Origins of the 2,4-Dinitrotoluene Pathway

Abstract

The degradation of synthetic compounds requires bacteria to recruit and adapt enzymes from pathways for naturally occurring compounds. Previous work defined the steps in 2,4-dinitrotoluene (2,4-DNT) metabolism through the ring fission reaction. The results presented here characterize subsequent steps in the pathway that yield the central metabolic intermediates pyruvate and propionyl coenzyme A (CoA). The genes encoding the degradative pathway were identified within a 27-kb region of DNA cloned from Burkholderia cepacia R34, a strain that grows using 2,4-DNT as a sole carbon, energy, and nitrogen source. Genes for the lower pathway in 2,4-DNT degradation were found downstream from dntD, the gene encoding the extradiol ring fission enzyme of the pathway. The region includes genes encoding a CoA-dependent methylmalonate semialdehyde dehydrogenase (dntE), a putative NADH-dependent dehydrogenase (ORF13), and a bifunctional isomerase/hydrolase (dntG). Results from analysis of the gene sequence, reverse transcriptase PCR, and enzyme assays indicated that dntD dntE ORF13 dntG composes an operon that encodes the lower pathway. Additional genes that were uncovered encode the 2,4-DNT dioxygenase (dntAaAbAcAd), methylnitrocatechol monooxygenase (dntB), a putative LysR-type transcriptional (ORF12) regulator, an intradiol ring cleavage enzyme (ORF3), a maleylacetate reductase (ORF10), a complete ABC transport complex (ORF5 to ORF8), a putative methyl-accepting chemoreceptor protein (ORF11), and remnants from two transposable elements. Some of the additional gene products might play as-yet-undefined roles in 2,4-DNT degradation; others appear to remain from recruitment of the neighboring genes. The presence of the transposon remnants and vestigial genes suggests that the pathway for 2,4-DNT degradation evolved relatively recently because the extraneous elements have not been eliminated from the region.

Dinitrotoluenes are intermediates in the synthesis of explosives and expanded polyurethane foam (53). Bacteria degrade 2,4-dinitrotoluene (2,4-DNT) by a pathway involving two oxygenase reactions that lead to the removal of the nitro substituents (65) (Fig. 1). The resulting methylhydroxyquinone is reduced to yield 2,4,5-trihydroxytoluene, which subsequently undergoes meta cleavage. The reactions following ring cleavage have not been previously characterized.

FIG. 1.

Pathway for degradation of 2,4-DNT in Burkholderia cepacia R34.

There are few naturally occurring nitroaromatic compounds so environmental exposure to nitroarenes has been limited until the development of modern synthetic chemistry (44). De novo evolution of genes for nitrotoluene degradation during the short period seems unlikely. Instead, it is more plausible that the pathways evolve by recruiting genes that encode degradative enzymes for other compounds to assemble a functional pathway (47). Cloning and analysis of the genes encoding the nitroarene pathways makes it possible to distinguish between the two adaptation mechanisms.

Burkholderia cepacia R34 was isolated from surface water collected at the Radford Army Ammunition plant in West Virginia (43). Strain R34 grows using 2,4-DNT as a sole carbon and nitrogen source using a pathway like that from Burkholderia sp. strain DNT (43, 65) (Fig. 1). Previous molecular genetic studies of the enzymes for the catabolism of nitrotoluenes were limited to genes encoding individual steps of the degradative pathways (21, 22, 28, 48, 68). The studies provided clues to the relationships among the nitrotoluene pathway enzymes and other degradative enzymes but less insight about the evolution of the entire pathway. We reported that the meta-ring cleavage enzyme from strain R34 shares limited identity with its counterpart in strain DNT (28). The present paper provides the nucleotide sequence of a contiguous region encoding the pathway for 2,4-DNT degradation and additional elements proposed to contribute to regulation and evolution of the pathway. Inferences from molecular genetic data and results from enzyme assays allowed identification of the metabolic steps following ring cleavage. In addition, the analysis provided insight into the mechanisms involved in recruitment of the genes, assembly of the pathway, and progenitors of the enzymes used in nitroarene degradation.

(Preliminary reports of this work have been presented [B. E. Haigler, C. C. Somerville, J. C. Spain, and R. K. Jain, Abstr. 97th Gen. Meet. Am. Soc. Microbiol., abstr. Q-343, p. 512, 1997; G. R. Johnson, R. K. Jain, B. E. Haigler, and J. C. Spain, Abstr. 98th Gen. Meet. Am. Soc. Microbiol., abstr. Q-83, p. 435, 1998].)

MATERIALS AND METHODS

Bacterial strains and plasmids.

Bacterial strains and plasmids used in the study are listed in Table 1.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Relevant propertiesa	Reference or source
Strains
Burkholderia cepacia R34	2,4-DNT degrading strain	43
Escherichia coli HB101	Cloning host	36
Escherichia coli JM109	Cloning host, lacIq	82
Escherichia coli DH5α	Cloning host	23
Escherichia coli Top10F′	Cloning host, lacIq	Invitrogen (Carlsbad, Calif.)
Plasmids
pCP13	E. coli cloning vector (cosmid), Tetr	9
PGEM7F′	E. coli cloning vector, Ampr	Promega (Madison, Wis.)
pBluescript II SK(+)	E. coli cloning vector, Ampr	Stratagene (La Jolla, Calif.)
pK18	E. coli cloning vector, Kanr	54
pK19	E. coli cloning vector, Kanr	54
pUC18	E. coli cloning vector, Ampr	82
pJS311	216-kb plasmid from strain R34, dnt genes	28
pJS312	50-kb plasmid from strain R34	28
pJS314	pCP13 containing 27-kb DNA fragment from pJS312 R34	This work
pJS324	ORF12, dntAaAbAcAd, dntB; pGEM7f containing 14.8-kb EcoRI fragment from pJS314	This work
pJS325	dntD, dntE, ORF13, dntG; 12.2-kb EcoRI fragment from pJS314	28
pJS326	dntB, ORF3; 8.6-kb HindIII fragment from pJS314	This work
pJS329	dntAaAbAcAd, dntB; pGEM7f containing 7.7-kb NsiI fragment from pJS324	This work
pJS332	dntAaAbAcAd; pGEM7f containing 5.1-kb NsiI-EcoRV fragment from pJS329	This work
pJS333	dntD, pGEM7f containing 1.5-kb XhoI::ClaI fragment from pJS328.	28
pJS336	dntD, dntE, ORF13, dntG, ORF9, ORF10; pGEM7f containing 8.2-kb EcoRI::ClaI fragment from pJS325	This work
pJS336ΔBamHI	ORF13, dntG, ORF9, ORF10; deletion derivative of pJS336	This work
pJS338	dntB; pGEM7f containing 3.2-kb BamHI fragment from pJS324	This work
pJS1362	dntE, pK18 containing 1.8-kb region from pJS328 that encodes methylmuconic acid semialdehyde dehydrogenase	This work
pJS1363	ORF13, dntG, pK18 containing 3.0-kb BamHI::PstI fragment from pJS336	This work
pJS1363ΔXmaI	dntG, deletion derivative of pJS1363	This work
pJS1364	ORF9, ORF10; pK19 containing 3.4-kb PstI::EcoRI fragment from pJS336	This work
pJS1364ΔHindIII	ORF10; deletion derivative of pJS1364	This work

^aTet, tetracycline; Amp, ampicillin; Kan, kanamycin.

