Hypoxia and acidification antagonistically suppress nitrification but enhance its N₂O production in coastal waters

Abstract

The concurrent intensification of hypoxia and acidification poses a growing threat to estuarine-coastal ecosystems globally, yet their interactive impacts on nitrification, a key process in the nitrogen (N) cycle and a substantial source of nitrous oxide (N₂O), remain poorly understood. This knowledge gap hinders reliable projections of future N cycling and climate feedbacks in these critical ecosystems. Here, we demonstrate that along the estuarine-coastal continuum, the co-occurrence of hypoxia and acidification suppresses nitrification rates but stimulates nitrification-derived N₂O emissions—although their combined effect is lesser than the sum of their individual effects. Metatranscriptomic analysis provided molecular evidence for these isotopically traced responses and uncovered critical microbial coping strategies through transcriptional adjustments in metabolic pathways for energy conservation and stress tolerance. This previously overlooked interaction disrupts microbial N cycling with potential impacts on N₂O emissions and climate feedbacks, highlighting the necessity to revise predictive models by incorporating such multistressor dynamics.

INTRODUCTION

Since the mid-20th century, intensified anthropogenic activities, including greenhouse gas emissions, global warming, and excessive nutrient loading, have driven a persistent decline in dissolved oxygen (DO) levels in both open ocean and coastal waters (1, 2). The oceanic oxygen (O₂) inventory has already decreased by ∼2% (2), with projections indicating a potential 10% reduction by the end of the present century, even if anthropogenic carbon dioxide (CO₂) emissions were halted immediately (3, 4). Today, more than 500 hypoxic zones (DO ≤2 mg/liter) have been documented in estuarine and coastal regions around the world, with their frequency and spatial extent continuing to expand (1). Simultaneously, escalating CO₂ emissions have accelerated aquatic acidification, which is recognized alongside deoxygenation as a critical planetary boundary regulating Earth system stability (5).

These interacting stressors converge most intensely in estuarine and coastal ecosystems, which function as dynamic biogeochemical hotspots at the intersection of land and sea (6–9). Driven by intensive human activities, excessive nutrient loading, particularly inorganic nitrogen (N), fuels eutrophication, which stimulates O₂-consuming microbial respiration and leads to hypoxia when combined with water column stratification (1, 8, 10). Coupled with the mineralization of terrestrially derived organic matter, this nutrient-stimulated benthic respiration generates CO₂ internally, markedly accelerating acidification beyond the background rate fueled by atmospheric CO₂ uptake alone (7, 11, 12). Consequently, coastal waters are acidifying more rapidly than the open ocean, having experienced a pH drop of ∼0.21 units in recent decades—more than twice the open-ocean average (13, 14). These stressors may profoundly alter the ecosystem structure and function, imposing severe physiological stress on marine organisms and disrupting microbially mediated biogeochemical processes, particularly those involved in N transformation. Owing to their intricate drivers, hypoxia and acidification frequently co-occur in estuarine and coastal ecosystems (12, 15, 16). Notably, microbial pathways shift from aerobic to anaerobic processes under O₂-limited conditions, leading to the accumulation of acidifying compounds, such as low-molecular-weight organic acids, which further depress pH and create feedback loops that amplify both stressors (17). This mechanistic interdependence necessitates a research paradigm shift from investigating the impacts of hypoxia and acidification in isolation to deciphering their coupled effects and underlying interactive mechanisms on biogeochemical processes in estuarine and coastal ecosystems.

Nitrification (NH₃ → NH₂OH → NO₂⁻ → NO₃⁻) is a central process in the N cycle mediated by diverse nitrifiers, including ammonia-oxidizing bacteria (AOB), ammonia-oxidizing archaea (AOA), complete ammonia oxidizers (comammox), and nitrite-oxidizing bacteria (NOB) (18). This process links reduced and oxidized N pools, thereby mitigating reactive N pollution and alleviating eutrophication when coupled with denitrification or anaerobic ammonium (NH₄⁺) oxidation (anammox) (18, 19). However, nitrification also represents a major biological source of nitrous oxide (N₂O), a potent greenhouse gas that holds the potential to deplete stratospheric ozone (20, 21). Acidification has been shown to decrease nitrification rates while boosting N₂O emissions in estuarine and marine waters (13, 22–24). Similarly, hypoxia impairs ammonia (NH₃) oxidation and alters N transformation pathways because of O₂ limitation (25–27). However, existing studies largely investigated these stressors in isolation. It therefore remains unclear how the frequently co-occurring phenomena of acidification and hypoxia interact to affect nitrification and its associated N₂O production, creating a critical gap in predicting the future of coastal N cycling.

In this study, we used ¹⁵N isotope tracing to investigate the interactive effects of hypoxia and acidification on nitrification and associated N₂O emissions in the Yangtze Estuary and adjacent coastal waters. Metatranscriptomic analysis was further conducted to elucidate the metabolic response of nitrifying microbes. These integrated methodologies provide mechanistic insights into how microbially mediated N transformation processes respond to co-occurring stressors, offering a scientific basis for predicting the evolution of estuarine and coastal ecosystems under the background of future climate change and anthropogenic disturbances.

