Unprecedented N₂O production by nitrate-ammonifying Geobacteraceae with distinctive N₂O isotopocule signatures

ABSTRACT

Dissimilatory nitrate reduction to ammonium (DNRA), driven by nitrate-ammonifying bacteria, is an increasingly appreciated nitrogen-cycling pathway in terrestrial ecosystems. This process reportedly generates nitrous oxide (N₂O), a strong greenhouse gas with ozone-depleting effects. However, it remains poorly understood how N₂O is produced by environmental nitrate-ammonifiers and how to identify DNRA-derived N₂O. In this study, we characterize two novel enzymatic pathways responsible for N₂O production in Geobacteraceae strains, which are predominant nitrate-ammonifying bacteria in paddy soils. The first pathway involves a membrane-bound nitrate reductase (Nar) and a hybrid cluster protein complex (Hcp–Hcr) that catalyzes the conversion of NO₂⁻ to NO and subsequently to N₂O. The second pathway is observed in Nar-deficient bacteria, where the nitrite reductase (NrfA) generates NO, which is then reduced to N₂O by Hcp–Hcr. These enzyme combinations are prevalent across the domain Bacteria. Moreover, we observe distinctive isotopocule signatures of DNRA-derived N₂O from other established N₂O production pathways, especially through the highest ¹⁵N-site preference (SP) values (43.0‰–49.9‰) reported so far, indicating a robust means for N₂O source partitioning. Our findings demonstrate two novel N₂O production pathways in DNRA that can be isotopically distinguished from other pathways.

IMPORTANCE

Stimulation of DNRA is a promising strategy to improve fertilizer efficiency and reduce N₂O emission in agriculture soils. This process converts water-leachable NO₃⁻ and NO₂⁻ into soil-adsorbable NH₄⁺, thereby alleviating nitrogen loss and N₂O emission resulting from denitrification. However, several studies have noted that DNRA can also be a source of N₂O, contributing to global warming. This contribution is often masked by other N₂O generation processes, leading to a limited understanding of DNRA as an N₂O source. Our study reveals two widespread yet overlooked N₂O production pathways in Geobacteraceae, the predominant DNRA bacteria in paddy soils, along with their distinctive isotopocule signatures. These findings offer novel insights into the role of the DNRA bacteria in N₂O production and underscore the significance of N₂O isotopocule signatures in microbial N₂O source tracking.

INTRODUCTION

Nitrous oxide (N₂O) is a powerful greenhouse gas and the dominant ozone-depleting substance throughout the 21st century (1, 2). The largest current emissions of N₂O into the atmosphere are driven by microbial activities in agricultural soils, exacerbated by the application of nitrogen fertilizers (3). Denitrification and nitrification are well-established as the dominant processes, accounting for approximately two-thirds of all soil-derived N₂O emissions (4). However, another microbial process, dissimilatory nitrate reduction to ammonium (DNRA), has also been implicated in contributing to global N₂O emission (5, 6). DNRA, also termed nitrate (NO₃⁻) ammonification, is a nitrogen (N)-retaining process in which water-leachable NO₃^- is anaerobically reduced to soil-adsorbable NH₄⁺ via the intermediate nitrite (NO₂⁻). This reduction is catalyzed by cytoplasmic and/or periplasmic NO₃⁻ reductases (Nar and/or Nap) and cytochrome c₅₅₂ and/or NADH-dependent NO₂⁻ reductases (NrfA and/or NirB) in a two-step process (Fig. 1)(7, 8).

Fig 1.

Simplified diagram showing N₂O production pathways in the nitrogen cycle under both aerobic and anaerobic conditions. SP_bD, SP_fD, SP_NI, and SP_DN: ¹⁵N-site preference of N₂O derived from the bacterial denitrification, fungal denitrification, nitrification, and DNRA processes, respectively. The value ranges of SP_bD, SP_fD, and SP_NI are retrieved from Ref. (9). Key enzymes of the nitrogen cycle are indicated as nitrate reductase (Nar/Nap), nitrite reductase (Nrf/NirB/NirS/NirK), NO dioxygenase (Hmp/Fdp), nitric oxide reductase (Nor), nitrous oxide reductase (Nos), nitrogenase (Nif), ammonia monooxygenase (AMO), hydroxylamine oxidoreductase (HAO), and nitrate oxidoreductase (Nxr). Dotted lines represent the pathways with predicted enzymes and intermediates.

The phenotype of N₂O production in DNRA was first documented in 1981, where 163 out of 168 studied nitrate-ammonifying soil bacteria were found to produce N₂O (10). Subsequent research has identified many nitrate-ammonifiers, primarily affiliated with Proteobacteria and the genus Bacillus, as N₂O producers (11–13). The widely accepted pathway for N₂O production in these bacteria involves the generation of N₂O from NO₂⁻ via a two-step process, with NO as an intermediate. This process is catalyzed by Nar and NO detoxification-related enzymes, including flavodiiron proteins (Fdp), flavorubredoxin (NorVW), and flavohemoglobin (Hmp), as demonstrated in Nar-containing bacteria Escherichia coli and Salmonella typhimurium (Fig. 1) (14–18). However, several studies have reported contradictory findings. For instance, the Hmp and NorV mutations in E. coli did not impair NO reduction (19). Additionally, a recent study involving soil DNRA bacteria suggested that N₂O production from Nap and Nar was unlikely, with the NO₂⁻-to-NH₄⁺ reaction being a more plausible N₂O source (20). This raises questions about the mechanism of N₂O generation in Nar-containing ammonifiers. To date, there has been no attempt to explore the N₂O generation mechanism in Nap-driven nitrate-ammonifiers (Nar-absent), although their phenotype of N₂O production from NO₃⁻ has been observed in the DNRA model organism Wolinella succinogenes (21). Therefore, a comprehensive understanding of DNRA-derived N₂O production requires further investigation of DNRA bacteria that use Nar versus Nap enzymes for NO₃⁻ reduction.

The analysis of N₂O isotopocule signatures (δ¹⁵N^bulk, δ¹⁸O, and ¹⁵N-site preference [SP]) is a promising tool for tracing the N₂O origins and quantifying the contributions of distinct production processes, as they rely on the natural abundance of N and O isotopes without perturbing in situ conditions (22). Particularly, SP refers to the difference in the ¹⁵N isotopic composition of ^αN compared with the ^βN in the linear N₂O molecule (^βN-^αN-O)(22). Unlike conventional isotopic modes like δ¹⁵N^bulk and δ¹⁸O, SP is independent of the concentrations and isotope ratios of substrates, making it a unique indicator for distinguishing N₂O production pathways (9). Although the N₂O isotopocule signatures of most known N₂O generation pathways have been documented (Fig. 1) (9, 23), those associated with DNRA remain unexplored. N₂O production via DNRA and other pathways, especially denitrification, often occurs simultaneously under anaerobic conditions (6, 24), but few studies have evaluated the contribution of DNRA to N₂O emissions due to the lack of a means for distinguishing DNRA-derived N₂O from bulk N₂O emissions. Characterizing the N₂O isotopocule signatures associated with DNRA could enable the partitioning of bulk N₂O emissions into contributions from DNRA and other pathways.

Paddy soils are the largest anthropogenic wetlands on Earth and contribute greatly to global N₂O emissions (25, 26). These soils are rich in DNRA drivers due to the reducing conditions caused by waterlogging during rice cultivation (27). Geobacteraceae is a strictly anaerobic and ubiquitous bacterial group in terrestrial ecosystems (28) and one of the dominant nitrate-ammonifying groups in paddy soils (27, 29). Recently, we isolated dozens of Geomonas and Oryzomonas strains (both belonging to the family Geobacteraceae) from paddy soils and nearby sediments (30–34). They were genome-annotated as different types of nitrate-ammonifiers, as Geomonas strains mainly contain only Nar or both Nar and Nap for NO₃⁻ reduction, while Oryzomonas strains only possess Nap (Fig. 2). The co-presence of Nar- and Nap- driven strains in these closely related genera provides an ideal model for investigating the diverse features of DNRA, including the N₂O generation mechanisms.

