PAHs) have toxic, mutagenic, and carcinogenic properties, and therefore, their long persistence in the environment is of major concern. Among the PAHs, fluorene (FN) has been classified as one of 16 priority pollutants by the U.S. Environmental Protection Agency because of its toxicity to organisms and abundance in the environment (18). For the purpose of bioremediation, the bacterial degradation of FN has been studied for many years (2, 3, 8, 9, 10, 11, 19, 26, 27), and three major degradative pathways have been proposed (Fig. 1). Two of these pathways are initiated by a dioxygenation at the 1,2 (Fig. 1A) or 3,4 (Fig. 1B) position. The third route is initiated by monooxygenation at the C-9 position to give 9-fluorenol (VI), which is then dehydrogenated to 9-fluorenone (VII). Angular dioxygenation of 9-fluorenone then leads to 1,1a-dihydroxy-1-hydro-9-fluorenone (DHF, VIII) (6, 7, 23). Dehydrogenation of DHF produces a 2′-carboxy-2,3-dihydroxybiphenyl (CDB, IX; this metabolite is spontaneously transformed to 8-hydroxy-3,4-benzocoumarin [HBC, X]), which is metabolized by the reactions analogous to those of biphenyl degradation, leading to the formation of phthalate (XI) (2, 9, 26). However, although many FN-utilizing bacteria have been isolated and characterized, very little is known about the specific enzymes involved in the catabolism of FN and especially the genes coding for these enzymes.

FIG. 1.

Proposed pathways of fluorene catabolism by aerobic bacteria. Metabolites I to XI were reported previously (2, 9). Compound IX was indirectly identified because of its spontaneous transformation to compound X (26). The chemical structures of the other metabolites shown have not been characterized yet. Compound designations: I, 2-indanone; II, 3-isochromanone; III, 1-indanone; IV, 3,4-dihydrocoumarin; V, salicylic acid; VI, 9-fluorenol; VII, 9-fluorenone; VIII, DHF; IX, CDB; X, HBC; XI, phthalic acid; XII, 1,2-dihydro-1,2-dihydroxy-9-fluorenone. The corresponding enzymes in Terrabacter sp. strain DBF63 are as follows: DbfA, 9-fluorenone 1,1a-dioxygenase (fluorene 9-monooxygenase, DF 4,4a-dioxygenase); FlnB, DHF dehydrogenase; FlnD, CDB 1,2-dioxygenase; FlnE, 2-hydroxy-6-oxo-6-(2′-carboxyphenyl)-hexa-2,4-dienoate hydrolase.

Terrabacter sp. strain DBF63 was originally isolated from a soil sample as a bacterium capable of utilizing dibenzofuran (DF) and FN as the sole source of carbon and energy (17, 19). Recently, novel terminal oxygenase genes of angular dioxygenase (dbfA1 and dbfA2), whose products can catalyze the angular dioxygenation of DF with the complementation of the nonspecific electron transport system of Escherichia coli, were isolated by a PCR-based strategy (16). The facts that pht genes for phthalate to protocatechuate cluster together with the dbfA1A2 genes (12) and that both the dbfA1A2 and pht genes were expressed in FN-grown DBF63 cells (20) suggested that the whole island could be involved in FN metabolism. The present report is the first on the characterization of the upper metabolic pathway genes for FN.

Nucleotide sequencing of the downstream region of dbfA1A2 genes.

We determined the nucleotide sequences of the 6,525-bp BamHI region (Fig. 2, shaded box) as described previously (12, 16), and seven ORFs (open reading frames; ORF3, ORF15, ORF16, ORF17, ORF18, ORF19, and ORF20) were found in that DNA region (Fig. 2). The deduced amino acid sequences of the respective ORFs were compared with those registered in the databases (Table 1).

FIG. 2.