RESULTS AND DISCUSSION

Responses of nitrification rates to hypoxia and acidification

To investigate the effects of hypoxia and acidification on nitrification activity, near-bottom water samples were collected from three representative sites along the Yangtze River estuary-coastal continuum (fig. S1 and table S1). These samples were incubated under controlled DO and pH gradients established via bubbling with CO₂-O₂-N₂ gas mixtures (table S2). Across all three sites, nitrification rates consistently declined with increasing acidification (Fig. 1A and figs. S2 and S3). In particular, a reduction in nitrification rates ranging from 9.7 to 29.8% was observed when the pH dropped by ∼0.3 units (P < 0.05). This suppression of nitrification rates under reduced pH levels is consistent with previous research in open ocean and estuarine waters (13, 22, 23). These common findings can be explained by two key mechanisms: reduced substrate availability and intracellular pH stress. In particular, acidification shifts the NH₃-NH₄⁺ equilibrium toward NH₄⁺, thereby limiting the supply of NH₃, the primary substrate for NH₃-oxidizing microorganisms (28). Concurrently, the low extracellular pH imposes proton pressure on cells, destabilizing the proton gradient and forcing nitrifiers to expend additional energy to maintain intracellular pH homeostasis (13, 29).

Fig. 1. Effects of hypoxia and acidification on nitrification rates and nitrification-derived N₂O production rates.

(A) Percentage changes in nitrification rates along the acidification gradient under different DO conditions relative to the control treatment (unacidified and normoxic conditions). (B) Percentage changes in nitrification-derived N₂O production rates along the acidification gradient under different DO conditions relative to the control treatment. Data represent the mean responses calculated from three sampling sites with triplicate incubation replicates (n = 9). Error bars indicate the standard deviation of the nine measurements. Shaded areas represent 95% confidence intervals. Equations and P values for the fitted curves are provided in table S3.

Similarly, the results showed that nitrification rates declined significantly with decreasing DO concentrations (Fig. 1A and figs. S2 and S3). A reduction of 10.9 to 23.1% (P < 0.05) was observed as the DO level dropped from normoxic (7.8 to 8.8 mg/liter) to moderately hypoxic (2.1 to 2.7 mg/liter). This inhibition was markedly more pronounced under severe hypoxia (0.7 to 0.9 mg/liter), where rates were reduced by 21.3 to 42.5% (P < 0.05). This sensitivity to hypoxia mainly stems from the aerobic metabolism of nitrifiers, wherein O₂ plays a dual role as both a substrate for ammonia monooxygenase and terminal electron acceptor in the NH₃ oxidation pathway, such that reduced O₂ availability directly suppresses nitrification rates (27, 29, 30). These findings further suggest that although certain AOA taxa (e.g., Nitrosopumilus maritimus SCM1), possibly in addition to a small number of AOB, can generate trace O₂ via nitric oxide dismutation under anoxic conditions (30, 31), diminished DO levels nevertheless lead to decreased nitrification rates.

Notably, despite the frequent co-occurrence of hypoxia and acidification in estuarine and coastal ecosystems (fig. S4), their combined impact on nitrification processes remains poorly understood (12, 15, 16). Our findings demonstrated that simultaneous exposure to both stressors resulted in significantly greater nitrification suppression than individual exposure; however, the combined effect was generally less than the sum of their individual inhibitions, revealing an antagonistic interaction (figs. S2 and S5). Under moderate hypoxia (2.1 to 2.7 mg/liter), a pH decline of ∼0.3 units led to combined nitrification inhibition of 21.1 to 35.4% (P < 0.05), while under severe hypoxia (0.7 to 0.9 mg/liter), the inhibition was higher at 30.9 to 52.8% (P < 0.05). However, these ranges generally fell below the theoretical additive inhibition ranges of 20.6 to 52.9% under moderate hypoxia and 31.0 to 60.3% under severe hypoxia, respectively (fig. S5). This antagonistic interaction contrasts with the predominantly additive or synergistic negative effects reported for marine microbiota under hypoxia and acidification (16). We hypothesize that this mechanistic distinction may stem from hypoxia-induced metabolic suppression in nitrifiers, which indirectly alleviates the energetic burden required for cellular homeostasis (e.g., proton pumping) under acidification, thereby partially offsetting the combined inhibitory effect of hypoxia and acidification. Nitrification represents a key step linking reduced and oxidized N pools, playing a crucial role in supplying substrates for downstream denitrification and anammox processes (18, 32) and consequently performing the essential function of removing anthropogenic reactive N inputs from estuarine and coastal ecosystems (33). However, our findings highlight the risk that the globally widespread and increasingly frequent co-occurrence of hypoxia and acidification may suppress nitrification rates, ultimately compromising the N removal capacity of these vital ecosystems and exacerbating eutrophication.