Fig 2.

Phylogenomics and nitrogen metabolism-related gene inventory of Geobacteraceae species. The tree is a maximum-likelihood (ML) phylogeny based on 92 concatenated core proteins in the UBCG database (35). Bootstrap values shown in circle shapes at branching nodes were calculated using 100 replicates. The scale bar represents 0.1 substitutions per amino acid position. The studied genera Geomonas and Oryzomonas were labelled by gray color. The nitrogen metabolism-related gene inventory is shown in the heatmap and the copy numbers of every functional gene are denoted by the blue bar. The accession numbers of genome sequences used in tree construction and detailed information of the strain habitat are listed in Table S4.

In this study, we investigated the N₂O production pathways and the key enzymes involved in strictly anaerobic nitrate-ammonifiers Geobacteraceae and synchronously measured the isotopocule signatures of DNRA-derived N₂O. We hypothesized that these nitrate-ammonifiers would exhibit novel N₂O production pathways or N₂O-generating enzymes, given the absence of fdp, norVW, and hmp genes related to NO detoxification reported in enteric facultative anaerobes (16–18), and thus possibly create distinctive N₂O isotopocule signatures. This study provides a comprehensive description of DNRA-derived N₂O production in environmental microorganisms, enhancing our understanding of N₂O emission pathways and the role of DNRA microorganisms in greenhouse gas production.

RESULTS AND DISCUSSION

DNRA activity and N₂O production in representative Geomonas and Oryzomonas strains

Geomonas and Oryzomonas species were genome-annotated as nitrate-ammonifiers, as they contain the key marker genes nar/nap and nrfA of the DNRA, except for two strains, Geomonas bremensis R1 and Geomonas bemidjiensis DSM 16622^T, which lack nar and nap genes and are solely nitrite-ammonifiers (Fig. 2). These strains exhibit various combinations of functional genes related to N metabolism, but these combinations are conserved within phylogenetic clusters at the species level (Fig. 2). We selected three representative species, including two Nar-containing bacteria Geomonas sp. Red32 and Geomonas terrae Red111 and one Nar-absent but Nap-containing bacterium Oryzomonas rubra Red88 to assay DNRA activities using either NO₃⁻ (8 mM) or NO₂⁻ (2 mM) as substrates.

G. terrae Red111 and O. rubra Red88 were observed to utilize NO₃⁻ for growth, producing NH₄⁺, NO₂⁻, and N₂O as end products, thereby demonstrating DNRA activity (Fig. 3a through c). In contrast, Geomonas sp. Red32 produced only NO₂⁻ and N₂O without NH₄⁺ production or significant biomass increase (Fig. 3a and d), likely due to its weak NO₂⁻-reducing activity, which may be attributed to the single copy of the nrfA gene (Fig. 2). When NO₂⁻ was the substrate, G. terrae Red111 and O. rubra Red88 completely consumed NO₂⁻, producing NH₄⁺ and N₂O along with bacterial growth (Fig. 3e through g). In contrast, Geomonas sp. Red32 only partially consumed NO₂⁻, producing N₂O but no NH₄⁺ (Fig. 3e and h). None of the studied strains produced NO₃⁻ from the added NO₂⁻ (Fig. 3f through h), indicating that NO₂⁻ oxidation did not occur in any of the cultures. Notably, the N₂O produced by these nitrate-ammonifiers accounted for only 1% or less of consumed NO₃⁻ and 1%–3% of consumed NO₂⁻, far less than the main product NH₄⁺ (>60% of substrates for G. terrae Red111 and O. rubra Red88) and the accumulated NO₂⁻ during NO₃⁻ reduction (3%–20%) (Fig. 3i). This suggests that N₂O production during DNRA is a byproduct rather than a key intermediate or main product.

Fig 3.

Dissimilatory NO₃⁻ and NO₂⁻ reduction by three Geobacteraceae strains. (a–h) Dynamics of growth conditions and concentrations of NO₃⁻, NO₂⁻, NH₄⁺, and N₂O in the cultures after NO₃⁻ (8 mM, left panel) or NO₂⁻ (2 mM, right panel) additions for Geomonas terrae Red111, Oryzomonas rubra Red88, and Geomonas sp. Red32. The start time was from the bacterial inoculation for the NO₃⁻ reduction process and from the NO₂⁻ addition for the NO₂⁻ reduction process. The values are shown as mean ± standard deviation (SD). (i) The relative abundance of N compositions in the cultures at the final reaction stage during NO₃⁻ and NO₂⁻ reduction by the three strains.

The facultative anaerobic nitrate-ammonifiers have been reported to produce N₂O at levels accounting for 3%–36% of consumed NO₃⁻ (10, 36, 37), indicating that these Geobacteraceae strains have a relatively low N₂O production ratio from NO₃⁻ transformation. However, such ratios are comparable to those observed in the strict anaerobe Wolinella succinogenes, which converted approximately 0.15% of NO₃⁻ to N₂O via DNRA (21). This finding indicates a difference in N₂O production capacities between strictly and facultative anaerobic nitrate-ammonifiers, possibly resulting from the difference in the pathways or enzymes involved in N₂O production. Moreover, among the biological processes of N₂O production, denitrification exhibits the highest N₂O production ratio from NO₃⁻, as N₂O is a key intermediate and can be the sole product, accounting for all consumed NO₃⁻, depending on culture conditions and denitrifier types (38). Previous studies have reported that ammonia-oxidizing bacteria (AOB) converted 0.3%–10% of substrate NH₃ to N₂O and its precursor NO during both NH₃ oxidation to NO₂⁻ and NO₂⁻ reduction to N₂O (39). In contrast, ammonia-oxidizing archaea (AOA) and comammox bacteria yield approximately 0.08% N₂O from substrate NH₃ (40, 41), and anammox bacteria converted 0.1% of substrates NO₂⁻ and NH₄⁺ to N₂O (42). These findings suggest that nitrate-ammonifiers produce N₂O at levels comparable to AOB, and much higher than AOA and comammox and anammox bacteria. Given the wide distribution and diverse species of nitrate-ammonifying bacteria (6), DNRA has great potentials as a contributor to greenhouse gas emission and should not be overlooked.

N₂O production coupled to NO₂⁻ and NO accumulation during DNRA

To determine the direct substrate contributing to N₂O during DNRA, we conducted a comparative analysis of N₂O production across different strains and substrates based on prior culture experiments. First, N₂O production from NO₃⁻ in G. terrae Red111 and O. rubra Red88 suggested five potential sources for N₂O: NO₃⁻, NO₂⁻, NH₄⁺, and the intermediate reactions between NO₃⁻ and NO₂⁻, and between NO₂⁻ and NH₄⁺ (Fig. 3b and c; Path_1 in Fig. 4a). Given that N₂O production also occurred from NO₂⁻ in G. terrae Red111 and O. rubra Red88 (Fig. 3f and g), we ruled out NO₃⁻ and the intermediate processes between NO₃⁻ and NO₂⁻ as N₂O sources (Path_2 in Fig. 4a). Then, Geomonas sp. Red32 partially consumed NO₃⁻ and NO₂⁻ with concurrent N₂O generation, but no NH₄⁺ was detected in the medium (Fig. 3d and h), eliminating NH₄⁺ and the intermediate process between NO₂⁻ and NH₄⁺ as N₂O sources (Path_3 in Fig. 4a). Collectively, NO₂⁻ remains the most likely substrate for N₂O (or its precursor) production during DNRA (Fig. 4a).

