Degradation of nitramines (R–N(R′)NO2; R′ = H or alkyl) has been well studied in the context of the environmental degradation of explosive cyclic nitramines . The cyclic nitramines hexahydro-1,3,5-trinitro-1,3,5-triazine (commonly called RDX), octogen (HMX), and hexanitrohexaazaisowurtzitane (CL-20) are compounds found in military grade explosives and propellants. Contamination of these cyclic nitramines in soil and groundwater is concerning due to their toxicity and potential carcinogenicity . Biotic and abiotic degradation of cyclic nitramines often produce linear nitramine byproducts. For example, degradation of RDX and HMX by microbes or alkaline hydrolysis forms the linear nitramine 4-nitro-2,4-diazabutanal (NDAB) . Linear nitramines are also produced during the process of amine-based carbon dioxide capture technologies . These linear nitramines from these reactions pose their own health and environmental consequences . Therefore, strategies to remediate linear nitramines are needed.

Compared to cyclic nitramines , there is far less known regarding the environmental biodegradation pathways of linear nitramine contaminants. Biodegradations of NDAB by the fungus Phanerochaete chrysosporium and the bacterium Methylobacterium sp. strain JS178 have been reported . Initiation of the P. chrysosporium degradation was attributed to a manganese peroxidase, however, the mechanism of degradation is unclear. Linear nitramines, produced by carbon capture, were shown to be biodegraded in soil and water . Nitramines with hydroxy groups were best degraded in this study, including diethylnitramine, 2-methyl-2-(nitroamino)-1-propanol, and 2-nitroaminoethanol (2-NAE). While little is known regarding the degradation of these anthropogenic linear nitramines, we can glean insight into their reactivities from recent studies regarding the enzymatic degradation of N-nitroglycine (NNG), a naturally occurring linear nitramine.

Scheme 1: NNG degradation by reduced NnlA.

While the activity of NnlA is established less is known about its physiological function and, for that matter, the physiological function of its substrate NNG. This compound is one of the few known nitramine natural products and the only one produced by bacteria instead of fungi . Its only known natural sources are strains of Streptomyces bacteria . The abundance and distribution of these NNG producers and of NNG is unknown. Additionally, NNG’s physiological function is unknown, but it is toxic to plants, mice, and Gram-negative bacteria . While there is no direct evidence of the mechanism of this toxicity, NNG has been shown to competitively inhibit succinate dehydrogenase, a component enzyme of the Krebs cycle . Therefore, NNG may be a toxin released to kill or outcompete nearby bacteria or other organisms for limited resources. In such a context, the physiological function of NnlA could be to protect bacteria from toxic NNG exposure. Alternatively, NnlA could be a promiscuous nitramine degrader that allows bacteria to use alternative nitrogen sources. In fact, the RDX-degrading enzyme, XplA, is remarkably conserved amongst RDX-degrading microbes (>99% identity across several species) . Based on these observations, it has been proposed that XplA evolved within the past 100 years in response to the rise in RDX contamination. Given that Variovorax sp. strain JS1663 was isolated from a nitramine-contaminated sludge, it should be considered if NnlA is promiscuous and can degrade both natural nitramines, such as NNG, and anthropogenic nitramines, such as 2-NAE.

We have previously identified several Vs NnlA homologs in sequence databases , however, the NNG degradation activities of these homologs have not been tested. Doing so will differentiate between these two hypotheses by testing if NnlA homologs with NNG degradation activity are highly conserved or if they are found in bacteria found in widespread classes and environments. Herein, we report the characterization of five NnlA homologs. It is shown that all five homologs exhibited NNG degradation activity. Isolation and characterization of four of these homologs showed that all contain heme, the reduction of which is required for NNG degradation activity. In addition, we show that NnlA cannot degrade 2-NAE. Combined with previous substrate scope studies, this result strongly suggests that NnlA is specific for NNG. The implications of our results in understanding the environmental abundance and physiological function of NNG are discussed below.

