INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Sphingomonas aromaticivorans F199 was isolated from sediments collected 410 m below the land surface near Allendale, S.C., in 1988 (4, 21). It was established that this bacterium possessed the novel ability to degrade a variety of aromatic compounds including toluene, all isomers of xylene, p-cresol, naphthalene, biphenyl, dibenzothiophene, fluorene, salicylate, and benzoate (20, 22). In recent years, there have been many reports of other Sphingomonas strains that are capable of degrading aromatic compounds (12, 17, 30, 36, 40, 44, 45, 62-65, 68-70, 88, 91). Studies of Sphingomonas strains suggest that members of this genus are well adapted for the degradation of high-molecular-weight polycyclic aromatic hydrocarbons and other aromatic contaminants. The inability to detect Sphingomonas biodegradative genes via hybridization with catabolic genes from phylogenetically distinct bacteria suggested that biodegradative genes from Sphingomonas sp. evolved independently from phylogenetically distinct bacteria such as those within the genus Pseudomonas (46, 47). Some Sphingomonas strains are further distinguished in that the genes necessary for degradation of one type of aromatic compound are distributed into multiple operons that also possess genes for the degradation of other aromatic compounds (107). This unusual gene arrangement suggests that a highly complex regulatory network is responsible for the expression of aromatic degradative pathways in some Sphingomonas spp.

S. aromaticivorans F199 was shown to possess two plasmids (20, 22), which are designated pNL1 (~180 kbp) and pNL2 (~480 kbp). We reported earlier that catechol meta ring cleavage activity, a central step in the catabolism of aromatic rings, was associated with the smaller plasmid, pNL1 (86), and we described a physical map for this plasmid. To further probe the catabolic functions and accessory genes encoded on pNL1, we undertook the complete sequencing and annotation of this plasmid. This approach has allowed a thorough genetic analysis of pNL1-associated catabolic genes, a comparison of these genes with analogous chromosomally located ones in S. yanoikuyae B1, and the development of hypotheses regarding functions of pNL1-encoded genes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and culture conditions. S. aromaticivorans F199, originally isolated by our laboratory, is also maintained in the U.S. Department of Energy's Subsurface Microbial Culture Collection at Florida State University (3). Sphingomonas sp. strain S-88 (ATCC 31554) mutant m260 was obtained from Thomas Pollock (77). Its resistance to bacitracin and inability to degrade aromatic compounds examined in this study was useful for plasmid transfer experiments. S. paucimobilis ATCC 298377 was purchased from the American Type Culture Collection. Bacteria were maintained on either full-strength (Escherichia coli) or half-strength (Sphingomonas) Luria-Bertani medium (LB).

Transposon mutagenesis. A portion (50 µl) of an overnight culture of the S. aromaticivorans F199 recipient was spotted and allowed to dry on half-strength LB agar. An equivalent amount of overnight cultures of the donor strains, E. coli with either SB354 or pRK2013, were then overlaid, dried, and incubated overnight at 30°C. The resulting bacterial spots were resuspended in 1 ml of saline, and 100-µl aliquots were plated onto 0.5× LB containing kanamycin to select for transposition and polymyxin B to select against the E. coli donor strains. Sphingomonas colonies are readily distinguished from E. coli colonies because of their characteristic yellow pigmentation. The resulting Sphingomonas insertion mutants were screened for loss of dioxygenase activity by placing a small crystal of indole in the lid of the petri dish; the colonies were then examined for the absence of blue coloring (i.e., lack of indigo formation). Conditions for testing the ability to grow on aromatic compounds or to produce colored intermediates from them by Sphingomonas strains have been described elsewhere (20).

Conjugal transfer of pNL1. Recipient Sphingomonas sp. S-88 m260 and donor S. aromaticivorans F199 strains were mated as described above. Exconjugants were isolated on 0.5× LB plates containing bacitracin to select against donor cells and kanamycin to select for transfer of transposon-mutagenized pNL1. After incubation, a few crystals of indole were added to the lid of the plate and further incubated to allow the development of blue color. The presence of pNL1 in exconjugants was confirmed by plasmid isolation (39) and analysis by pulse field gel electrophoresis (1% agarose, 0.5× TBE, 6 V/cm, included angle of 120, 0.475-s initial switch time, 21.79-s final switch time, linear ramping factor, 20.5-h run time) with the Bio-Rad Chef system (Bio-Rad, Hercules, Calif.).