Physical map of the 14.7-kb DNA region containing fluorene degradation genes, flnRB-dbfA1A2-flnED1-ORF16, in Terrabacter sp. strain DBF63. The pentagons in the physical map indicate the sizes, locations, and directions of transcription of the ORFs. The shaded box represents the region whose nucleotide sequences were determined in this study. The solid lines indicate the DNA region inserted in one vector or two compatible vectors under control of the lac promoter (pUC119 [Ap^r] for pDF27; pBluescript II KS(−) [Ap^r] for pDF104 and pDF109; pSTV29 [Cm^r] for pDFS110 and pDFS112). Restriction site designations: A, ApaI; B, BamHI; Bg, BglII; N, NotI; S, SacI; Sp, SphI; P, PstI.

TABLE 1.

Homology search analyses of the fln genes and the neighboring region

Genea	Probable function or product	Mol wt	Homologous protein	Source	Identity (%)b	Accession numberc
flnRd (ORF5)	Regulatory protein	32,437	TfdT	Burkholderia sp. strain NK8	38	AB050198
			TfdT	Ralstonia eutropha JMP134	32	U16782
			GltC	Listeria innocua	32	NC_003212
flnB (ORF4)	DHF dehydrogenase	38,510	PhbB	Pseudomonas sp. strain 61-3	33	AB014757
			CymB	Pseudomonas putida F1	32	U24215
			Oxydoreductasee	Bacillus halodurans	31	NC_002570
flnE (ORF3)f	meta-cleavage product hydrolase	35,518	ORF4 product	Rhodococcus sp. strain YK2	97	AB070456
			OhpC	Rhodococcus sp. strain V49	38	AF274045
			CmtE	Pseudomonas putida F1	36	U24215
flnD1 (ORF15)	Extradiol dioxygenase large subunit	31,563	Edi4	Rhodococcus sp. strain YK2	94	AB070456
			CarBb	Sphingomonas sp. strain GTIN11	37	AF442494
			CarBb	Pseudomonas resinovorans CA10	36	AF061751
ORF16g		20,451	ORF6 producth	Rhodococcus sp. strain YK2	95	AB070456
			ORF7 producti	Rhodococcus sp. strain YK2	66	AB070456
			PimFi	Streptomyces natalensis	40	AJ278573
flnC (ORF17)	Short-chain dehydrogenase-reductase	25,932	Dehydrogenasee	Pseudomonas aeruginosa PAO1	53	NC_002516
			ChnA	Acinetobacter sp. strain SE19	49	AF282240
			MlhAe	Rhodococcus erythropolis DCL14	48	AJ292535
ORF18	Unknown	24,665
ORF19j	Transposase	13,561	Protein Ae	Corynebacterium glutamicum	50	X69104
			Transposasee	Rhodococcus sp. strain 124	47	AF452376
			Transposasee	Corynebacterium glutamicum	46	NC_003450
ORF20j	Transposase	33,719	Transposasee	Corynebacterium glutamicum	59	NC_003450
			Protein Be	Corynebacterium glutamicum	58	X69104
			Transposasee	Micrococcus sp. strain 28	54	AY034092

^aAccession number AB095015.

^bHomology searching was performed with the SwissProt and GenBank protein databases or the DDBJ, EMBL, and GenBank nucleotide sequence databases with the BLAST program (version 2.0.10) (1). Multiple sequence alignment was carried out with Clustal W version 1.6 (25).

^cAccession number in the SwissProt and DDBJ, EMBL, and GenBank databases.

^dThe ORF is encoded in the complementary strand.

^eHypothetical protein.

^fThe C-terminal region of the ORF3 product (FlnE) was longer than that of any other meta-cleavage product hydrolase by approximately 30 amino acids.

^gThe ORF16 product is the modular protein of the extradiol dioxygenase small subunit and a putative ferredoxin (see text).

^hThe ORF product was compared with only the 90-amino-acid-long N-terminal portion of the ORF16 product.

ⁱThe ORF product was compared with only the 100-amino-acid-long C-terminal protein of the ORF16 product. This region was revealed to be longer than other related [3Fe-4S] types of ferredoxin by approximately 30 amino acids.

^jORF19, ORF20, and their flanking regions had the general features of IS3-related elements with 28-bp, imperfectly matched, terminal inverted repeats. We designated it ISTesp3.

