Iron-Dependent Enzyme Catalyzes the Initial Step in Biodegradation of N-Nitroglycine by Variovorax sp. Strain JS1663

ABSTRACT

Nitramines are key constituents of most of the explosives currently in use and consequently contaminate soil and groundwater at many military facilities around the world. Toxicity from nitramine contamination poses a health risk to plants and animals. Thus, understanding how nitramines are biodegraded is critical to environmental remediation. The biodegradation of synthetic nitramine compounds such as hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) has been studied for decades, but little is known about the catabolism of naturally produced nitramine compounds. In this study, we report the isolation of a soil bacterium, Variovorax sp. strain JS1663, that degrades N-nitroglycine (NNG), a naturally produced nitramine, and the key enzyme involved in its catabolism. Variovorax sp. JS1663 is a Gram-negative, non-spore-forming motile bacterium isolated from activated sludge based on its ability to use NNG as a sole growth substrate under aerobic conditions. A single gene (nnlA) encodes an iron-dependent enzyme that releases nitrite from NNG through a proposed β-elimination reaction. Bioinformatics analysis of the amino acid sequence of NNG lyase identified a PAS (Per-Arnt-Sim) domain. PAS domains can be associated with heme cofactors and function as signal sensors in signaling proteins. This is the first instance of a PAS domain present in a denitration enzyme. The NNG biodegradation pathway should provide the basis for the identification of other enzymes that cleave the N—N bond and facilitate the development of enzymes to cleave similar bonds in RDX, nitroguanidine, and other nitramine explosives.

IMPORTANCE The production of antibiotics and other allelopathic chemicals is a major aspect of chemical ecology. The biodegradation of such chemicals can play an important ecological role in mitigating or eliminating the effects of such compounds. N-Nitroglycine (NNG) is produced by the Gram-positive filamentous soil bacterium Streptomyces noursei. This study reports the isolation of a Gram-negative soil bacterium, Variovorax sp. strain JS1663, that is able to use NNG as a sole growth substrate. The proposed degradation pathway occurs via a β-elimination reaction that releases nitrite from NNG. The novel NNG lyase requires iron(II) for activity. The identification of a novel enzyme and catabolic pathway provides evidence of a substantial and underappreciated flux of the antibiotic in natural ecosystems. Understanding the NNG biodegradation pathway will help identify other enzymes that cleave the N—N bond and facilitate the development of enzymes to cleave similar bonds in synthetic nitramine explosives.

INTRODUCTION

N-Nitroglycine ([NNG] nitraminoacetic acid) is a nitramine natural product of the soil bacterium Streptomyces noursei (1) that is toxic to plants (2), mice (3), and Gram-negative bacteria (1). NNG interferes with the Krebs cycle by competitively inhibiting succinate dehydrogenase (3). While over 200 naturally occurring nitro compounds have been identified (4), natural N-nitrated compounds have been reported only rarely, and nothing is known about their biosynthesis or biodegradation (4). NNG was discovered in 1968 and was the first reported natural nitramine (1). Other nitramine natural products include β-nitramino-l-alanine and N-nitroethylenediamine (NEDA) isolated from the mushrooms Agaricus silvaticus (5) and Agaricus subrutilescens (6), respectively. These three nitramines are structurally similar (7) and are suspected mutagens (8). They also are structural analogs of the synthetic energetic materials nitroguanidine and dinitrourea. Agaricus subrutilescens also produces two other nitramines, l-4-nitramino-2-aminobutanoic acid and the γ-glutamyl peptide of NEDA (6, 9), but little is known about their biological activities. Synthetic cyclic nitramines such as hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX), octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX), and 2,4,6,8,10,12-hexanitro-2,4,6,8,10,12-hexaazaisowurtzitane (CL-20) are explosives used by the military in high-yield munitions. RDX, HMX, and CL-20 are toxic to biological systems and human health (10), and RDX is also listed as a possible human carcinogen by the U.S. EPA (10). Other synthetic nitramines include dinitrourea and nitroguanidine. Nitroguanidine is used as an insensitive military ammunition, a propellant for modern airbags, and an insecticide (12, 13). These toxic and recalcitrant chemicals enter the environment through the discharge of waste waters from manufacturing processes (14). Synthetic nitramines have been shown to be degraded or transformed by sludges, mixed cultures, or specific isolates (10, 15–26). The products of secondary nitramine degradation are often primary nitramines, similar to NNG. For example, a cytochrome P450 system catalyzes the initial reaction in bacteria that use RDX as the sole source of nitrogen (27–33). The enzyme cleaves nitro groups from RDX in a pathway that releases nitrite and yields linear nitramine products.