Construction and isolation of small and large insert libraries. The procedure for isolation of pNL1 has been described elsewhere (86). For construction of the large insert library, a partial Sau3A digestion of pNL1 was used to produce fragments of approximately 40 kb, which were then dephosphorylated and ligated into the BamHI site of the sCos-1 cosmid vector (16). Ligation mixtures were packaged into lambda particles by using the Gigapack II packaging extract (Stratagene, La Jolla, Calif.), infected into E. coli DH5, and plated onto LB agar supplemented with 100 µg of kanamycin per ml. Cosmids were isolated from 6-ml overnight cultures of the resulting kanamycin-resistant colonies by using a modified boiling lysis procedure (55). Briefly, bacterial pellets were dissolved in 300 µl of solution A (8% sucrose; 5% Triton X-100; 50 mM Tris, pH 8.0; 50 mM EDTA) and 10 µl of solution B (10 mg RNase per ml; 1 mg of lysozyme per ml; 50 mM Tris, pH 8) and boiled for 1 min. After removal of the cellular debris by centrifugation, 10 µl of a 10-mg/ml concentration of pronase was added to the supernatant and incubated for 30 min at 65°C to remove any remaining bacterial proteins. The cosmid DNA was precipitated by adding 20 µl of 5 M ammonium acetate and 900 µl of isopropanol-1 mM phenylmethylsulfonyl fluoride and was then resuspended in 75 µl of TE (pH 8.0). Restriction enzyme digestion of approximately 300 cosmid clones with NotI and SspI demonstrated uniform coverage of pNL1.

Sequencing. Plasmid DNA was purified with the Qiagen QIAwell 96 Ultrawell plasmid kit (Qiagen, Inc., Valencia, Calif.). Cosmid DNA was purified by standard alkaline lysis (55), denatured at room temperature for 10 min in 0.5 M NaOH, neutralized with 1.2 M (final concentration) sodium acetate, and ethanol precipitated. For sequencing reactions, 500 ng of plasmid DNA or 1 µg of cosmid DNA were used for both the cosmid and the plasmid libraries, and primers complementary to the T3 and T7 vector promoter sequences were used to generate the sequence from the insert ends by using dideoxy chain-termination sequencing reactions (83) with Perkin-Elmer/Applied Biosystems Dye Terminators and Ampli-Taq DNA polymerase. For gap closure, primers were designed from the ends of assembled sequences by using Sequencher software (Gene Codes Corp., Ann Arbor, Mich.), MacVector sequence analysis software (Eastman Kodak, Rochester, N.Y.) and Gap4 from the Staden software package. Primers were synthesized by using standard phosphoramidite chemistry on the Applied Biosystems model 392 DNA synthesizer. Sequencing reactions were analyzed with an ABI 377 DNA sequencer.

Sequence assembly. Sequences were assembled by using tools from the Staden Sequence Analysis software package (10). Sequences were initially processed with PREGAP to automate the generation of compressed files for assembly and to mark vector and other non-pNL1 sequences that appear as a result of library construction. These files were assembled and manually edited by using GAP4. Upon completion, a FASTA file was generated for automated sequence analysis.

Automated sequence analysis. Automated sequence analyses were completed by using MAGPIE (multipurpose automated genome project investigation environment) (27). Initial assignments of putative open reading frames (ORFs) were made by using the criteria that (i) the start codon was ATG, GTG, or TTG; (ii) the stop codon was TAA, TAG, or TGA; and (iii) the ORF size was between 150 and 30,000 bp in length. Default MAGPIE settings were used to characterize the 1,346 potential ORFs identified. Data collected was manually surveyed to identify the ORFs most likely to encode a gene. The position of each start codon was estimated manually by identification of the Shine-Dalgarno sites within 15 bp of a potential start codon, the position of upstream genes, and a comparison to the start position of the gene homologues. Protein sequences were aligned with CLUSTALW. ORFs predicted to encode genes were adjusted in size to reflect the manually predicted start position and realigned with GenBank sequences by using BLASTP2 and TBLASTN.

Nucleotide sequence accession number. The sequence of the S. aromaticivorans F199 plasmid pNL1 has been deposited with GenBank under accession number AF079317.