Fig 4.

Comparison of N₂O production and SP values showing the unique N₂O production from NO₂⁻ via the intermediate NO. (a) A comparison diagram showing the possible N₂O production source on DNRA. Three pathways, labeled by Path_1, Path_2, and Path_3, represent the experimental results from the NO₃⁻ reduction by all studied Geomonas strains except for Red32, NO₂⁻ reduction by all studied Geomonas strains except for Red32, and NO₃⁻ reduction by strain Red32, respectively. The red dots on the arrows indicate the possible source of N₂O production. (b and c) Dynamics of NO₂⁻ and N₂O concentrations during the NO₃⁻ reduction process driving by Geomonas terrae Red111 with (gray squares) and without (black circles) acetate (5 mM) addition. Arrows indicate the time when the extra acetate was added to the medium. (d and e) N₂O production from the NO₃⁻ reduction process driven by Geomonas strain Red111 and Oryzomonas strain Red88 across time with (gray squares) and without (black circles) c-PTIO additions. c-PTIO was added to the medium at the same time to bacteria inoculation. CK indicates control experiments, asterisks indicate significant difference, *** P < 0.001, *P < 0.05, ns indicates insignificant difference, P > 0.05. The values are shown as mean ± standard deviation (SD). (f and g) The comparison of SP values between different genera (Geomonas and Oryzomonas) and substrates (NO₃⁻ and NO₂⁻). Med and n indicate the median and number of every data set, respectively.

To verify the relationship between NO₂⁻ and N₂O in DNRA, we compared the production and consumption rates of N₂O and other N compounds in cultures. N₂O production was slower than NO₃⁻ consumption and NH₄⁺ production (Fig. 3b and c) but comparable to NO₂⁻ accumulation with significantly positive linear relations (R²>0.7, P < 0.0001) (Fig. S1). Additionally, more carbon sources (high C:N ratios) were found to reduce the NO₂⁻ accumulation during NO₃⁻ reduction in G. terrae Red111 (Fig. S2). In a further study, the absence of NO₂⁻ accumulation halted N₂O production (Fig. 4b and c). These results indicate that NO₂⁻ is the direct substrate contributing to N₂O production (or its precursor) during DNRA in Geobacteraceae strains, consistent with previous findings that N₂O is generated from NO₂⁻ in nitrate-ammonifying Enterobacteriaceae (15, 16, 36).

Previous studies have reported that NO is produced before N₂O in nitrate-ammonifying Enterobacteriaceae (14–18), raising the question of whether intermediates, such as NO and hydroxylamine, exist between NO₂⁻ and N₂O in Geobacteraceae strains. We measured hydroxylamine concentrations in the cultures and used NO inhibitors (c-PTIO and L-NMMA) and donors (SNP) for treatments. c-PTIO, a NO scavenger, removes NO from the medium, while L-NMMA, a NO synthase inhibitor, prevents endogenous NO generation from amino acids (43, 44). Hydroxylamine was undetected during bacterial growth, but c-PTIO (100 µM) significantly inhibited N₂O production (P < 0.05; Fig. 4d and e). Notably, c-PTIO had no impact on bacterial growth or the DNRA rate, even at high concentrations of 300 µM (P > 0.05, Fig. S3), and L-NMMA (100 µM) did not affect N₂O production (P > 0.05, Fig. S4). These results indicate that c-PTIO inhibits N₂O production by reducing exogenous NO (derived from supplied NO₃⁻) rather than by affecting bacterial activity. Furthermore, G. terrae Red111 and O. rubra Red88 converted approximately half of the supplied extra NO (80 µM SNP solution) to N₂O (Fig. S5), demonstrating their ability to reduce NO to N₂O. Altogether, we believe that NO is an intermediate between NO₂⁻ and N₂O, contributing partially or entirely to N₂O production during DNRA. Notably, c-PTIO reduced N₂O emission by 39.9% in Oryzomonas strains, but only by 19.8% in Geomonas strains (Fig. 4b and c). This difference suggests that Oryzomonas and Geomonas strains probably contain different pathways or enzymes for converting NO₃⁻ to N₂O.

The process proposed for N₂O production from NO₂⁻ via NO during DNRA is plausible for Geobacteraceae strains. However, three questions remain unanswered: 1) Is NO₂⁻ the sole substrate contributing to N₂O in all studied strains? 2) Are there other N₂O production pathways besides the NO-mediated one? 3) Does N₂O reduction occur during N₂O generation in the studied strains? To address these questions, we measured and analyzed SP values, which depend solely on the structures of enzymatic precursors right before N₂O production (22). The SP values of N₂O produced from NO₃⁻ ranged from 44.8‰ to 47.7‰ in 3-day cultures and from 40.1‰ to 46.6‰ in 7-day cultures, while those from NO₂⁻ ranged from 43.4‰ to 49.9‰ in 3-day cultures and from 43.0‰ to 47.3‰ in 7-day cultures (Table 1; Table S1). Comparing different substrates, the median SP values for NO₂⁻- and NO₃⁻- derived N₂O were 46.2‰ and 46.4‰, respectively, with no significant difference (P > 0.05) (Fig. 4f). Similarly, the SP values were insignificant (P > 0.05) among strains in the two genera Geomonas and Oryzomonas, with close medians of 46.4‰ and 46.2‰, respectively (Fig. 4g). These similar SP values indicate a shared N₂O production pathway in these strains, regardless of whether NO₂⁻ or NO₃⁻ is used as the substrate. Additionally, to prove the uniqueness of the N₂O production pathway, we compared the SP values of N₂O collected at different reaction times, 3 and 7 days. Geomonas sp. Red32 and O. rubra Red88 showed constant SP values for both NO₃⁻ and NO₂⁻ reduction processes, whereas G. terrae Red111 only showed consistent SP values for NO₂⁻ reduction (Table 1). It is known that mixed N₂O production/reduction pathways can alter their relative contribution ratios over time, thereby changing SP values. However, the consistent SP values observed in this study refute this proposal and demonstrate a single, consistent pathway for N₂O generation in the studied strains, except for NO₃⁻ reduction driven by G. terrae Red111, where mixed processes for N₂O production may occur (see the last section for explanation). These findings prove a unique N₂O production pathway (NO₂⁻-NO-N₂O) without N₂O reduction in the studied nitrate-ammonifiers, addressing the questions above.

TABLE 1.