Effects of hypoxia and acidification on nitrification-derived N₂O emissions

In contrast to their inhibitory effects on nitrification rates, acidification and hypoxia significantly enhanced nitrification-derived N₂O emissions in estuarine and coastal ecosystems (Fig. 1 and figs. S2 and S6). A pH decline of ∼0.3 units stimulated nitrification-related N₂O production rates by 28.3 to 73.5% (P < 0.05). Likewise, a shift from normoxic (7.8 to 8.7 mg/liter) to moderately hypoxic conditions (2.1 to 2.8 mg/liter) resulted in an increase of 53.4 to 96.5% (P < 0.05). These response patterns align with previous reports, demonstrating that environmental stressors such as hypoxia, acidification, and toxic pollution tend to suppress nitrification activity while simultaneously enhancing the production and release of N₂O as a by-product (13, 23, 34, 35). Our findings further indicated that despite the fundamentally distinct N₂O production mechanisms among different NH₃-oxidizing microorganisms (36–39) and the significantly divergent nitrifying microbial communities along the estuary-coastal continuum (fig. S7), their responses to acidification and hypoxia followed similar trends.

Our interactive simulation experiments revealed that the coupled stress of acidification and hypoxia exerted an antagonistic (subadditive) stimulatory effect on nitrification-derived N₂O emissions (figs. S5 and S6). When acidification (a pH decline of ∼0.3 units) was superimposed on different hypoxic levels, N₂O emissions increased by 86.4 to 148.6% under moderate hypoxia (2.1 to 2.8 mg/liter) and 100.2 to 162.0% under severe hypoxia (0.7 to 0.9 mg/liter) (P < 0.05). Notably, these combined effects were generally lower than the sum of the promotional effects of the individual stressors (fig. S5). Furthermore, we found that the DO level modulated the promotional effect of acidification-driven N₂O emissions released from nitrification (fig. S8). In particular, the stimulation efficiency of N₂O emissions by acidification decreased under lower DO conditions. Similarly, the promotional efficiency of reduced DO also exhibited a declining trend with the intensification of acidification (fig. S8). These findings underscore the need to incorporate the coupled effects of acidification and hypoxia, rather than simply summing their individual impacts, when developing predictive models for future coastal and estuarine N₂O emissions.

Assuming the representativeness of our results and that half of estuarine-coastal N₂O emissions are derived from nitrifiers (Supplementary Text) (40–44), Monte Carlo simulations yielded an estimated median global increase of 0.96 Tg N₂O-N per year [interquartile range (IQR): 0.28 to 2.45 Tg N₂O-N per year], driven by coastal acidification (average pH decline of ∼0.21) in nonhypoxic waters and by co-occurring hypoxia-acidification (average pH decline of ∼0.56 and DO ≤2 mg/liter) in hypoxic waters (Fig. 2, figs. S10 and S11, and Supplementary Text). This increment corresponds to 14.3% of the current global anthropogenic N₂O emissions (6.7 Tg N₂O-N per year) (45). Under a future scenario in which the hypoxic area doubles and the hypoxic season is extended by 1 month (46, 47), emissions solely derived from hypoxic regions are projected to increase by a median of 0.05 Tg N₂O-N per year (IQR: 0.02 to 0.14 Tg N₂O-N per year), contributing an additional 0.8% to global anthropogenic N₂O emissions (Fig. 2 and table S7). However, these estimates are subject to inherent uncertainty, particularly because the short-term incubation design captures acute physiological responses to environmental perturbations, which may not fully reflect longer-term ecological outcomes involving microbial selection and adaptation. How nitrifying communities respond to these stressors over extended timescales remains a key topic for future investigation. Moreover, it should be noted that assuming that half of estuarine-coastal N₂O emissions are generated by nitrification is necessarily a simplification, given that N₂O production pathways and emission magnitudes vary widely across estuarine ecosystems. While this assumption provides a reasonable starting point (40, 41), a more systematic quantification of the relative contributions of nitrification across diverse estuarine-coastal ecosystems represents an important direction for future work. Furthermore, the quantitative understanding of the interactive effects of acidification and hypoxia remains limited, and our parameterization scheme therefore relies primarily on response coefficients. Given that estuarine systems around the world differ substantially in terms of nitrifier community composition, nutrient regimes, and other environmental factors, extrapolating these findings to other estuaries carries additional uncertainty. Nevertheless, these estimates offer a quantitative baseline for exploring the effects of co-occurring hypoxia and acidification under current knowledge constraints.

Fig. 2. Global estimation of N₂O emissions from nitrification processes in estuarine and coastal waters under acidification and hypoxia.

(A) Global distribution of estuarine and coastal sites with DO concentrations below 2 mg/liter (<63 μM). Pie charts use double visual encoding, where the angular extent of red sectors reflects the proportion of months under hypoxic conditions, while the overall chart size conceptually represents the hypoxic area (table S4). Red triangles indicate sites where hypoxia was detected. Data source: World Resources Institute and references (1, 73). (B) Fitted gamma distribution of N₂O flux (μmol m⁻² day⁻¹). Values represent the median (25th to 75th percentiles). Data are shown in table S5 and fig. S9. d, day. (C) Fitted lognormal distribution of hypoxic duration. Values represent the median (25th to 75th percentiles). Data are shown in table S6. (D) Fitted normal distributions of hypoxia-induced and multiyear pH decline. Dashed lines and adjacent values represent mean levels. (E) Probability density functions of increased N₂O emissions [Tg N₂O-N per year (yr)] due to coastal acidification derived via Monte Carlo uncertainty propagation. Values represent the median (25th to 75th percentiles). (F) Probability density functions of increased N₂O emissions (Tg N₂O-N per year) under combined acidification and hypoxia derived via Monte Carlo uncertainty propagation. Values represent the median (25th to 75th percentiles). (G) Probability density functions of enhanced N₂O emissions (Tg N₂O-N per year) in hypoxic zones under four scenarios: hypoxia only, acidification only, combined acidification-hypoxia, and a future projection with a doubled hypoxic area and a 1-month extended duration. Values represent the median (25th to 75th percentiles).