Variovorax sp. strain JS1663 was isolated for its ability to grow using NNG as the sole carbon, nitrogen, and energy sources. Experiments were performed to identify the products of NNG catabolism and to establish the biodegradation pathway. The initial step is a previously unreported β-elimination reaction catalyzed by an iron-dependent enzyme that has an amino acid sequence highly divergent from previously characterized enzymes. Enzymes involved in the catabolism of natural nitramines may be of importance in the adaptation of microbes to degrade xenobiotics. In addition, the synthesis and biodegradation of nitramines could play an important role in soil chemical ecology.

RESULTS

Isolation and identification of NNG-degrading bacteria.

Selective enrichment with NNG as the sole carbon source yielded an isolate that grew on NNG as the sole source of carbon and nitrogen The 16S rRNA gene sequence of strain JS1663 is most similar to that of Variovorax sp. strain RA8 (99% identity over 1,526 nucleotides [nt]) (34), Variovorax defluvii strain 2C1-b (99% identity over 1,498 nt) (35), and Variovorax paradoxus strain B4 (98% identity over 1,543 nt) (36). However, the JS1663 and V. paradoxus genome sequences share only 83% average nucleotide identity, which suggests that the strains belong to different species within the Variovorax genus (37).

Growth on NNG and preliminary experiments with cell extracts.

Strain JS1663 released stoichiometric amounts of nitrite during aerobic growth on NNG as the source of carbon (Fig. 1). The disappearance of NNG and release of nitrite were complete before growth began, which indicated that an unidentified denitrated intermediate accumulated and served as the growth substrate. Since nitrite accumulated during the disappearance of NNG, we hypothesized that the initial cleavage reaction might be between the amino and nitro groups, yielding glycine as an intermediate. However, glycine supported only weak growth (data not shown). The isolate grew well with succinate as the carbon source and NNG as the nitrogen source. To assess the growth kinetics, cells were grown on succinate with ammonium chloride as the nitrogen source, and were washed and suspended in medium with NNG as the growth substrate. NNG degradation was initially undetectable, but began after a 1- to 2-h acclimation period (data not shown), which indicated that the enzyme(s) involved in the denitration of NNG was inducible. Intact cells from several experiments catalyzed the release of nitrite from NNG with specific activities ranging from 15.0 to 41.2 nmol · min⁻¹ · mg⁻¹ protein. Extracts prepared from NNG-grown cells catalyzed the release of nitrite from NNG with specific activities of 8.0 to 10.2 nmol · min⁻¹ · mg⁻¹ protein. The results obtained from measurements of nitrite release were similar to the rates of NNG disappearance. The optimal pH was 7.0, and no added cofactors were necessary for the activity. The activity was abolished upon dialysis and restored by the inclusion of ferrous chloride (0.2 mM) in the reaction mixture.

FIG 1.

Growth of Variovorax sp. strain JS1663 on NNG as the sole carbon, nitrogen, and energy sources. Squares, optical density at 600 nm (right axis); circles, nitrite concentration (μM); and diamonds, NNG concentration (μM) (left axis).

Fosmid library and sequencing.

Four strains from the Escherichia coli library of 576 fosmid clones produced nitrite in the presence of NNG. The position of the sequences in strain JS1663 was determined by using the Basic Local Alignment Search Tool (BLAST). The four clones overlapped in a 28-kbp sequence that comprised 34 coding DNA sequences (CDS). One gene (A8M77_12210) encoded a putative glutaminase protein. Two other genes were predicted to encode acyl coenzyme A (acyl-CoA) dehydrogenase (A8M77_12295) and CoA-transferase (A8M77_12290) enzymes. Finally, two adjacent genes were predicted to encode a hypothetical PAS-domain protein (A8M77_12345) and a C4-dicarboxylase transporter (A8M77_12340) (Fig. 2).

FIG 2.

Map of the gene cluster encoding NNG lyase. GenBank locus A8M77_12355 encodes a hypothetical protein, A8M77_12350 encodes a YjgF superfamily protein, A8M77_12345 (nnlA) in black encodes the NNG lyase, A8M77_12340 encodes a putative C₄-dicarboxylate transporter, A8M77_12335 encodes a YolD-like protein, A8M77_12330 encodes a LysR-type regulatory protein, and A8M77_12325 encodes a thioesterase protein.