Oxygenase activity measurements. Two cosmids were selected from the E. coli XL1-Blue large insert library which contained either bphC (cosmid 6) or xylE (cosmid 18) for oxygenase activity measurements. The sCOS-1 T3 primer target site in these clones is positioned 5,887 and 288 bp upstream of the start codons of bphC and xylE, respectively. Overnight cultures were harvested and washed in 20 mM sodium phosphate buffer (pH 7.5). After the addition of a final 0.1-mg/ml concentration of DNase, the cells were lysed by the application of two passes through a French pressure cell at 16K lb/in². Cellular debris was removed by centrifugation (JA-20; 10,000 rpm for 25 min). The protein content was estimated with a Bio-Rad protein assay kit. Enzymatic production of colored intermediates was measured spectrophotometrically by monitoring the increase in the absorbance at the corresponding wavelength of each meta cleavage product formed from the following substrates: catechol, 375 nm; 3-methyl catechol, 388 nm; 4-methyl catechol, 382 nm, and 2,3-dihydroxybiphenyl, 434 nm. The reaction mixture contained 0.1 M sodium phosphate (pH 8) and between 0.5 and 1 mg of cellular lysate. The reaction was initiated by the addition of the appropriate substrate at a final concentration of 0.4 mM for catechol-type substrates or 0.5 mM for 2,3-dihydroxybiphenyl.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Association of conjugative and biodegradative functions with pNL1. Attempts to transfer native pNL1 to nonaromatic degrading S. paucimobilis and Sphingomonas sp. strain S88m260 in conjugation experiments were unsuccessful. In order to provide a more robust selectable marker on the plasmid, S. aromaticivorans F199 was subjected to transposon mutagenesis. A single mutant, designated F199 tn349, with an altered phenotype that was linked to plasmid function, was identified. S. aromaticivorans F199 tn349 was unable to convert indole to indigo, a property commonly affiliated with oxygenases, including styrene monooxygenase (71), xylene monooxygenase (11), toluene dioxygenase (37), and naphthalene dioxygenase (13). This mutant was also unable to grow on naphthalene or biphenyl and could not produce orange metabolites from dibenzothiophene or yellow metabolites from fluorene or biphenyl. It retained its ability to grow on p-cresol, m-xylene, salicylate, and benzoate and to produce yellow metabolites from catechol.

Nucleotide composition and restriction sites. The circular plasmid pNL1 from S. aromaticivorans F199 was 184,457 bp in length. The overall G+C content was 62% and the A+G content was 49%; this value is consistent with an earlier report of a G+C content of 62.9 to 65.4 mol% for S. aromaticivorans F199 total DNA (5), which suggests that this plasmid was not laterally transferred from another genus.

View larger version (29K): [in this window] [in a new window]   FIG. 1.   Physical map of pNL1. Fragments predicted by computational analysis of the pNL1 sequence for NotI and SspI digests are shown in the outer and inner rings, respectively. The positions of selected genes are shown outside the rings, and putative regulators encoded by orf7 (A), orf158 (B), orf569 (C), orf597 (D), orf758 (E), and orf994 (F) are shown in circles on the perimeter of the outer ring. Also depicted are the regions with homology to S. yanoikuyae B1 and Sphingomonas sp. strain RW1 (bidirectional arrows) and the regions encoding genes associated with aromatic degradation, conjugation, and plasmid partitioning and replication.

ORF analysis. A graphical representation of the ORFs predicted to encode genes is depicted in Fig. 2. Similarities between representative homologs and pNL1 ORFs are detailed in Table 1. Based on gene homolog analysis, functions associated with the plasmid were divided into three general categories: catabolism of aromatic compounds, plasmid replication and partition, and conjugation. The genes associated with each of these general categories are discussed in detail below, with emphasis on those associated with the catabolism of aromatic compounds.

View larger version (45K): [in this window] [in a new window]   FIG. 2.   Graphical depiction of ORFs. Putative ORFs are depicted as boxes placed either above (frames 1 to 3) or below (frames 4 to 6) the axis. The numbers correspond to MAGPIE number assignments for each of 1,394 potential ORFs found on pNL1. Red arrows indicate putative transcripts (1 to 11) which encode enzymes involved in biphenyl, m-xylene, and naphthalene degradation. Colors are used to highlight putative regulators (red); aromatic oxygenase subunits (purple); enzymes in biphenyl (green), naphthalene (gray), m-xylene (yellow), and p-cresol (turquoise) pathways; possible aromatic transport proteins (light blue); transposases and recombinases (medium blue); plasmid partitioning and replication (light orange); and conjugative genes (dark orange). Hatched boxes depict ORFs with no detected homologs outside pNL1.

Ring-hydroxylating dioxygenases. Seven homologs to both large and small substrate binding components of ring hydroxylating dioxygenases were identified (bphA1[a-f]-bphA2[a-f]). Only a single homolog for each the ferredoxin (bphA3) and the ferredoxin reductase (bphA4) components were found. A homolog to XylZ, possessing both ferredoxin and ferredoxin reductase domains, was not found. The Cys-X₁-His-X₁₇-Cys-X₂-His Rieske-type (2Fe-2S) cluster binding site motif is conserved in each of the seven pNL1-encoded large oxygenase components. All except BphA1e also possess a potential mononuclear non-heme iron coordination site consensus sequence, E-X_3/4-D-X₂-H-X_4/5-H, defined by Jiang et al. (38). In BphA1e (S. aromaticivorans F199 and S. yanoikuyae B1) and OrfG1 from Sphingomonas sp. RW1 (1), two aspartate residues are separated by only one amino acid (D-X₁-D-X₂-H-X₄-H). It is unclear whether these latter proteins can accommodate the Fe(II) ligand.