Metabolic response mechanisms of nitrifiers under hypoxia and acidification

In response to environmental stress, nitrifiers can modulate their gene expression via intracellular signaling networks (13, 48), thereby regulating nitrification rates and N₂O production. Metatranscriptomic sequencing is a powerful tool for profiling microbial gene expression responses under the stresses of hypoxia and acidification. However, in complex environments such as estuaries and coastal waters, the low nitrifier abundance (<5%; fig. S12A) coupled with high levels of environmental interference impedes the acquisition of high-quality transcripts, constraining the in-depth analysis of their metabolic response mechanisms. To overcome this challenge, we established a continuous-flow environmental simulation system inoculated with samples from site S2 to enrich for nitrifying microorganisms and create a simplified, controllable model community. This approach generated a simplified but controllable nitrifying community model that was capable of robustly simulating the stress responses of natural nitrifying communities. After about 70 days of enrichment, nitrifiers accounted for ∼20% of the total microbial community, dominated by Nitrosomonas, Nitrosopumilus, Nitrospira, and Nitrospina—the predominant nitrifying taxa detected in the environmental samples (fig. S12A). Under reduced DO and low-pH conditions, the enrichment cultures also exhibited significantly reduced nitrification rates coupled with elevated N₂O emissions (fig. S12B), validating the patterns detected in the original samples (Fig. 1).

Metatranscriptomic analyses revealed that hypoxia and acidification substantially perturbed the expression of genes in nitrifiers that underpin key processes, including N transformation, energy metabolism, and CO₂ fixation (Figs. 3 and 4, fig. S13, and table S9). Consistent with the suppression of nitrification rates under acidification-hypoxia (Fig. 1A and fig. S2), we detected a concomitant down-regulation in the expression of the ammonia monooxygenase gene (amoA), which is pivotal for NH₃ oxidation. In particular, the expression of bacterial amoA was reduced by 28.3, 20.7, and 60.9% under acidification, hypoxia, and both stressors combined, respectively (Fig. 3A), whereas the corresponding decrease in archaeal amoA expression was markedly less severe (11.8, 8.0, and 2.1%; Fig. 3B), suggesting that AOA had higher tolerance toward these imposed stressors. This variation is likely due to the stronger NH₃ affinity of AOA (24, 49, 50), which confers a competitive advantage under acidification-induced decreases in NH₃ availability. Consistent with this, under acidification, the expression of low-affinity Rh-type NH₄⁺ transporters of AOB showed an increasing trend, likely as a compensatory response to maintain NH₃ uptake, whereas the expression of MEP-type high-affinity transporters of AOA displayed a decreasing trend, reflecting an energy-conserving strategy enabled by their intrinsically high NH₃ affinity (fig. S13) (51–53). Moreover, AOA have been reported to exhibit a higher affinity for O₂ than AOB (29), which likely contributes to their greater resilience under hypoxic conditions. Certain AOA (e.g., N. maritimus SCM1) may even produce trace O₂ via nitric oxide dismutation under anoxia (31), representing a distinct survival strategy. Despite these adaptive traits, the results showed that overall nitrification rates were still significantly suppressed by acidification and hypoxia (Fig. 1A and fig. S2), underscoring the severity of these combined stressors. Notably, although the metatranscriptomic data indicated that AOA showed higher tolerance at the transcriptional level, the inhibition of nitrification was not alleviated at the AOA-dominated coastal area (Fig. 1A and fig. S7). This discrepancy is likely because enrichment culture transcriptomes reflect the intrinsic physiological potential of AOA, whereas in situ nitrification rates reflect how that potential is modulated by local environmental conditions, such as ambient NH₄⁺ availability (22).

Fig. 3. Response of nitrifiers to hypoxia and acidification in estuarine and coastal waters.

Schematic model illustrating the impact of hypoxia and acidification on the expression of genes involved in stepwise ammonia oxidation, N₂O production, energy generation, and cytosolic pH homeostasis in nitrifiers. Fold changes in gene expression were calculated by comparing manipulated treatments to the ambient control. Roman numerals denote enzyme complexes I [NADH (reduced form of NAD⁺) dehydrogenase], III (cytochrome bc₁ complex), IV (cytochrome c oxidase), and V (ATP synthase) of the respiratory chain. Dotted purple arrows represent electron flow; “E” represents unidentified enzymes. (A) Schematic model for nitrifying bacteria. The dotted line represents the membrane, indicating that nitrite oxidation and ammonia oxidation typically occur in different organisms, except in comammox Nitrospira. Solid blue arrows denote the NH₂OH oxidation pathway, and solid purple arrows indicate the nitrifier denitrification pathway that contributes to N₂O production. (B) Schematic model for AOA. Definitions of the abbreviations are shown in table S8. FC, fold change.

Fig. 4. Expression of carbon fixation genes in nitrifying microbes under hypoxia and acidification.