Identification of the gene encoding NNG lyase.

Three gene clusters were subcloned from fosmid 8G8 into E. coli expression vectors containing the P_BAD arabinose-inducible promoter and an N-terminal hexahistidine affinity tag (see Table 2). Cells expressing the three gene clusters were incubated with NNG, and only cells containing the pDG698 vector produced nitrite. To determine which of the two JS1663 genes heterologously expressed by this strain was required for NNG cleavage, the A8M77_12345 gene was independently expressed using vector pDG708. E. coli cells containing pDG708 released nitrite from NNG, which established the identity of the N-nitroglycine lyase (nnlA) gene. The heterologously expressed NNG lyase catalyzed the transformation of NNG with concomitant release of nitrite, consistent with the activity in the wild-type organism.

TABLE 2.

Strains isolated or constructed in this work

Strain	Description
Variovorax sp. JS1663	NNG-degrading isolate
E. coli EPI300-T1R(pCC1Fos-8G8)	JS1663 fosmid 8G8
E. coli EPI300-T1R(pCC1Fos-7C1)	JS1663 fosmid 7C1
E. coli EPI300-T1R(pCC1Fos-4G10)	JS1663 fosmid 4G10
E. coli EPI300-T1R(pCC1Fos-4C1)	JS1663 fosmid 4C1
E. coli NEB5α(pBAD-HisA)	Vector control strain (plasmid from Invitrogen)
E. coli NEB5α(pDG692)	JS1663 A8M77_12210 gene in pBAD-HisA
E. coli NEB5α(pDG695)	JS1663 A8M77_12295 to A8M77_12290 genes in pBAD-HisA
E. coli NEB5α(pDG698)	JS1663 A8M77_12345 to A8M77_12340 genes in pBAD-HisA
E. coli NEB5α(pDG708)	JS1663 A8M77_12345 in pBAD-HisA

Purification of the NNG lyase.

Recombinant Variovorax sp. JS1663 NnlA was expressed in E. coli DH5α as an N-terminal His₆ fusion protein and purified using affinity chromatography. NnlA was produced at a high level in E. coli, yielding ∼20 mg of >95% pure protein per liter of culture. The molecular mass was estimated to be 19 kDa using SDS-PAGE, which was similar to its predicted mass of 20.2 kDa (pI = 5.8).

Cofactors, stoichiometry, and substrate specificity.

The purified NnlA protein catalyzed the stoichiometric conversion of NNG to glyoxylate and nitrite (Fig. 3) with a specific activity of 1,185 ± 16 nmol · min⁻¹ · mg⁻¹. The results provide strong evidence that the degradation mechanism involves a β-elimination reaction that releases nitrite from NNG (Fig. 4). No cofactor was required when reactions were performed with whole cells or cell extracts (data not shown), but when purified protein was used as a catalyst, only the reaction mixtures containing ferrous ammonium sulfate or ferrous chloride released nitrite. The requirement for iron(II) indicates that the protein is an iron-dependent enzyme. The enzyme was not active in the presence of other metals, including zinc, calcium, cobalt, nickel, iron(III), copper, magnesium, and manganese, or in the presence of cofactors such as flavin adenine dinucleotide (FAD), flavin mononucleotide (FMN), NAD, or NADH (data not shown). Glyoxylate production was also confirmed by high-resolution liquid chromatography-mass spectrometry (LC-MS) with comparison to an authentic standard. The reaction product produced a peak at 72.9929 m/z (mass error, ∼4.5 ppm), which corresponds to the glyoxylate anion. The enzyme was not active against 4-nitro-2,4-diazabutanal, NEDA, RDX, HMX, or nitroguanidine, which suggests that it is specific for NNG. The above compounds also did not inhibit lyase activity in the presence of NNG.

FIG 3.

Transformation of NNG by purified NnlA JS1663. Squares, N-nitroglycine concentration (μM) (left axis); circles, nitrite concentration (μM); and triangles, glyoxylate phenylhydrazone concentration (μM) (right axis). NnlA protein purified from cells of E. coli NEB 5α(pDG708) (see Table 2) was incubated with NNG and Fe(II) at 30°C. All assays were performed in duplicates with a standard deviation less than 10%.

FIG 4.

Proposed β-elimination reaction in the biodegradation of NNG.

Sequence analysis of NNG lyase.