Media and growth conditions.

Strain R34 was grown at 30°C using Trypticase soy medium (Difco Inc., Detroit, Mich.) or modified Stanier's base medium (MSB) (65) supplemented with succinate (0.5%) and other carbon sources as noted in the text. Escherichia coli strains were grown in Luria-Bertani medium (Difco, Inc.) or Terrific broth (73). Recombinant E. coli strains were tested for growth on 2,4,5-trihydroxytoluene by auxanography as described by Parke and Ornston (50). E. coli cultures were incubated at 37°C for molecular genetic procedures and at 30°C for use in enzyme and auxanography assays. Ampicillin (100 μg/ml), tetracycline (15 μg/ml), or kanamycin (40 μg/ml) was included in media when needed to select recombinant strains.

Recombinant DNA techniques.

Standard molecular genetic techniques were used for DNA isolation, cloning, and analysis of recombinant bacteria (59). Plasmid DNA was transferred into competent E. coli strains using a chemical transformation method (24) or electroporation using the general method described previously (3) and parameters recommended by the manufacturer (Hoefer Scientific, San Francisco, Calif.).

Cloning and localization of 2,4-DNT pathway genes.

Total plasmid DNA was isolated from strain R34 by using the method described by Kado and Liu (29), partially digested with restriction enzyme EcoRI, and then ligated with cosmid vector pCP13. The ligated vector mix was packaged into lambda phage using the Gigapak II Plus system from Stratagene Inc. (La Jolla, Calif.). The resulting phage mix was used to transfect E. coli HB101 cells. Recombinant strains were screened for DNT-dioxygenase, methylnitrocatechol (MNC) monooxygenase, and trihydroxytoluene (THT) oxygenase activity by using methods described by Suen and Spain (69). Subclones were derived from cosmids that appeared to carry genes for 2,4-DNT degradation enzymes, transferred to E. coli JM109 cells, and then tested for enzyme activity as described before in order to identify coding regions (69).

DNA sequencing and sequence analysis.

Plasmid DNA for sequencing reactions was isolated using P100 midi-prep tips (Qiagen, Inc., Valencia, Calif.). Plasmid subclones, restriction endonuclease-based deletion derivatives, and unidirectional nested deletion derivatives of plasmids pJS324, pJS325, and pJS326 were used as sequencing templates. Primers complementary to vector and internal sequences were used in dideoxy-chain termination sequencing reactions (60). Sequencing reactions were carried out using the AutoRead sequencing reagents according to the manufacturer's suggested methods, and the products were analyzed using an ALFexpress DNA sequencing system (Amersham Pharmacia Biotech, Piscataway, N.J.). The nucleotide (nt) sequence of the regions encoding the 2,4-DNT dioxygenase and methylnitrocatechol monooxygenase (nt 4341 to 14868) was determined by the Genetic Engineering Core Facility, University of Illinois.

The Lasergene software package (DNAStar Inc., Madison, Wis.) was used for initial alignment of DNA sequences and assisting primer design for PCR and sequencing reactions. Sequence databases were searched using the BLASTX, BLASTP, and BLASTN programs (1) via the National Center for Biotechnology Information website (Bethesda, Md.). Multiple-sequence alignments were done using the ClustalX program (75).

Enzyme assays.

E. coli strains used in enzyme assays were grown to mid-log phase in Luria-Bertani broth containing appropriate antibiotics, and then the synthesis of recombinant proteins was induced by the addition of IPTG (1 mM). The optimal induction time was found by determining the specific activity of the enzyme of interest or the relative synthesis of recombinant protein by using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Strain R34 was grown in MSB with 2,4-DNT (3 mM) as a sole carbon and energy source or in MSB with 2,4-DNT and 0.3% succinate to supplement growth.

Cell extracts were used in enzyme assays except where noted. The cells were harvested by centrifugation, washed with cold potassium phosphate buffer (20 mM, pH 7.5), suspended in a minimal volume of cold buffer, and then passed twice through a French pressure cell (1,360 Pa) for cell lysis. The lysate was clarified by centrifugation (21,800 × g, 30 min, 4°C) before use in enzyme assays. Protein concentration was measured by using the bicinchoninic acid protein reagent (Pierce Inc., Rockford, Ill.).

Salicylate hydroxylase activity was assayed using whole cells as described by Fuenmayor et al. (17). Reverse-phase high-performance liquid chromatography (HPLC) was used to measure salicylate and gentisate concentrations after incubation of cells with salicylate (1 mM). HPLC was done using a model 1050 chromatography system (Agilent, Santa Clarita, Calif.) with an Allsphere octyl C-8 column (250 by 4.6 mm) (Alltech Associates Inc., Deerfield, Ill.). The mobile phase was 0.1% trifluoroacetic acid-acetonitrile (60:40), 1.0 ml/min. The eluent was monitored at 210, 290, and 325 nm.

Benzenetriol oxygenase activity was measured spectrophotometrically (decrease A₂₉₀, increase A₃₄₅) (8). The benzenetriol oxygenase was partially purified from E. coli JM109 (pJS326) cells by using a series of ammonium sulfate precipitation steps. Cell extracts were obtained as above except that the initial cell lysate was subjected to high-speed centrifugation (100,000 × g, 1.5 h, 4°C) before ammonium sulfate treatment. Benzenetriol oxygenase activity precipitated between 45 and 55% ammonium sulfate.

2,4-Dihydroxy-5-methyl-6-oxo-2,4-hexadienoate (DMOH) isomerase/4-hydroxy-2-keto-5-methyl-6-oxo-3-hexenoate (HKMOH) hydrolase activity was initially detected by measuring the decrease in absorbance at 350 nm due to transformation of DMOH in phosphate buffer (50 mM, pH 7.5). Cell extracts were dialyzed for 12 h against 2 liters of cold phosphate buffer to remove diffusible cofactors. Qualitative assays of substrate transformation were monitored by scanning the UV absorbance in reaction mixtures from 220 to 425 nm. Time course assays of DMOH isomerase/HKMOH hydrolase activity were done in 5-ml reaction mixtures containing DMOH and cell extract from E. coli JM109 (pJS1363ΔXmaI) or strain R34. Samples were collected from the reaction mixture and treated with H₂SO₄ (final concentration, 0.05 N) to stop the reaction. Pyruvate and methylmalonic acid semialdehyde production were measured by HPLC using a model 1050 chromatography system (Agilent) with an Alltech model OA-1000 column (300 by 6.5 mm) (Alltech Associates Inc.). The mobile phase was 0.01 N H₂SO₄, 0.7 ml/min. The eluent was monitored at 210 and 240 nm. Pyruvate concentrations were determined by reference to a standard (Chem Service, West Chester, Pa.). The methylmalonic acid semialdehyde used as a standard was produced by hydrolysis of sodium α-formylpropionate diethyl acetal (34). In all assays, DMOH was produced from 2,4,5-trihydroxytoluene (THT) using excess THT oxygenase (DntD) supplied in cell extracts of E. coli (pJS333).