View larger version (40K): [in this window] [in a new window]   FIG. 3.   CLUSTALW dendrogram of representative large (above) and small (below) oxygenase component amino acid sequences. The GenBank accession numbers associated with these sequences are as follows: Acinetobacter calcoaceticus ADP1 (2996621-2996622), P. putida mt-2 plasmid pWWO (139861-139862), Pseudomonas sp. strain U2 plasmid pWWU2 (2828018-2828019, 2828015-2828016), Sphingomonas sp. strain RW1 (3618270), Burkholderia sp. strain DNT (1477921), Cycloclasticus sp. strain 1P032 (3170521), P. aeruginosa PaK1 (12555669-1255670), P. putida OUS82 (1073044-1073045), and P. aeruginosa JB2 (3643998-3643999). Sequences from S. yanoikuyae B1 were derived from the dissertation thesis of E. Kim (45).

Biphenyl, naphthalene, and m-xylene catabolic gene homologs. Homologs to genes associated with m-xylene, biphenyl, and naphthalene model degradative pathways were identified on pNL1 (Fig. 4). Based on gene spacing and the occurrence of terminator-like structures, we predict that genes encoding enzymes associated with these pathways are distributed among at least 11 transcriptional units (Fig. 2).

View larger version (27K): [in this window] [in a new window]   FIG. 4.   Hypothetical naphthalene, biphenyl, and toluene degradative pathways encoded on pNL1. Pathway intermediates include naphthalene (n1), cis-1,2-dihydroxy-1,2-dihydronaphthalene (n2), 1,2-dihydroxynaphthalene (n3), 2-hydroxy-2-(2-oxo-3,5-cyclohexadienyl)-buta-2,4-dienoate (n4), 2-hydroxychromene-2-carboxylate (n5), salicylaldehyde (n6), salicylate (n7), biphenyl (b1), cis-2,3-dihydrodiol (b2), 2,3-dihydroxybiphenyl (b3), 2-hydroxy-6-oxo-6-phenylhexa-2,4-dienoic acid (b4), cis-2-hydroxy-penta-2,4-dienoate (b5), 4-hydroxy-2-oxovalerate (b6), pyruvate (b7), acetaldehyde (b8), toluene (t1), m-methylbenzyl alcohol (t2), m-methyl benzaldehyde (t3), benzoate (t4), 1,2-dihydroxycyclohexa-3,5-dienecarboxylate (t5), and catechol (t6). The enzymes predicted to produce these intermediates are dioxygenase (bphA1fA2fbphA3bphA4), cis-dihydrodiol dehydrogenase (bphB), catechol dioxygenase (bphC), 2-hydroxychromene-2-carboxylate dehydrogenase (nahD), 1,2-dihydroxybenzylpyruvate aldolase (nahE), salicylaldehyde dehydrogenase (nahF), salicylate hydroxylase (nahG), salicylate-5-monooxygenase (nagG), 2-hydroxy-6-oxo-6-phenylhexa-2,4-dienoate hydrolase (bphD), 2-oxopent-4-enoate hydratase (bphE), 4-hydroxy-2-oxo-valerate aldolase (bphF), xylene monooxygenase (xylAM), benzyl alcohol dehydrogenase (xylB), benzaldehyde dehydrogenase (xylC), toluate oxygenase (xylXYbphA3bphA4), and dihydroxy cyclohexadiene carboxylate dehydrogenase (xylL). R1 = R2 = R3 = H for toluene degradation, R1 = R3 = H and R2 = CH3 for m-xylene degradation, and R1 = R2 = H and R3 = CH3 for p-xylene degradation.

p-Cresol catabolic gene homologs. The prototypical toluene catabolic pathway from P. mendocina KR1 proceeds through a p-cresol intermediate and involves the action of three enzymes to generate protocatechuate from p-cresol (99). The conversion of p-cresol to p-hydroxybenzaldehyde is a two-step process that is catalyzed by a two-component p-cresol methylhydroxylase. Two gene homologs, designated pchFa and pchFb, to the p-cresol methylhydroxylase flavoprotein component and one cytochrome subunit homolog, pchFc, are found on a single cluster in pNL1 (Fig. 2). Subsequent conversion of p-hydroxybenzaldehyde to p-hydroxybenzoate in P. mendocina KR1 is catalyzed by p-hydroxybenzaldehyde dehydrogenase. NahF clusters most closely with p-hydroxybenzaldehyde dehydrogenase (unpublished results; GenBank no. 995954) and presumably is capable of catalyzing the aldehyde dehydrogenase reactions in both the naphthalene and the p-cresol pathways. Homologs to known proteins responsible for catabolism of p-hydroxybenzoate were not found on pNL1.

