Reduced methane emissions in transgenic rice genotypes are associated with altered rhizosphere microbial hydrogen cycling

Shi, Ling-Dong; Ercoli, Maria Florencia; Kim, Junhyeong; de Araujo Junior, Artur Teixeira; Estera-Molina, Katerina; Soni, Subah; Weitz, Tracy Satomi; Shigenaga, Alexandra M.; Dukovski, Ilija; Sachdeva, Rohan; Turumtay, Halbay; Louie, Katherine B.; Bowen, Benjamin P.; Kosina, Suzanne M.; Scheller, Henrik V.; Pett-Ridge, Jennifer; Segrè, Daniel; Northen, Trent R.; Ronald, Pamela C.; Banfield, Jillian F.

doi:10.1038/s41467-026-68640-9

Download PDF

Article
Open access
Published: 26 January 2026

Reduced methane emissions in transgenic rice genotypes are associated with altered rhizosphere microbial hydrogen cycling

Nature Communications volume 17, Article number: 2028 (2026) Cite this article

6510 Accesses
20 Altmetric
Metrics details

Subjects

Abstract

Rice paddies significantly contribute to atmospheric methane (CH₄). Here, we show that two independent rice genotypes overexpressing genes for PLANT PEPTIDES CONTAINING SULFATED TYROSINE (PSY) reduce cumulative CH₄ emissions by 38% (PSY1) and 58% (PSY2) over 70 days of growth compared with controls. Genome-resolved metatranscriptomic data from PSY rhizosphere soils reveal lower ratios of gene activities for (mostly hydrogenotrophic) CH₄ production versus consumption, decreased activity of H₂-producing genes, and increased activity of bacterial H₂ oxidation pathways. Metabolic modeling using metagenomic and metabolomic data predicts elevated H₂ oxidation and suppressed H₂ production in the PSY rhizosphere. Assembled genomes of rhizosphere H₂-oxidizing bacteria are enriched in genes utilizing gluconeogenic acids compared with H₂-producing counterparts, and their activities are likely stimulated by elevated levels of gluconeogenic acids, primarily amino acids, in PSY root exudates. Overall, our study indicates that decreased CH₄ emissions are due to a lower amount of H₂ available for hydrogenotrophic methanogenesis and provides a powerful strategy to mitigate CH₄ emissions from increasingly widespread rice cultivation.

Conserved genotype-independent rhizobacteria promote maize growth

Article Open access 31 December 2025

Metabolic coupling between soil aerobic methanotrophs and denitrifiers in rice paddy fields

Article Open access 24 April 2024

Rhizosphere-triggered viral lysogeny mediates microbial metabolic reprogramming to enhance arsenic oxidation

Article Open access 30 April 2025

Introduction

Methane (CH₄) is a greenhouse gas with a 27–30 times greater warming effect than carbon dioxide (CO₂) over a 100-year period. Atmospheric CH₄ has recently increased to a record-high concentration of 1933 ppb (May, 2025)¹. One of the most important CH₄ sources is the waterlogged, anaerobic soils of rice paddies, which support the growth of methanogenic archaea that catalyze the final step in anaerobic decomposition of organic matter, generating CH₄^2,3. Although produced CH₄ can be partially oxidized in situ through aerobic and/or anaerobic methane oxidation^4,5, the imbalance between CH₄ production and oxidation leads to rice paddies contributing 7-17% of global CH₄ emissions^6,7. Emissions are predicted to further increase due to the expansion and intensification of rice cultivation needed to feed the fast-growing human population⁸.

Numerous approaches have been explored to mitigate CH₄ emissions from rice paddies, primarily focusing on farming practices such as managing organic matter, water, and nitrogen⁹. Genetic optimization of rice plants, for example, allocating less photosynthate to roots, is an alternative CH₄ mitigation strategy. Transgenic introduction of a single transcription factor gene, barley SUSIBA2, into rice plants confers more nutrients to aboveground biomass and less to roots, therefore reducing CH₄ formation¹⁰. Similarly, a loss-of-function allele of rice gs3 (a gene on chromosome 3 controlling grain size) allocates more carbon to the grain and less to the roots and causes lower production of CH₄¹¹.

Another approach to reduce CH₄ emissions is to alter the rice root phenotype. Through programmed cell death, rice roots develop aerenchyma, which facilitates gas diffusion and enables efficient oxygen supply from the aerial parts of the plant to the submerged roots and surrounding soil. Simultaneously, plants deposit suberin and lignin to form complete apoplastic barriers in the outer root layers to reduce radial oxygen loss^12,13. It has been reported that rice genotypes with a higher capacity for internal oxygen diffusion grow larger root systems in waterlogged soils and are associated with lower CH₄ emissions due to their stronger ability to oxygenate the soil^14,15. Previously, we and others have shown that small tyrosine-sulfated peptides, such as the PLANT PEPTIDES CONTAINING SULFATED TYROSINE (PSY) family, play central roles in coordinating root development. Acting as extracellular signals, these peptide hormones synchronize cellular activities across tissues by binding to leucine-rich repeat receptor-like kinases (LRR-RLKs), thereby triggering downstream signaling cascades¹⁶. A critical post-translational modification, sulfation of a tyrosine residue by TYROSYLPROTEIN SULFOTRANSFERASE (TPST), is essential for receptor recognition and signaling efficiency^17,18,19,20. In Arabidopsis thaliana, overexpression of PSY genes enhances root growth by promoting cell elongation and increasing the size of mature cortical cells^21,22,23. These peptides exert their effects through three known LRR-RLK receptors, PSYR1, PSYR2, and PSYR3, recently identified as direct mediators of PSY signaling^22,24. Notably, PSY1 signaling in Arabidopsis activates genes involved in secondary metabolism that regulate the extent of cell elongation and transition to maturation. Some of these metabolites are subsequently secreted into the rhizosphere as part of the root exudate profile^21,23,25.

Given that methanogenic archaea inhabit the rice root rhizosphere and that substrates for CH₄ production (e.g., H₂/CO₂ and acetate) ultimately derive from root exudates²⁶, we hypothesized that modulation of the rice root phenotype would affect microbiome activity and CH₄ cycling. To this end, we generated rice plants with altered seminal roots by overexpressing the rice PSY genes (OsPSY1 and OsPSY2) and cultivated them in a greenhouse, along with the control Kitaake genotype, in soils from rice paddies. Plant physiological traits, root exudate chemistry, CH₄ emissions, rhizosphere microbial community composition, and in situ gene activities were compared over 70 days of rice growth.

Results

Generation of rice plants overexpressing rice PSY1 and PSY2 genes

To identify putative rice PSY genes, we surveyed the Nipponbare and Kitaake genomes using bioinformatic approaches as previously described^25,27, and identified nine candidate PSY genes (Supplementary Fig. 1a). Sequence alignment of the mature peptide domain revealed high conservation between Arabidopsis (AtPSYs) and rice (OsPSYs) peptides (Supplementary Fig. 1b), suggesting that OsPSY peptides may regulate rice root growth similarly to AtPSY peptides^{21,22,23,25,28}. Supporting this hypothesis, we demonstrated that Kitaake seedlings, grown on solid media or hydroponically, developed longer seminal roots when treated with synthetic OsPSY1 (Fig. 1a, b). Publicly available transcriptomic datasets^29,30 reveal that OsPSY1, also known as CHALKY GRAIN 5 (OsCG5)³¹, is predominantly expressed in spikelets, with lower expression levels in other rice tissues, including roots (Supplementary Fig. 1c–e). We hypothesize that the root growth effect observed after synthetic peptide treatment may result from the presence of leucine-rich repeat receptor-like kinases (LRR-RLKs) in root tissues that recognize OsPSY1. It has been recently shown in Arabidopsis, that PSY peptides directly bind to LRR-RLKs from subfamily XI to trigger a signaling cascade that regulates root development²². The conservation of LRR-RLKs across plant species suggests that functionally similar receptors likely exist in rice and may mediate comparable developmental responses. In contrast to OsPSY1, other members of this gene family, such as OsPSY2, displayed more ubiquitous expression across different rice organs (Supplementary Fig. 1d). Specifically, OsPSY2 expression in rice roots increased progressively in the differentiation zone (DZ), a pattern previously noted for AtPSY1 promoter activity²³ (Supplementary Fig. 1e).

**Fig. 1: Overexpression of OsPSY1 and OsPSY2 enhanced primary root growth in rice.**

To further investigate the roles of OsPSY1 and OsPSY2 in rice development, we independently overexpressed each gene in the Kitaake genetic background. We generated >50 independently transformed lines using the constitutive UBIQUITIN (Ubi) promoter to drive the expression of OsPSY1 and OsPSY2 (Ubi::OsPSY1 and Ubi::OsPSY2). We self-pollinated the resulting transgenic plants and conducted analyses on the progeny derived from four independently transgenic lines, including Ubi::OsPSY1#7, Ubi::OsPSY1#9 (hereafter referred to as PSY1-9), Ubi::OsPSY1#17 (PSY1-17), Ubi::OsPSY1#32, Ubi::OsPSY2#8 (PSY2-8), Ubi::OsPSY2#19, Ubi::OsPSY2#21 and Ubi::OsPSY2#34 (PSY2-34).