The Variovorax sp. JS1663 NnlA is distantly related to several proteins from diverse bacteria (Fig. 5), but none have been previously characterized by experimental analysis. The nnlA gene is not conserved in the genomes of other characterized Variovorax spp. NnlA is the first example of a denitration enzyme with the PAS (Per-Arnt-Sim) domain structure (38). The NnlA homologs all belong to the PAS superfamily, but their functions are unknown. Most proteins with PAS domains function as ligand binding sensors, which are often coupled with signaling enzymes to effect signal transduction (39, 40). Although NnlA requires iron(II) for activity, it does not appear to have any known iron binding motifs in the amino acid sequence.

FIG 5.

Phylogeny of proteins related to NNG lyase from disparate bacteria. No function has been previously attributed to any of these PAS superfamily proteins. The scale bar indicates 0.3 amino acid replacement per position. Numbers associated with branches indicate bootstrap support percentages.

The nnlA gene in strain JS1663 is associated with a cluster of 7 genes transcribed in the same orientation, potentially constituting an operon (Fig. 2). The first gene encodes a hypothetical protein that is not conserved in the genomes of other strains containing genes similar to nnlA. The second gene (A8M77_12350) encodes a 154-amino-acid YjgF superfamily protein that is conserved in Microbispora rosea. One member of this largely uncharacterized superfamily, RidA, catalyzes the hydrolytic deamination of an imine (41). The third gene is nnlA (A8M77_12345). The fourth gene encodes a 427-amino-acid putative C₄-dicarboxylate transporter (A8M77_12340) with widespread homologs among bacteria, but no Pseudovibrio homolog. It might be involved in transporting NNG into the cell, since NNG is a C₄-dicarboxylate analog. However, it was not required for NNG lyase activity in the expression experiments described above. The fifth gene encodes a YolD-like protein that is not conserved in other nnlA-bearing organisms. The sixth gene encodes a LysR-type regulatory protein, which has numerous homologs in other bacteria. Finally, the seventh gene encodes a thioesterase protein with numerous homologs. Therefore, only the YjgF-type protein and NnlA have congruent distributions in other bacterial genome sequences.

Because glyoxylate is a product of NNG degradation, the JS1663 genome was examined for the presence of genes that encode homologs of the enzymes required for glyoxylate metabolism in E. coli (42). Genes consistent with glyoxylate metabolism were readily identified. The first three are in a single gene cluster: glyoxylate carboligase (gcl) (A8M77_03645), tartronate semialdehyde reductase (gxlR) (A8M77_03635), hydroxypyruvate isomerase (hyi) (A8M77_03640), and glycerate kinase (glxK) (A8M77_27040). The nnlA gene (A8M77_12345) is not located nearby. Surprisingly, Variovorax sp. JS1663 exhibited negligible growth on glyoxylate as the sole carbon source (data not shown), perhaps due to transport limitations.

DISCUSSION

Variovorax sp. JS1663 can utilize the natural nitramine NNG as the sole nitrogen, carbon, and energy sources. Results with wild-type cells indicate that the initial reaction catalyzed by the inducible NNG lyase (NnlA) eliminates nitrite from NNG. Heterologously expressed nnlA encodes an enzyme that catalyzes the same reaction in whole cells, crude cell lysates, and purified protein. No paralogs of nnlA were detected in the genome of JS1663, so the above results provide strong evidence for the physiological role of NnlA. Knockouts will be required to determine other potential functions of the enzyme. The most likely mechanism is a β-elimination reaction that proceeds by the abstraction of a proton alpha to the glycine carboxyl group and the elimination of the nitro group from NNG as nitrite, which is a good leaving group. The other product of the elimination reaction is presumed to be glyoxylate imine (Fig. 4). The reaction is analogous to a number of enzyme-catalyzed β-elimination reactions, such as those catalyzed by fumarase (43) and aconitase (44). Additional characterization of the structure of the enzyme will be required to elucidate the details of the reaction mechanism. The imine is likely to undergo spontaneous hydrolysis to glyoxylate and ammonia in vitro. The presence of the A8M77_12350 gene in the same operon as the gene encoding NnlA suggests that the hydrolysis of the glyoxylate imine may be under enzymatic control of an enamine deaminase-related protein in vivo, but it is clear from experiments with purified enzyme that the reaction can proceed spontaneously as well.