Analysis of mRNA by RT-PCR.

Cultures of B. cepacia R34 were grown in MSB containing succinate and 2,4-DNT to mid-log phase. Total RNA was isolated using the RNeasy system from Qiagen. The purified RNA was treated with RNase-free DNase to eliminate any contaminating DNA prior to the reverse transcriptase (RT) reaction. RT-PCR was done using the Access RT-PCR system from Promega (Madison, Wis.) according to the manufacturer's instructions. Primers for RT-PCR were designed to amplify the following: a 501-bp region within dntAc (dntAc_up, 5′-TCT TGC GGA AAA CTT TGT AGG TGA-3′; dntAc_down, 5′-TCG TTG TCG TCG CTT TCC CAG TAT-3′), a 383-bp region from dntAd to dntB (dntAdB_up, 5′-CTA TGC AAC GCG AGA AGA CAA ATG-3′; dntAdB_down, 5′-CAA GCT ACC CCC GAC GAT GAG AAC-3′), a 1,391-bp region from ORF8 to dntD, (ORF8dntD_up, 5′-TCC GGC GGC CAG CAG CAG ATG-3′; ORF8dntD_down, 5′-TCC GGG AAA AGG GCA AGC GAG TGA-3′), a 533-bp region within dntD, (dntD_up, 5′-GCT CTG GGG CGT GGT TAC TGC TAC-3′; dntD_down, 5′-GTG GTG GAC GCG GGG ACT C-3′), a 511-bp region from dntD to dntE (dntDE_up, 5′-GTC CCC GCG TCC ACC ACT TC-3′; dntDE_down, 5′-CGG CGG CGT TCA CCT CTT-3′), and an 845-bp region from dntE to dntG (dntEG_up 5′ AGC TGG GCG ATC TCG GTC CTT CTT ACG 3′; dntEG_down 5′ GCG ACG CTT CCC ATC CAA C 3′). Products from RT-PCR were analyzed by electrophoresis in a 1.5% agarose gel.

Chemicals.

2,4,5-Trihydroxytoluene was provided by R. Spanggord (SRI International, Menlo Park, Calif.). Sodium α-formylpropionate diethyl acetal was kindly provided by R. Harris (Indiana University School of Medicine, Indianapolis).

Nucleotide sequence accession number.

The nucleotide sequence described here is available in GenBank under accession no. AF169302.

RESULTS

Cloning 2,4-DNT catabolic genes.

Strain R34 carries two large plasmids, pJS311 (216 kb) and pJS312 (50 kb) (28). The region encoding the 2,4-DNT pathway was localized to a 50-kb region of pJS311 by analyzing plasmid DNA from strains that had lost the ability to grow with 2,4-DNT (28). No DNT dioxygenase, MNC monooxygenase, or THT dioxygenase activity was detectable in the mutant strain, indicating that the genes encoding the three enzymes are within the deleted region. The deleted region was isolated on a cosmid subclone as described in Materials and Methods. Three recombinant strains showed activity for all three enzymes, and the cosmids carried by the strains were designated pJS314, pJS315, and pJS316.

Restriction endonuclease digests of the three cosmids showed identical EcoRI restriction patterns, indicating that each held the same region of DNA from strain R34. The fragments were subcloned from pJS314 into vector pGEM7f and transferred to E. coli JM109 in order to identify the loci encoding the individual enzymes. The THT oxygenase gene was located within the fragment cloned in pJS325 (28). A second large restriction fragment (14.8 kb) was subcloned in pGEM7f and designated pJS324 (Fig. 2). Results of the enzyme assays showed that pJS324 included genes for the DNT dioxygenase and MNC monooxygenase. Subclones were derived from pJS324, transferred to E. coli JM109, and then tested for enzyme activity. The results indicated that the DNT dioxygenase genes were within the 5.6-kb DNA insert in pJS332. The MNC monooxygenase was encoded within the 3.2-kb BamHI restriction fragment subcloned in pJS338 (Fig. 2). Nucleotide sequence analysis allowed identification of other structural genes within the regions.

FIG. 2.

Physical map of region encoding 2,4-DNT pathway genes. DNA inserts from subclones used to localize structural genes and synthesis of 2,4-dihydroxy-5-methyl-6-oxo-2,4-hexadienoic acid for enzyme assays are shown below the map.

DNT dioxygenase genes.

The genes encoding the DNT dioxygenase (dntAaAbAcAd) were contained within a 5.8-kb region of pJS324 (Fig. 2). The DNT dioxygenase is related to the three-component Rieske non-heme iron oxygenases (25). The hydroxylating dioxygenase comprises an oxidoreductase (DntAa), an iron-sulfur ferredoxin protein (DntAb), and a terminal oxygenase center, DntAc (α-subunit) and DntAd (β-subunit), containing a Rieske-type iron-binding site in the α-subunit. The deduced amino acid sequences of the individual polypeptides were closely related to those of other nitroarene dioxygenases (48, 68) and the naphthalene dioxygenase from Pseudomonas sp. strain U2 (17) (Table 2), and the genes for the enzymes also have the same organization in the respective operons. Significant identity was also shared with other naphthalene and PAH dioxygenase genes (6, 12, 35, 64, 71) (Table 2). However, analysis of multiple-sequence alignments revealed that the oxygenase subunits from archetypal naphthalene dioxygenases form a separate phylogenetic group from the nitroarene dioxygenases and the naphthalene dioxygenase from strain U2.

TABLE 2.