Expression analysis of root tissues showed that progeny derived from the Ubi::OsPSY1 and Ubi::OsPSY2 transgenic lines accumulated significantly higher levels of OsPSY1 or OsPSY2 transcripts compared with Kitaake (Fig. 1c, d). Consistent with the synthetic peptide treatment analysis, rice plants expressing higher levels of OsPSY1 or OsPSY2 exhibited longer seminal roots (Fig. 1e, f) and also carried the transgene (Supplementary Fig. 1f, g).

Given that in Arabidopsis, AtPSY1 regulates root growth by controlling the extent to which cells elongate before reaching their final differentiated size^22,23, we measured the length of mature cortical cells in the DZ of PSY1-9 and PSY1-17 genotypes. Consistent with the phenotypes observed in Arabidopsis, we found that the roots of rice genotypes accumulating higher levels of OsPSY1 exhibited longer mature cortical cells (Fig. 1g, h). Additionally, the amount of lignin in PSY1-17 and PSY2-34 plants was significantly lower compared with Kitaake (Fig. 1i). A similar phenotype was observed in the roots of Arabidopsis plants with higher levels of AtPSY1, where lignin deposition in the tracheary elements of the protoxylem occurred further away from the root tip compared with the wild type²³. Collectively, these results suggest that PSY peptides from different species may regulate root development through similar molecular mechanisms.

Finally, we assessed aerenchyma formation in a selected subset of the rice PSY1 (PSY1-9 and PSY1-17) and PSY2 (PSY2-8 and PSY2-34) genotypes. We found that in all the transgenic lines tested, the percentage of aerenchyma formation was significantly higher compared with Kitaake (Fig. 1j–l). However, no differences in aerenchyma were observed between the PSY1 and PSY2 lines (Fig. 1j–l). Given the observed phenotypic modifications in roots of rice seedlings, particularly the longer seminal roots and enhanced aerenchyma formation, we hypothesized that PSY genotypes may impact the rhizosphere microbial community and associated CH₄ cycling.

Growth of PSY rice genotypes in agronomic soils mitigated CH₄ emissions compared with controls

To assess the effect of PSY overexpression on CH₄ emissions, we cultivated three independent transgenic rice lines—PSY1-9, PSY1-17, and PSY2-34—along with the control Kitaake plants in a greenhouse using soils collected from a rice paddy in Davis, CA. We also included bulk soil controls that were not planted. Throughout the experiments, all rice lines developed similarly to Kitaake, with heading observed at 55 days after being transferred to the greenhouse (Fig. 2a). By 70 days post-transplant, both PSY1 and PSY2 genotypes were slightly shorter than Kitaake (Fig. 2b and Supplementary Fig. 2a), but they produced significantly more tillers (Fig. 2c and Supplementary Fig. 2b), resulting in comparable dry weights of stems and leaves per plant at the end of the plant life cycle (Fig. 2d). PSY1 genotypes showed partially filled panicles, fewer seeds, and lower thousand-grain weight (TGW) per plant, compared with Kitaake (Fig. 2e–g and Supplementary Fig. 2c). These phenotypes could be attributed to abnormal panicle development, as we found that the tips of the panicles in PSY1 genotypes often contained smaller and unfilled seeds, a phenotype not seen in PSY2 plants (Fig. 2h). Consistently, no significant differences in seed production occurred in the PSY2 genotypes compared with Kitaake (Fig. 2e–h).

Fig. 2: Kitaake plants with elevated expression of OsPSY1 or OsPSY2 exhibited reduced CH4 emissions. — **Fig. 2: Kitaake plants with elevated expression of *OsPSY1* or *OsPSY2* exhibited reduced CH₄ emissions.**

We also investigated whether the longer seminal root phenotype observed in young seedlings persisted in older PSY plants. We found that PSY1-17 plants developed longer roots 30 days after transplanting, although their root dry biomass was significantly lower than that of Kitaake plants (Fig. 2i and Supplementary Fig. 2d, e). Similar differences in root dry biomass were observed at the end of the plant life cycle (Fig. 2j). Additionally, lignin content in the roots of adult plants revealed that, consistent with findings in young seedlings, PSY1-17 plants accumulated less lignin than Kitaake (Fig. 2k). Although PSY2 plants also had reduced lignin content, their root biomass remained comparable to that of Kitaake (Fig. 2j, k).

Despite these phenotypic differences, no significant differences in soil CH₄ emissions were observed between the PSY1-9, PSY1-17, and PSY2-34 genotypes (Fig. 2l). For these analyses, CH₄ emissions were measured in situ using a Picarro gas analyzer, and soil samples were collected for microbiological analyses throughout the rice growth period (i.e., 15, 30, 55, and 70 days after transplanting). Negligible CH₄ emissions were observed from the bulk soil controls, while differentiated emissions were recorded from pots with PSY and Kitaake rice plants over the 10-week growth period (Fig. 2l and Supplementary Fig. 3a). All PSY rice genotypes emitted substantially less CH₄ compared with Kitaake, with reductions of up to ~30% observed after heading (55 and 70 days; Fig. 2l and Supplementary Fig. 3a). Notably, cumulative CH₄ emissions throughout the growth period for PSY1 and PSY2 genotypes showed an average decrease of ~38% and ~58% relative to Kitaake, respectively (Supplementary Fig. 3b). Given the crucial role of microorganisms in the production and consumption of CH₄, we hypothesized that distinct microbial communities might accumulate in the rhizospheres of the different rice genotypes, potentially affecting the rice CH₄ emissions. Because both PSY1 and PSY2 genotypes displayed comparable reductions in CH₄ emissions, we focused on PSY1 genotypes for further study.

Rice PSY genotypes affected microbial activity but not microbial community structure

To explore the difference in root microbiomes, we extracted and sequenced total DNA and RNA from 72 soil samples (3 replicates for each of the bulk soils and 5 replicates for each planted with the PSY1 rice genotypes and Kitaake control, at each collection time point) (Supplementary Data 1). We used assembled ribosomal protein S3 (rpS3) sequences to perform an initial census of microbial diversity, including organism type and abundance³², in 16 representative samples obtained by combining replicates. In total, 4,087 rpS3 sequences were identified and grouped into 1,656 non-redundant, approximately species-level clusters. Phylogenetic analysis based on rpS3 sequences classified all detected microorganisms into at least 4 archaeal and 14 bacterial phyla (Fig. 3a). By mapping metagenomic and metatranscriptomic reads to rpS3 genes, the most abundant and active organisms were Proteobacteria, Chloroflexi, Acidobacteria, Bacteroidetes, Gemmatimonadetes, and Actinobacteria (Fig. 3b, c), consistent with previous soil surveys^33,34,35. Between 1 and 5% of sequences were assigned to the “Candidate Phyla Radiation” (CPR) bacteria and DPANN archaea (Diapherotrites, Parvarchaeota, Aenigmarchaeota, Nanoarchaeota and Nanohaloarchaeota), both of which were recently reported in paddy soils^36,37.

**Fig. 3: Microbial community structure and activity.**

Non-metric multidimensional scaling analyses (NMDS) show clear differences in community structure and microbial activity between soils collected from the rhizosphere and unplanted bulk soil (Fig. 3d, e), indicating a strong effect from the rice plants³⁸. For the planted groups, cultivation time had a large and statistically significant effect on the community structure (dissimilarity score of up to 0.539), whereas rice genotype did not (Fig. 3d). Notably, however, microbial activity was significantly influenced by both time and genotype with comparable contributions (dissimilarity scores of 0.265 and 0.195, respectively; Fig. 3e).

Lower activity of CH₄-producing genes in microbiomes associated with PSY1 rice genotypes

CH₄ fluxes from rice paddy soils reflect the combined impact of the activities of methanogens and methanotrophs. A search for methyl-coenzyme M reductases (McrA) that participate in methanogenesis and anaerobic CH₄ oxidation identified multiple homologs assigned to Methanobacterium, Methanomassiliicoccus, Methanoregula, Methanocella, and Methanosarcina (Fig. 4a). None of the recovered McrA sequences are similar to those found in anaerobic CH₄-oxidizing archaea, suggesting that anaerobic methanotrophs were either absent or too rare to be detected. Many particulate methane monooxygenases (PmoB) genes that enable aerobic CH₄ oxidation were identified and taxonomically assigned to several bacterial genera (Fig. 4b). The soluble form of methane monooxygenase was not detected. These genes are typically only expressed under copper-limiting conditions³⁹ (e.g., <0.1 μM versus 0.4 μM in our fertilized water; Supplementary Data 2).

Fig. 4: Taxonomic affiliation and in situ activity of genes involved in CH4 production and oxidation. — **Fig. 4: Taxonomic affiliation and in situ activity of genes involved in CH₄ production and oxidation.**

In general, the transcriptional activity of mcrA increased with cultivation time in the three rice genotypes. For example, the activity of Methanocella mcrA in the Kitaake genotype increased by approximately one order of magnitude in the last growth stage (70 days, Fig. 4c). The transcriptional activity of pmoB also increased with cultivation time, and we observed more active terminal oxidases in microbes associated with the PSY1 genotypes late in the experiment (Supplementary Fig. 4), suggestive of higher O₂ concentrations. However, the ratio of mcrA to pmoB remained stable over the whole growth period, except for Kitaake (Fig. 4c). Given the importance of the balance between activities that produce and consume CH₄, we calculated the fold changes of mcrA/pmoB ratios for each group relative to that at the corresponding first sampling point (i.e., 15 days after transplanting), and observed notable increases in Kitaake but negligible changes in PSY1 genotypes (Fig. 4c). This result indicates that, over the rice growth period, mcrA activity was promoted much more than pmoB in Kitaake, leading to a higher ratio of CH₄ production to consumption, as compared to the PSY1 genotypes.