Isotopocule signatures of N₂O produced from NO₃⁻ and NO₂⁻ reduction processes by Geobacteraceae strains and abiotic reactions^a

Substrate	Strainb	Incubation time (d)c	δ18O (‰)	δ15Nbulk (‰)	SP (‰)
NO3-	Geomonas sp. Red32	3	16.2 (0.1)	−39.8 (0)	46.5 (0.1)
(8 mM)		7	14.6 (0.4)	−27.5 (1.2)	45.4 (1.4)
	G. diazotrophica Red69	3	26.4 (4.1)	−29.6 (4.5)	45.5 (0.7)
		7	23.6 (3.9)	−28.2 (4.9)	41.9 (0.9)
	G. terrae Red111	3	15.2 (1.2)	−19.7 (2.1)	47.0 (0.3)
		7	9.6 (0.7)	−23.0 (3.4)	41.1 (1.4)
	G. silvestris Red330	3	18.6 (4.1)	−21.1 (7.0)	46.8 (1.3)
		7	15.5 (1.5)	−18.0 (7.2)	46.5 (0.1)
	O. rubra Red88	3	23.4 (0.8)	−31.4 (3.7)	46.3 (0.6)
		7	19.2 (1.5)	−22.4 (5.4)	46.4 (0.3)
NO2-	Geomonas sp. Red32	3	27.8 (0.2)	−23.7 (0)	46.8 (0.2)
(2 mM)		7	27.0 (1.0)	−21.3 (0.1)	46.1 (1.0)
	G. diazotrophica Red69	3	28.8 (0.1)	−23.7 (0)	45.8 (0.7)
		7	28.6 (0)	−23.5 (0.3)	46.0 (0.1)
	G. terrae Red111	3	27.5 (0.3)	−24.3 (0.1)	47.2 (0.7)
		7	29.5 (0)	−23.9 (0.1)	46.3 (0.2)
	G. silvestris Red330	3	28.5 (0.4)	−22.5 (0.5)	46.5 (0.2)
		7	28.6 (0.2)	−21.6 (0.5)	46.4 (0.2)
	O. rubra Red88	3	28.8 (0.4)	−21.6 (0.7)	44.8 (2.0)
		7	28.8 (0.3)	−21.0 (0.1)	44.4 (2.0)
	G. bemidjiensis DSM 16622	3	28.6 (0.6)	−22.9 (0.2)	47.4 (2.3)
	Geob. anodireducens JCM 30203	3	28.7 (0.7)	−23.0 (0)	46.7 (0.7)
	Abiotic	1	20.1 (0.2)	−28.8 (0.4)	20.0 (1.2)

^a All N₂O isotope values are the isotopic composition of the N₂O produced, the δ¹⁵N and δ¹⁸O values are the difference from those of the substrate NO₃⁻ (δ¹⁵N_NO3- = 2.4 ± 0.1‰, δ¹⁸O_NO3- = 17.9 ± 0.3‰) or NO₂⁻ (δ¹⁵N_NO2- = −1.8 ± 0.3‰, δ¹⁸O_NO2- = 3.7 ± 0.2‰). The values in parentheses represent the standard deviation obtained from replicate experiments (refer to the raw data in Table S1).

^b Abiotic represents the N₂O production from a chemical reaction without bacteria inoculation under low pH conditions (pH < 2).

^c Sample collection at different incubation times, the start time is from that the substrate NO₃⁻ or NO₂⁻ was injected into the cultures.

Enzymatic pathways of N₂O production during DNRA

To elucidate the enzymatic pathways of nitrate-ammonifying N₂O production, we cultured G. terrae Red111 (containing both Nar and Nap) and O. rubra Red88 (containing only Nap). Initially, abiotic sources of N₂O were excluded, as little or no N₂O production was observed in bacteria-free and autoclaved cultures (Fig. 5a and b), consistent with the differing SP values in nitrate-ammonifiers from the abiotic reaction (Table 1). In contrast, N₂O production was observed in filtered cultures, where live bacteria had been removed, but free and partial periplasmic enzymes were present, suggesting the enzymatic reactions responsible for N₂O production during DNRA (Fig. 5a and b). Growth curves and fluorescence microscopy revealed that kanamycin (100ug/mL) reduced biomass and increased cell mortality of G. terrae Red111 and O. rubra Red88 in non-concentrated cultures (Fig. S6). Consequently, kanamycin was used to prepare cell lysates. Kanamycin addition significantly increased N₂O production in G. terrae Red111 (P < 0.01) but decreased it in O. rubra Red88 (P < 0.01) (Fig. 5a and b). This increase in G. terrae Red111 possibly resulted from the release of more intracellular enzymes, whereas the decrease in O. rubra Red88 may be attributed to the cessation of extracellular/periplasmic enzyme synthesis, suggesting different enzyme profiles between these strains, consistent with their varying responses to c-PTIO (Fig. 4b and c). Moreover, kanamycin slightly increased N₂O production with SNP addition in both strains (P > 0.05) (Fig. S5), suggesting the intracellular enzymes in both strains facilitating the reduction of NO to N₂O. The addition of oxygen significantly decreased N₂O production in both strains with NO₂⁻ or SNP (P < 0.001, Fig. 5a and b; Fig. S5), implying that an oxygen-sensitive enzyme is responsible for N₂O production, aligning with the anaerobic metabolism of DNRA.

Fig 5.

Comparisons of N₂O productions, gene expressions, and gene structures of Geobacteraceae strains. (a and b) N₂O production from the NO₂⁻ reduction process driven by strains Red111 and Red88 under different treatments: Abiotic, bacteria-free medium; CK, bacteria-containing medium; Filter, filtered bacteria-containing medium; Autoclaved, autoclaved bacteria-containing medium; Oxygen, bacteria-containing medium gassed air; Kan+, bacteria-containing medium supplemented with 100 µg/mL kanamycin; Kan + Oxy, bacteria-containing medium gassed air and supplemented with 100 µg/mL kanamycin. The default medium is NNFM_N. (c) The expression levels of nitrogen metabolism related genes from transcriptome analysis in Geomonas sp. Red32 with NO₃⁻ (8 mM, left panel) and NO₂⁻ (2 mM, right panel) inductions. Red colors represent the upregulated genes, while blue colors represent the downregulated genes. Asterisk indicates significant difference, adjusted P < 0.05. (d and e) The expression levels of representative genes related to DNRA and nitrosative stress processes from RT-qPCR in strains Red111 and Red88 with NO₃⁻ (8 mM, right panel) and NO₂⁻ (2 mM, left panel) inductions for 4 and 24 h. (f and g) N₂O production from the NO₂⁻ reduction process driven by strains Red111 and Red88 over time with (grey squares) and without (black circles) chlorate additions. (h) structure comparison of DNRA-related genes (nap and nrf clusters) in three Oryzomonas strains, Red88, Red96, and Red100. Arrows indicate open reading frames and their orientations in the genomes. (i) N₂O production from the NO₂⁻ reduction process driven by nar/nap-absent strains Geomonas bemidjiensis DSM 16622 and Geobacter anodireducens JCM 30203 at different culture times. Asterisks indicate significant difference, *** P < 0.001, **P < 0.01, ns indicates insignificant difference, P > 0.05.

To identify the key enzymes catalyzing N₂O production, we quantified the expression levels of genes related to DNRA and nitrosative stress processes under NO₃⁻, NO₂⁻, and SNP treatments in G. terrae Red111 and O. rubra Red88. The target genes were identified based on transcriptomic results from Geomonas sp. Red32, which showed high conversion ratios from NO₃⁻ and NO₂⁻ to N₂O (Fig. 3d and h) and was selected for transcriptomic study (Table S2). A total of 1,699 and 1,134 differentially expressed genes were detected in Geomonas sp. Red32 under NO₃⁻ and NO₂⁻ treatments, respectively (Fig. S7). Genes related to nitrogen fixation and ammonium assimilation were globally downregulated under both NO₃⁻ and NO₂⁻ treatments, whereas genes involved in DNRA and nitrosative stress were upregulated (Fig. 5c; Table S3). Given that Geomonas sp. Red32 produced N₂O following NO₃⁻ or NO₂⁻ addition (Fig. S8), the upregulated genes in the DNRA and nitrosative stress pathways are connected to N₂O production.