Purified NnlA requires iron(II) for denitration activity, but it does not seem to contain a known motif for iron-sulfur cluster binding. The lack of stimulation by reductants such as ascorbate or dithiothreitol suggested that the metal is not redox active. While further studies will be required to identify the role of the loosely bound iron(II) cofactor, several possible roles can be envisioned. The iron could bind the NNG carboxylate, stabilizing the Michaelis complex and enforcing substrate selectivity. Alternatively, iron could bind the nitramine group, facilitating the elimination of nitrite and the hydrolysis of the imine intermediate as proposed for nitrile hydratase enzymes (45, 46). Finally, the iron could coordinate with the nitrite leaving group to facilitate its expulsion, similar to the elimination step in aconitase (47). The low specific activity of the purified enzyme indicates that further efforts are required to stabilize the enzyme, to optimize reaction conditions, and to understand its interactions with iron(II).

The degradation of the synthetic nitramine RDX is catalyzed by the cytochrome P450 XplA and its associated reductase, XplB (27, 33). These electron transfer proteins catalyze the reductive denitration of RDX, resulting in subsequent ring cleavage and the formation of nitrous oxide, ammonia, formaldehyde, and a linear nitramine 4-nitro-2,4-diazabutanal (18, 33, 48). The degradation of the nitramine ring cleavage product, nitramine 4-nitro-2,4-diazabutanal (18, 28, 49), is proposed to involve amidohydrolases from the purine catabolic pathway (49). The above mechanisms are completely different from that of NNG lyase degradation. NnlA is not a heme-containing oxygenase and does not require additional electron donors or other electron transfer components to catalyze the reaction.

This preliminary report establishes the biodegradation of NNG and the identity of the enzyme that catalyzes the initial reaction(s). Understanding the biodegradation pathway and characterizing the NnlA enzyme reported here could be useful in the development of enzymes to cleave similar bonds in RDX, HMX, nitroguanidine, or dinitrourea. The detailed mechanism of the initial reaction, the role of iron in the reaction, and the nature of the intermediate that accumulates in the medium during the initial stages of growth are under investigation. The ecological role of the bacteria and their limitation of the effects of NNG on targeted organisms will be the subject of future studies.

MATERIALS AND METHODS

Chemicals.

NNG was from AKos Consulting & Solutions (Deutschland GmbH, Germany). Glyoxylic acid (97%), RDX, HMX, and nitroguanidine were from Sigma-Aldrich (St. Louis, MO). Phenylhydrazine hydrochloride (98.5%) was from Tokyo Chemical Industry (Tokyo, Japan). NEDA and 4-nitro-2,4-diazabutanal were synthesized as previously described (49, 50).

Enrichment, growth, and whole-cell biotransformations of NNG-degrading bacteria.

Variovorax sp. JS1663 was obtained by selective enrichment from activated sludge provided by Holston Army Ammunition Plant based on its ability to grow on NNG as the sole nitrogen, carbon, and energy sources. The sludge samples (10 ml) were inoculated into one-quarter-strength minimal salts medium ([MSB] 40 ml) (51) supplemented with NNG (100 μM) as the sole carbon source and were incubated at 30°C with shaking. The disappearance of NNG was monitored by high-performance liquid chromatography (HPLC). After 3 additions of NNG (100 μM), the samples were serially diluted in MSB containing NNG (500 μM) and incubated at 30°C with shaking. Following the disappearance of NNG, samples were serially diluted and spread on one-quarter-strength Trypticase soy agar (TSA). Colonies that appeared after 4 days of incubation and showed phenotypic differences were tested for the ability to degrade NNG (400 μM) in one-quarter-strength MSB. Isolates were routinely grown in one-quarter-strength TSA or one-quarter-strength MSB supplemented with NNG (400 μM) at 30°C and pH 7.0. Growth curves were initiated by the addition of the isolate at an optical density at 600 nm (OD₆₀₀) of 0.001 to 0.005. The wild-type whole-cell biotransformations were carried out at 30°C in potassium phosphate buffer (pH 7.2, 50 mM) containing 0.4 to 1.2 mg protein/ml and NNG (0.5 mM) in a total assay volume of 1 ml. When larger volumes of induced biomass were required, strain JS1663 was grown on succinate, harvested by centrifugation, washed with one-quarter-strength MSB, and suspended in one-quarter-strength MSB supplemented with NNG to a final A₆₀₀ of 0.5 to 1.0. After incubation for 4 to 6 h, cells were harvested by centrifugation, washed in 0.02 M phosphate buffer, and lysed by passage through a French pressure cell. Exudates were centrifuged at 100,000 × g, and supernatants were stored on ice until used for enzyme assays.