Experiments in which FN was biotransformed to phthalate by a recombinant E. coli clone.

pDF104 harboring ORF4-dbfA1A2-ORF3-ORF15-ORF16 and pDFS110 harboring ORF16-ORF17 under the control of the lac promoter were constructed (Fig. 2; detailed information on the plasmid construction in this study is available on request), both plasmids were introduced into the same E. coli JM109 cells, and an FN biotransformation experiment was performed with the recombinant E. coli cells. Biotransformation experiments with E. coli clones and gas chromatography-mass spectrometry (GC-MS) analysis of the extracts from reaction mixtures were performed as described previously (12, 16). GC-MS analysis of the methylated extract revealed the presence of a product with a molecular ion M⁺ at m/z 194 [6] and fragmentation ions at m/z 163 [100], m/z 135 [5], m/z 133 [7], m/z 104 [6], m/z 92 [11], m/z 77 [38], and m/z 76 [24] (the retention time [RT] was 6.0 min; relative intensities expressed as percentages are given in brackets) (Fig. 2). The mass spectral properties were identical to those of authentic dimethylated phthalate.

On the other hand, we also constructed pDF109, in which a 3.8-kb SacI-NotI fragment harboring ORF3 (flnE) and ORF15 (flnD1) was deleted from pDF104 (Fig. 2). We then performed an experiment to biotransform FN by using E. coli cells harboring both pDF109 and pDFS110. GC-MS analysis of the methylated extract revealed the presence of another product with a molecular ion M⁺ at m/z 226 [100] and fragmentation ions at m/z 211 [2], m/z 197 [4], m/z 195 [2], m/z 183 [44], m/z 168 [3], m/z 155 [17], m/z 139 [9], m/z 127 [34], m/z 113 [3], m/z 101 [9], m/z 77 [18], and m/z 63 [18] (the RT was 11.5 min), consistent with the notion that these are the methyl derivatives of HBC previously reported (9) (Fig. 2). We did not detect phthalate in this reaction mixture. Considering that CDB is spontaneously transformed to HBC, these results indicate that the genes located downstream from ORF5 are the FN metabolic genes in strain DBF63 and that ORF3 (flnE) and ORF15 (flnD1) are involved in the degradation of CDB to phthalate.

9-Fluorenol dehydrogenase.

We previously demonstrated that DbfA1A2 could catalyze the monooxygenation of FN to produce 9-fluorenol with ferredoxin and reductase from E. coli (16). In this study, since two genes encoding putative short-chain dehydrogenase-reductase (SDR) family proteins, ORF4 (designated flnB) and ORF17 (designated flnC), were found within the FN degradative gene cluster (Fig. 2), we determined the 9-fluorenol-dependent reduction of NAD⁺ to assess the 9-fluorenol dehydrogenase activities of these enzymes. The 1,074- and 759-bp DNA fragments containing ORF4 (flnB) and ORF17 (flnC), respectively, were amplified by PCR with the forward primers containing an efficient Shine-Dalgarno (SD) sequence for E. coli (24) (detailed information on the primer sequences in this study is available on request). The PCR products were cloned with the pT7Blue (R) vector (Novagen), and the nucleotide sequences of PCR products were confirmed by sequencing. The clones were cut at both the HindIII and EcoRI sites (derived from the primer), and then the fragment was cloned between the HindIII and EcoRI sites of pBluescriptII KS(−) to give pDF106 (containing flnB) and pDF108 (containing flnC). Preparation of E. coli cell extracts was performed as described previously (21), and the 9-fluorenol dehydrogenase activities were measured as described by Trenz et al. (26). As a control, E. coli JM109 harboring pBluescriptII KS(B) was used. Compared to the NAD⁺ reduction by the cell extracts of E. coli JM109 harboring pBluescriptII KS(B) (4.22 ± 0.912 U mg of protein⁻¹; the nonspecific dehydrogenase derived from E. coli acted on 9-fluorenol to produce 9-fluorenone), the cell extracts of E. coli harboring pDF106 (flnB) showed threefold higher activity with 9-fluorenol (12.8 ± 2.00 U mg of protein⁻¹). By contrast, the cell extracts of E. coli harboring pDF108 (flnC) showed the same level of 9-fluorenol dehydrogenase activity (4.64 ± 0.402 U mg of protein⁻¹) as that harboring pBluescriptII KS(−). This tendency was observed in several independent experiments. These results indicated that FlnB catalyzed the dehydrogenation of 9-fluorenol to produce 9-fluorenone.