Membrane proteins. Several pNL1 ORFs, distributed among the catabolic genes, have extensive homology to bacterial efflux pump genes. Typical efflux pump transport systems are composed of a cytoplasmic and outer membrane protein and a membrane fusion protein (74). Sequence similarity analysis suggests that ORF140, ORF132, and ORF121 encode the outer membrane, membrane fusion, and cytoplasmic membrane components of a novel efflux pump. Psort analysis of ORF140 identified a possible signal cleavage site and supports its localization in the outer membrane. Tmpred analysis of the putative cytoplasmic component of the pNL1 encoded efflux pump, ORF121, predicts the presence of 14 transmembrane helices, a feature common to Family 1 transport exporters such as EmrB (73). A signature for the lysR family of regulatory proteins is present at the N terminus of ORF121 and, like ORF140, it has a membrane lipoprotein lipid A signature.

Other ORFs found in the aromatic catabolic region of pNL1. Homologs to two additional ORFs, bphK and orf1158, also cluster with aromatic catabolic genes, but their function in aromatic catabolism has not been established. BphK is a glutathione S-transferase, and its homologs are found in polycyclic aromatic catabolic gene clusters of Pseudomonas sp. strain DJ77 (unpublished data), S. paucimobilis epa505 (54), Pseudomonas pseudoalcaligenes KF707 (49), Pseudomonas sp. strain LB400 (34), S. yanoikuyae B1 (45), and Cycloclasticus oligotrophus RB1 (95). Homologs to ORF1158 are found in the B. cepacia Pc701 plasmid-encoded 4-methylphthalate catabolic cluster (82) and in S. yanoikuyae B1 (ORF2) (107). The latter ORFs are also similar to the pyridoxal phosphate biosynthetic protein (PdxA).

Plasmid partition-replication. Two RepA-like proteins are encoded by the bidirectionally transcribed genes orf211 (repAa) and orf218 (repAb). Their deduced protein sequences are dissimilar and consequently cluster with different families of RepA proteins. The intergenic space between repAa and repAb has six 17-bp direct repeats with the consensus sequence 5'-rsCGATGAwyTCvGwkC. Five repeats are located on the strand that encodes repAb, and one is on the repAa-encoding strand. An additional repeat with this consensus is found at the end of repAb. In the intergenic space between repAa and repB are three inverted repeats with the consensus sequence 5'-AGTTGCCACGTGGCAAC.

Integration-recombination-associated genes. pNL1 ORFs predicted to function in DNA integration or recombination include a resolvase family site-specific recombinase (orf277), an excisionase (orf338), a phage-type integrase-recombinase (orf338), and two transposons (tnpABC and isAab). Since orf277 and orf271 are homologous to adjacent genes in M. tuberculosis (76), it is probable that both ORFs are functionally related. Sequences from the region encoding isAab resemble the ISRm2011-2 transposase of R. meliloti (84). Like ISRm2011-2, there is a potential frameshifting window (5'-AAAAAAAG) near the end of IsAa that would allow its fusion with IsAb to form the mature transposase IsA. IsA is flanked by inverted repeat sequences: 5'-CGCGCTAGAGCGGTTTTCGA ending 79 bp upstream of isAa and 5'-TCGAAAATCGCTCTAGCGCG starting immediately after the TGA stop codon in isAb. These repeats are flanked by an upstream CG and a downstream GC putative target of the duplication sequence.

Conjugative genes. Three gene clusters encode homologs to E. coli F plasmid genes required for conjugative sex pilus formation and mating-pair stabilization (24). The first cluster includes homologs to the F plasmid genes traL, traE, traK, and traB; to dsbC (61); and to S. typhimurium plasmid R64 trbB (26). DsbC catalyzes disulfide bond formation in some periplasmic proteins. Like DsbC, ORF883 possesses an active site typical of disulfide isomerases (FsdfrCgyC) and is predicted to reside in the periplasm. The second cluster includes homologs to E. coli F plasmid genes traC, trbI, traW, traU, traN, traH, and traG. Homology to the entire TrbI is localized within the N-terminal portion of ORF851. The C-terminal portion is homologous to TraF from E. coli RK2 (97) and to various type-I leader peptidases. The third cluster encodes an F plasmid-like TraD and a protein that is similar to TrwC from E. coli R388 (53) and to the N-terminal portion of F plasmid TraN. A total of six additional ORFs are also present in these clusters but have no identifiable homolog.