In G. terrae Red111 and O. rubra Red88, most genes associated with DNRA and nitrosative stress pathways were significantly upregulated, except for the genes napA, norB, and qnor in G. terrae Red111 and norB and qnor in O. rubra Red88, which exhibited low or negative expression levels (Fig. 5d and e; Fig. S9). Denitrifying N₂O production driven by cNor (encoded by norBC) and qNor has been reported with SP values less than 0‰ (23, 45), significantly lower than the SP values observed for DNRA-derived N₂O. These findings excluded the possibility of cNor and qNor involved in NO reduction in Geobacteraceae strains. Interestingly, the hcp-hcr gene cluster (encoding the hybrid cluster protein complex, Hcp–Hcr) in the nitrosative stress group showed high expression levels under all conditions (Fig. 5c through e; Fig. S9). Hcp–Hcr is an unusual redox endoenzyme complex with oxygen-sensitive hydroxylamine and NO reductase activity (46), and its function in NO detoxication to N₂O has been confirmed by Hcp mutants in Methylobacter species and E. coli (47, 48). Moreover, hcp-hcr is the only remaining gene cluster related to NO detoxification present in all Geobacteraceae genome sequences (Fig. 2). Given the constant SP values suggesting similar enzymes catalyzing NO to N₂O in different Geobacteraceae strains, we propose the Hcp–Hcr complex as the primary driver of NO reduction to N₂O in these strains. This finding is distinct from the reported enteric nitrate-ammonifiers that use NO reductase (NorVW) or NO dioxygenase (Hmp or Fhp) for NO detoxication (16, 17).

For the enzyme catalyzing NO₂⁻ to NO, we first assume Nar in Geomonas strains, due to its high expression and central role in the gene co-expression network under the NO₂^- treatment in Geomonas sp. Red32 (Fig. 5c and d; Fig. S10). Although Nar is known for catalyzing NO₃⁻ to NO₂⁻, its role in catalyzing NO₂⁻ to NO is less established. To investigate this, we added the Nar inhibitor chlorate (ClO₃⁻) to the cultures (49). ClO₃⁻ showed little effect on bacterial growth and NO reduction to N₂O but completely inhibited NO₃⁻ reduction in Geomonas diazotrophica Red69 (Nar-containing but Nap-absent) and G. terrae Red111 (Nar- and Nap-containing), with no effect on Nap-driven O. rubra Red88 (Fig. S11 and S12). This indicates the Nar-driven type (Nap deactivated) of G. terrae Red111 related to DNRA. ClO₃⁻ addition further completely inhibited N₂O production from NO₂⁻ in G. terrae Red111 but had no effect on O. rubra Red88 (Fig. 5f and g), underscoring the critical role of Nar in N₂O production. After numerous unsuccessful attempts to construct genetic mutants of Geobacteraceae strains, we moved to use another nitrate-ammonifying strain, E. coli MG1655, as an alternative and constructed mutant strains MG1655 ∆narG and MG1655 ∆napA. Compared with the wild type, strain MG1655 ∆narG produced much less N₂O from NO₂⁻, while MG1655 ∆napA produced similar amounts (Fig. S13). Nar is an oxygen-sensitive membrane-bound enzyme in which the catalytic subunit faces the cytoplasm (50), meeting the proposed characteristics of intracellular enzymes. Thus, Nar is proposed as the enzyme responsible for NO₂⁻ to NO conversion in Nar-containing Geomonas strains.

For Nar-deficient Oryzomonas strains, Nap and NrfA are the potential candidates for catalyzing NO₂⁻ to NO, as both are periplasmic enzymes and showed high expression levels in O. rubra Red88 under both NO₃⁻ and NO₂⁻ treatments (Fig. 5e). To determine the responsible one, we cultured the other two bacteria Oryzomonas japonica Red96 and Oryzomonas sagensis Red100 and compared their NO₃⁻ and NO₂⁻ reduction products. All three strains could produce N₂O from NO₃⁻ reduction, but only O. rubra Red88 produced N₂O from NO₂⁻ reduction (Fig. 3c and g; Fig. S14), indicating that the target enzyme is present in all three strains but in difference from O. rubra Red88 to O. japonica Red96 and O. sagensis Red100. Gene cluster comparisons revealed that the napDAB cluster is structurally conserved and shares high sequence similarities (>95%) among the three strains, whereas the nrfAH cluster was distinctly different, as O. rubra Red88 harbors two paralogs of nrfAH (nrfAH1_Red88 and nrfAH2_Red88), and O. japonica Red96 and O. sagensis Red100 harbor only one (nrfAH1_Red88 orthologous) (Fig. 5h; Fig. S15). Moreover, the gene nrfAH2_Red88 showed higher gene expression than nrfAH1_Red88 under NO₃⁻ and NO₂⁻ treatments (Fig. 5e), highlighting the difference of NrfA from O. rubra Red88 to O. japonica Red96 and O. sagensis Red100. These comparisons indicate that NrfA is more likely than Nap to be involved in N₂O production. In addition, two other Geobacteraceae strains, G. bemidjiensis DSM 16622 and Geobacter anodireducens JCM 30203, which lack both nar and nap genes and are incapable of NO₃⁻ reduction, also produced N₂O from NO₂⁻ reduction with comparable N₂O ratios and SP values to other studied nitrate-ammonifiers (Table 1; Fig. 5i; Fig. S16). This finding excluded the necessity of Nap in N₂O production. With these results and the reported crystal structure of NrfA that predicted NO occurrence during NO₂⁻ reduction to NH₄⁺ (51), NrfA is supposed as the target enzyme catalyzing NO production in Nap-driven nitrate-ammonifiers.

Although Geobacteraceae strains contain one or more copies of the nrfA operon, their phenotypes with respect to NO₂⁻ reduction are not consistent. For example, Geomonas sp. Red32, O. japonica Red96 and O. sagensis Red100 each contain a single copy of nrfA, but only Geomonas sp. Red32 cannot produce NH₄⁺ from NO₃⁻. This variability in NrfA functionality suggests evolutionary divergence among these strains. Phylogenetic analyses based on NrfA sequences reveal that the NrfA proteins of Geobacteraceae strains are distributed within Clade I and the outgroup (52), displaying phylogenetic differences at the single species level (Fig. S17). Notably, in Oryzomonas strains (Red96, Red100, and Red88-NrfA1), NrfA forms a robust branch within Clade I-I (bootstrap value 0.9) and shares <75% similarity at the amino acid level with other NrfA enzymes. This sequence specificity of NrfA in Clade I-I likely underlies the differential activities observed between Oryzomonas and Geomonas strains. Most Geomonas strains contain both Nar and NrfA enzymes; however, there is currently no evidence supporting the involvement of these Nrf enzymes in N₂O production. Previous studies have shown that nitrate-ammonifying strains deleting nrfA (Nar-containing type) do not influence N₂O production (16, 17). Given that the Nar complex contains specialized NO₂⁻/NO₃⁻ membrane transporters (NarK)(53), we assume that Nar transforms pericellular NO₂⁻ in high concentration into bacterial cells, thereby alleviating NO₂⁻ pressure on NrfA proteins and inhibiting NO release from NrfA enzymes in Nar-containing bacteria. Although the enzymatic combinations Nar-Hcp and NrfA-Hcp related to DNRA-derived N₂O production were functionally identified in Geobacteraceae strains, these gene combinations narG-hcp and nrfA-hcp are widely distributed across the bacterial domain, including within the phyla Pseudomonadota, Bacteroidota, Desulfobacterota, and Bacillota (Fig. S18). Thus, this study reveals a previously unrecognized, yet potentially widespread, N₂O production pathway in various bacteria.