Angular dioxygenase for 9-fluorenone.

We previously reported that the angular dioxygenase component, DbfA1A2, converted 9-fluorenone to a cis-diol compound with the ferredoxin and reductase component from E. coli (16). However, the cis-diol compound has not been determined to be DHF (Fig. 1, compound VIII) or 1,2-dihydro-1,2-dihydroxy-9-fluorenone (XII) (16, 19) because the E. coli clone harboring the dbfA1A2 genes did not yield enough of the compound for nuclear magnetic resonance (NMR) analysis. When the resting cells of E. coli harboring pDF32 (dbfA1A2) (100-ml culture) were incubated with 9-fluorenone (0.1%, wt/vol), the accumulation of a small amount of the cis-diol compound was observed by reverse-phase high-pressure liquid chromatography (HPLC) at an RT of 6.2 min (the HPLC conditions are described below). By contrast, the resting cell reaction with E. coli harboring both pDF32 (dbfA1A2) and pDFS112 (ORF16), in which the 1.48-kb SmaI fragment containing flnC was deleted from pDF110 (Fig. 2), gave a peak area of the cis-diol compound four to five times that obtained with E. coli harboring only pDF32 (dbfA1A2) (data not shown).

With E. coli containing both pDF32 and pDFS112, we tried to isolate the cis-diol compound. The recombinant E. coli cells were harvested by centrifugation of 8 liters of a 12-h culture. Collected cells were washed twice with 50 mM sodium-potassium phosphate buffer (pH 7.0) and resuspended in 2 liters of the same buffer. The washed cell suspension with 9-fluorenone (1 g/liter) was incubated for 18 h on a rotary shaker at 30°C (120 rpm). After incubation, the cells and the remaining 9-fluorenone were removed by centrifugation. First, rough purification of the compound by silica gel column chromatography was performed as described by Selifonov et al. (23), with some modifications. The residue was redissolved in an appropriate volume of methanol and further purified by HPLC. HPLC analysis was performed with a Waters 600E chromatograph (Waters Co., Milford, Mass.) equipped with an octadecyl silane 4253-D column (10 by 250 mm; Senshu Scientific Co., Ltd., Tokyo, Japan). The mobile phase was water-methanol (40:60, vol/vol), and the flow rate was 3 ml min⁻¹. The elutions were monitored at 254 and 275 nm by a Waters 996 photodiode array detector and collected every minute. The solvent of fractions collected at an RT of 6 to 7 min was evaporated under reduced pressure at 25°C. Finally, 3.1 mg of the yellowish solid was obtained from 8 liters of the E. coli culture. As this compound was chromatographically pure (thin-layer chromatography, HPLC) and showed the presence of a major UV absorption maximum (314 nm [in methanol]), the compound was subjected to NMR analysis. For ¹H-NMR analysis, the purified DHF sample was dissolved in an appropriate volume of CDCl₃. The ¹H-NMR spectrum of DHF was recorded with a JNM-A500 spectrometer (JEOL, Ltd., Tokyo, Japan) operating at 500 MHz with tetramethyl silane as the internal standard. A NAORAC H5X/FG probe was used. The 500-MHz ¹H-NMR spectrum of the compound showed the proton signals in the aromatic, olefinic, sp.³-CH(OH) regions to be as follows (parts per million): 4.708 (ddd, H-1), 5.914 (dd, H-2), 6.189 (ddd, H-3), 6.570 (d, H-4), 7.433 (ddd, H-7), 7.673 (dd, H-6), 7.707 (dd, H-5), and 7.813 (dd, H-8). These ¹H-NMR proton signals were identical to those of DHF previously published by Selifonov et al. (23). GC-MS analysis of the TMS derivative of the compound revealed a molecular ion M⁺ at m/z 358 [21] and fragmentation ions at m/z 343 [4], m/z 270 [11], m/z 253 [30], m/z 223 [2], m/z 180 [2], m/z 147 [34], and m/z 73 [100] (the RT was 10.5 min). This mass spectral pattern was almost identical to that of the cis-diol compound previously reported by our group (16). From these results, we concluded that the cis-diol compound previously reported is DHF and that DbfA1A2 can catalyze the angular dioxygenation of 9-fluorenone.