Similarity of gene products from 2,4-DNT gene cluster to selected homologs^a

Gene product	Source	Function	Length (no. of aa)	% Identity	Alignment	Source or reference
DntAa	Burkholderia cepacia R34	2,4-DNT dioxygenase	328
NtdAa	Pseudomonas sp. strain JS42	2-Nitrotoluene dioxygenase	328	99	1-328—1-328	48
NagAa	Ralstonia sp. strain U2	Naphthalene dioxygenase	328	99	1-328—1-328	17
DntAa	Burkholderia sp. strain DNT	2,4-DNT dioxygenase	346	98	1-293—1-293	68
NahAa	Pseudomonas putida 9816	Naphthalene dioxygenase	328	66	1-328—1-293	64
ORF1a	Burkholderia cepacia R34		302
NagG	Ralstonia sp. strain U2	Salicylate hydroxylase	423	99	1-295—1-295	17
ORF2	Pseudomonas sp. strain JS42		186	99	1-179—1-179	48
ORF2	Burkholderia sp. strain DNT		423	93	1-295—1-295	68
OhbB	Pseudomonas aeruginosa JB2	o-Halobenzoate dioxygenase	421	79	10-292—7-290	Accession no. AAC69484
OhbB	Pseudomonas aeruginosa 142	o-Halobenzoate dioxygenase	428	51	20-257—12-245	77
IpbAa	Pseudomonas putida RE204	Isopropylbenzene dioxygenase	459	37	30-226—41-236	14
ORF1b	Burkholderia cepacia R34		120
NagG	Ralstonia sp. strain U2	Salicylate hydroxylase	423	100	1-120—304-423	17
ORF2	Burkholderia sp. strain DNT		423	100	1-120—304-423	68
OhbB	Pseudomonas aeruginosa JB2	o-Halobenzoate dioxygenase	421	81	1-118—304-423	Accession no. AAC69484
OhbB	Pseudomonas aeruginosa I42	o-Halobenzoate dioxygenase	428	36	3-119—294-410	77
IsbAa	Pseudomonas putida RE204	Isopropylbenzene dioxygenase	459	25	3-120—325-451	14
ORF2	Burkholderia cepacia R34		161
ORFX	Burkholderia sp. strain DNT		161	100	1-161—1-161	68
NagH	Ralstonia sp. strain U2	Salicylate hydroxylase	161	98	1-161—1-161	17
OhbC	Pseudomonas aeruginosa JB2	o-Halobenzoate dioxygenase	157	57	1-161—1-157	Accession no. AAC69484
OhbA	Pseudomonas aeruginosa I42	o-Halobenzoate dioxygenase	176	26	8-161—27-176	77
DntAb	Burkholderia cepacia R34	2,4-DNT dioxygenase	104
DntAb	Burkholderia sp. strain DNT	2,4-DNT dioxygenase	104	86	1-104—1-104	68
NtdAb	Pseudomonas sp. strain JS42	2-Nitrotoluene dioxygenase	104	86	1-104—104	48
NagAb	Ralstonia sp. strain U2	Naphthalene dioxygenase	104	84	1-104—1-104	17
NahAb	Pseudomonas putida 9816	Naphthalene dioxygenase	104	74	1-103—1-103	64
DntAc	Burkholderia cepacia R34	2,4-DNT dioxygenase	447
DntAc	Burkholderia sp. strain DNT	2,4-DNT dioxygenase	451	93	1-447—6-451	68
NagAc	Ralstonia sp. strain U2	Naphthalene dioxygenase	447	91	1-447—1-447	17
NtdAc	Pseudomonas sp. strain JS42	2-Nitrotoluene dioxygenase	447	90	1-447—1-447	48
NahAc	Pseudomonas putida 9816	Naphthalene dioxygenase	449	83	1-447—1-449	64
DntAd	Burkholderia cepacia R34	2,4-DNT dioxygenase	194
DntAd	Burkholderia sp. strain DNT	2,4-DNT dioxygenase	194	98	1-194—1-194	68
NagAd	Ralstonia sp. strain U2	Naphthalene dioxygenase	194	92	1-194—1-194	17
NtdAd	Pseudomonas sp. strain JS42	2-Nitrotoluene dioxygenase	194	92	1-194—1-194	48
NahAd	Pseudomonas putida 9816	Naphthalene dioxygenase	194	78	1-194—1-194	64
DntB	Burkholderia cepacia R34	4-Methyl-5-nitrocatechol oxygenase	560
DntB	Burkholderia sp. strain DNT	4-Methyl-5-nitrocatechol oxygenase	548	53	15-547—12-547	22
PheA	Plasmid pEST126	Phenol monooxygenase	610	31	5-546—24-598	46
TfdB	Pseudomonas putida (pEST401)	2,4-dichlorophenol hydroxylase	585	31	15-527—9-563	32
TfdBII	Ralstonia eutropha JMP134	Chlorophenol monooxygenase	586	30	13-547—7-584	39
ORF3	Burkholderia cepacia R34	Intradiol dioxygenase	278
HadC	Ralstonia pickettii DTP0602	Hydroxyquinol-1,2-dioxygenase	315	65	1-264—1-264	72
TfdC	Alcaligenes eutrophus	Chlorocatechol 1,2-dioxygenase	255	29	22-278—2-250	18
DntD	Burkholderia cepacia R34	Trihydroxytoluene oxygenase	314			28
DntD	Burkholderia sp. strain DNT	Trihydroxytoluene oxygenase	315	60	1-313—1-314	21
BphC	Pseudomonas stutzeri A401	2,3-Dihydroxybiphenyl dioxygenase	363	38	6-308—28-329	70
HpcA	Brucella melitensis	Homoprotocatechuate dioxygenase	331	25	14-308—30-309	11
XylE	Pseudomonas putida pWWO	Catechol dioxygenase	307	21	23-279—20-285	41
DntE	Burkholderia cepacia R34	Methylmalonate semialdehyde dehydrogenase	508
MmsA	Rattus norvegicus	Methylmalonate semialdehyde dehydrogenase	535	60	9-494—34-519	31
MmsA	Pseudomonas aeruginosa	Methylmalonate semialdehyde dehydrogenase	497	55	15-497—5-487	66
CmtC	Pseudomonas putida F1	p-Cumic aldehyde dehydrogenase	494	34	18-489—19-490	13
StyD	Pseudomonas sp. strain Y2	Phenylacetaldehyde dehydrogenase	502	31	16-487—22-498	78
ORF13	Burkholderia cepacia R34	Putative NAD(P)H-dependent reductase	188
ORF9	Burkholderia cepacia R34	Putative NAD(P)H-dependent reductase	189	62	5-184—7-186
LanZ4	Streptomyces cyanogenus	Putative reductase	200	38	42-156—55-163	80
SsuE	Pseudomonas putida S-313	NADH-dependent FMN reductase	200	29	5-135—3-133	Accession no. O85762
DntG	Burkholderia cepacia R34	2,4-Dihydroxy-5-methyl-6-oxo-2,4-hexadienoate isomerase/4-hydroxy-2-keto-5-methyl-6-oxo-3-hexenoate hydrolase	281
HpcE	Methanococcus jannaschii	2-Hydroxyhepta-2,4-diene-1,7-dioate (HHDD) isomerase	237	45	45-256—36-236	7
HpcE	Brucella melitensis	HHDD isomerase/5-carboxymethyl-2-oxo-3-hexene- 1,7-dioate decarboxylase	311	38	1-279—25-309	11
DitD	Pseudomonas abietaniphila	Putative isomerase	293	35	40-255—60-274	40
ORF10	Burkholderia cepacia R34	Putative maleylacetate reductase	379
TftE	Burkholderia cepacia AC1100	Maleylacetate reductase	352	65	8-338—4-332	10
ClcE	Pseudomonas sp. strain B13	Maleylacetate reductase	352	53	7-336—2-329	30
TfdF	Alcaligenes eutrophus (pJP4)	Maleylacetate reductase	354	52	16-338—12-332	52
ORF12	Burkholderia cepacia R34	Putative regulatory protein	301
NahR	Pseudomonas putida G7	Regulatory protein	300	61	1-300—1-300	63
OhbR	Pseudomonas putida JB2	Putative regulatory protein	310	57	1-301—1-301	Accession no. AAC69484

^aDeduced amino acid sequences of gene products from R34 were compared to previously described peptides using the BlastP search algorithm. Length indicates size of peptide (in amino acids). % Identity, percent of identical amino acids over the aligned region. Alignment, first range indicates portion of the R34-derived peptide and second range indicates portion of homolog used in alignment.

Additional terminal oxygenase genes.