To select Vs NnlA homologs to test for NNG degradation activity, we performed a BLAST search of the Vs NnlA amino acid sequence in the NCBI database. This search resulted in retrieval of 99 homologous amino acid sequences with an E-value of less than 1 × 10−10. A subset of 55 of these sequences was selected with a query cover of greater than 88%. From these sequences, we selected five homologs of the nnlA gene [Pseudovibrio denitrificans JCM 1230 (Pd), Pseudovibrio japonicus strain KCTC 12861 (Pj), Pseudonocardia spinosispora DSM 44797 (Ps), Mycobacterium sp. 1465703.0 (Ms), Microbispora rosea subsp. nonnitritogenes strain NRRL B-2631 (Mr)], which were synthesized and cloned into E. coli recombinant expression vectors. These homologs ranged in amino acid sequence identity from 46 to 76% compared to Vs NnlA. Additionally, these homologs along with Vs NnlA are found in bacteria that span a wide range of bacterial classes (Alphaproteobacteria, Betaproteobacteria, and Actinomycetia).

A preliminary screen of these homologs for NNG degradation activity was performed to identify homologs for further characterization. The H73A Vs NnlA variant, previously shown to lack NNG degradation activity, was used as a negative control . E. coli transformants containing the expression vectors were incubated at 37 °C overnight in diluted lysogeny broth (LB) containing NNG and IPTG, the latter component was used to induce NnlA expression.

Figure 1: Nitrite concentrations observed in cultures of E. coli transformed with NnlA homologs or variants grown in the presence of NNG. Cells were incubated overnight in 1/5 LB, 20 µM IPTG, 3 mM NNG with appropriate antibiotics, incubated overnight at 37 °C.

Figure 2: Molecular mass determination of purified NnlA homologs by A) SDS-PAGE or B) analytical size exclusion chromatography. Homolog labeled in figure. Mobile phase and sample buffer is 100 mM tricine 100 mM NaCl buffer at pH 7.5. Dashed grey lines represent elution volumes of molecular mass standards. Theoretical molecular weights are as follows Vs NnlA: 21,869 Da; Mr NnlA: 20638 Da; Pd NnlA: 18,473 Da; Ms NnlA : 18,239 Da; and Ps NnlA: 19,042 Da.

Figure 3: UV–vis absorption spectra of purified NnlA homologs. All spectra were measured in 100 mM tricine, 100 mM NaCl buffer at pH 7.5. NnlA homolog concentrations were Vs (10 μM), Mr (19 μM), Pd (18 μM), Ms (13 μM), and Ps (4 μM). Inset: Q-band region of the UV–vis spectra for concentrated NnlA homolog samples: Vs (90 μM), Mr (170 μM), Pd (160 μM), Ms (38 μM), and Ps (28 μM).

Table 1: Iron analyses of purified NnlA homologs.

aRef.

Figure 4: Representative LC–MS EICs monitoring molecular anions of NNG (m/z 119.01 ± 100 ppm) and glyoxylate (m/z 72.99 ± 100 ppm) in samples containing 350 μM NNG, 10 μM dithionite, and 5 μM of indicated NnlA homolog. Samples were incubated for approximately one hour at 21 °C in deoxygenated 30 mM tricine buffer at pH 7.5.

Table 2: Nitrogen mass balance resulting from NNG degradation by NnlA.

aReaction conditions: 5 μM NnlA, 10 µM sodium dithionite, 350 μM NNG in 30 mM tricine buffer at pH 7.5 and room temperature in anaerobic glovebox. Mean values with standard deviations from triplicate reactions are shown.

Figure 5: Alpha-fold model of Vs NnlA dimer (cyan) overlayed with Pseudomonas aeruginosa (grey; PDB: 3VOL ) to estimate the position and orientation of the heme cofactor. Conserved residues in the distal heme pocket are labeled. Conserved basic residues outside of the heme pocket are colored in magenta. Nitrogen, oxygen, and iron atoms are colored blue, red, and orange, respectively. Figure generated using PyMOL.