(A) Calvin-Benson-Bassham (CBB) cycle. (B) Reductive tricarboxylic acid (rTCA) cycle. (C) 3-HP/4-HB cycle. From left to right, the three columns represent acidified, hypoxic, and coupling treatments. (D to F) Density plots for the CBB, rTCA, and 3-HP/4-HB cycles. Fold changes in gene expression were calculated by comparing manipulated treatments to the ambient control. Definitions of the abbreviations are provided in table S8.

Consistent with the transcriptional down-regulation of amoA, the expression of the hydroxylamine dehydrogenase gene (hao) was down-regulated by 16.7, 10.0, and 18.5% under acidification, hypoxia, and combined stress, respectively. Similarly, the nitrite oxidoreductase genes (nxrA and nxrB) exhibited decreased expression, with nxrA expression declining by 1.6, 4.5, and 5.7% under acidification, hypoxia, and combined stress, respectively, while nxrB expression dropped by 11.5, 28.9, and 29.7%, respectively (Fig. 3A). Overall, transcriptional profiling revealed a general down-regulation of functional genes mediating the stepwise oxidation of NH₃ to NO₃⁻ across all stress conditions.

In contrast, we observed that transcripts encoding nitric oxide reductase (nor; N₂O-forming) were up-regulated in AOB under acidification and hypoxia (Fig. 3A). In particular, acidification, hypoxia, and combined treatment resulted in 0.6-, 1.1-, and 2.4-fold increases in norC expression, respectively, and 0.6-, 1.5-, and 2.5-fold increases in norQ expression, respectively (Fig. 3A and fig. S13). These patterns provide molecular evidence that sheds light on the observed stimulation of N₂O emissions by these stressors (Fig. 1B). However, the expression of the nitrite reductase gene nirK in AOB was down-regulated under acidification (reduced to 93.7%), which contrasted with its 1.7-fold up-regulation under hypoxia. In addition, this hypoxic up-regulation was mitigated under concomitant acidification (1.3-fold), which implied that the nitrifier-denitrification pathway for N₂O production was enhanced by low DO but suppressed under acidification (34). Therefore, the acidification-induced stimulation of N₂O emissions may primarily arise from the impact of acidification on the NH₂OH oxidation pathway. Acidification likely inhibits the complete oxidation of NH₂OH to NO₂⁻, leading to the accumulation of the intermediate NO (54), which in turn up-regulates nor gene expression and boosts N₂O emissions. The presumed accumulated NO may also directly react with NH₂OH via the NO-driven hybrid pathway, further contributing to N₂O production. In addition, the HNO₂-driven hybrid pathway may be enhanced under acidification, as reduced pH shifts the NO₂⁻/HNO₂ equilibrium toward HNO₂, thus accelerating the chemical reaction between NH₂OH and HNO₂ (55). Under reduced DO, the potentially elevated NO availability associated with nirK up-regulation in AOB may similarly promote the NO-driven hybrid pathway, competitively consuming NH₂OH to produce N₂O. Although previous reports implicated cytochrome P460 in the production of N₂O from NH₂OH (21, 56), its expression decreased under acidification and hypoxia in the present study (Fig. 3A), suggesting that it did not contribute to the observed increase in N₂O emissions.

It should be noted that hypoxia can stimulate N₂O production not only via nitrifier denitrification but also by denitrifying bacteria. In this study, however, the use of ¹⁵NH₄⁺ isotope tracing ruled out contributions from processes other than NH₃ oxidation. Moreover, although coupled nitrification-denitrification might occur under hypoxia, its contribution would be negligible because the produced ¹⁵NO₃⁻ accounted for only a minor portion of the total NO₃⁻ pool (<1%; table S10). In addition, denitrification is known to be inhibited at oxygen concentrations above ∼6 μM (57, 58), and the oxygen levels in our hypoxic treatment consistently remained above this threshold (>20 μM). Nevertheless, we conducted direct measurements of denitrification activity using ¹⁵NO₃⁻ tracing, which yielded rates below the detection limit (Supplementary Text).

While archaeal N₂O production has been commonly attributed to abiotic hybrid reactions between NH₂OH and NO or HNO₂ (38, 59), presumptive enzymes, including copper-containing NH₂OH oxidoreductases and putative nitrosyl oxidases, have been proposed to mediate N₂O production in AOA (31, 39). However, transcripts for these candidate enzymes were not detected in our experiments (Fig. 3B). The expression of archaeal nirK [proposed to oxidize NH₂OH to N₂O or participate in nitrifier-denitrification (60)] was down-regulated under all stress conditions, suggesting its limited contribution to enhanced N₂O production. In contrast, transcripts of a putative archaeal nitric oxide reductase nor gene rose 0.2-, 0.9-, and 1.3-fold under acidification, hypoxia, and combined stress, respectively (Fig. 3B). However, the mechanistic basis for AOA-driven N₂O production remains limited, and thus, its response mechanism to acidification and hypoxia requires further elucidation.