Distinctive isotopocule signatures of DNRA-derived N₂O with the role in quantifying N₂O production processes

The isotopocule signatures of DNRA-derived N₂O (except abnormal values from mixed processes) range from 43.0‰ to 49.9‰ for SP, −39.9‰ to −5.8‰ for δ¹⁵N^bulk, and 14.1‰ to 30.5‰ for δ¹⁸O (Table 1; Fig. 6a through c). These values are notably distinctive compared to those from other N₂O generation processes (9), particularly the SP values, which surpass those reported for any other processes (Fig. 6a and b), suggesting that SP is a robust and independent tool for identifying DNRA-derived N₂O. In contrast, the δ¹⁵N^bulk and δ¹⁸O values (isotopic fractionation from substrates) of the produced N₂O show greater variability (δ¹⁵N^bulk = −24.6 ± 6.0‰, δ¹⁸O = 23.6 ± 6.3‰) than the SP values and overlap with values from other known N₂O production processes (Table 1; Fig. 5a through c), indicating their limited effectiveness as indicators for N₂O partitioning. However, the combined δ¹⁸O/δ¹⁵N^bulk plot enables clear differentiation of DNRA-related N₂O from other processes, including nitrification, nitrifying-denitrification, and fungal denitrification (Fig. 6c). This highlights the utility of dual isotope plots in distinguishing N₂O sources. Previous studies have noted that the N₂O reduction process mediated by N₂O reductase (NosZ) can elevate SP values appreciably (54), which may obscure the identification of high SP value processes. Our results show that the slope of SP/δ¹⁵N^bulk for residual N₂O caused by N₂O reduction from nitrification and fungal denitrification sources overlaps with the DNRA zone (Fig. 6a), while this slope of SP/δ¹⁸O does not overlap with the DNRA zone (Fig. 6b). Thus, SP/δ¹⁸O provides a more precise means of distinguishing DNRA-derived N₂O from other pathways, offering enhanced resolution compared to SP alone. Indeed, high SP values over 45‰ have been reported in a nitritation-anammox reactor (55), which could be attributed to DNRA-derived N₂O as characterized in this study.

Fig 6.

Comparisons of N₂O isotopocule signatures from different N₂O production reactions. (a–c) SP/δ¹⁵N^bulk, SP/δ¹⁸O, and δ¹⁸O/ δ¹⁵N^bulk plots showing the isotopocule signatures of N₂O from various nitrogen metabolism processes. nD, nitrifier denitrification; bD, bacterial denitrification; fD, fungal denitrification; cD, chemical denitrification; NI, nitrification; DN, DNRA. The black arrows denote the N₂O reduction (N₂O-R) by bacterial denitrification based on early studies [slope ε(SP) / ε(δ¹⁵N) = 0.96, slope ε(SP) / ε(δ¹⁸O) = 0.45, slope ε(δ¹⁸O) / ε(δ¹⁵N) = 2.21]. The value ranges (except for DNRA obtained in this study) are retrieved from Ref (9). (d) SP value comparisons among strains Red69/Red111 (7 days), strain Red330 (7 days), and all Geomonas strains (3 days) with the substrate NO₃⁻. The raw data are presented in Table S1.

N₂O isotopocule signatures also serve as tools for identifying and quantifying mixed N₂O production processes through isotope mass balance. For instance, the N₂O production from NO₃⁻ reduction by G. terrae Red111 at 7-day culture suggests mixed processes, given its lower SP values (41.1‰–41.9‰) than others (Fig. 6d). To elucidate the contributing processes, we introduced another two strains, G. diazotrophica Red69 and Geomonas silvestris Red330, and measured their N₂O isotopocule signatures. Both G. terrae Red111 and G. diazotrophica Red69 showed similar decreases in SP value over prolonged culture periods, whereas G. silvestris Red330 did not (Table 1), despite all strains showing similar NO₃⁻/NO₂⁻ reduction rates (Fig. S18). This discrepancy indicates the differences between G. silvestris Red330 and those two strains causing the decrease in SP values. Genetic comparison revealed that G. diazotrophica Red69 and G. terrae Red111 contain norBC clusters, while G. silvestris Red330 does not (Fig. 2). cNor has been documented to catalyze NO to N₂O during denitrification, with known SP values ranging from 7.5‰ to 3.7‰ (9). Given that NO is a precursor to N₂O in DNRA, the partial involvement of cNor in G. diazotrophica Red69 and G. terrae Red111 likely contributes to the observed SP value reductions over time. Quantitative analysis of mixed N₂O production processes revealed that DNRA-related enzymatic reactions account for 93.9% and 94.0% of N₂O production during NO₃⁻ reduction at the 7-day culture of G. diazotrophica Red69 and G. terrae Red111, respectively. This approach has also been applied to quantify four mixed N₂O⁻ production/reduction processes using N₂O isotope mass balance (56, 57), displaying its significant potentials for environmental research. However, estimating pathways based on only two or three N₂O isotopic dimensions can lead to an underdetermined set of equations. Thus, additional measurements, such as microbial ecological data or pathway modelling approaches, are necessary for accurate determination of extra pathways.

Previous studies on pure cultures of nitrifiers and fungal denitrifiers have demonstrated that N₂O formed from asymmetrical precursors (N-N-O) typically exhibits high SP values (>20‰) owing to sequential binding mechanisms, whereas low SP values around 0‰ occurring in bacterial denitrification involve dimerization of NO (23, 58). The sequential reaction of asymmetrical precursors is thus supposed to account for the synthesis of DNRA-derived N₂O molecules. Although the mechanism for the high SP values in DNRA-derived N₂O is not yet fully elucidated, we hypothesize that specific enzyme activities or catalytic centers govern the sequential binding of NO, resulting in distinct fractionations during the addition of the first or second NO precursor. Further studies involving enzymatic structures is needed to bridge this gap. Coincidentally, the predicted equilibrium SP value of N₂O at room temperature was reported as 45‰ based on theoretical modeling (59), comparable to that of DNRA-derived N₂O. We thus believe that the high SP values of DNRA-derived N₂O may provide novel insight into N₂O formation mechanisms in follow-up studies.

Our results elucidate the N₂O production pathways (NO₂⁻-NO-N₂O) in both Nar- and Nap-type DNRA bacteria within the Geobacteraceae family. We identified two novel enzymatic pathways for N₂O generation in nitrate-ammonifiers: 1) Nar and the Hcp–Hcr complex catalyzed the NO₂⁻-NO-N₂O process in Nar-type nitrate-ammonifiers; 2) NrfA and the Hcp–Hcr complex catalyzed the NO₂⁻-NO-N₂O process in Nap-type nitrate-ammonifiers (Fig. 7). Furthermore, DNRA-derived N₂O exhibits the highest SP values reported to date across various N₂O production pathways in pure culture experiments. Cross plots of SP with δ¹⁵N^bulk and δ¹⁸O values clearly distinguish DNRA-derived N₂O from other known production pathways. These results are crucial for precisely modeling global greenhouse emissions. Nevertheless, the limited nitrate-ammonifiers in one group of Geobacteraceae are insufficient to obtain a comprehensive understanding of DNRA-derived N₂O production, as environmental nitrate-ammonifiers exhibit a range of enzyme combinations related to N metabolism. On this basis, large-scale nitrate-ammonifier screening with N₂O production pathway parsing will be considered in future studies.

Fig 7.

Schematic description of DNRA-derived N₂O production. (a) N₂O production pathway in Nar-driven Geomonas strains. (b) N₂O production pathway in Nap-driven Oryzomonas strains. Red arrows with relative thin lines indicate the N₂O production as a byproduct from NO₂⁻ during the DNRA process. Key enzymes are indicated as nitrate/nitrite antiporter (NarK), respiratory nitrate reductase (NarGHI), periplasmic nitrate reductase (NapAB), cytochrome c nitrite reductase (NrfA), and hybrid cluster protein complex (Hcp–Hcr). The enzymes in different colors denote different enzyme types, green labels the exoenzyme (Nap) or the outer membrane protein (Nrf), brown labels the endoenzyme (NarGHI and Hcp–Hcr complex), and blue labels the membrane transport proteins (NarK and others). The shown ranges of SP values were summarized from the NO₃⁻/NO₂⁻ reduction process driven by Geomonas or Oryzomonas strains in this study.