DHF dehydrogenase.

Previously, Trenz et al. (26) indicated that the NAD⁺-dependent dehydrogenase with four identical 40-kDa subunits catalyzed the conversion of DHF to CDB; however, the gene sequence of this enzyme has not been reported. In order to elucidate whether flnB or flnC takes part in the conversion of DHF to CDB, we performed experiments in which purified DHF (approximately 0.7 mM final concentration) was biotransformed by E. coli JM109 cells harboring pBluescriptII KS(−) (control), pDF106 (flnB), or pDF108 (flnC), respectively. By GC-MS analysis, we identified HBC as a metabolite in E. coli cells harboring flnB. On the other hand, biotransformation experiments with E. coli JM109 cells harboring pBluescriptII KS(−) or pDF108 (flnC) did not produce HBC. As HBC was usually isolated as a lactone form of CDB (9, 26), this result indicates that FlnB is a DHF dehydrogenase. The deduced amino acid sequences of FlnB showed homology not to cis-dihydrodiol dehydrogenases involved in aromatic compound degradation but to the group of SDR family proteins (Table 1). Involvement of SDR-type enzymes in the metabolic steps, instead of the ordinary cis-dihydrodiol dehydrogenase, may be one of the features of bacterial 9-fluorenone degradation via angular dioxygenation. However, the C-terminal region of FlnB was approximately 70 amino acids longer than those of other, related, SDR-like enzymes (data not shown). By contrast, the role of another SDR family protein, FlnC, is still unknown.

Novel two-subunit extradiol dioxygenase.

The 932-, 1,495-, and 1,201-bp DNA fragments containing ORF15 (flnD1), both ORF15 and ORF16, and both ORF15 and the 5′-terminal 273 bp of ORF16, respectively, were amplified by PCR with the forward primers containing an efficient SD sequence for E. coli (24). The DNA fragments were cut at both the HindIII and EcoRI sites (derived from the primer) and then cloned between the HindIII and EcoRI sites of pUC119 to give pDF501 (containing flnD1), pDF502 (containing flnD1 and ORF16) and pDF503 (containing flnD1 and the 5′-terminal 273-bp portion of ORF16) (Fig. 3). First, we constructed only pDF501 and performed preliminary biotransformation experiments. However, E. coli JM109 harboring pDF501 could not transform 2,3-dihydroxybiphenyl (2,3-DHB) to 2-hydroxy-6-oxo-6-phenylhexa-2,4-dienoic acid (6-phenyl-HODA), and the reaction mixture did not show a yellow color (data not shown). As several enzymes belonging to class III of the extradiol dioxygenases are known to consist of two proteins (large and small subunits) (5, 15), we examined whether ORF15 is one of the subunits of extradiol dioxygenase. When an experiment in which 2,3-DHB was biotransformed was performed with pDF502 containing ORF15-ORF16, the reaction mixture exhibited a yellow color derived from 6-phenyl-HODA, indicating that the ORF16 product is necessary for the activity. However, we did not understand how the active extradiol dioxygenase is constituted, because ORF16 has a modular organization of an unknown ORF and a ferredoxin-encoding gene.

FIG. 3.