Metatranscriptomic data further revealed coping strategies in the electron transport chain and carbon assimilation of nitrifiers in response to acidification-hypoxia stress (Figs. 3 and 4). Under reduced pH conditions, the up-regulation (1.8-fold) of the V-type adenosine triphosphatase (ATPase) gene [encoding an adenosine 5′-triphosphate (ATP)–dependent proton pump] in nitrifying bacteria implies potential cytoplasmic acidification, thereby necessitating enhanced proton efflux to maintain pH homeostasis (13, 61). In addition, more energy might also be allocated for the translocation of substrates across membranes, as suggested by the up-regulation of gene transcripts encoding the corresponding ATP-binding cassette transporters (17.2 to 68.4%; Fig. 3). However, the total ATP production was likely reduced, as transcripts of genes encoding bacterial F-type ATPases, which functioned as proton motive force (PMF)–driven ATP synthases (61), were down-regulated under acidification. Consequently, less ATP would have been available to fuel chemoautotrophic growth, which was consistent with the ubiquitous down-regulation of gene transcripts involved in the CO₂ fixation of nitrifying bacteria (the Calvin-Benson-Bassham cycle for Nitrosomonas and the reductive tricarboxylic acid cycle for Nitrospira) under acidified conditions (reduced to 37.5 to 99.5% and 36.3 to 97.4%; Fig. 4). However, nitrifying bacteria generally displayed the up-regulated expression of genes encoding electron transport chain complexes under hypoxia (Fig. 3). Under hypoxic conditions, the decreased availability of O₂ (the terminal electron acceptor) disrupts electron flow, thereby impairing the PMF and inhibiting ATP production (62, 63). Therefore, the up-regulation of genes encoding electron transport chain complexes represents a cellular adaptation to sustain electron flow for proton pumping, thereby maintaining the PMF to support essential cellular processes.

Moreover, under energy shortage, cells may prioritize the maintenance of essential biosynthetic capacity as a fundamental trade-off (64). This hypothesis of prioritization may explain the observed up-regulation in the expression of carbon fixation–related genes in nitrifying bacteria under hypoxic conditions (Fig. 4, A and B). Alternatively, or possibly concurrently, this up-regulation might serve as a response to electron accumulation resulting from a bottleneck in the electron transport chain under hypoxia (63). To alleviate this stress, nitrifying bacteria may up-regulate carbon fixation pathways, thereby creating an alternative electron sink to dissipate the reducing equivalents (64). However, the fact that nitrification rates were concurrently suppressed under hypoxia suggested that the sustainability of such a response was uncertain, as CO₂ fixation was unlikely to be maintained without the energy supply provided by nitrification. Under combined acidification and hypoxia, most genes related to carbon fixation in nitrifying bacteria were down-regulated, revealing the suppression of their carbon assimilation activity (Fig. 4, A and B). In contrast, gene transcripts involved in the 3-hydroxypropionate/4-hydroxybutyrate (3-HP/4-HB) carbon fixation pathway in AOA were up-regulated not only under hypoxia but also under acidification and combined stress conditions (Fig. 4C). The consistent up-regulation of transcripts encoding the energetically efficient 3-HP/4-HB pathway under all tested stressors suggests its possible involvement in the metabolic resilience of AOA (65, 66). Genes encoding complexes I and III and A-type ATPase were also up-regulated in AOA under the same stress conditions (Fig. 3B), a pattern that could help maintain the PMF as well as ATP synthesis. However, whether this coordinated transcriptional response reflects enhanced functional carbon fixation or merely represents a transcriptional stress signature remains to be explored using pure cultures. Moreover, it is increasingly recognized that AOA may have alternative energy-yielding pathways beyond NH₃ oxidation, such as the use of organic substrates or H₂ (29). The potential involvement of such pathways in the up-regulation of carbon fixation genes under stress conditions warrants further investigation.

This study reveals that the intensifying co-occurrence of acidification and hypoxia disrupts coastal N cycling through a previously unrecognized, subadditive antagonistic interaction that suppresses nitrification rates while enhancing N₂O production (fig. S14). The results of metatranscriptomic analysis indicate that this response is linked to profound transcriptional reprogramming in nitrifiers that involves the down-regulation of key nitrification genes, the concurrent up-regulation of N₂O-producing genes, and transcriptional adjustments in electron transport and carbon assimilation pathways as an acclimation strategy. Scaling these findings on the basis of Monte Carlo simulations, we estimate that this interplay contributes a median increase of 0.96 (IQR: 0.28 to 2.45) Tg N₂O-N per year to global estuarine-coastal emissions, representing concerning positive climate feedback. Furthermore, the suppression of nitrification by the co-occurrence of these stressors is expected to impair the coupled nitrification-denitrification process, thereby weakening the N removal capacity of these ecosystems and exacerbating coastal eutrophication. These results provide crucial insights into the underlying mechanisms through which co-occurring acidification and hypoxia affect nitrification and associated N₂O emissions in estuarine-coastal ecosystems. This mechanistic understanding underscores the need for predictive models to integrate such multistressor interactions, which is essential for predicting the evolution of N cycling and its climatic feedbacks under intensifying human disturbances and climate pressures.