PSY1 rice genotypes supported higher activities of microbial H₂-consuming genes but lower activities of H₂-producing genes

Expressed mcrA transcripts were predominantly from Methanocella, Methanoregula, and Methanobacterium (>90%; Fig. 4c). Archaea of these genera all use H₂, instead of acetate or methylated compounds, to produce CH₄⁴⁰. Thus, we identified genes involved in H₂ production and consumption and tracked their expression levels as a function of genotype and time. Two types of hydrogenases that catalyze the reversible conversion between H₂ and protons/electrons⁴¹, namely [NiFe] and [FeFe] hydrogenases, were identified and divided into four and three subtypes, respectively, using protein phylogeny (Fig. 5a, b). The most active clade was [NiFe] Group 1 hydrogenases that oxidize H₂ (Fig. 5c). The transcriptional activities of this group were generally higher in the two PSY1 genotypes than in the Kitaake control, particularly in the first and last growth stages where the differences were significant (15 and 70 days; Fig. 5d). The [NiFe] Group 4 and [FeFe] Group A1 (further classified by HydDB⁴²) are predicted to produce H₂⁴³. Transcript quantification shows [NiFe] Group 4 hydrogenases were more active than [FeFe] Group A1 hydrogenases, and the combined activity of these two H₂-producing hydrogenases was significantly higher in Kitaake compared with the two PSY1 genotypes, especially in the second and last growth stages (30 and 70 days; Fig. 5d). Based on the transcriptomic profile analysis, we hypothesize that the microbiomes of the PSY1 genotypes might produce less and consume more H₂ than those in Kitaake. As a result, the pool of hydrogen available for CH₄ production in soils from the PSY genotypes would be expected to be lower than that of the Kitaake control.

**Fig. 5: Classification and activity of hydrogenases.**

H₂ may also be produced as a byproduct of N₂ fixation (one molecule of H₂ per N₂ fixed). The nitrogenase gene (nifH) was used as a marker of N₂ fixation activity and associated H₂ production^44,45. The transcriptional activity of nifH generally increased with cultivation time and was linked to Delta-, Gamma-, and Alphaproteobacteria (Supplementary Fig. 5). Like H₂-producing hydrogenases, nifH activity was significantly higher in Kitaake than in the PSY1 genotypes, especially in the last two growth stages (55 and 70 days).

Altered PSY1 root exudates modulate activities of H₂-metabolizing microorganisms

To explore the influence of rice phenotypes on rhizosphere microbiome activity, we performed metabolomic analysis on root tissues and exudates from axenic rice seedling roots to identify compounds secreted by the PSY1 genotypes. Metabolites were extracted using methanol and profiled on an HPLC-MS platform. Using gradient-based hydrophilic interaction liquid chromatography (HILIC) in positive-ion mode, we identified a total of 3612 features across all exudate samples, of which 2047 were shared between the Kitaake and PSY1 genotypes (Supplementary Fig. 6a). Principal component analysis (PCA) showed a clear separation between PSY1 and Kitaake exudate samples (Supplementary Fig. 6b). Among the shared features, 168 (8.3%) were significantly differentially abundant in PSY1 exudates ( | log₂FC | >0.5, p < 0.05), including 146 novel and 15 depleted features (Supplementary Fig. 6c). A subset of these differentially abundant features was putatively identified and classified using Global Natural Products Social Molecular Networking (GNPS) (Supplementary Data 3–5). Most of the differentially abundant metabolites in PSY1 root exudate samples belong to the same compound classes, including ‘amino acids’, ‘alkaloids’, and ‘shikimates and phenylpropanoids’ (Supplementary Fig. 6d). Similarly, in root tissues, we identified 2,237 shared features, with PCA indicating an even stronger genotypic separation (Supplementary Fig. 6e, f). Of these, 272 (12%) were differentially abundant in PSY1 tissues, featuring 208 new and 80 depleted metabolite signals (Supplementary Fig. 6g), again reflecting shifts in similar compound classes (Supplementary Fig. 6h). Overall, more features increased than decreased in PSY1 samples. Similar results were obtained from HILIC in negative ionization mode and reverse phase (C18) in both positive and negative ionization modes and are detailed in Supplementary Data 3–5, and Supplementary Figs. 7–9. Altogether, our analysis indicated significant changes in the abundance of compounds in the PSY1 root exudates compared with Kitaake. Approximately 90% of the metabolites enriched in the PSY1 metabolome are predicted to be involved in microbial gluconeogenic pathways (Supplementary Data 6). We also noted the presence of two putative flavonoids, isorhamnetin 3-rutinoside and malvidin-3-glucoside, that differentially accumulated in exudate samples from the PSY1 genotypes but were absent from Kitaake, a feature observed in previous studies in Arabidopsis²³.

To assess whether these metabolic differences reflected changes in the total amount of carbon released, we measured the concentration of non-purgeable organic carbon (NPOC) in the root exudate samples. No significant differences in total carbon content were detected between Kitaake and PSY genotypes, indicating that PSY overexpression primarily alters the composition, instead of the overall quantity, of secreted metabolites (Supplementary Fig. 10).

To determine whether the altered exudate composition of PSY genotypes can directly influence microbial activity, we conducted soil incubation assays using paddy soil from the same rice field used in our greenhouse experiments. Exudates collected from PSY1- and PSY2-overexpressing rice seedlings were added as amendments, with exudates from Kitaake and sterile water as controls. CH₄ emissions were then monitored over time. No CH₄ flux was detected in water-treated soils, confirming that exudates were necessary to stimulate methanogenesis. Soils treated with Kitaake exudates exhibited consistently higher CH₄ emissions than those treated with PSY1 or PSY2 exudates, mirroring trends observed in the greenhouse-grown plant experiments (Supplementary Fig. 11).

Next, we analysed all the microbial genomes reconstructed from the rhizosphere (n = 296) and found 17 had transcriptional activities that were significantly upregulated in soils of rice PSY genotypes relative to the Kitaake control (log2FoldChange ≥ 1, p ≤ 0.05; Supplementary Fig. 12). The genomes of five of the 17 encode H₂-consuming hydrogenases, including those affiliated with Pyrinomonadaceae, Burkholderiaceae, Haliangiaceae, Rhodocyclaceae, and Dissulfurispiraceae, but none contain H₂-producing enzymes (Supplementary Data 7). This result suggests that H₂-consuming bacteria may respond to the PSY rice genotype-derived exudates differently than organisms that produce H₂. We therefore surveyed the entire H₂-consuming (n = 55) and H₂-producing (n = 36) populations and calculated glycolytic sugar versus gluconeogenic acid preference values (SAP) using the method described previously⁴⁶. We found that the H₂-consuming organisms have much higher preferences for gluconeogenic acids than the H₂-producing organisms, and their genomes have higher GC contents, likely reflecting this metabolic preference⁴⁶ (Fig. 6a, b; Supplementary Data 7). Consequently, H₂-consuming organisms had higher transcriptional activities in association with the PSY1 genotypes, whereas their H₂-producing counterparts were more active in the Kitaake rhizosphere (Fig. 6c).

Fig. 6: Metabolic preference and activity of H2-consuming and H2-producing genomes. — **Fig. 6: Metabolic preference and activity of H₂-consuming and H₂-producing genomes.**

To further investigate how rice physiological traits affect microbial H₂ cycling, we built metabolic models using representative, high-quality genomes (completeness >90%, contamination <5%). We then predicted H₂ fluxes in response to variations in PSY1-secreted exudate compounds and O₂ concentrations inferred from aerenchyma densities of the different rice genotypes (Supplementary Data 8). The only identified compounds under-secreted by PSY1 compared with Kitaake were two glycerolipids, putatively 1-oleoyl-2,3-di-linoleoyl-sn-glycerol and 1,2-di-hexadecanoyl-3-(9Z,12Z,15Z-octadecatrienoyl)-sn-glycerol, and one dipeptide containing leucine (Supplementary Data 3). However, none of the modeled H₂-related organisms can metabolize these compounds. As a result, the PSY1 under-secreted metabolites are predicted to have no effect on H₂ fluxes. In contrast, when over-secreted compounds increase from 0.01 to 0.05-fold, all three modeled H₂-producing bacteria from different phyla (Bacteroidia, Ignavibacteriota, and Moranbacteria) show the same trend, with maximal produced H₂ decreasing (Fig. 7a–c). Beyond this range, H₂ production is predicted to plateau. Microbial growth increased with increasing concentration of over-secreted compounds and plateaued when the H₂ production plateaued, except for Bacteroidia_37, which continually increased its biomass over the whole simulation (Supplementary Fig. 13a–c). The results suggest that the modeled H₂-producing bacteria can use over-secreted PSY1 compounds to support growth, but not via pathways that generate H₂. Increased O₂ availability has little or no effect on H₂ production (Fig. 7a–c).