Comparison of the DNA region containing ORF15-ORF16 from Terrabacter sp. strain DBF63 with that containing edi4-ORF6-ORF7 from Rhodococcus sp. strain YK2 (14). The genetic organization of strain YK2 was drawn on the basis of the sequence data from the DDBJ, EMBL, and GenBank nucleotide sequence databases under accession number AB070456. Pentagons represent the locations, lengths, and directions of transcription of ORFs. The solid lines indicate the DNA region inserted into the pUC119 vector under control of the lac promoter. The nucleotide sequences of the border region between ORF6 and ORF7 from strain YK2 and of the corresponding region from strain DBF63 are shown. The insert DNAs of pDF501 to pDF503 were amplified by PCR (detailed information on the primer sequences in this study is available on request).

Detailed comparison of the DNA region containing ORF15 and ORF16 of strain DBF63 with that containing edi4, ORF6, and ORF7 of another DF degrader, Rhodococcus sp. strain YK2 (accession no. AB070456) (14; Fig. 3), revealed that although the sequence identity was more than 92%, they had different genetic organizations of the ORF16 portion (Fig. 3). The insertion of a single C base, at position 266 from the first base of the ORF16 start codon, caused the frame shift, and ORF16 of strain DBF63 was predicted to be formed as a fusion of two genes highly homologous to ORF6 and ORF7 from strain YK2 (Fig. 3). Hence, we hypothesized that the unknown N-terminal 90-amino-acid sequence of the ORF16 product is involved in the enzyme activity. We constructed pDF503 (Fig. 3) by PCR with the primer sets for amplifying both ORF15 and the 5′-terminal 273-bp region of ORF16, in whose reverse primer designation the single C base at position 266 is deleted and as a result there is a stop codon (Fig. 3). By a biotransformation experiment with E. coli harboring pDF503, we also confirmed the extradiol dioxygenase activity for 2,3-DHB by the yellow color of 6-phenyl-HODA.

The extradiol dioxygenase activities for 2,3-DHB and 3-methylcatechol in the cell extracts of E. coli carrying flnD were assayed as described by Iwata et al. (15). Cell extracts from E. coli harboring pDF502 and pDF503 showed the extradiol dioxygenase activity for 2,3-DHB, and their enzymatic activities for 2,3-DHB were 208 ± 7.81 and 365 ± 4.36 U μg of protein⁻¹, respectively. Compared to those activities, almost negligible activities were observed with both extracts from E. coli harboring pUC119 (5.76 ± 1.53 U μg of protein⁻¹) and pDF501 (2.67 ± 1.15 U μg of protein⁻¹). In addition, the activities of these extracts for 3-methylcatechol were the same as those of the control (E. coli harboring pUC119, 4.33 ± 1.53 U μg of protein⁻¹, E. coli harboring pDF501, 4.67 ± 2.89 U μg of protein⁻¹, E. coli harboring pDF502, 2.57 ± 0.577 U μg of protein⁻¹, E. coli harboring pDF503, 1.31 ± 0.608 U μg of protein⁻¹). These results demonstrated that the ORF15 product and the N-terminal portion of the modular protein (ORF16 product) consist of a novel two-subunit extradiol dioxygenase. However, the enzymatic activity of the extract from E. coli harboring pDF502 was about 57% of that of the extract from E. coli harboring pDF503, probably owing to the unusual three-dimensional conformation of the ORF16 product. We also tried several methods to prepare CDB as a substrate for FlnD, e.g., accumulation of CDB by resting cells of FN-grown strain DBF63 with the extradiol dioxygenase inhibitor 3-chlorocatechol, but until now, all failed (data not shown). However, FlnD can catalyze the meta cleavage of CDB as described in biotransformation experiments. We designated ORF15 flnD1, encoding a large subunit of a two-subunit class III extradiol dioxygenase.

Since we were interested in the modular nature of ORF16, we sought to investigate whether this allele exists in other strains belonging to the genus Terrabacter. PCR experiments were performed with total DNAs of another two DF-degrading Terrabacter strains, DFA1 and DFA10, isolated from activated sludge (our unpublished data; the 16S rRNA gene sequences of strains DFA1 and DFA10 were 100 and 98% identical to that of strain DBF63, respectively). We can successfully amplify the whole DNA region of ORF16 homologues, and sequence analysis revealed that ORF16 homologues from strains DFA1 and DFA10 were 100 and 99% identical to the ORF16 sequence of strain DBF63, respectively (data not shown). Thus, in our survey, this modular nature of ORF16 was found in not only strain DBF63 but also other Terrabacter strains. It will be interesting to investigate how ORF16 really works in the FN or DF degradation of strain DBF63.

meta-cleavage product hydrolase.