Within the operon encoding the initial dioxygenase, three additional open reading frames (ORFs) were identified and designated ORF1a, ORF1b, and ORF2. Significant identity was shared between ORF1a/ORF1b and a group of terminal oxygenase α-subunit proteins (Table 2), in particular the salicylate hydroxylase from strain U2 (17). Alignment of the ORF1a and ORF1b amino acid sequences with their homologs revealed that the two peptides were similar to the N- and C-terminal portions of the oxygenase α-subunits, respectively. The result suggested that a frameshift mutation had disrupted the ORF following amino acid 295 and led to a truncated peptide at amino acid 302 (Table 2). The next ORF (ORF1b) begins just after the ORF1a stop codon and encodes a 120-amino-acid peptide. Again, the alignment indicated strong conservation of sequence (Table 2); however, it is questionable whether the peptide is synthesized. No putative ribosome-binding site is present upstream of the ATG start codon, though translational coupling could allow synthesis of the second peptide. The deduced amino acid sequence of ORF2 was similar to those of terminal oxygenase β-subunits. The relative identity shared with counterparts from other operons was like that of the ORF1a/ORF1b sequence (Table 2).

Assays were done with extracts from E. coli JM109(pJS332) to find whether ORF1a, ORF1b, and ORF2 encode a functional salicylate hydroxylase (Fig. 2). No transformation of salicylate to gentisate or reduction in the initial concentration of salicylate was found during the assays. The result is consistent with the frameshift mutation eliminating the function of the putative salicylate hydroxylase. Cells from the same cultures exhibited DNT dioxygenase activity, which indicated that the other gene products in the region were synthesized to yield a functional hydroxylating dioxygenase.

MNC monooxygenase and benzenetriol dioxygenase.

Functional mapping of the dnt gene cluster showed that the MNC monooxygenase gene (dntB) was downstream of the dntA gene cluster (Fig. 2). Analysis of the nucleotide sequence for this region revealed a 1,683-bp ORF that was designated dntB. The deduced amino acid sequence of DntB shared identity with FAD-dependent monooxygenases, including the MNC monooxygenase from strain DNT, as well as chlorophenol monooxygenases (Table 2). Although the overall identity is limited among this group of monooxygenases, the sequence motifs and residues that are associated with the hydroxylase activity are conserved in DntB. The sequence between residues 18 and 45 is similar to the βαβ fold responsible for ADP binding within flavin-containing monooxygenases (25, 81). Further comparisons indicated that the probable FAD-binding site (15, 25) was between amino acids 307 and 317. The region downstream from the FAD-binding motif retained the residues conserved among phenol hydroxylases (33).

Sequence comparisons showed significant differences between the MNC monooxygenases from strains DNT and R34, despite the fact that they catalyzed equivalent reactions in 2,4-DNT degradation. The deduced amino acid sequences share only 53% identity. Previous comparisons of the enzymes (45) showed that the MNC monooxygenase from strain R34 is not subject to the substrate inhibition found with the MNC monooxygenase from strain DNT (22). Another curious difference between the two MNC monooxygenases was found in the G+C content of the coding regions. Nucleotide sequence comparisons revealed that the G+C content of the genes for the two peptides was 57.5 and 71.8 mol% for strain R34 and strain DNT, respectively. The disparity in G+C content suggests that the two dntB genes evolved in very different hosts, although both were isolated from Burkholderia strains, a genus that has G+C compositions of 64 to 70%.

An 837-bp ORF (designated ORF3) lies downstream from dntB. The G+C content of ORF3 is 57.3%, very similar to that of neighboring dntB, which is consistent with the two genes evolving together in the same host. The deduced amino acid sequence of ORF3 shares significant identity with intradiol ring cleavage enzymes (Table 2). The greatest identity was shared with the hydroxyquinol (1,2,4-benzenetriol) dioxygenase from the chlorophenol degradation pathway of Ralstonia pickettii DTP0602 (72). Whole cells and cell extracts from E. coli JM109(pJS326) catalyzed the transformation of 1,2,4-benzenetriol but not the intermediate of the 2,4-DNT pathway, 2,4,5-trihydroxytoluene. The partially purified benzenetriol oxygenase was incubated with 1,2,4-benzenetriol and monitored spectrophotometrically for the appearance of product. The product spectrum had two absorbance maxima, a broad peak centered at 345 nm and the second at 240 nm, consistent with intradiol cleavage of 1,2,4-benzenetriol to yield maleylacetate (8). meta-ring fission of 1,2,4-benzenetriol yields a product with absorbance maxima of 340 and 280 nm (21). The results indicate that ORF3 encodes an intradiol ring fission enzyme that is not used in the degradation of 2,4-DNT.

Lower pathway genes.

A previous report described the characterization of the 2,4,5-THT oxygenase (ring cleavage enzyme) of the 2,4-DNT pathway. The THT oxygenase shares sequence identity with the family of meta-ring fission enzymes, including homogentisate dioxygenases from the phenylalanine pathway and ring fission dioxygenases that have an unknown degradative role in biphenyl-degrading bacteria (28, 70) (Table 2). E. coli JM109(pJS336) (Fig. 2), which contains the THT oxygenase gene (dntD) and the region downstream from dntD, grows on minimal media with 2,4,5-trihydroxytoluene as a sole carbon source. The result indicated that the DNA fragment cloned in pJS336 includes genes for enzymes that catalyze transformation of the ring fission product to compounds for intermediary metabolism in E. coli. The nucleotide sequence of the region downstream from dntD was determined and analyzed to provide insight into the metabolic steps following ring fission in the 2,4-DNT pathway. Five ORFs were identified downstream from dntD on pJS336 and were designated dntE, ORF13, dntG, ORF9, and ORF10 (Fig. 2).

The dntE gene product was similar to the CoA-dependent methylmalonate semialdehyde dehydrogenases from the valine degradation pathway in microbes and mammals (31, 66). The deduced peptide sequence contains the amino acid residues associated with NAD⁺ binding, the nicotinamide ring binding motif (GSTRVG) at amino acids 236 to 241, and the adenine ribose-binding residue (Glu-187). The catalytic glutamic acid and cysteine residues (Glu-274 and Cys-292) (26) lie within the region that matches the consensus sequence of the dehydrogenase active site (31).

The next ORF, ORF13, did not share significant identity with any previously identified proteins but appears to be distantly related to putative NADH-dependent reductases (Table 2). The biochemical role of the ORF13 homologs in Table 2 is not well defined. Their reported function appears to be based upon their sequence similarity to prototypical NADH-dependent reductases.