Fig. 7: Predicted H2 flux by metabolic modeling using representative high-quality genomes. — **Fig. 7: Predicted H₂ flux by metabolic modeling using representative high-quality genomes.**

The H₂-consuming organisms from Acidobacteriota, Proteobacteria, and Deinococcota are modeled to oxidize more H₂ (and accordingly yield more biomass) in the presence of higher O₂ concentrations that are associated with higher aerenchyma formation in PSY1 rice genotypes (Fig. 7d–f and Supplementary Fig. 13d–f). The PSY1 over-secreted exudate compounds have a negligible influence on H₂ oxidation, yet still contribute substantially to the accumulation of biomass of H₂-consuming organisms (Supplementary Fig. 13d, e). These modeling results support the metagenomics-based hypothesis that PSY1 rice genotypes enhance the growth and activity of H₂-consuming organisms due to their distinct aerenchyma characteristics and exudate profiles.

Discussion

Rice is the most important staple food for half the world’s human population. It is estimated that the global rice production will continue to increase from 506 million tonnes in 2020 to 567 million tonnes by 2030⁴⁷. During cultivation, rice paddies emit millions of tons of CH₄ annually and contribute ~10.1% of global agricultural greenhouse gas emissions⁴⁸. Here, we tested whether overexpression of the rice genes, OsPSY1 or OsPSY2, could affect root growth, exudate composition, activities of rhizosphere microbiomes and CH₄ emissions, potentially providing a route to decrease the climate impact of rice production. To our knowledge, this is the first study to use an integrated multi-omic approach to elucidate how specific rice genotypes mitigate CH₄ emissions. The research highlights how the intersecting activities of H₂-metabolizing, CH₄-producing and other microbes in rhizosphere communities modulate CH₄ cycling (Fig. 8).

Fig. 8: Proposed mechanism for the role of PSY rice genotypes in mitigating CH4 emissions. — **Fig. 8: Proposed mechanism for the role of PSY rice genotypes in mitigating CH₄ emissions.**

The CH₄ production in our study was derived mainly from hydrogenotrophic methanogens ( > 90%; Fig. 4c), consistent with extensive prior research^11,49,50. Previous rice field studies using isotope-labeled substrates revealed that the presence of electron acceptors such as sulfate and nitrate channelled acetate towards CO₂, and that CH₄ was mainly produced from H₂/CO₂-dependent methanogenesis^51,52. Dependence on H₂ is likely explained by the insufficient acetate concentrations, which are insufficient for methane production. Acetoclastic methanogens (e.g., Methanosarcina) require 0.2–1.2 mM acetate to activate CH₄ formation^53,54, whereas hydrogenotrophic methanogens (e.g., Methanobacterium) only need 2.8–10 Pa H₂^55,56. Low acetate concentrations may be due to the rapid in situ utilization by co-existing organisms that outcompete methanogens for the acetate. One possible route for acetate consumption is syntrophic acetate oxidation (SAO) that uses the reverse Wood-Ljungdahl (WL) pathway (sometimes combined with the glycine cleavage system) and hydrogenases to produce CO₂ and H₂, the substrates for hydrogenotrophic methanogenesis^57,58,59. However, none of our hydrogenase-harboring microbial genomes encode complete WL pathways, although some unclassified Firmicutes members that may include SAO bacteria were detected at low abundance ( < 1%; Fig. 3). An alternative way for acetate consumption is the reduction of nitrate and sulfate, observed in previous studies^51,52. This activity is indicated by the expression of nitrate and sulfate reductases in our system (Supplementary Figs. 14 and 15). The substrates (i.e., ~2.0 mM of both nitrate and sulfate) were periodically added to water used to fertilize the rice plants during growth (Supplementary Data 2). Given the absence of WL pathways in hydrogenase-containing microorganisms, the homoacetogenesis process that relies on H₂ oxidation to produce acetate appears to be negligible in our study.

Although for all wild-type and transgenic rice genotypes the dominant pathway for CH₄ production was H₂-based methanogenesis, PSY1 and PSY2 genotypes mitigated the cumulative CH₄ emissions by up to 58%, relative to the Kitaake control (Supplementary Fig. 3). The most straightforward distinction between the wild-type and transgenic genotypes is the lower expression of mcrA and the lower ratio of transcripts for CH₄ production versus oxidation (mcrA/pmoB) in microbiomes associated with the PSY1 genotypes (Fig. 4c). The much lower expression of Methanocella mcrA in the PSY1 genotypes compared to the Kitaake control (Fig. 4c) in the last growth stage is in line with the lower measured CH₄ emissions (Fig. 2l). This difference may be partially attributed to higher O₂ concentrations in the PSY rhizosphere due to more aerenchyma in the PSY compared with Kitaake roots (Fig. 1j–l). Consistent with this, the activity of terminal oxidases in microbes associated with the PSY1 genotypes was higher than for microbes associated with the Kitaake control (Supplementary Fig. 4). Although Methanocella cells encode genetic machinery to resist oxidative stress and can persist in aerated environments⁶⁰, their methanogenesis activity would be inhibited by oxygen exposure until conditions become anoxic⁶¹.

Another factor that may have affected CH₄ production is the availability of the methanogenesis substrate, H₂. In rice paddies, H₂ is typically produced during microbial fermentation, e.g., of plant-derived sugars, and can be consumed via diverse processes⁶². We found much higher activities of H₂-consuming hydrogenases but lower activities of H₂-producing counterparts associated with the PSY1 genotypes (Fig. 5). This likely reduced the H₂ available for hydrogenotrophic methanogenesis, compared to that with the Kitaake.

Activities of the root-associated H₂-metabolizing microorganisms can be altered by rice root characteristics. As larger root systems can be associated with increased CH₄ emissions in paddy fields⁶³, rice breeding efforts for mitigating CH₄ emissions have aimed to generate plants with smaller root systems^10,11. However, overexpression of rice PSY genes led to longer roots (Fig. 2k) and only the PSY1 genotype displayed reduced root biomass (Fig. 2i), compared with the control. The distinct developmental phenotypes observed in rice plants overexpressing OsPSY1 and OsPSY2 may be driven by differences in receptor binding and tissue-specific expression. Similar behavior has been observed in other small peptide families, such as the CLE family⁶⁴. Still, both OsPSY1 and OsPSY2 genotypes appear to share common molecular mechanisms contributing to comparable reductions in CH₄ emissions (Fig. 2l).

Given the similar reduction in CH₄ emissions despite differences in PSY1 and PSY2 root biomass, we considered whether the enrichment of PSY exudates in gluconeogenic acid metabolites (mostly amino acids) could alter rhizosphere microbial H₂ metabolism, with implications for CH₄ production (Supplementary Data 6). H₂-consuming microorganisms that are more invested in the use of these compounds than H₂-producing bacteria had significantly higher activities associated with the PSY1 genotypes, suggestive of lower hydrogenotrophic methanogenesis in the PSY1 rhizosphere (Fig. 6). Consistent with this hypothesis, soil incubations with exudates collected from PSY rice roots exhibited lower CH₄ emissions than those amended with exudates from Kitaake. While our soil exudate incubation experiments clearly show the effect of rice root exudates on microbial shifts and associated CH₄ emissions, it is unclear whether this is caused by a specific exudate compound or by a combination of multiple compounds. It is also possible that root-associated environmental changes (e.g., altered redox conditions) may affect the rhizosphere microbiome. Future research is warranted to investigate these possibilities.

Our metabolic modeling results also indicate that PSY1 over-secreted compounds suppress H₂ production activity (Fig. 7). Intriguingly, the modeling predicts the increase in biomass of H₂-producing bacteria despite the decreased H₂ generation (Supplementary Fig. 13). This may be explained by the high relative abundance of amino acids in PSY1 exudates (Supplementary Data 6). Root-associated H₂-producing bacteria may partly rely on extracellular amino acids that can be converted to pyruvate and acetyl-CoA, and thus rely less on fermentation of sugars to form H₂^65,66. The rice-derived amino acids could also alleviate the necessity for microbial nitrogen fixation (Supplementary Fig. 5), reducing the formation of byproduct H₂⁶⁷. Conversely, microbial H₂ consumption was barely affected by the PSY1 over-secreted compounds (Fig. 7). Overall, our metabolic modeling predicts that the exudate profiles of PSY1 rice genotypes may inhibit the activities of H₂-producing microorganisms and stimulate those of the H₂-consuming counterparts. This result suggests that in the PSY1 rhizosphere, the bacterial metabolic activities further diminish the pool of H₂ available for methanogenesis, thus mitigating overall CH₄ emissions (Fig. 8). Future studies could isolate and co-cultivate the organisms used in the models to validate the inferred microbial responses to root-derived environmental changes.

It has been demonstrated that exudation rate and composition vary throughout plant development⁶⁸. In this study, we collected exudates from seedlings grown under axenic hydroponic conditions. This method offers several advantages: precise control over the chemical milieu, adequate recovery of fresh exudates, avoidance of mineral sorption biases, and minimization of microbial degradation⁶⁹. However, hydroponic growth can alter exudate profiles compared with soil systems, often increasing total carbon release but reducing metabolite diversity due to the absence of rhizosphere feedbacks⁷⁰. Given these methodological trade-offs, future studies could integrate both axenic hydroponic and soil-based exudate sampling across multiple developmental stages to gain a more comprehensive and ecologically relevant understanding of how PSY influences plant–microbe interactions and biogeochemical cycles.