Regulatory genes. Six regulatory genes were identified on pNL1. ORF597 is a member of the Ros-MucR family and shares the conserved C₂H₂ zinc finger-like domain, C-X₂-C-X₃-(F,M)-X₅-H-X₄-H associated with this family. ORF569 belongs to the ECF (extracytoplasmic functions) sigma-54 subfamily of regulators (Prosite no. PDOC00814). The ECF sigma factors constitute a diverse group of alternative sigma factors that have been demonstrated to regulate gene expression in response to environmental conditions. ORF758 is a gntR family regulator (Prosite no. PDOC00042) with similarity to regulators that cluster with aromatic catabolic genes in Acinetobacter sp. strain ADP1 (unpublished results; GenBank no. AF009672), P. pseudoalcaligenes KF707 (unpublished results; GenBank no. 1389649), and B. cepacia LB400 (14) and with the regulator for the pNL1 ORF24 homolog, NtaA (94). These similarities suggest that ORF758 may have a role in the regulation of aromatic degradative genes.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This is the first report of a complete sequence of a conjugative plasmid that encodes pathways for the complete catabolism of aromatic organic compounds. Nearly one-half of the pNL1 DNA encodes genes likely to be involved in either catabolism or transport of aromatic compounds. The remainder of the plasmid appears to encode functions associated with plasmid replication, maintenance, or transfer. A remarkable conservation in both sequence and gene order is evident between pNL1-encoded aromatic catabolic genes and their homologs in other Sphingomonas strains. DNA sequences from S. agrestis HV3 and Pseudomonas sp. strain DJ77 possess high homology to pNL1. A 4,012-bp DNA fragment encoding cmpFEXC from S. agrestis HV3 plasmid pSKY4 is 90% identical to the pNL1 region encoding xylF-xylE-cmpX-xylG (105). DNA fragments from Pseudomonas sp. strain DJ77 are 88% identical over a 2,363-bp region encoding homologs (phnDEF) to xylF-xylE-cmpX (sequence assembled from GenBank entries PSU83881, PSU88298, and PSU83882) and 91% identical over a 1,520-bp region encoding homologs to xylG-xylJ (48). The high degree of sequence similarity suggests that either strain DJ77 is a member of the genus Sphingomonas or that lateral transfer of genes occurred between Sphingomonas and Pseudomonas species. The genus Sphingomonas is a relatively recent addition to the list of accepted genera (103), and resulted in the renaming of several bacteria (i.e., P. paucimobilis, Flavobacterium capsulatum, and Beijerinkia sp. strain B1).

Thirty-five genes, extending from bphA2e to nahD are found in the same order and transcriptional direction of the chromosome of S. yanoikuyae B1 (107), an organism that degrades a range of aromatic compounds similar to that of S. aromaticivorans F199. The presence of five additional genes (bphA2c, bphA2d, cmpX, xylT, and bphA2e) not listed in the preliminary report of the S. yanoikuyae B1 gene order were confirmed (106). The primary difference between these regions exists at the ends of the 39-kb region. At one end, downstream of the S. yanoikuyae nahD gene, is an insertion sequence element. The bphX gene at the other end of the S. yanoikuyae B1 sequence is a putative membrane protein that is similar to other putative membrane proteins associated with catabolic genes including the P. fluorescens IP01 cumene (29) and the P. putida F1 toluene (96) degradative gene clusters. Absent from the sequenced S. yanoikuyae B1 region were homologs to the putative membrane proteins, ORF1038 and ORF1042, located between xylM and bphA2 in pNL1. Another notable difference between the pNL1 and S. yanoikuyae B1 catabolic sequences was the distance between the 3' ends of bphB and xylA. The gap between these genes in pNL1 was 60 bp, and in S. yanoikuyae B1 it was 555 bp.

Although DNA sequences from S. yanoikuyae B1 and S. aromaticivorans F199 are similar, significant differences in the catabolic activity of these two strains has been noted. S. yanoikuyae B1 is capable of growth on both monocyclic (toluene, p-ethyltoluene, 1,2,4-trimethyl benzene, and m- and p-xylene) and polycyclic aromatics (biphenyl, naphthalene, phenanthrene, and anthracene) (107). S. aromaticivorans F199 can also grow on toluene, all xylene isomers, naphthalene, and biphenyl (22), but in our studies growth was not as robust as that of S. yanoikuyae B1 on these compounds. Another difference noted was in the ability to express bphC and xylE in E. coli without the need for vector-encoded promoter sequences. E. coli containing cosmid clones of S. yanoikuyae B1 encoding xylE or bphC could not be detected by screening for catechol ring cleavage activity (47). In addition, expression of xylE from pGEM7Zf() was dependent on the vector-encoded lac promoter. In contrast, cosmid clones encoding pNL1 xylE or bphC genes were readily detected in E. coli by noting the yellow ring cleavage product that appeared after colonies were sprayed with catechol (86). In sequencing these cosmids, we found that vector-encoded promoters were not required for the expression of ring cleavage activity. Furthermore, the absence of the putative promoter region upstream of xylF in cosmid 18 (the present study) suggests that a secondary promoter recognized by the E. coli transcriptional machinery is located within the C terminus of xylF.