The 774-bp ORF that follows ORF13 was designated dntG. The deduced amino acid sequence of DntG shared significant identity with the class of fumarylacetoacetate (FAA) hydrolases from the phenylalanine/tyrosine degradation pathway and cycloisomerases (tautomerases) from various lower pathways (Table 2). The crystal structure and reaction mechanism of the FAA hydrolase is known (76). Interpretation of the consensus sequence for the DntG sequence aligned with several FAA hydrolase sequences revealed catalytically and structurally significant residues of DntG. The carboxyl oxygens from Asp56, Glu123, Glu125, and Glu154, should coordinately bind the calcium ion cofactor that is involved in substrate binding and catalysis. The spacing of the four acidic amino acids in DntG is perfectly conserved with the other hydrolases, although aspartate is usually found in the final position rather than glutamate, like in DntG. The FAA hydrolase class of enzymes uses a Glu-His-water triad in the reaction mechanism (76). The histidine serves as the general base to form the nucleophilic water molecule that is used in the reaction, and the carboxy group from glutamate stabilizes the histidine in its basic form. In the DntG sequence, His-82 lies within the active site predicted from sequence comparisons and should act as the general base in catalysis. The glutamine residue at position 161 of DntG is also expected to participate in stabilization of the nucleophilic water. The stabilizing glutamate residue of the triad is not obvious from sequence comparisons alone, but four conserved acidic residues in the C-terminal portion of the peptide, Asp236, Asp260, Glu265, and Glu267, are the best candidates for the role based on the crystal structure of FAA hydrolase.

The remaining ORFs in pJS336 lie 1.8 kb downstream from dntG (Fig. 2). ORF9 appears to be homologous with ORF13, sharing 69% identity over a 133-amino-acid region of the two peptides, greater than any other gene product in the public databases. ORF10 is 68 bp downstream from ORF9. ORF10 encodes a 452-amino-acid polypeptide that is similar to several maleylacetate reductases (Table 2). Maleylacetate reductases are typically found in ortho-ring fission pathways, such as those for degradation of chloroaromatic compounds and the proposed dioxin pathway (2). The anomalous association with the meta-ring fission enzyme and significant distance from dntG suggest that ORF9 and ORF10 are not associated with the 2,4-DNT pathway and instead are artifacts remaining from assembly of the region.

Subclones and deletion derivatives of pJS336 were made and used as sources for cell extracts in enzyme assays in order to find which of the gene products downstream from dntD catalyze the step following ring fission (Fig. 3). Cell extracts from strains that carried dntG catalyzed rapid disappearance of the 2,4,5-THT ring fission product in the absence of added cofactors. The putative aldehyde dehydrogenase encoded by dntE did not transform the intermediate despite addition of NAD or an NAD-regenerating system. Thus, oxidation of the aldehyde moiety to yield the corresponding dicarboxylic acid does not appear to follow ring fission as it does in some meta-ring fission pathways (19). The two putative NADH-dependent reductases were not involved in the step following ring fission, since the deletion of ORF13 or ORF9 (Fig. 3, pJS1363ΔXmaI and pJS1364ΔHindIII) had no apparent effect in the reaction. Similarly, no change was seen at A₃₅₀ with additions of cell extract from E. coli(pJS1364) or E. coli(pJS1364ΔHindIII), so the ORF10 gene product had no measurable activity against the THT ring fission product.

FIG. 3.

Localization of gene product that catalyzes transformation of 2,4-dihydroxy-5-methyl-6-oxo-2,4-hexadienoic acid. Physical map of the DNA insert in pJS336 and subclones derived from the region is shown. The column on the right side indicates whether cell extracts from the corresponding recombinant strain consumed the 2,4,5-THT ring fission product.

Further experiments were done to define the reaction catalyzed by DntG. The identity that DntG shares with isomerases and hydrolases suggested that DntG might display dual catalytic abilities like one of its homologs, the bifunctional HHDD isomerase/OPET decarboxylase from the homoprotocatechuate degradation pathway (58). The spectrophotometric scans of the reaction mixture containing the ring fission product and DntG support the dual catalysis hypothesis. There was an initial shift in λ_max of 288 to 283 nm, and then as the reaction progressed to completion, the compound at 283 nm was transformed to compounds with little UV absorbance (Fig. 4). The stepwise change and absence of an isosbestic point in the spectra indicated that the ring fission product is not directly converted to the final compounds.

FIG. 4.

Conversion of 2,4-dihydroxy-5-methyl-6-oxo-2,4-hexadienoic acid to pyruvate and methylmalonate semialdehyde. The sample and reference cuvettes contained 2 μl of cell extract from E. coli JM109(pJS333) (120 μg of protein) and 1 μl of cell extract from E. coli JM109(pJS1363ΔXmaI) (56 μg of protein). The sample cuvette contained 200 nmol of 2,4,5-THT, which was converted to 2,4-dihydroxy-5-methyl-6-oxo-2,4-hexadienoic acid by the THT oxygenase from E. coli JM109(pJS333) extract. Spectra were recorded prior to and 20, 40, 60, 80, 100, 120, 140, 160, 180, and 200 s after addition of E. coli JM109(pJS1363ΔXmaI) cell extract. The arrow indicates the first spectrum that was collected.

The products of the DntG-catalyzed reaction were analyzed using HPLC. Pyruvate (retention time, 6.1 min) was produced during the reaction in stoichiometric amounts from the ring fission product (Fig. 5). The results are consistent with previous work using strain DNT (21). The HPLC analysis revealed a second product that eluted as a broad peak at 8 min. We hypothesized that the second product was methylmalonic acid semialdehyde (2-methyl-3-oxo-propionic acid) (Fig. 6) by analogy to reactions catalyzed by the fumarylacetoacetate hydrolase of the tyrosine pathway and fumarylpyruvate hydrolase from the gentisate degradation pathway (83). Because hydrolysis of the ring fission product yields pyruvate, the remainder of the molecule should yield methylmalonic acid semialdehyde. The second compound coeluted with a methylmalonic acid semialdehyde standard and provided an identical spectrum (not shown). The rate at which the second compound accumulated was parallel to that of pyruvate (Fig. 5), consistent with it being a product of the hydrolysis reaction along with pyruvate. Similar results were found with cell extracts from strain R34 grown with 2,4-DNT.

FIG. 5.

Production of pyruvate and methylmalonate semialdehyde from 2,4-dihydroxy-5-methyl-6-oxo-2,4-hexadienoic acid by cell extracts of E. coli JM109(pJS1363ΔXmaI). The reaction was initiated with addition of cell extract (17 μg of protein/ml) to 200 μM substrate. The substrate was produced enzymatically from 2,4,5-THT with cell extracts from E. coli JM109(pJS333) (70 μg of protein/ml). Time course reactions were done in triplicate; the error bars show standard deviations for each time point. The methylmalonate semialdehyde quantitation is the peak area for the chromatogram showing absorbance at 240 nm (▴). ▪, pyruvate concentration (μM).

FIG. 6.

Proposed pathway for degradation of 2,4,5-THT in the 2,4-DNT pathway of B. cepacia R34. Compounds in brackets are proposed and have not been confirmed experimentally.

The methylmalonic acid semialdehyde that results from hydrolysis of 4-hydroxy-2-keto-5-methyl-6-oxo-3-hexenoic acid (Fig. 6) would be the expected substrate for the methylmalonic acid semialdehyde dehydrogenase homolog encoded by dntE. Oxidative decarboxylation of methylmalonic acid semialdehyde yields propionyl-CoA in the valine degradation pathway. The nucleotide sequence downstream of dntD from strain DNT (accession number AF076848) contains an incomplete ORF similar to dntE. The identical organization suggests that the dehydrogenase step is conserved in the pathways from both 2,4-DNT-degrading strains. The metabolism of methylmalonic acid semialdehyde in the 2,4-DNT pathway has not been confirmed experimentally.