We sought to identify the heme binding site, but AlphaFold does not model this. However, this AlphaFold model was predicted to bind a heme cofactor by the consensus modeling tool COACH . This protein–ligand model exhibited steric clashes with the heme and protein side chains (data not shown), limiting the use of this model to predict the heme environment. Nevertheless, this protein–ligand model heme binding between the β-sheet and an α-helix based on similarity to the oxygen-sensing dimeric DosH protein . DosH is also a heme-binding PAS-domain containing protein, further validating the assignment of NnlA as a heme-binding PAS domain protein.

Figure 6: Phylogenetic tree of NnlA homologs with accession numbers. Branch lengths correspond to amino acid substitutions per position. Numbers at nodes indicate Shimodaira–Hasegawa-like approximate likelihood ratio test (SH-aLRT) support (%) and ultrafast bootstrap support (%).

Scheme 2: Acid–base equilibrium of 3-nitropropionate (3-NP) vs N-nitroglycine (NNG).

This antibiotic activity may also require further modification of NNG or its incorporation into a larger natural product. NNG is a non-proteinogenic amino acid, similar to other such N–N containing compounds such as piperazic acid and hydrazinoacetic acid . These precursors are incorporated into larger NPs by non-ribosomal peptide synthases or polyketide synthases, and NNG may have a similar fate .

Another possibility is that NNG has several physiological functions and fates. For example, a natural product nitronate intermediate was recently shown to have two fates within Streptomyces achromogenes var. streptozoticus NRRL 3125 . This nitronate intermediate was shown to be O-methylated to form O-methylnitronate, and subsequently incorporated into enteromycin. Alternatively, the intermediate could be denitrified by a nitronate monooxygenase (NMO) to produce NO2−. NnlA could replace NMO as the denitrifying enzyme in NNG producing bacteria, however, a BLAST search of NnlA in the genome of Streptomyces noursei, an NNG-producing bacterium, did not reveal any NnlA homologs. Interestingly, four NMOs are annotated in the S. noursei genome. These enzymes could protect S. noursei from NNG toxicity during its biosynthesis. Meanwhile, we posit that NnlA protects non-NNG producing bacteria from exposure. In vivo experiments comparing the toxicity of NNG towards wild-type cells expressing NnlA and NnlA knockout mutant strains would test this hypothesis.

In this study the NNG degradation activity of five Vs NnlA homologs was screened in E. coli transformants, providing evidence that all five degrade NNG. Of these, four were fully isolated and characterized. Each isolated homolog exhibited similar oligomerization and heme occupancy as Vs NnlA. In addition, we confirmed by in vitro assays that initiation of NNG degradation activity by the NnlA requires reduction of the heme, verifying the necessity of the heme for NnlA activity. The nitrogen mass balance was consistent with NNG degradation to NO2− and NH4+ as shown for Vs NnlA. It was also shown that NnlA cannot degrade the hydroxylated linear nitramine, 2-NAE. The combined data indicate that NnlA homologs specific for NNG degradation activity are found in diverse bacteria and environments. These results suggest the natural product NNG may also be found in diverse environments. The reactivity of the nitramine functionality begs for further studies confirming the natural abundance and physiological functions of NNG and other nitramine natural products.

Isopropyl β-ᴅ-1-thiogalactopyranoside (IPTG) and 5-aminolevulinic acid (5-ALA) were purchased from Gold Biotechnology. NNG was purchased from AAblocks. 2-NAE was purchased from Toronto Research Chemicals. General buffers and media components were purchased from Fisher Scientific or VWR. Stock dithionite concentrations were determined by UV–vis absorbance at 318 nm (ε318 = 8000 M−1cm−1). Water used for all solutions was of 18.2 MΩ·cm resistivity from a Barnstead Nanopure (Thermo Fisher Scientific). Solvents for LC–MS experiments were of at least HPLC grade and contained 0.1% vol/vol formic acid.

The vectors to express Vs NnlA and H73A Vs NnlA were previously reported . Pd, Mr, Ms, Ps, and Pj nnla genes were synthesized as E. coli codon-optimized constructs and cloned into the NdeI and XhoI restriction sites of pET-28a(+)-TEV by GenScript.