Besides probable differences in gene expression, the presence or absence of specific membrane proteins may also account for the differences in levels of catabolic activity observed in these two Sphingomonas strains. Membrane proteins have been implicated in the transport of aromatic compounds across bacterial cell walls. Functions ascribed to these proteins include the export of toxic aromatic compounds out of the cell (2a, 3, 34, 36a, 39) or the uptake of nontoxic aromatics from the environment (9, 67, 78, 100). Aromatic compounds can be toxic due to the adverse effect of accumulation of the hydrophobic compounds in the cellular membrane. Consequently, bacterial growth on aromatic compounds is typically achieved by allowing the substrate to volatilize from a separate phase and diffuse into the medium rather than by direct addition to the growth medium. Some aromatic compounds are not toxic to the cell but are sparingly soluble in water, making them poorly available to bacteria. Therefore, successful growth on aromatic compounds requires both mechanisms for exclusion of toxic aromatics from the cellular interior and uptake of nontoxic aromatic substrates into the cytoplasm where they can be metabolized. A homolog to the functionally uncharacterized membrane protein, BphX, is absent on pNL1 and may play an important role in modulating growth on aromatic compounds which somehow enables S. yanoikuyae B1 to grow better in laboratory conditions than S. aromaticivorans F199.

The presence of putative efflux pump genes within the aromatic catabolic region of pNL1 suggests that it may function in efflux of aromatic substrates or intermediates associated with pNL1-encoded degradative pathways. Several reports on bacterial efflux pumps associated with solvent (toluene) export and resistance have been published prior to this work (42, 47a, 52a, 79a). The P. putida S12 and putative pNL1 efflux pumps are distinct from each other in that they are members of the major facilitator superfamily and the resistance-nodulation-cell division family, respectively. Both families are predicted to utilize proton motive force to drive substrate export (73). It would be interesting to determine if other Sphingomonas strains contain similar efflux pump genes.

The export and catabolism of aromatic compounds are two biochemical mechanisms for solvent resistance. Physical mechanisms that lead to increases in cell membrane rigidity are also thought to be important in solvent tolerance (79). This alteration is achieved through desaturation of cis fatty acids and subsequent conversion to the trans isomer (33, 35, 98). ORF15 and ORFs that cluster with bphD are similar to fatty acid biosynthetic pathway enzymes and may prove to function in altering the cellular membrane in response to the presence of aromatic compounds.

The occurrence of multiple oxygenase binding subunit homologs also appears to be characteristic of Sphingomonas sp. Although it is tempting to speculate that S. aromaticivorans F199 is able to mix and match these oxygenase binding subunits to theoretically form as many as 49 different four-component enzymes, the fact that (i) every large subunit gene is adjacent to a small subunit gene, (ii) members of each large and small subunit pair align best to subunits of the same enzyme, and (iii) all but two of the oxygenase subunit sets are on different transcripts suggests that this does not normally occur. There have been, however, several reports on the construction of hybrid aromatic oxygenases which were in some cases functionally active (72, 89, 93). This suggests that if S. aromaticivorans F199 simultaneously expresses several sets of oxygenase binding subunits along with BphA3 and BphA4, novel combinations of oxygenase binding subunits and electron transfer components may occur that result in the formation of hybrid enzymes. Experimentation is needed to test this hypothesis and to measure the ability of hybrid enzymes from this bacterium to oxidize aromatic compounds.

Except for the apparent absence of a nahG or xylZ homolog, homologs to all genes found in model degradative pathways for biphenyl, m-xylene, and naphthalene were found on pNL1. We presume that the xylZ ferredoxin and ferredoxin reductase activities are provided by bphA3 and bphA4 for the benzoate dioxygenase encoded by pNL1. The existence of similar four-component benzoate dioxygenases have been described in S. yanoikuyae B1 (107) and P. aeruginosa (81). Since pNL1 conferred the ability to grow with salicylate as the sole carbon and energy source to Sphingomonas sp. strain S-88 m260, the genes for catabolism of salicylate are most likely encoded on pNL1. Salicylate can be degraded to gentisate by some microorganisms via catalysis by salicylate-5-monooxygenase (NagG) (25). The BphA1c-BphA2c and BphA1d-BphA2d oxygenase binding subunits cluster most closely with NagG-NagH, to functionally uncharacterized proteins in dinitrotoluene degradative gene clusters, and to the o-halobenzoate dioxygenase large subunit, OhbB (Fig. 4). Since bphA1cA2c clusters with nahD, it is a prime candidate for an oxygenase catalyzing the degradation of salicylate. Although the gentisate pathway has been described, the sequence for genes in this pathway is limited to two short peptide stretches of 15 and 26 amino acids from gentisate 1,2-dioxygenase (32). No significant homology with these peptides was found among sequences encoded by pNL1. Until more of the gene sequence information is available from the gentisate catabolic pathway, it will not be possible to rule out a gentisate pathway encoded on pNL1.

The putative replication-partitioning region of pNL1 is characteristic of iteron plasmids. The six repeats found between the orf211 and orf218 repA genes are likely iteron sequences to which the RepA binds to regulate plasmid replication. The occurrence of two different repA genes on the same plasmid is unusual and may provide a mechanism for regulating the copy number of pNL1 in response to different environmental stimuli. Bacterial homologs to eukaryotic group II-associated maturases have been previously associated with conjugative elements (60, 66, 104). To our knowledge, this is the first evidence for group II introns on a conjugative plasmid. The unusual distribution of the three putative exons near matRa suggest that both maturase-associated splicing and possibly other novel recombinatorial mechanisms are utilized to generate alternate proteins from pNL1.