While our greenhouse-based experiments using agronomic soil collected from a rice basin provide clear evidence that rice genotypes overexpressing PSY genes reduce CH₄ emissions, controlled environments do not fully replicate the complexity of field conditions, where factors such as temperature fluctuations and seasonal dynamics may play major roles. We consider our study a critical first step in uncovering mechanistic links between specific plant genotypes, root exudate chemistry, and CH₄ cycling. To validate and extend these findings to agronomically relevant conditions, field trials incorporating multi-omics profiling, similar in scope to our current approach, are necessary. Such future research will be essential to determine the generalizability and robustness of PSY-mediated reduced CH₄ emissions across various soil types and growing conditions.

Methods

Plant materials and growth conditions

Oryza sativa ssp. Japonica cultivar Kitaake was used throughout this study as the wild-type (wt) background. Rice seeds were dehulled, surface-sterilized in 20% bleach for 30 minutes, and thoroughly washed with autoclaved water. The seeds were grown in 1× Murashige and Skoog (1× MS) salt mixture with 1% sucrose (pH 5.65), either as a solid medium containing 0.3% Gellex (Gellan Gum, Caisson Laboratories) or hydroponically in flasks with 100 mL of the same media without Gellex. For solid medium conditions, seeds were sown on plates, sealed with Micropore tape for gas exchange, and placed vertically for 7 days. For hydroponic conditions, seeds were first germinated on plates with water for 2 days in darkness, then transferred to flasks for another 5 days. The incubators used for rice germination and growth were set with a 14-h-light/10-h-dark photoperiod at 28 °C/24 °C.

For synthetic peptide treatments in hydroponic conditions, peptide (or water for untreated plants) was added to the MS media before transferring the plants. The length and concentration of the peptide treatment are specified in the figure legends. The synthetic OsPSY1 peptide was obtained from Pacific Immunology (Ramona, CA, USA), is tyrosine sulfated, and was diluted in ddH2O to a final concentration of 1 mM. For each experiment, MS media was freshly prepared and cooled for 1 hour in a 55–60 °C water bath after autoclaving before adding chemicals.

For primary root elongation, seedlings grown hydroponically or in solid media were photographed 7 days after germination, and the primary root length was measured with Fiji Is Just ImageJ⁷¹.

Sequence alignment and peptide conservation analysis

To assess the conservation of the PSY peptides across species, we extracted the mature peptide sequences from Arabidopsis thaliana and Oryza sativa PSY family members. The peptide sequences were defined as the 13-amino acid mature domain, including the conserved aspartate-tyrosine (DY) motif, following the strategy described previously²⁵. A sequence logo was generated using WebLogo web interface (http://weblogo.threeplusone.com)⁷², which provides a graphical representation of the multiple sequence alignment, where each position’s stack height indicates sequence conservation, and symbol heights within the stack reflect the relative frequency of individual amino acids at that position.

Plasmid construction and rice transformation

DNA constructs were created with the Gateway cloning technology⁷³. The genomic OsPSY1 and OsPSY2 sequences were amplified using the primers described in Supplementary Data 9. These sequences were then recombined with pENTR^TM/D-TOPO (Invitrogen, Cat#45-0218) to yield pTOPO_OsPSY1 and pTOPO_OsPSY2. The latter vectors were used in a Gateway LR cloning (Gateway® LR Clonase^TM II Plus Enzyme Mix, Invitrogen; Cat#:12538-120) with pC1300-Ubi-Nos to yield Ubi::OsPSY1 and Ubi::OsPSY2 constructs. The generated vectors were transferred to Agrobacterium tumefaciens strain EHA105 and then introduced into the rice lines via Agrobacterium-mediated transformation. Rice transformation was performed following previously published protocols⁷⁴. Post-transformant screening was based on hygromycin resistance and PCR determination of T-DNA insertion. Out of 52 positive lines for Ubi::OsPSY1 and 53 for Ubi::OsPSY2, four homozygous lines were selected for characterization. Eventually, three homozygous lines, Ubi::OsPSY1#9 (PSY1-9), Ubi::OsPSY1#17(PSY1-17), and Ubi::OsPSY2#34 (PSY2-34), were selected for detailed studies.

Gene expression analysis by quantitative PCR (qPCR)

Total RNA (1 mg) was extracted from the roots of seven-day-old rice seedlings grown hydroponically using TRIzol reagent (Invitrogen) and treated with the TURBO DNA-free kit (Ambion) to remove residual genomic DNA. cDNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). The cycle threshold (Ct) value was measured on a Bio-Rad CFX96 Real-Time System coupled to a C1000 Thermal Cycler (Bio-Rad) using the iTaq Universal SYBR Green Supermix (Bio-Rad). Normalized relative quantities (NRQs) were obtained using the qBase method⁷⁵, with LOC_Os08g18110 (putative alpha-soluble NSF attachment protein) and LOC_Os11g26910 (putative SKP1-like protein 1B) as reference genes for normalization across samples⁷⁶. NRQ values were normalized to the mean value obtained in wild-type plants. NRQ Melting curve analyses at the end of the process and “no template controls” were performed to ensure product-specific amplification without primer–dimer artifacts. Primer sequences are given in Supplementary Data 9. Three biological replicates were analyzed.

Cellular analysis using LSCM

Laser scanning confocal microscopy (LSCM) was performed throughout the study using a Plan Apochromat ×63/1.40 oil and ×20/0.75 CS2 lens on a Leica TCS SP8 microscope.

Cell wall staining, seedling fixation and staining were performed using an adapted Clearsee protocol⁷⁷. Briefly, the primary root of 7-day-old seedlings was sectioned into three short segments with a length of 0.5 cm starting from the root tip. The root segments were fixed for 1 h at room temperature in 10% neutral buffered formalin in PBS, using six-well plates, then washed five times for 1 min with PBS 1×. For root transverse section analysis, the 1 cm root fragments collected 4 cm from the root tip were embedded in 4% agarose and sectioned using a Leica automated vibrating blade microtome (VT1200 S). Once fixed, root fragments and transverse sections from those fragments were cleared in Clearsee solution for at least 24 hours under mild shaking. Fixed and cleared samples were incubated overnight in a Clearsee solution supplemented with 0.1% Calcofluor White. After 12 hours, the staining solution was removed, and samples were rinsed once in fresh Clearsee solution, then washed twice for at least 120 minutes in a renewed Clearsee solution with gentle shaking. Roots were carefully placed on a microscope slide with ClearSee and covered with a coverslip. Excitation and detection windows were set as follows: Calcofluor white for excitation at 405 nm, and detection between 415 and 570 nm. Central longitudinal section images were acquired from root complete fragments. Mature cortical cell length was assessed in 10 consecutive cells in the second cortical layer. Transverse section images were acquired to quantify the percentage aerenchyma formation, expressed as a percentage of the total cortical cell area.

Soil collection and growth conditions in the greenhouse experiments

All data presented here were generated from two greenhouse studies carried out at the University of California—Davis in the fall of 2022 and the spring-summer of 2024. Soil from a rice field in Davis (38.543224 degrees North and -121.810536 West) was collected in July 2022 and in July and December 2023 using a front-end loader to gather down to a depth of ~6–10 inches. The soil at the field site belongs to the Capay soil series, a fine, smectitic, thermic Typic Haploxerert. Soil samples were analyzed at the UC Davis Analytical Lab for chemical content (Supplementary Data 10). Soil pH was 6.59 on average. For both experiments, all soils were transported back to the greenhouse, crushed to a 2 mm particle size, thoroughly homogenized, and stored until planting. Then, the soil was transferred into new 5.5 × 5.5 inch pots (2000 g per pot), which were then placed into five 23 gallon tubs per genotype (18 pots each). Prior to seedling transplantation, enough water was added to the tubs to submerge the soils.

Seeds were germinated in solid MS media, and root growth was followed as previously described for 7 days. Axenic seedlings were then transplanted into the watered pots (one seedling per pot) in the greenhouse, where they were irrigated with deionized water every other day to keep the soil submerged. Fertilized water [N, 77 ppm (parts per million); P, 20 ppm; K, 75 ppm; Ca, 27 ppm; Mg, 17 ppm; S, 68 ppm; Fe, 1.5 ppm; Cu, 0.02 ppm; Mn, 0.5 ppm; Mo, 0.01 ppm; Zn, 0.1 ppm; B, 0.5 ppm] replaced the distilled water in the tubs after microbiome samples were collected on day 15, 30, 55 and 70 after transferring to the pots. Each tub contained only one plant genotype to avoid root exudate compositional changes. We also kept a separate tub with unplanted pots and bulk soil controls. Tubs were distributed in the greenhouse benches following a completely randomized block design. All weeds were manually removed from the pots when identified.

Plant developmental stage and height were measured on the microbiome samples collection days. Roots and shoots were collected 100 days after transferring plants to pots, and dry biomass was measured. Seed number and one thousand seed weight were also calculated for each plant.

Cell wall residue preparation

Lyophilized rice roots used for cell wall residue preparation method were adapted from Fonseca Garcia et al. ⁷⁸. Samples for lignin analysis were collected from 7-day-old rice seedlings or from root systems of 90-day-old plants grown in agricultural soil. Samples from older root systems were collected in the same root-shoot junction location as the one used for rhizosphere sample collection.

The fresh root biomass was lyophilized for 3 days, and then ball-milled biomass was extracted sequentially by sonication (15 min) with 80% ethanol (three times), acetone (one time), chloroform–methanol (1:1, v/v, one time) and acetone (one time).