We investigated by E. coli biotransformation experiments whether the ORF3 product possesses hydrolase activities for meta-cleavage products of DHB or 2′-substituted DHB (2′-aminobiphenyl-2,3-diol). 2′-Aminobiphenyl-2,3-diol was prepared as described previously (15). The 1,018-bp DNA fragment containing ORF3 (flnE) was amplified with the forward primer containing an efficient SD sequence for E. coli (24). The DNA fragments were cut at both the HindIII and EcoRI sites (derived from the primer) and then cloned between the HindIII and EcoRI sites of pUC119 to give pDF27, and those of pSTV29 were modified to give pDFS27 (Fig. 2). E. coli JM109 was transformed with both plasmids pDF502 and pDFS27. As a control, E. coli JM109 harboring both plasmids pDF502 and pSTV29 was used. Biotransformation experiments and GC-MS analysis showed that only E. coli cells with both pDF502 and pDFS27 could transform DHB to benzoate. Also, 2′-aminobiphenyl-2,3-diol was transformed to anthranilate (data not shown).

Enzymatic hydrolase activity of the ORF3 product was assayed as described by Nojiri et al. (21). Since the half-life of 6-(2′-aminophenyl)-HODA (meta-cleavage product of 2′-aminobiphenyl-2,3-diol) was too short (21), we could not measure the decrease in its absorbance. The enzymatic activities of the extracts from E. coli harboring pDF27 for 2-hydroxy-6-oxohepta-2,4-dienoic acid (6-methyl-HODA) and 6-phenyl-HODA were 0.391 ± 0.0512 and 3.73 ± 0.101 U μg of protein⁻¹, respectively. No hydrolase activities were observed in the control experiments (0.0474 ± 0.0205 U μg of protein⁻¹). After hydrolase activity measurement, the products in the reaction mixture were extracted with ethyl acetate and analyzed by GC-MS. As a result, benzoic acid was detected only from the reaction mixture with E. coli harboring pDF27 with 6-phenyl-HODA. This indicates that ORF3 encodes the meta-cleavage product hydrolase for 6-(2′-substituted phenyl)-HODA, and thus we designated ORF3 flnE.

The meta-cleavage product hydrolases belong to the α/β-hydrolase family, and most of them are classified into two major groups, I and III, proposed by Hernáez et al. (13). However, FlnE did not form a cluster with the group I proteins including BphD but formed a branch with the group IV proteins including OhpC and CmtE (data not shown). OhpC and CmtE are involved in the degradation of monocyclic compounds with carboxyl group substitution, i.e., 3-(2-hydroxybiphenyl)propionate (22) and p-cumate (4), respectively. There may be some evolutionary relationship among these enzymes.

In conclusion, we characterized the dbf-fln genes of Terrabacter sp. strain DBF63, which are involved in the degradation of FN to phthalate (flnB-dbfA1-dbfA2-flnE-flnD1-ORF16), and several genes were found to be unusual in their sequence length, low homology to the known proteins, modular nature, and so on. This new information adds some knowledge to the genetics of bacterial PAH degradation, because although the genes involved in the upper pathway of naphthalene, phenanthrene, and anthracene metabolism have been studied extensively in recent decades, there have been no reports about the complete upper metabolic pathway genes for the other 16 priority PAHs. For better understanding of the pathway and mechanism of FN catabolism by strain DBF63, biochemical studies with the purified enzymes in each catabolic step are under way.

Nucleotide sequence accession numbers.

The nucleotide sequence data reported here have been submitted to the DDBJ, EMBL, and GenBank nucleotide sequence databases and assigned accession number AB095015.