MATERIALS AND METHODS

Bacterial strains and cultivation

Nine bacterial strains of the family Geobacteraceae and a reference strain E. coli K-12 MG1655 were cultured for this study. Among these, Geomonas sp. Red32 was recently isolated from paddy soils and has been deposited in the Japan Collection of Microorganisms (JCM 33031) and Marine Culture Collection of China (MCCC 1K03692). G. bemidjiensis DSM 16622 was obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (Braunschweig, Germany), while Geobacter anodireducens JCM 30203 was sourced from the Japan Collection of Microorganisms (Tsukuba, Japan). Others, including three Geomonas strains Red69, Red111, and Red330 and three Oryzomonas strains Red88, Red96, and Red100, were described in our previous works (30–32, 34) and reactivated for this study. E. coli K-12 MG1655 was acquired from the National BioResource Project (NBRP)-E. coli (Mishima, Japan). Gene knockout mutants of E. coli K-12 MG1655 were constructed as described previously (60) and hereafter termed as MG1655 ∆nar and MG1655 ∆nap, representing the markerless narG and napA deletion mutants, respectively.

Unless otherwise specified, the nine Geobacteraceae strains were routinely cultured in 50 mL glass serum bottles containing 20 mL of anoxic nitrogen-free freshwater medium (NFFM, pH 6.8) supplemented with 10 mM acetate as the electron donor and carbon source and 10 mM fumarate (NFFM_F) or nitrate (NFFM_N) as the electron acceptor. Cultures were incubated at 30°C without shaking. The composition of NFFM was consistent with that described previously (34), except that NH₄Cl was omitted due to the N₂-fixing capability of these strains. The only non-N₂-fixing strain, Geob. anodireducens JCM 30203, was cultured equally with the addition of 5 mM NH₄Cl to the medium. Serum bottles containing NFFM were neck-sealed with butyl-rubber stoppers and aluminum crimps, with the gaseous phase replaced by N₂/CO₂ (80:20, v/v). E. coli K-12 MG1655 and its mutants were routinely cultured using aerobic LB broth (Nihon Pharmaceutical, Japan) at 37°C with shaking at 120 rpm and anaerobically cultured when monitoring NO₂⁻ reduction using LB broth supplemented with 2 mM NO₂⁻ .

Measurement of NO₃⁻ and NO₂⁻ reduction to ammonium

NO₃⁻ reduction experiments were conducted using the strains cultured in NNFM_N, while NO₂⁻ reduction experiments involved culturing the strains in NNFM_F for 2 days until the biomass reached an optical density of >0.02 (OD₆₀₀) and then 2 mM NaNO₂ was injected into the bottles. The sampling was started from the time of NaNO₂ addition. For sample collection, 1 mL of culture was withdrawn from each bottle at designated time intervals, and the bacterial biomass was measured at OD₆₀₀. After that, the samples were filtered through 0.22 µm syringe filters prior to preserved in a freezer (−20°C) for subsequent measurement. The concentrations of NO₃⁻ and NO₂⁻ were measured using an ion chromatograph (Dionex ICS-900) equipped with a Dionex IonPac AS12A analytical column (Thermo Fisher Scientific, USA). NH₄⁺ concentrations were determined using the indophenol-blue method followed by colorimetry (61). The produced N₂O in the headspace gas of the cultural bottles (1 mL) was measured using a gas chromatograph equipped with a Porapack Q column and an electron capture detector (GC-ECD, GC-2014, Shimadzu, Japan). The temperature of the injector, detector, and column were 90°C, 345°C, and 80°C, respectively, with an Ar/CH₄ carrier gas flow rate of 150 mL/min. The N₂O dissolved in the liquid medium was calculated using the Bunsen absorption coefficient (α = 0.544, 25°C)(62). The effect of carbon-to-nitrate (C:N) ratios on DNRA activities and bacterial growth was investigated using G. terrae Red111 in NNFM_N for 1 week. Different C:N ratios were prepared using a constant NO₃⁻ concentration of 10 mM as the nitrogen source and varying acetate concentrations (5–40 mM) as the carbon source.

Control experiments to explore N₂O production pathways

To assess the contribution of NO to N₂O, a NO scavenger, 2-(4-carboxyphenyl)−4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (c-PTIO, Santa Cruz Biotechnology, USA), was applied during bacterial culture in NNFM_N. The effect of c-PTIO on bacterial growth was evaluated using G. terrae Red111 at various concentrations (0, 50, 100, and 300 µM). NO inhibition tests were performed using 100 µM c-PTIO and quantified by the comparison of N₂O concentration with c-PTIO-free cultures. L-N^G-monomethyl arginine acetate (L-NMMA, Dojindo, Japan), used to investigate the role of endogenous NO in N₂O production, was applied at a work concentration of 100 µM. Hydroxylamine concentrations were measured as described previously (63). The effect of kanamycin (100 μg/mL) on bacterial growth was monitored using growth curves and cell viability assays, with the latter performed via fluorescence microscopy (Carl Zeiss Axio Scope A1, Germany) following bacterial staining with a live/dead cell viability assay kit (Bestbio, China). For treatments under filtered, autoclaved, oxygen-containing and kanamycin-supplemented conditions, cultures were prepared using NNFM_N inoculated with target strains, and then dispensed into 30 serum vials (5 mL culture in 20 mL vials, five replicates/treatment) of each strain once the biomass reached an OD₆₀₀ >0.02. Filtered cultures were treated using 0.22 µm syringe filters, autoclaved cultures were prepared by autoclaving at 121°C for 20 min, and kanamycin-supplemented cultures were prepared by injecting kanamycin solution to a final concentration of 100 µg/mL. The gaseous phase in the vials was then replaced with N₂/CO₂ (80:20, v/v), except for the oxygen-containing cultures. After that, 2 mM NO₂⁻ or 80 µM sodium nitroprusside (SNP, Sigma, USA; NO donor) was injected into sub-cultures or cell-free media as required, which were then cultured at 30 °C for 3 days before N₂O sampling. The effect of sodium chlorate (NaClO₃; 20 mM) on Nar, Nap, and both Nar and Nap containing strains, G. diazotrophica Red69, O. rubra Red88, and G. terrae Red111, respectively, was examined using NNFM_N and NNFM_F supplemented with 2 mM NO₂⁻, the consumed and produced N compounds were measured as described above. ClO₃⁻ concentrations were determined using an ion chromatograph as described above.

Isotopic analysis of N₂O, NO₃⁻, and NO₂⁻

The N₂O isotopocule ratios were measured using an automated system containing an autosampler (GX-222 XL liquid handler; Gilson, USA), a customized purge and trap system, a gas chromatographs (6850; Agilent Technologies, USA) equippled with Pola PLOT column, and an isotope ratio mass spectrometer (IRMS, MAT 253, Thermo Fisher Scientific, Germany) (64). High-purity N₂O (Showa Denko Co., Ltd., purity >99.999%), which was calibrated previously with respect to the international isotopic standards air-N₂ for δ¹⁵N and Vienna standard mean ocean water (VSMOW) for δ¹⁸O, were diluted with high-purity N₂ and measured parallelly to correct and calibrate the ratios. Values of δ¹⁵N^bulk, δ¹⁵N^α, and δ¹⁸O were calculated as described previously (65). Values of δ¹⁵N^β and SP were obtained as follows: δ¹⁵N^β =2δ¹⁵N^bulk - δ¹⁵N^α; SP = δ¹⁵N^α - δ¹⁵N^β. The typical analytical precisions are 0.2‰ for δ¹⁵N^α and 0.1‰ for δ¹⁵N^bulk and δ¹⁸O. Abiotic control N₂O samples were obtained from bacteria-free NNFM supplemented with 2 mM NaNO₂, acidified to pH <2 using 6M HCl.