MATERIALS AND METHODS

Study area and sample collection

The Yangtze River estuary and adjacent coastal waters were selected as the study region because of the severe local co-occurrence of hypoxia and acidification (17). Sampling was conducted at three representative sites (S1 to S3) along the estuary-coastal continuum (fig. S1). Near-bottom water samples were collected in October 2023 during a research cruise supported by the National Natural Science Foundation of China using a conductivity-temperature-depth rosette profiler equipped with Niskin-X bottles. Subsamples were immediately filtered through sterile 0.22-μm polycarbonate membranes (Millipore) to quantify the dissolved inorganic carbon, NO₃⁻, NO₂⁻, and NH₄⁺ contents. Filters were stored at −20°C for subsequent DNA extraction, while the remaining unfiltered water was preserved in darkness on ice and transported to the laboratory for incubation experiments and physicochemical analyses. Detailed environmental parameters for each sampling site are presented in table S1.

Incubation experiments

Freshwater samples were subjected to a full-factorial design comprising five pH reduction gradients (ΔpH from −0.1 to −1.0) and three DO treatments (normoxic: 7.8 to 8.8 mg/liter; moderately hypoxic: 2.1 to 2.8 mg/liter; severely hypoxic: 0.7 to 1.0 mg/liter). For nitrification rate measurements, near-bottom water samples were transferred in triplicate into 120-ml glass bottles, which were immediately sealed with butyl rubber stoppers. For the parallel assessment of N₂O production rates, 150-ml aliquots were introduced in triplicate into 250-ml glass bottles and similarly sealed. To simulate different levels of acidification and hypoxia, samples were bubbled for at least 30 min with premixed N₂-O₂-CO₂ gas mixtures, which allowed for the independent control of dissolved CO₂ (governing pH) and O₂ concentrations. Subsequently, ¹⁵N-labeled NH₄Cl (>98 atom % ¹⁵N) was added to the samples at <20% of the ambient NH₄⁺ concentration. Samples were incubated in the dark at in situ temperature on an orbital shaker (∼30 rpm) for 24 hours (table S1), and the DO and pH levels were measured at the end of the incubation period to validate the targeted conditions (table S2).

Measurements of nitrification and nitrification-derived N₂O production rates

After incubation, samples for nitrification rate analysis were immediately filtered through 0.22-μm syringe filters (Millipore). A portion of the filtered water was used to determine the NO_x⁻ (NO₂⁻ and NO₃⁻) concentrations, while another portion was analyzed to obtain the δ¹⁵N ratios of NO_x⁻ according to the denitrifier method (67). The nitrification rates were calculated using the following equation (68)

$$R_{\text{nitrification}}=\frac{(R_t{{\text{NO}}_x}^{-}\times {\left[{{\text{NO}}_x}^{-}\right]}_t)-(R_{t_0}{{\text{NO}}_x}^{-}\times {\left[{{\text{NO}}_x}^{-}\right]}_{t_0})}{t-t_0}\times \frac{1}{F}$$ (1)

where $R_{\text{nitrification}}$ is the nitrification rate (nmol liter⁻¹ hour⁻¹); $R_{t_0}{{\text{NO}}_x}^{-}$ and $R_t{{\text{NO}}_x}^{-}$ represent the atom % ¹⁵N in NO_x⁻ at the beginning ($t_0$) and end ($t$) of incubation, respectively; and ${\left[{{\text{NO}}_x}^{-}\right]}_{t_0}$ and ${\left[{{\text{NO}}_x}^{-}\right]}_t$ indicate the corresponding NO_x⁻ concentrations, respectively. F is the fraction of labeled NH₄⁺, which is calculated as

$$F=\frac{\left[{}^{15}\text{NH}_4^{+}\right]}{\left[{}^{14}\text{NH}_4^{+}\right]+\left[{}^{15}\text{NH}_4^{+}\right]}$$ (2)

where $\left[{}^{14}\text{NH}_4^{+}\right]$ and $\left[{}^{15}\text{NH}_4^{+}\right]$ refer to the ambient and added ¹⁵N-labeled NH₄⁺ concentrations, respectively.

The application of ¹⁵NH₄⁺ tracing enabled the differentiation and quantification of nitrification-derived N₂O production rates (69). At the end of the incubation period, microbial activity was terminated by adding a saturated ZnCl₂ solution. Headspace gas samples were collected and analyzed to obtain the N₂O concentrations using a gas chromatograph (GC-2014, Shimadzu, Kyoto, Japan). Isotopic compositions [m/z (mass/charge ratio) = 44, 45, and 46] were determined using an isotope ratio mass spectrometer (Delta V Advantage, Thermo Fisher Scientific). The nitrification-derived N₂O production rates were calculated on the basis of the accumulation of ⁴⁵N₂O and ⁴⁶N₂O during the incubation period according to the following equation (69)

$$R_{{\mathrm{N}}_2\mathrm{O}}=\frac{1}{F}\times (\frac{{}^{45}\mathrm{N}_2{\mathrm{O}}_t-{}^{45}\mathrm{N}_2{\mathrm{O}}_t^0}{t-t_0}+2\times \frac{{}^{46}\mathrm{N}_2{\mathrm{O}}_t-{}^{46}\mathrm{N}_2{\mathrm{O}}_t^0}{t-t_0}\times \frac{1}{F})\times \frac{1}{V}$$ (3)