Identification of ancillary ORFs.

The nucleotide sequence of the region lying between dntD and the boundary of pJS325 (Fig. 2) contains four ORFs (ORF5 to ORF8) encoding peptides that shared significant similarity to ATP-binding cassette (ABC) transport proteins (16). ORF5 and ORF6 appear to be the channel-forming polypeptides based on their similarity to proteins from ABC transport systems and hydropathy plots of the amino acid sequences. ORF7 and ORF8 resemble ATPase components of ABC transport systems, and the amino acid sequence of each contains the well-conserved ATP-binding motif (79) and significant identity to the hydrophilic region of ATPase proteins from other ABC transport systems. No sequence encoding a potential periplasmic protein for solute binding, another gene product often associated with operons encoding ABC transport systems (42), was present upstream of ORF5.

A divergently transcribed ORF lies 120-bp upstream from dntAa (Fig. 2) that is similar to members of the LysR family of transcriptional regulators (61). The location and sequence similarity suggests that the gene product acts as a transcriptional regulator of the dntA genes, but adequate characterization has not been done. The ORF was designated ORF12. ORF12 and NahR from different naphthalene-degrading strains (6, 62) share about 60% identity (Table 2). Similar LysR-type regulators are found adjacent to the dioxygenase operons of strain DNT (accession number AF076848), U2 (83), JS42 (48), and JS765 (D. J. Lessner, R. E. Parales, and D. T. Gibson, Abstr. Am. Soc. Microbiol. Conf. Biodegradation Biotransformation Biocatalysis, 2001, abstr. 53, p. 38). The 2,4-DNT dioxygenase specific activity increases in strain R34 when 2,4-DNT or salicylate is present in minimal succinate media. The induced activity level suggests that transcription of the dioxygenase genes is positively regulated, but the ORF12 gene product has not been directly implicated in the regulation.

A 1,422-bp ORF adjacent to ORF12 was designated ORF11. The deduced amino acid sequence of ORF11 shared identity with methyl-accepting chemotaxis proteins, such as CheC, the aspartate chemoreceptor from E. coli. The highest degree of identity is in the peptide sequence that contains the conserved signaling domain (amino acids [aa] 288 to 332), which interacts with other proteins of the chemotaxis complex (5, 67). Four potential methylation and demethylation sites were revealed (aa 223, 228-229, 235-236, and 417-418) from interpretation of multi-sequence alignments (74). The region predicted to contain the sensor domain of the naphthalene chemoreceptor (NahY) from Pseudomonas putida G7 (20) shares no significant similarity with the putative receptor encoded by ORF11.

Two regions of DNA sequence within the cloned fragment are related to insertion sequence elements. ORF4, downstream from the benzenetriol oxygenase gene (Fig. 2), shares limited identity with transposases from various bacteria (Table 2) (4, 84). The region including nucleotides 1 to 2160 is related to the IS21 family of insertion sequences (55). Two imperfect direct repeats (22 out of 25 identical) lie in the 5′ end of the homologous region (nucleotides 2133 to 2156 and 2131 to 2107). The region does not encode a complete transposase; however, the deduced amino acid sequence from nucleotides 1802 to 447 shared 63% identity to the central portion of the transposase (IstA) from IS1631 (27). The deduced amino acid sequence for nucleotides 414 to 2 is similar to that of the N-terminal portion of the ATP-dependent binding protein (IstB) from IS1631 (71% identity) (27), but a frameshift mutation seems to have introduced a stop codon at nt 321 to 319.

Analysis of operon organization by RT-PCR.

RT-PCR was used to determine which of the colinear genes found in the region were transcribed as operons (Fig. 7). The results showed that the genes for the initial hydroxylating dioxygenase (dntAa-dntAd) and the methylnitrocatechol monooxygenase (dntB) were not cotranscribed. No product was detected using the dntAdB_up and dntAdB_down primers. The control reaction using primers within dntAc provided the expected 501-bp product. In the lower pathway, the genes encoding the meta-ring cleavage dioxygenase (dntD) and methylmuconic acid semialdehyde dehydrogenase (dntE) were cotranscribed and the primers corresponding to the two regions allowed detection of the anticipated cDNA. Likewise, the dntE_dntG primers allowed detection of the corresponding cDNA. The results indicate that dntD is the initial gene of the lower pathway operon that comprises four identified genes. ORF8 and dntD were not cotranscribed, so the dntDEFG transcript must originate from a promoter lying between dntD and ORF8 (Fig 2).

FIG. 7.

Summary of RT-PCR experiments. Gray bars below map represent potential amplicons resulting from the RT-PCRs. X's crossed through gray boxes indicate primer pairs that did not provide products in RT-PCRs. Bold arrows indicate transcripts deduced from sequence analysis, protein assays, and RT-PCR results.

DISCUSSION

Oxidative growth with 2,4-DNT requires two novel oxygenases to catalyze the hydroxylation reactions that lead to release of nitro substituents from the aromatic ring. The oxygenolytic removal of the nitro substituent is electrochemically similar to oxygenolytic dehalogenation. However, the 2,4-DNT dioxygenase is closely related to the naphthalene dioxygenases, enzymes that have not demonstrated dehalogenation or denitration activity in previous investigations (56). The second step of the 2,4-DNT pathway also involves an oxygenolytic denitration reaction when methylnitrocatechol is transformed to 2-hydroxy-5-methylquinone. The second reaction is analogous to those catalyzed by the similar chlorophenol hydroxylases that remove the electron-withdrawing halogen during hydroxylation. The flavin-containing pentachlorophenol monooxygenase (PcbB) from strain UG30 catalyzes oxygenolytic denitration of 4-nitrocatechol to yield 1,2,4-benzenetriol (38). The similarities between the monooxygenase-catalyzed reactions are intriguing considering the presence of the benzenetriol oxygenase gene (ORF3) adjacent to dntB (Fig. 2). However, PcbB and DntB share little sequence identity (<22%) and there is no ORF3 homolog near pcbB. Therefore, the biochemical analogy is not matched by strong similarities in the gene sequences.