For protein expression, plasmids were electroporated into E. coli BL21(DE3) cells and protein was expressed and purified by immobilized metal affinity chromatography (IMAC) as previously described for Vs NnlA . The only modification was that plasmids using pET-28a(+)-TEV required 50 µg/mL kanamycin instead of 100 µg/mL ampicillin. IMAC purified and concentrated protein were exchanged into 100 mM tricine, 100 mM NaCl buffer at pH 7.5 and stored at −60 °C.

Total iron concentrations in protein samples were quantified using an iron assay that allows for release and subsequent detection of heme-ligated iron . Protein concentration was determined using bicinchoninic acid protein quantification assay (Pierce). The oligomeric state was determined by processing the protein through Superdex 200 Increase 10/300 GL analytical size exclusion column with 100 mM tricine with 100 mM NaCl at pH 7.5 as the mobile phase. Protein size exclusion chromatography standards (BioRad) were used to determine molecular masses.

LC–MS analysis was performed using an Agilent 1260 LC stack equipped with a Zorbax RX-C18 column (5 μm, 4.6 × 150 mm) and connected to an Agilent 6230 TOF mass spectrometer with electrospray ionization (ESI). Analyses used an isocratic mixture containing 65% water, 25% acetonitrile, and 10% isopropanol at a flow rate of 0.5 mL/min. The mass spectrometer was run in the negative ion mode with a probe voltage of 4,500 V and a fragmentation voltage of 175 V. To monitor NNG, 2-NAE, and glyoxylate, extracted ion chromatograms were obtained at m/z 119.0, 105.0, and 73.0, respectively.

Ammonium concentrations were determined using a glutamate dehydrogenase assay (Sigma-Aldrich) kit using the manufacturer’s instructions. Nitrite concentrations were determined by reacting 25 μL aliquots of reaction sample with 25 μL of deoxygenated Griess reagent R1 (1% sulfanilamide in 5% H3PO4) followed by addition of 25 μL of deoxygenated Griess reagent R2 (0.1% naphthylethylenediamine dihydrochloride in water). The absorbance was read at 548 nm using an Infinite M200 Plate Reader (Tecan). Nitrite concentrations were determined by comparison of A548 nm to a nitrite standard curve.

Transformation of NnlA homologs were obtained as described above. The cells were then plated on LB agar plates containing ampicillin (100 µg/mL) or kanamycin (50 µg/mL) as appropriate and incubated overnight at 37 °C. Three colonies from each plate were picked with a toothpick and then resuspended in 0.2 mL of sterile water. Thirty microliter aliquots of suspended cells were used to inoculate 100 µL of selective growth media (1/5 LB media, 20 µM isopropyl β-ᴅ-thiogalactopyranoside (IPTG), 300 µM 2-NAE or 3 mM NNG, and antibiotic) in a 96-well plate. After overnight incubation at 37 °C, the cells were pelleted by centrifugation and the nitrite quantified by Griess assay as described above.

Triplicate samples containing 2 mM 2-NAE or 350 µM NNG, 40 µM titanium citrate with or without 20 µM Vs NnlA in deoxygenated 23 mM tricine at pH 7.5 were incubated overnight at 21 °C.

Homologs of the Variovorax nnlA gene, including those described here, were identified by sequence similarity searches and their predicted amino acid sequences were used to infer phylogenetic relationships. An alignment of 11 amino acid sequences with 28 to 87% identity was prepared using MUSCLE software (ver. 5.1) and trimmed to 148 positions in conserved blocks using Gblocks (ver. 0.91b) . A maximum likelihood tree was inferred using IQ-TREE (ver. 2.2.2.6) with the LG+G4 substitution model .

The world map in the graphical abstract was supplied by simplemaps.com. This content is not subject to CC BY 4.0.

JDC and AAH were supported by funding from the United States Army Research Office award #W911NF2010286. BMR and ML were supported by a supplement to this parent grant provided by the Army Education Outreach Program (AEOP). This work was supported in part by the Strategic Environmental Research and Development Program (SERDP) under project WP20-1151. Oak Ridge National Laboratory is managed by UT-Battelle, LLC, for the U.S. Department of Energy under contract no. DE-AC05-00OR22725.

The data that supports the findings of this study is available from the corresponding author upon reasonable request.