The δ¹⁵N_NO3- and δ¹⁸O_NO3- values of the substrate NO₃⁻ in the medium were measured by the denitrifier method, where NO₃⁻ was converted into N₂O by a denitrifying bacterium Pseudomonas aureofaciens ATCC 13985. The δ¹⁵N_NO2- and δ¹⁸O_NO2- values of the substrate NO₂⁻ in the medium were measured by chemical conversion of NO₂⁻ into N₂O using the azide method. Detailed descriptions of these approaches and the procedures used for calibration of multiple standards (NO₃⁻, NO₂⁻, and N₂O) for isotopomer ratios are indicated in Ref (66). The isotopologues of the substrate NO₃⁻ were determined as δ¹⁵N_NO3- of 2.4 ± 0.1‰ and δ¹⁸O_NO3- of 17.9 ± 0.3‰, while those of the substrate NO₂⁻ were δ¹⁵N_NO2- of −1.8 ± 0.3‰ and δ¹⁸O_NO2- of 3.7 ± 0.2‰. These isotopologues of the substrates were then used for correcting the absolute δ¹⁵N^bulk and δ¹⁸O values of N₂O.

To clarify the mixed N₂O production processes during NO₃⁻ reduction by strains Red69 and Red111, a quantitative estimation was performed using isotope mass balance. The fractional contributions (f value) to total N₂O production based on SP were expressed as:

$${\displaystyle \mathrm{S}{\mathrm{P}}_{7-\mathrm{d}\mathrm{a}\mathrm{y}}=\mathrm{S}{\mathrm{P}}_{3-\mathrm{d}\mathrm{a}\mathrm{y}}\times {\mathit{f}}_{\mathrm{D}\mathrm{N}\mathrm{R}\mathrm{A}}+\mathrm{S}{\mathrm{P}}_{\mathrm{c}\mathrm{N}\mathrm{o}\mathrm{r}}\times (1-{\mathit{f}}_{\mathrm{D}\mathrm{N}\mathrm{R}\mathrm{A}})}$$

where SP_7-day and SP_3-day represent the measured SP values of strains Red69 and Red111 in NO₃⁻ reduction at the culture time of 7 and 3 days, respectively. SP_cNor denotes the SP values of denitrifying N₂O produced from cNor (−5.9 ± 2.1‰)(67), and f_DNRA indicates the relative contribution of DNRA-mediated process.

Phylogenomic analysis and comparative genomics

The reference genome sequences utilized in this study were retrieved from the NCBI database with accession numbers provided in Table S4. The phylogenomic tree was constructed using the up-to-date bacterial core gene set (UBCG) pipelines equipped with RAxML tool based on the amino acid sequences of 92 concatenated core genes (35). The presence and absence of functional genes related to the N metabolism in Geobacterales strains were determined through genome annotations by BlastKOALA server of the KEGG database (68) and RAST server (ClassicRAST annotation scheme) of the SEED database (69). In cases of inconsistencies between the two annotation servers, manual examination was performed using BLASTp against the KEGG GENES and NCBI-nr databases. For NrfA-based phylogenetic tree construction, amino acid sequences were retrieved from sequenced genomes available in the NCBI database, following the guidance in Ref (52, 70). MEGA X (71) and IQ-TREE (72) were used to align the sequences and generate the trees, respectively. AnnoTree v1.2, a tool used for exploration of functional genes across microbial tree of life, was used to search narG-hcp and nrfA-hcp gene combinations across the bacterial domain (Table S5)(73). All phylogenetic trees were polished by the interactive tree of life (iTOL) v5 (74).

Transcriptome analysis and real-time quantitative PCR

Geomonas sp. Red32 was inoculated into nine serum bottles (5% inoculum size, v/v) containing 20 mL NFFM_F and cultured for 2 days at 30°C. Then, NO₃⁻ and NO₂⁻ stock solutions were individually injected to three of the serum bottles to a final concentration of 10 and 2 mM, respectively. The remaining three serum bottles supplemented with the same amount of autoclaved water were the control group. After a further 4-h incubation, biomass was harvested by centrifugation at 10,000×g for 10 min at 4°C and then frozen in liquid nitrogen prior to storage in a freezer at −80°C. Total RNA was extracted using the total RNA extraction kit (Isogen II, Nippon Gene, Japan) and subsequently sent to Novogene Co. Ltd. (Beijing, China) for transcriptome sequencing on the Illumina NovaSeq 6000 platform. Differential gene expression analysis was performed using Bowtie2 (75) and DESeq2 (version 1.24.0) (76). Gene expression was quantified using the transcripts per million (TPM) method. Differentially regulated genes were defined with the criteria of a more than twofold change in expression and a false discovery rate-adjusted P < 0.05. Co-expressional networks were conducted by Cytoscape (version 3.8.2)(77) based on the Pearson correlation coefficient (Pearson’s r) calculated using R (version 4.0.4).

cDNA was synthesized using the ReverTra AceTM qPCR RT Master Mix with gDNA Remover (Toyobo, Japan). Real-time quantitative PCR (RT-qPCR) was performed using the StepOnePlus Real-Time PCR System (Applied Biosystems, USA) using a 20 µL reaction mixture containing 10 µL of PowerUp SYBR Green PCR Master Mix (2Χ, Thermo fisher), 1 µL of the primer (10 μΜ), 1 µL of the cDNA, and 8 µL of distilled water. The reaction condition was the default program of the instrument with the Fast-cycling model. The primer pairs for RT-qPCR were listed in Table S6. Fold changes of gene expression were calculated by comparative C_T method. The reference gene rpoB was used to normalize the expression levels of target genes in this study (reference gene selection see Supplementary method). The consistency assessment of transcriptional and RT-qPCR results was described in the Supplementary method.

Statistical analyses

Statistical analyses and graphical visualizations were performed using the computing environment R v4.0.4 and GraphPad Prism v8.0.2. Significant differences (P- values) in N₂O/NH₄⁺ productions, bacterial biomass, and chlorate consumptions were assessed using the Student’s t-test for comparisons between two groups, or one-way ANOVA followed by LSD significant difference test for comparisons involving more than two groups. For RT-qPCR results and N₂O isotopocule signatures, statistical significance was determined using the Mann–Whitney U tests for two-group comparisons or Kruskal–Wallis H tests for comparisons involving more than two groups.

DATA AVAILABILITY

The transcriptomic data generated in this study are publicly available in the NCBI under the BioProject accession number PRJNA801072.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mbio.02540-24.

Supplemental Material. mbio.02540-24-s0001.pdf.

Supplemental methods; Figures S1 to S21.

Table S1. mbio.02540-24-s0002.xlsx.

Detailed isotopocule signatures of N₂O produced from NO₃^- and NO₂^- reduction processes by Geobacterales strains and abiotic reactions.

Table S2. mbio.02540-24-s0003.xlsx.

Summary statistics of transcriptomic data and mapping results.

Table S3. mbio.02540-24-s0004.xlsx.

Detailed transcriptomic data focus on the nitrogen metabolism process of strain Red32 under NO₃^- and NO₂^- treatments.

Table S4. mbio.02540-24-s0005.xlsx.

Detailed information of species in the family Geobacteraceae.

Table S5. mbio.02540-24-s0006.xlsx.

Hit numbers and hit ratios of narG-hcp and nrfA-hcp gene combinations in the Bacteria domain.

Table S6. mbio.02540-24-s0007.xlsx.

Primers used in this study.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.