The basic organization is identical among the naphthalene and nitroarene dioxygenase operons that have been characterized (Fig. 8). The operons encoding archetypal naphthalene dioxygenases, represented by the nah genes of P. putida 9816, lack the ORF1 and ORF2 homologs found in the nag operon (17) and the operons that encode nitroarene dioxygenases (37, 48, 68). In strain U2, nagG and nagH encode the α and β subunits of the salicylate hydroxylase in the naphthalene pathway. The oxygenase components use the ferredoxin and oxidoreductase proteins from the operon to yield the functional salicylate hydroxylase (17). In each of the nitroarene dioxygenases described to date (2,4-DNT dioxygenase (strain DNT [68]), 2-nitrotoluene dioxygenase (strain JS42 [48]), nitrobenzene dioxygenase (strain JS765 [37]) and the 2,4-DNT dioxygenase from strain R34), there are apparent homologs of one or both oxygenase subunits of the salicylate hydroxylase from strain U2 (Fig. 8). However, in each case, mutations have eliminated enzyme function. In strain R34, a frameshift mutation truncated ORF1a compared to the functional homolog from strain U2. ORF2 from strain DNT yields a gene product comparable in size to the functional homolog, but two frameshift mutations lead to a 20-amino-acid region of dissimilarity. The homolog from strain JS42 is also truncated compared to NagG, and the β-subunit gene is absent in JS42 (Fig. 8). In the archetypal naphthalene dioxygenases, such as those from strain G7 or 9816 (64), there is no remnant of the additional oxygenase proteins. The inactivation and deletion of the additional ISP genes suggest that the nitroarene dioxygenase operons are evolving towards elimination of the extraneous oxygenase subunit genes to yield a compact operon. If the suggestion is true, then it follows that the nag operon is the progenitor for both the archetypal naphthalene dioxygenases and the nitroarene dioxygenases described to date. Another possibility that has been suggested is that oxygenase genes for the salicylate hydroxylase were acquired through an insertion event to provide the alternate naphthalene pathway encoded by the nag gene cluster, and the nitrotoluene dioxygenases then diverged from the nag genes (17).

FIG. 8.

Genetic organization of various multicomponent dioxygenases related to the 2,4-DNT dioxygenase from strain R34.

The regions downstream from the initial hydroxylating dioxygenases of the naphthalene and nitrotoluene operons differ considerably (Fig. 8). A gene encoding cis-dihydrodiol dehydrogenase follows each of the naphthalene dioxygenase gene clusters. The sequence for the 2-nitrotoluene dioxygenase from strain JS42 is also followed by an ORF similar to the genes encoding dihydrodiol dehydrogenases, but it does not yield a functional enzyme (48). A putative transposase borders the 3′ end of the 2,4-DNT dioxygenase from strain DNT (Fig. 8). Strain R34 appears to have a more efficient organization than the other characterized nitroarene-degrading strains. The gene adjacent to the dioxygenase operon encodes the enzyme used in the next step of the pathway rather than an unnecessary dehydrogenase.

Although the 2,4-DNT degradation pathways of strains R34 and DNT are identical, the sequence and organization of genes encoding the enzymes that catalyze the reactions following initial hydroxylation differ considerably. The disparity in sequence and G+C content of the two dntB genes opposes the hypothesis that recent horizontal gene transfer from a common ancestor contributed to the ability to degrade 4-methyl-5-nitrocatechol. In strain R34, the monooxygenase is linked to the initial dioxygenase although the genes for the two enzymes are not cotranscribed. For strain DNT, the relative locations of the dntA and dntB are not known but it is clear that the two clusters are separated by more than 8 kb (69); furthermore, the regions downstream of the dntB genes do not share significant similarity. Likewise, the sequence, G+C content, and organization of the extradiol ring cleavage dioxygenases (DntD) from strain R34 and strain DNT are very different. One gap remains in the genetics of 2,4-DNT degradation. The gene encoding the methylhydroxyquinone reductase for the 2,4-DNT pathway has not been identified for either strain.

The two regions resembling portions of insertion sequences might be associated with initial positioning and assembly of the dnt structural genes, subsequent failed transposition events, or remnants from insertion into another transposon. The transposase sequences found in the region (Fig. 2) shared no significant identity, indicating that the elements represent independent transposition events since it is rare that individual arms of transposable elements carry nonhomologous transposase genes. Preliminary characterization showed that the 2,4-DNT pathway was spontaneously lost from strain R34 under nonselective conditions (28). Only a portion of the machinery necessary for transposition was identified, so the instability and deletion of the phenotype in strain R34 was not likely to be mediated by either of the described regions but by elements outside of the sequenced DNA fragment.

Parke et al. (49) suggest that genetic characterization of catabolic pathways will uncover associations between catabolic gene clusters and transport systems that play a role in degradative pathways. The suggestion was supported in the characterization of the catabolic plasmid, pNL1, from Sphingomonas aromaticus F199 (57). In pNL1, several putative membrane proteins homologous to transport proteins were identified and proposed to play a role in aromatic compound catabolism. The putative ABC transport system formed from the ORF5 to ORF8 gene products might be related to nitroarene catabolism. Alternatively, the transport system could have been recruited with the lower pathway operon, since each gene cluster has homologs in amino acid transport or degradation. However, since genes encoding ABC transport proteins compose as much as 5% of a prokaryotic genome (51), it is also reasonable that the close association is a coincidence and unrelated to 2,4-DNT transport and degradation.

The dissimilarity in the receptor regions (periplasmic domains) of ORF11 and the naphthalene chemoreceptor is incongruous with the high identity shared by the naphthalene and DNT dioxygenases. The difference suggests that ORF11 evolved for chemotaxis to another compound and is not a remnant from an earlier naphthalene-sensing cascade. There are two possibilities: the ORF11 gene product has evolved to mediate chemotaxis toward nitroarenes and enhance their biodegradation or, once again, its association with the dnt genes is coincidental.

Regulation of gene expression might be an important adaptation for bacteria to grow with 2,4-DNT, but the mechanism that strain R34 uses to control expression of the multiple transcripts is not known. The juxtaposition of the LysR-type regulatory protein (ORF12) found upstream from the dntA genes suggests that it might regulate their transcription but has not been characterized adequately. No additional regulatory genes were found in analysis of the regions around the methylnitrocatechol monooxygenase or lower pathway genes. Transcription of the other operons could be unregulated, controlled by other global regulators, or dependent on the promoter/operator region of each operon.

Inferences from the comparison of the structural genes of the 2,4-DNT pathway suggest that the pathway came together from three sources. The initial dioxygenase appears to have originated from a naphthalene degradation pathway like that of strain U2 (17). A large portion of the salicylate hydroxylase oxygenase component is retained but is not functional. The MNC monooxygenase was probably derived from a pathway for degradation of chloroaromatic compounds. The presence of the vestigial (with respect to 2,4-DNT degradation) ortho-ring fission dioxygenase is consistent with its recruitment from a pathway for chloroaromatic compounds. The true ring fission enzyme for 2,4-DNT degradation has a different origin. The sequence of DntD is quite dissimilar to all other described meta-ring fission enzymes, including those from naphthalene and chloroarene degradative pathways. The distinctive sequence of the ring cleavage enzyme reflects the substrate specificity observed for the THT oxygenase (28). The distant relationship between homogentisate dioxygenase and DntD and the association with homologs from amino acid metabolism (dntE and dntG) indicate that the lower pathway operon arose from a gene cluster for amino acid degradation.

The disparate origins of the various dnt and associated genes described in this study are consistent with the difficulties that bacteria face to achieve efficient metabolism of synthetic compounds like 2,4-DNT. The organization of the pathway genes suggests there is a progression towards a compact region encoding the entire pathway. In that progression, remnants from assembly persist, such as the benzenetriol oxygenase (ORF3), putative maleylacetate reductase (ORF10), and putative transposase (ORF4). No role in nitroarene degradation is apparent for the remnants; their presence might indicate an intermediate point in the evolution of an optimal system or perhaps some of the proteins could be used in other pathways when another substrate is available.