TGA lignin

Aliquots of 15 mg of cell wall residue were weighed into Eppendorf tubes and mixed with 1 ml of 3 N HCl and 0.1 ml thioglycolic acid (adapted after Suzuki et al.⁷⁹). Samples were incubated at 80 °C for 4 hours and repeatedly mixed. Samples were rapidly cooled on ice and centrifuged for 10 min at 15,000 × g. The supernatant was discarded. Pellets were washed three times with distilled water (1 ml). Thereafter, pellets were incubated with 1 ml of 1 N NaOH for 16 hours on a shaker at room temperature. The suspension was centrifuged for 10 min at 15,000 × g. The supernatant was carefully transferred into a 2-ml Eppendorf tube. The resulting supernatant was combined with the first alkaline supernatant and mixed with 0.2 ml concentrated HCl. Samples were incubated for 4 hours at 4 °C to precipitate the lignothioglycolate derivates. The samples were centrifuged, the supernatant discarded, and the pellet solubilized in 1 ml of 1 N NaOH. Absorbance of the resulting solution was measured at 280 nm. Results were expressed as an absorbance per mg cell wall.

Measurement of methane emissions

Gas flux measurements of CH₄ were made in real-time in a transparent dynamic flow chamber using Picarro G2508 Gas Analyzer. In the fall 2022 experiment, fluxes at a total of four timepoints (15 d, 30 d, 55 d, 70 d) were measured across the growing season for 5 biological replicates corresponding to Kitaake, PSY1-17, PSY1-9, and bulk soil sample types. In the spring-summer 2024 experiment, CH₄ fluxes at a total of 12-timepoints were measured across the growing season for five biological replicates corresponding to Kitaake, PSY1-17, PSY2-34, and bulk soil sample types. Each replicate constituted a single rice plant except for bulk soil samples, which represented a negative control. Outlier detection was performed using the median absolute deviation (MAD) approach. Measurements deviating by more than three MADs from the median within a given time point and genotype were classified as outliers and excluded. This resulted in a reduced number of biological replicates for some timepoints. Because plants were destructively sampled for microbiome analyses at specific timepoints, repeated measurements over time corresponded to the same tub rather than the same individual plant. Accordingly, area under the curve (AUC) was calculated for each tub by integrating methane flux values across time. At the start of every measurement, a gas-tight chamber was placed over a rice plant with minimal disturbance. The chamber was left on for at least 5 minutes for flux analysis. Gas fluxes were calculated by fitting linear regression lines across the measurement time window.

The conversion to mass fluxes using the Ideal Gas Law equation:

$$F=\frac{P\times V}{R\times T}\times \frac{\frac{\Delta {{CH}}_{4}}{\Delta t}}{A}\times {10}^{-6}$$

where F is the CH₄ flux (mg m⁻² h⁻¹), P is the pressure, V is the volume, and A is the area of the sampling chamber. R is the ideal gas constant, and ΔCH₄/Δt is the slope of the gas concentration change over time fitted by a linear equation. T is the temperature in the sampling chamber (K).

Microbiome sample collection

Rhizosphere samples were collected 15, 30, 55, and 70 days after rice was transplanted into pots. Briefly, we removed rice roots from pots and manually shook them to remove excess soil from the roots, leaving only ~1 mm of soil attached. These roots were transferred to 50-mL Falcon tubes that contained 15 mL of Powerbead Solution (Qiagen), stirred vigorously with sterile forceps to obtain closely attached soil, and frozen immediately in liquid nitrogen after removing roots. Samples were stored in a −80 °C freezer before nucleic acid extraction.

Extraction and sequencing of total DNA and RNA

Approximately 2.0 g of rhizosphere soil was used for co-extraction of DNA and RNA by combining the Qiagen RNeasy PowerSoil Total RNA Kit and RNeasy PowerSoil DNA Elution Kit. Isolated nucleic acids were qualified and quantified using Nanodrop (Thermo Fisher) and Qubit (Invitrogen). For DNA samples, libraries were prepared using a Roche KAPA HyperPrep Kit and sequenced on Illumina NovaSeq6000 PE150 platform in Maryland Genomics. For RNA samples, ribosomal RNA (rRNA) was first depleted using the NEBNext rRNA Depletion Kit, then libraries were prepared using NEBNext Ultra II Directional RNA Library Prep Kit and sequenced as above. In total, 72 DNA samples and 72 RNA samples (three replicates for each of the bulk soils and five replicates for each planted with the PSY1 rice genotypes and Kitaake control, at each time point) were sequenced, generating 974.8 Gbp and 920.7 Gbp of metagenomic and metatranscriptomic raw read data. Details for each sample can be found in Supplementary Data 1.

Metaomics de novo assembly and binning

Raw 150-bp paired reads from metagenomic sequencing were processed using BBDuk contained in the BBTools suite (https://jgi.doe.gov/data-and-tools/software-tools/bbtools/). Biological replicates from the same samples were merged using BBMerge. Samples were individually assembled using metaSPAdes (v3.15.5) with default parameters⁸⁰. Scaffolds longer than 1 kb were binned by a combination of MetaBAT2 (v2.15)⁸¹, CONCOCT (v1.1.0)⁸², MaxBin2 (v2.2.7)⁸³, and VAMB (v3.0.2)⁸⁴, using differential abundance calculated by pairwise cross-mapping against all samples by BBMap. Resulting bins from the same samples were integrated and optimized using DAS Tool (v1.1.2)⁸⁵. Optimized bins derived from all samples were dereplicated at 98% whole-genome average nucleotide identity (ANI) using dRep (v3.4.0)⁸⁶, generating a set of approximately subspecies-level genomes. Quality of representative genomes was assessed using CheckM2 (v1.0.1)⁸⁷, and taxonomy was assigned by GTDB-Tk (v2.3.0)⁸⁸ based on the reference database version r214⁸⁹ and further converted to NCBI taxon. Only genomes with predicted completeness ≥50% and contamination ≤5% were used for downstream analyses (n = 296).

For metatranscriptomics, raw 150-bp paired reads were pre-treated using BBDuk, and rRNA was filtered by SortMeRNA (v4.3.6)⁹⁰. Clean reads primarily containing mRNA were individually assembled using rnaSPAdes (v3.15.5)⁹¹ to generate transcripts.

Gene prediction and annotation

Open reading frames (ORFs) in assembled scaffolds and transcripts were predicted by Prodigal (v2.6.3)⁹². Protein sequences from all samples were combined together and dereplicated at 95% identity along with 85% coverage to generate a non-redundant gene set, using Linclust integrated in MMseqs2^93,94. Predicted genes were basically annotated against KEGG⁹⁵, UniRef100, and UniProt⁹⁶ databases by USEARCH v10⁹⁷, and against the CAZy database⁹⁸ using the online server following the manual instruction⁹⁹. Protein domains were searched against Pfam (release 36.0)¹⁰⁰ using Astra (https://github.com/jwestrob/astra).

For the identification and clustering of rpS3 genes, the PF00189 HMM was first used to search against predicted proteins based on pre-built noise cutoffs using HMMER (v3.3)¹⁰¹. Corresponding nucleotide sequences of protein hits were pulled out and clustered at 99% ident¹⁰² using VSEARCH (v2.13.3)¹⁰³. Representative sequences from each cluster were blasted against the Genbank nr database to get preliminary taxonomic information¹⁰⁴. Precise taxonomy was further retrieved by phylogeny construction that involved alignment of sequences using MAFFT (v7.453)¹⁰⁵, trimming of alignments using trimAl (v1.4.rev15)¹⁰⁶, and generation of phylogenetic trees using IQ-TREE v1.6.12 with best-fit models automatically determined¹⁰⁷. Trees were decorated on the iTOL server for better visualization¹⁰⁸.

For genes closely related to CH₄ cycling, the PF02249 and PF02745 HMMs were used to recruit McrA-like sequences; the TIGR03079 was used to recruit PmoB-like sequences; and the PF02964 was used to recruit the gamma subunit of methane monooxygenase hydroxylase. Further identification and classification were performed based on phylogeny as described above.

For genes involved in H₂ production and consumption, three curated HMMs were used to target sequences homologous to [NiFe], [FeFe], and [Fe] hydrogenases, respectively⁴². Groups and subtypes of identified hydrogenases were distinguished based on protein phylogeny. [FeFe] Group A1 hydrogenases were further classified using the online tool HydDB, which integrates genetic context for accurate classification⁴². For genes related to nitrate reduction, proteins that contain the molybdopterin-binding domain or molybdopterin oxidoreductase domain were phylogenetically classified into distinct groups, including NarG and NapA exclusive groups. NifH proteins were identified using the TIGR01287 HMM that targets nitrogenase iron protein, included in all three nitrogenase forms (i.e., Mo-Fe, V-Fe, and Fe-only). Reductive DsrA was identified by DiSCo (v1.0.0)¹⁰⁹. Taxonomy of genes was identical to that of gene-located scaffolds, which was assigned if 50% of the co-existing proteins belonged to the same taxon.

Abundances of assembled genomes and genes

Metagenomic abundances of genomes and genes were calculated using CoverM (v0.6.1)¹¹⁰. Given genomes dereplicated at 98% ANI, a minimum read identity during mapping was set at 98% together with a minimum aligned per cent of 75% in the “genome” mode. For non-redundant genes at a 95% identity level, corresponding scaffolds were mapped by reads with a minimum identity of 95% and an alignment percentage of 75% using the “contig” mode. Abundances were considered to be zero if the covered fraction by any read was <50%. Normalized abundances were calculated by dividing coverage by total read length and multiplying by a factor of 1E11 for visualization.