where $R_{{\mathrm{N}}_2\mathrm{O}}$ is the nitrification-derived N₂O production rate (N₂O-N pmol liter⁻¹ hour⁻¹); ${}^{45}\mathrm{N}_2{\mathrm{O}}_t^0$ and ${}^{45}\mathrm{N}_2{\mathrm{O}}_t$ represent the concentrations of ⁴⁵N₂O at the beginning and end of incubation, respectively; ${}^{46}\mathrm{N}_2{\mathrm{O}}_t^0$ and ${}^{46}\mathrm{N}_2{\mathrm{O}}_t$ denote the concentrations of ⁴⁶N₂O at the beginning and end of incubation, respectively; V indicates the sample volume (liters); and F is the fraction of labeled NH₄⁺, as defined above. It should be noted that the ¹⁵N-NH₄⁺ tracing method cannot resolve the contribution of the hybrid pathway. This is partly because ⁴⁵N₂O can originate from either the hybrid pathway or nitrifier denitrification with mixed N sources, while ⁴⁶N₂O can be produced by hydroxylamine oxidation or via uniformly labeled hybrid formation and nitrifier denitrification.

Nitrifier enrichment

Although nitrifiers play a crucial role in the global N cycle, they constitute only a small fraction of the complex microbial communities in estuarine and coastal ecosystems. This makes it challenging to extract sufficient transcripts to decipher their metabolic response mechanisms under hypoxia and acidification. To address this challenge, we established continuous-flow environmental simulators to enrich nitrifiers. Freshwater samples (∼3 liters) from site S2 were added to a sterilized (121°C for 30 min) Minifors2 bioreactor with a working volume of 4 liters, and the temperature was maintained at 25°C using an automatic heating plate coupled with a cold circulating water bath. The saturated DO level was maintained by continuously bubbling air through sterile 0.22-μm filters. Air bubbles were gently dispersed by two stirrers (∼30 rpm) installed at the bottom and above the inlet airflow to ensure homogeneity and maintain the gas-liquid equilibrium. The pH was maintained at an ambient level of 7.9 using an automatic acid-base regulator through the addition of 0.2 M NaOH, which compensated for the H⁺ released during NH₃ oxidation (70). Sterilized site water containing NH₄⁺ was continuously supplied at a rate of ∼500 ml/day to maintain NH₄⁺ concentrations below 5 μM. Effluent samples (30 ml) were collected daily and immediately filtered via 0.22-μm membranes (Whatman), after which they were stored at −20°C for NH₄⁺, NO₂⁻, and NO₃⁻ analyses. Additional samples were centrifuged (20,000g, 5 min) for DNA extraction and subsequent pyrosequencing. After 70 days of enrichment, the biomass was harvested and subjected to 24-hour stress incubation at 25°C in darkness under four different conditions: control [pH 7.9, DO (8.2 mg/liter)], acidification [pH 7.6, DO (8.2 mg/liter)], hypoxia [pH 7.9, DO (0.9 mg/liter)], and combined acidification-hypoxia [pH 7.6, DO (0.9 mg/liter)]. Following incubation, samples were immediately snap-frozen in liquid nitrogen for subsequent total RNA extraction and metatranscriptomic analyses.

RNA extraction and metatranscriptomic analyses

Total RNA was extracted using the EZNA RNA kit (Omega Bio-tek, Norcross), and residual genomic DNA was removed using the Turbo DNA-free kit (Ambion). The absence of genomic DNA contamination was confirmed via polymerase chain reaction with primers 515F and 909R (35). The RNA purity, integrity, and concentration were assessed using the Agilent 2100 Bioanalyzer and Nanodrop 2000 (Thermo Fisher Scientific). To isolate mRNA, ribosomal RNA was selectively eliminated using the Ribo-Zero rRNA Removal Kit (Epicentre). Metatranscriptomic cDNA libraries were then constructed with the TruSeq RNA Sample Prep Kit (Illumina) and sequenced on the Illumina HiSeq 4000 platform. Raw reads were quality checked using FastQC and trimmed with SeqPrep and Sickle. Low-quality reads (quality score <20), reads shorter than 50 base pairs (bp), and sequences containing ambiguous bases (N) were discarded. Ribosomal RNA contamination was further screened and removed using SortMeRNA (version 4.3.6). Trinity (version 2.15.1) was used for the de novo assembly of high-quality mRNA reads, with contigs shorter than 300 bp excluded (71). Open reading frames longer than 100 bp were predicted using MetaGeneMark, and nonredundant contigs (95% identity and 90% coverage) were clustered using CD-HIT. Taxonomic classifications of transcripts were assigned on the basis of hits to the National Center for Biotechnology Information NR database (BLASTP, e-value ≤10⁻⁵). Functional annotations were determined on the basis of homology with proteins in the Kyoto Encyclopedia of Genes and Genomes database (72).

Statistical analysis

Data visualization and analysis were conducted in Origin 2024b and CorelDraw 2024. Statistical differences between treatments were assessed using a one-way analysis of variance (ANOVA) with Tukey’s post hoc tests in SPSS 27.0. Data distribution fitting and Monte Carlo simulations were performed using Python 3.9 with scientific computing libraries including NumPy, SciPy, and Pandas. Detailed descriptions of the Monte Carlo simulation and uncertainty analysis are provided in the Supplementary Text. P < 0.05 was considered to indicate statistical significance.

Supplementary Materials

The PDF file includes:

Supplementary Text

Figs. S1 to S16

Legends for tables S1 to S15

References

Other Supplementary Material for this manuscript includes the following:

Tables S1 to S15