Overall, the conjugative transfer genes of pNL1 are most similar to those from E. coli F plasmid (Fig. 5). The gene order and transcriptional direction of shared genes is the same in pNL1 and F, except that the traD and traI genes are inverted with respect to the rest of the conjugative genes. F plasmid gene homologs not found on pNL1 include the following: (i) the regulatory genes traJ, traY, and finO; (ii) the pilus synthesis and assembly genes traA, traV, traQ, and traX; (iii) the mating-signal gene traM, the origin-nicking gene traY, and the DNA nicking-unwinding gene traI* (the C-terminal transcript within TraI); (iv) the surface exclusion genes traS and traT; and (v) the functionally uncharacterized genes traP, traR, and artA and all trb genes save trbi and trbc. The F plasmid genes are the closest homologs to pNL1 genes except for traE (orf207), which is most similar to S. typhimurium pED208 traE (19); orf851, whose C-terminal region is homologous to the entire traF gene of plasmid RK2 (97); and orf704, which is most similar to E. coli R388 trwC (53).

View larger version (18K): [in this window] [in a new window]   FIG. 5.   Comparison of F plasmid and pNL1 transfer regions. Capital letters refer to tra genes, and lowercase letters refer to trb genes. The direction of gene transcription is indicated by arrows. Lightly shaded boxes depict shared genes, medium-shaded boxes are pNL1 ORFs encoding the ISRm2011-2-like transposase, and the black box is the putative pNL1 regulator encoded by orf758. I* refers to the alternative transcript within I. The F plasmid portion of this graphic is from Frost et al. (24).

The presence of multiple ORFs with homology to genes that function in DNA integration and recombination suggests that pNL1 sequences can be rearranged and/or integrated into other genetic elements. The ability to promote site-specific recombination between two closely linked sites on pNL1 is probably provided by the ORF181 invertase. Based on other invertase activities, the predicted outcome of ORF181 activity would be an inversion of a segment of pNL1, most likely in the immediate vicinity of ORF181. Site-specific insertion into a different DNA target is the predicted outcome for phage-type integrases encoded by ORF338, ORF277, and TnpABC. In fact, the localization of the S. yanoikuyae B1 aromatic degradative genes on the chromosome may have resulted from such a recombinational event, as early studies with S. yanoikuyae B1 (formerly Beijerinckia sp. strain B1) suggested that the aromatic degradative genes were plasmid encoded (50).

The sequences of proteins associated with aromatic catabolism in Sphingomonas species diverge considerably from those associated with other types of bacteria with similar capabilities. A large proportion of new aromatic catabolic genes sequenced were isolated from Pseudomonas species and, not surprisingly, turn out to be very similar to previously characterized genes. This becomes more apparent when doing a cluster analysis of protein families. Identification and characterization of aromatic catabolic genes from phylogenetically distinct bacteria is invaluable in computational predictions of gene functions associated with large sequences generated from microbial genome projects. For example, the characterization and sequencing of the bphDEF genes from Rhodococcus sp. strain RHA1 (56) allowed us to predict with more reliability that ORFs 1317, 1324, and 1331 encoded bphDEF, rather than xylFJK, whose products perform very similar functions and whose deduced amino acid sequences are very similar. The same rationale is also true of genes associated with plasmid replication and conjugation. Very few homologs to sequences in the corresponding region of pNL1 were found, making it more difficult to predict functions associated with them. This problem was further compounded by the shuffling of domains among different proteins associated with conjugation (i.e., ORF851 has domains associated with plasmid F TrbI and plasmid RK-2 TraF).

In summary, the analysis of pNL1 sequences suggests that this plasmid confers the ability to utilize and/or tolerate a variety of aromatic compounds to its host, S. aromaticivorans. The conferred functions related to aromatic compounds can be divided into three general categories: aromatic catabolism, transport, and cell membrane alteration. The unique codisbursement of genes associated with distinct aromatic catabolic pathways appears to be unique to Sphingomonas species and delineates a novel evolutionary path for biodegradative gene organization. The association of membrane proteins with export and/or import of pathway substrates or intermediates is suggested by the sequence annotation. They may provide a means to prevent accumulation of toxic aromatic compounds in the cytoplasm or to sequester them when present at low concentrations in the environment or when they are otherwise poorly available. A protective effect may also be conferred by pNL1-encoded enzymes that change the membrane fatty acid saturation. It should be emphasized that the many of these hypotheses are derived primarily from sequence comparisons and should be considered hypotheses until evaluated experimentally. Work is currently in progress to test some of these predictions.