Identification and sequence analysis of NNG-degrading bacteria.

The isolate was grown in one-quarter-strength TSA supplemented with NNG (400 μM), and DNA was extracted using a Puregene yeast/bacterial kit (Qiagen). The isolate was phylogenetically identified by Illumina sequencing of 1,500-bp 16S rRNA gene amplicons obtained with primers 27F and 1525R (52). Small subunit rRNA genes from the isolate were identified using METAXA2 (ver. 2.1.2) (53) and compared with genes from the Ribosomal Database Project. The alignment of protein sequences was performed using the MUSCLE (ver. 3.8.31) software (54). A phylogeny was inferred by a maximum likelihood method using RAxML (ver. 8.2.8) software (55) with the LG model of amino acid replacements, a gamma model of rate heterogeneity, and a maximum likelihood estimate of the alpha parameter. Phylogenetic bootstrap replicates were combined using the extended majority rule.

Analytical methods.

Concentrations of NNG were analyzed by HPLC with a Synergi 4-μm-pore-size Hydro-RP C₁₈ column (2 mm by 150 mm; Phenomenex, Torrance, CA). The mobile phase consisted of 98% water, 1% acetonitrile, and 1% trifluoroacetic acid, delivered at a flow rate of 0.4 ml · min⁻¹. NNG was monitored at 228 nm with a retention time of 2.15 min. Nitrite concentrations were determined using the Griess assay (56). Protein concentrations were determined with a Pierce bicinchoninic acid (BCA) protein assay reagent kit (Rockford, IL). All experiments were performed in duplicates.

NNG transformation and glyoxylate production was also confirmed by high-resolution liquid chromatography-mass spectrometry (LC-MS) with comparison to authentic standards. A Dionex u3000 HPLC was operated in line with an Orbitrap Pro mass spectrometer (Thermo Fisher). Analytes were separated on a column containing reversed-phase packing material (Kinetex 5-μm-pore-size C₁₈; Phenomenex) and separated by isocratic elution (65% water, 25% acetonitrile, and 10% isopropanol with 5 mM glacial acetic acid) over 10 min at 500 nl/min (split flow via MicroTee employed to achieve nanoliter flow rates).

Gene library construction and screening.

Genomic DNA of Variovorax sp. JS1663 cells was extracted with a Gentra Puregene yeast/bacterial kit and randomly sheared by vortexing for 2 min. The sheared DNA fragments were end-repaired to blunt ends and then ligated into CopyControl vector pCC1FOS, packaged into phage, and transferred into phage T1-resistant Escherichia coli strain EPI300-T1^R as described in the CopyControl fosmid library production kit (Epicentre Biotechnologies, Madison, WI) to create a genomic library. Clones were then incubated in 96-well plates with LB medium containing chloramphenicol (12.5 μg/ml), and 1× Fosmid CopyControl induction solution at 30°C with shaking at 240 rpm. After 16 h, the library was replicated into new 96-well plates that contained one-quarter-strength MSB, succinate (1.25 g/liter), chloramphenicol (12.5 μg/ml), 1× Fosmid CopyControl induction solution, NNG (100 μM), leucine (100 μg/ml), and thiamine (10 μg/ml) and was grown at 30°C with shaking at 200 rpm. After 40 to 45 h, clones were screened colorimetrically for nitrite release from NNG. Transductants that released nitrite from NNG were selected for fosmid DNA purification and Sanger sequencing of the terminal ends of the fosmid inserts.

Candidate genes identified through bioinformatics analysis of nucleotide sequences of fosmids showing NNG transformation were subcloned into the E. coli expression vector pBAD/HisA (Invitrogen) using the overlapping oligonucleotide primers listed in Table 1, fosmid or genomic DNA templates, Q5 High-fidelity DNA polymerase, and the NEBuilder HiFi DNA assembly cloning kit (New England BioLabs) according to the manufacturer's instructions. The resulting plasmid vectors (Table 2) were transformed into E. coli NEB5α or LMG194 strains for heterologous expression and screening.

TABLE 1.