Overall metatranscriptomic abundances of genomes were calculated using CoverM with the same cutoff parameters as above, followed by a normalization in which aligned read counts for each genome were divided by total read size and genome length, and multiplied by a factor of 1E15. To calculate the abundances of genes, corresponding scaffolds were first mapped by metatranscriptomic reads using Bowtie2 with “--end-to-end” “--sensitive” mode¹¹¹. Exactly aligned reads for genes inside scaffolds were counted using featureCounts (v2.0.6)¹¹². Final abundances, that said transcriptional activities, were normalized by dividing read counts by gene length and multiplying a factor of 1E11 for visualization.

Enrichment of genes/pathways in upregulated genomes

Based on calculated overall transcriptional abundances of genomes, DESeq2¹¹³ was used to identify significant differences in genome activities between the Kitaake and PSY1 genotypes, and between rice planted and unplanted groups. We consider genomes as “upregulated genomes” if: 1) they had significantly higher overall activities (log2FoldChange ≥ 1, p ≤ 0.05) in PSY1 rice genotypes compared with the Kitaake in ≥1 rice growth stage; 2) they never showed significantly higher overall activities in the Kitaake than PSY1 genotypes; and 3) they were more active in rice planted groups than in bulk soils. Finally, 17 of 296 genomes were identified as up-regulated genomes in PSY1 genotypes.

KEGG Orthology (KO) identifiers were assigned to predicted proteins in all genomes, using kofamscan (v1.3.0; https://github.com/takaram/kofam_scan) with KOfam set r2023-10-02. Only hits above score thresholds and with the highest score were selected for individual proteins. Next, we performed Fisher’s exact test to evaluate the over/under enrichment of identified KOs in up-regulated genomes (n = 17) relative to remaining genomes (n = 279). A total of 3509 KOs that appeared in ≥10 genomes were tested, and the resulting p values were adjusted using Benjamini–Hochberg method¹¹⁴. KOs with FDR ≤ 0.05 were considered significantly enriched (odds ratio > 1) or depleted (odds ratio <1) in up-regulated genomes.

Given KEGG metabolic pathways comprised of multiple KOs, we quantified pathways by summing up the abundances of all genes that are associated with related KOs. Differential abundances were calculated by subtracting abundances in the PSY1 genotypes from abundances in the Kitaake, shown as PSY1-17 and PSY1-9, respectively.

Calculations of genome-wide SAP values

We followed the previous paper to predict SAP values from genomes⁴⁶. Briefly, we first calculated the relative abundances (number of genes normalized by the total number of genes in genomes) of sugar and acid genes that are implicated in the corresponding degradation pathways, based on the list in the above reference. The generalized linear model was then used to predict SAP values for each genome, as:

$${SAP}=tanh \left(60.76\times S-20.21\times A\right)$$

where S and A indicate relative abundances of sugar and acid genes, respectively. Statistical difference in average SAP values between genome groups was calculated using the Wilcoxon rank sum test.

Rice roots and exudate collection

Samples for metabolomic analysis were collected from 7-day-old rice seedlings grown vertically on 1× MS solid media plates. Prior to collection, seedlings were removed from the plates and thoroughly washed with deionized water to remove any residual media from the roots. For root tissue samples, the entire root system of 20 seedlings was ground using a mortar and pestle, with a subsample transferred to a 2 mL tube (Eppendorf #022363352). Tissue fresh weight was recorded with pre-weighted tubes, and samples were stored at −80 °C until further processing. For root exudate samples, 15 seedlings were placed in 50 mL flasks containing ultra-pure water and held in place by sponges so that only the root tips were submerged. The flasks were incubated with shaking for 2 hours before the exudates were collected in 50 mL tubes (Fisher Scientific#14-959-49 A). Exudates were initially stored at −20 °C for a few hours, then transferred to −80 °C until further use.

Liquid chromatography sample preparation

To extract metabolites from 300 to 400 mg root tissue and agar control media, samples were first frozen, lyophilized dry (FreeZone 2.5 Plus; Labconco), then powderized by bead-beating (MiniBeadbeater96, BioSpec) for 5 seconds, 2×, using a 3.2 mm stainless steel bead. Next, 500 μL of 100% methanol was added to each powderized root sample and 1.5 mL to each powderized agar sample, briefly vortexed, then sonicated in an ice water bath for 10minutes. Samples were centrifuged (5 min, 11,000 × g) to pellet debris, supernatant removed, and transferred to a 2 mL Eppendorf, dried in a SpeedVac (SPD111V; Thermo Fisher Scientific). To extract metabolites from exudates collected in 40 mL water, samples were first frozen then lyophilized dry. Next, 3 mL of 100% methanol was added to each sample, briefly vortexed, then sonicated in an iced water bath for 10 minutes. Samples were centrifuged (5 min, 7000 × g) to pellet debris, supernatant removed and transferred to a 5 mL Eppendorf, then dried in a SpeedVac. Next, 1 mL of 100% methanol was added to each dried extract, briefly vortexed, then sonicated in an ice water bath for 10 minutes. Samples were centrifuged again, then supernatant removed and transferred to a 2 mL Eppendorf and dried in a SpeedVac. All dried extracts were stored frozen at −80 °C. The same extraction procedures were followed using empty 50 mL and 2 mL tubes to serve as extraction controls, and a 50 mL tube containing 50 mL water to serve as an exudate collection control (3 replicates each).

In preparation for LC-MS, samples were resuspended in 100% methanol containing a mix of isotopically labeled internal standards (https://doi.org/10.17504/protocols.io.kxygxydwkl8j/v1). Resuspension volume was varied for each root extract to normalize by amount of extracted biomass (1500 mg/mL), while all other samples were resuspended in 150 μL.

LC-MS/MS metabolomics analysis

Samples were run using both normal and reverse-phase chromatography. Each of these was performed using an Agilent 1290 LC stack, with MS and MS/MS data collected using an Exploris 120 Orbitrap MS (Thermo Fisher Scientific, San Jose, CA) for normal phase, and QExactive HF Orbitrap for reverse phase. Mass spectrometer parameters and gradients are provided in this protocols.io (https://doi.org/10.17504/protocols.io.kxygxydwkl8j/v1) for normal phase, and this protocols.io (https://doi.org/10.17504/protocols.io.ewov19r27lr2/v1) for reverse phase. For both normal phase and reverse phase, full MS1 spectra were collected at 60k resolution in both positive and negative ionization mode, with MS/MS fragmentation data acquired using stepped then averaged 10, 20, and 40 eV collision energies at 15 k resolution. For normal phase, the MS1 m/z range for data collection was 70-1050, while for reverse phase, both m/z 80–1200 and m/z 1000–3000 were used in separate runs to cover mass ranges for metabolites and peptides. Samples consisted of three biological replicates each and extraction controls, with sample injection order randomized by replicate and an injection blank of 100% methanol run between each sample, with the blank replaced by an injection of internal standard mix every third sample, as well as QC mix every 15 samples. The raw data from the metabolite LC-MS runs are provided. For untargeted data analysis, features with 0.8 min<RT < 18.5 min (HILIC column) or features with 0.5 min<RT < 9.5 min (C18 column), and maximum peak height fold-change between sample and extraction control >3 were selected. We performed a Feature-Based Molecular Networking (FBMN)¹¹⁵ workflow using MZmine-3.7.2¹¹⁶ and a Global Natural Products Social Molecular Networking (GNPS2; http://gnps2.org)¹¹⁷. The MZmine workflow was used to generate a list of features obtained from extracted ion chromatograms containing chromatographic peaks within a narrow m/z range and filtered to remove isotopes. For each feature, the most intense fragmentation spectrum was uploaded to GNPS for putative identification by comparison with mass spectra deposited in the database. Compound classes were attributed to identified compounds using NPClassifier¹¹⁸.

For targeted analysis, compound identification was performed by comparing detected m/z, retention time (RT) and fragmentation pattern to standards run in-house using the same LC-MS methods and assigning a confidence level based on these criteria. For each identified compound, the highest level of confidence (Exceeds Level 1)¹¹⁹ was achieved when the measured m/z agreed with the expected theoretical m/z with less than 5 ppm error, detected RT of the feature was within 0.5 minutes of theoretical, and the fragmentation pattern matched that of the standard. When MSMS was not collected, some compounds were identified based on only mz and RT (Level 1); mis-matching MSMS would invalidate an identification.

NPOC measurement

Frozen root exudates collected as described previously were thawed at 4 °C and then diluted to 8 mL using Milli-Q water. Samples were analyzed on a Shimadzu TOC-L CPH total organic carbon analyzer equipped with TNM-L and ASI-L modules using UHP synthetic air mix as carrier gas. Calibration curves for non-purgeable organic carbon were prepared using potassium hydrogen phthalate. Standards and samples were analyzed under the same settings; blanks and control samples were included for quality control purposes. Sample measurements were set up with 1 wash step, followed by automated acid addition at 1.5% 1 M HCl with a 2.5-minute sparging period at 80 mL/min to remove inorganic carbon and purgeable carbon. Acidified, sparged samples containing non-purgeable organic carbon were then injected at 50 μL for analysis using a max 4.5 min integration window for each carbon measurement; this was performed with 3–5 technical replicate injections with the max cv set at 2%. Measurements were corrected for dilution factors and then reported as ppm C (mg of C per L of exudate).