Oligonucleotides used for plasmid vector construction

Oligonucleotide	Sequence
pBAD-HisA_rev	ACCATGATGATGATGATGATG
pBAD-HisA_fwd	GAATTCGAAGCTTGGCTG
JS1663-12210_fwd	TCTCATCATCATCATCATCATGGTATGAACTTTCAGCCCCTC
JS1663-12210_rev	GCCAAGCTTCGAATTCTTATCAGAAGATCGACAGG
JS1663-12345-12340_fwd	TCTCATCATCATCATCATCATGGTATGAATCAAGTGAATACCGAAGAAC
JS1663-12345-12340_rev	CCAAGCTTCGAATTCTTATCAGCGGGCCGGATA
JS1663-12345_rev	AACAGCCAAGCTTCGAATTCTTATCAGTCGTCGTTCGAC
JS1663-12295-12290_fwd	TCTCATCATCATCATCATCATGGTATGGCCCACGCCGCCTTC
JS1663-12295-12290_rev	GCCAAGCTTCGAATTCTTACTAGATCACGCCCGCATCGC

Genome sequencing.

The Variovorax sp. JS1663 genome sequence was assembled from 19,839,672 Illumina MiSeq paired-end reads, with average lengths of 96 bases after trimming. The SPAdes genome assembler (ver. 3.6.2) was used to correct and assemble paired Illumina reads with a k-mer length of 99 and an average of 107-fold coverage (57). This assembly was corrected with Pilon software (ver. 1.3) using paired-end sequences assembled using PEAR (ver. 0.9.6) (58). Contigs with low sequencing coverage or a size less than 1 kbp were removed. The QUAST (ver. 3.0) and CheckM (ver. 1.0.5) software assessed de novo genome assembly quality (59, 60). The permanent draft assembly contained 140 contigs comprising 7,324,361 bp, with an N₅₀ of 115,706 bp, an L₅₀ of 20 contigs, and a G+C content of 68%. Coding DNA sequences were identified and annotated by the NCBI Prokaryotic Genome Annotation Pipeline and the RAST server (61).

Purification of NNG lyase.

E. coli pDG708 was grown in LB medium supplemented with ampicillin (100 μg/ml) at 37°C with shaking until the optical density at 600 nm reached 0.2. l-Arabinose (0.2%) was added, and the culture was incubated at 30°C for 12 h. Cells were harvested by centrifugation, washed twice with ice-cold potassium phosphate buffer (pH 7.0, 20 mM), suspended in Tris-HCl (50 mM, pH 7.2) containing NaCl (200 mM), and then passed twice through a French pressure cell (20,000 lb/in²). The cell lysate was clarified by centrifugation (20,000 × g at 4°C for 20 min) prior to loading onto a 5-ml His-Trap HP nickel column (GE Healthcare) pre-equilibrated with Tris-HCl (50 mM, pH 7.2) and NaCl (200 mM) and eluting with a gradient from 25 to 500 mM imidazole. Fractions containing NNG lyase activity were identified by detecting nitrite in reactions supplemented with ferrous ammonium sulfate (15 mM) and NNG (200 μM). Active fractions were buffer exchanged with potassium phosphate buffer (50 mM) to remove the imidazole with Amicon 10,000-molecular-weight-cutoff centrifugal filter units.

Enzyme assays.

NNG lyase activity was determined by measuring the disappearance of NNG and the formation of the products nitrite and glyoxylate. The disappearance of NNG was determined by monitoring NNG concentrations by HPLC. The formation of nitrite was determined using the Griess reaction (56). The formation of glyoxylate was determined by derivatization with phenylhydrazine and by detection of the glyoxylate phenylhydrazone spectrophotometrically (62). The enzyme assays were carried out at 30°C in potassium phosphate buffer (pH 7.2, 50 mM) containing 11.6 μg of protein/ml, NNG (0.5 mM), and ferrous ammonium sulfate (15 mM) in a total assay volume of 1 ml. Samples containing ferrous ammonium sulfate were clarified by centrifugation (17,000 × g for 5 min) prior to adding the Griess reagents due to extensive iron precipitation. The enzyme was active in the presence of ferrous ammonium sulfate with concentrations as low as 0.2 mM. Excess iron was routinely added to assay mixtures, because the iron precipitated quickly under oxic conditions in phosphate buffer (pH 7.2). Concentrations of substrates and products were determined by HPLC or colorimetric/spectrophotometric assays at appropriate intervals.

Accession number(s).

The data from this whole-genome shotgun project has been deposited at DDBJ/EMBL/GenBank under the accession no. LYMK00000000, version LYMK01000000.