Soil incubations with rice root exudates

To assess the direct impact of rice root exudates on CH₄ emissions, microcosm incubations were performed in wide-mouth glass mason jars (volume: 500 mL) fitted with gas-tight rubber septa. Each microcosm was prepared by mixing 10 g of air-dried agronomic soil, collected from the same rice basin used in the greenhouse experiments, with 10 mL of deionized water. The microcosms were incubated at 28 °C in the dark for 48 hours to allow microbial respiration to stabilize. Root exudates from Kitaake, PSY1-, and PSY2-overexpressing rice seedlings were collected as previously described, lyophilized, and reconstituted in 1 mL of Milli-Q water. Each microcosm was flushed with nitrogen gas for 1 minute to displace ambient oxygen, and then exudates were gently added to the soil, taking care to avoid physical disturbance. The jars were sealed immediately, and headspace gas samples were collected at 0, 5, 10, 20, 30, 60, 120, 180, 240, and 300 minutes post-addition. CH₄ concentrations in the headspace were quantified using a cavity ring-down spectrometer (Picarro G2201, Picarro Inc.), providing a direct measurement of microbial CH₄ emissions in response to rice root exudate chemistry.CH₄ concentrations in ppm were converted to nanograms of CH₄ using the ideal gas law, assuming a constant headspace volume of 500 mL throughout the experiment, with final values expressed in mg CH₄ per g of dried soil.

Construction and analysis of genome-scale metabolic models

We selected microbial genomes for metabolic modeling based on the following criteria: (1) genomes contain H₂-producing or H₂-consuming hydrogenases; (2) genomes are predicted to have completeness > 90% and contamination <5%; (3) genomes are affiliated with different phyla. Three H₂-producing genomes from Bacteroidia, Ignavibacteriota, and Moranbacteria, and three H₂-consuming genomes from Acidobacteriota, Proteobacteria, and Deinococcota were finally selected and annotated using the App “Annotate Genome/Assembly with RASTtk” (v1.073) in KBase¹²⁰. The resulting annotated genomes were used as inputs for the App “MS2 - Build Prokaryotic Metabolic Models with OMEGGA” (v.2.0.0) to build gap-filled metabolic models using media comprised of basic nutrients and PSY1 over-secreted compounds (Supplementary Data 8). Basic nutrients included glucose and glutamate as primary carbon sources, which were present and had no significant difference in concentrations across rice genotypes. Also included were standard essential factors for microbial growth (e.g., Fe²⁺). PSY1 over-secreted compounds were measured at significantly higher concentrations in root exudates from PSY1 rice genotypes compared with the Kitaake.

Dynamic flux balance analysis (dFBA) was performed for each microbial model individually, assuming well-mixed batch growth conditions, using the Computation Of Microbial Ecosystems in Time and Space (COMETS) platform (v2.12.2)¹²¹. We set the initial biomass at 1 × 10⁻⁶ g and used default simulation parameters. To compare results across varying conditions, we calculated the maximal potential of substrate turnover and biomass yield when microbial growth stopped and substrates no longer changed, for each condition. As glucose and glutamate in basic nutrients were likely the most abundant carbohydrate and amino acid secreted by rice roots^122,123, concentrations of PSY1 over-secreted compounds were varied from 0.01–1-fold as that of basic nutrients. O₂ is transported to the rhizosphere through rice aerenchyma, and its concentrations were set at 1–10 fold that of basic nutrients. H₂ concentrations were set at 0 when testing H₂ production potential and at 10-fold when testing H₂ consumption potential. For visualization purposes, we calculated the consumed H₂ amounts by subtracting the final values from the initial.

Statistical analysis

NMDS analyses for the rpS3 gene were performed and plotted using “vegan” and “ggplot2” packages in R (v4.2.2)¹²⁴. Differences and significance between groups were assessed by analysis of similarities and permutational multivariate analysis of variance tests. Outliers were detected and removed from replicates using the MAD test¹²⁵. Differential expressions of genomes between samples were calculated by DESeq2¹¹³.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All metagenomic and metatranscriptomic sequencing reads have been deposited in the NCBI Sequence Read Archive (SRA) with the project accession number: PRJNA1161606. Microbial metabolic models have been deposited in figshare (https://doi.org/10.6084/m9.figshare.29115692). Assembled microbial genomes are present in NCBI BioProject PRJNA1161606. They can also be accessed at: https://ggkbase.berkeley.edu/Assembled_genomes_from_rice_genotype_experiment. The metabolomics raw data can be accessed at: https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=3349af8926624635af24c6b01babebd1. Source data are provided with this paper.

References

Lan, X., Thoning, K. W. & Dlugokencky E. J. Trends in globally-averaged CH4, N2O, and SF6 determined from NOAA Global Monitoring Laboratory measurements. Version 2024-01, https://doi.org/10.15138/P8XG-AA10 (2022).
Liu, Y. L. et al. Rice paddy soils are a quantitatively important carbon store according to a global synthesis. Commun. Earth Environ. 2, 154 (2021).
Thauer, R. K., Kaster, A. K., Seedorf, H., Buckel, W. & Hedderich, R. Methanogenic archaea: ecologically relevant differences in energy conservation. Nat. Rev. Microbiol. 6, 579–591 (2008).
Article CAS PubMed Google Scholar
Raghoebarsing, A. A. et al. A microbial consortium couples anaerobic methane oxidation to denitrification. Nature 440, 918–921 (2006).
Article ADS CAS PubMed Google Scholar
Cai, Y., Zheng, Y., Bodelier, P. L. E., Conrad, R. & Jia, Z. Conventional methanotrophs are responsible for atmospheric methane oxidation in paddy soils. Nat. Commun. 7, 11728 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Liu, Y. C. & Whitman, W. B. Metabolic, phylogenetic, and ecological diversity of the methanogenic archaea. Ann. N. Y. Acad. Sci. 1125, 171–189 (2008).
Article ADS CAS PubMed Google Scholar
Bridgham, S. D., Cadillo-Quiroz, H., Keller, J. K. & Zhuang, Q. L. Methane emissions from wetlands: biogeochemical, microbial, and modeling perspectives from local to global scales. Glob. Change Biol. 19, 1325–1346 (2013).
Article ADS Google Scholar
Bin Rahman, A. N. M. R. & Zhang, J. H. Trends in rice research: 2030 and beyond. Food Energy Secur. 12, e390 (2023).
Qian, H. Y. et al. Greenhouse gas emissions and mitigation in rice agriculture. Nat. Rev. Earth Environ. 4, 716–732 (2023).
Article ADS CAS Google Scholar
Su, J. et al. Expression of barley SUSIBA2 transcription factor yields high-starch low-methane rice. Nature 523, 602 (2015).
Article ADS CAS PubMed Google Scholar
Kwon, Y. et al. Loss-of-function gs3 allele decreases methane emission and increases grain yield in rice. Nat. Clim. Change 13, 1329–1333 (2023).
Article ADS CAS Google Scholar
Yamauchi, T., Abe, F., Tsutsumi, N. & Nakazono, M. Root Cortex Provides a Venue for Gas-Space Formation and Is Essential for Plant Adaptation to Waterlogging. Front. Plant Sci. 10, 259 (2019).
Article PubMed PubMed Central Google Scholar
Pedersen, O., Sauter, M., Colmer, T. D. & Nakazono, M. Regulation of root adaptive anatomical and morphological traits during low soil oxygen. N. Phytol. 229, 42–49 (2021).
Article CAS Google Scholar
Chen, Y. et al. Rice root morphological and physiological traits interaction with rhizosphere soil and its effect on methane emissions in paddy fields. Soil Biol. Biochem. 129, 191–200 (2019).
Article CAS Google Scholar
Ding, H., Jiang, Y. & Cao, C. Deep rice root systems reduce methane emissions in rice paddies. Plant Soil 468, 337–352 (2021).
Article CAS Google Scholar
Ercoli, M. F. et al. An open source plant kinase chemogenomics set. Plant Direct 6, e460 (2022).
Article CAS PubMed PubMed Central Google Scholar
Komori, R., Amano, Y., Ogawa-Ohnishi, M. & Matsubayashi, Y. Identification of tyrosylprotein sulfotransferase in Arabidopsis. Proc. Natl. Acad. Sci. 106, 15067–15072 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Wang, J. et al. Allosteric receptor activation by the plant peptide hormone phytosulfokine. Nature 525, 265–268 (2015).
Article ADS CAS PubMed Google Scholar
Song, W. et al. Signature motif-guided identification of receptors for peptide hormones essential for root meristem growth. Cell Res. 26, 674–685 (2016).
Article CAS PubMed PubMed Central Google Scholar
Stewart, V. & Ronald, P. C. Sulfotyrosine residues: Interaction specificity determinants for extracellular protein–protein interactions. J. Biol. Chem. 298, 102232 (2022).
Article CAS PubMed PubMed Central Google Scholar
Amano, Y., Tsubouchi, H., Shinohara, H., Ogawa, M. & Matsubayashi, Y. Tyrosine-sulfated glycopeptide involved in cellular proliferation and expansion in Arabidopsis. Proc. Natl. Acad. Sci. USA 104, 18333–18338 (2007).
Article ADS CAS PubMed PubMed Central Google Scholar
Ogawa-Ohnishi, M. et al. Peptide ligand-mediated trade-off between plant growth and stress response. Science 378, 175–180 (2022).
Article ADS CAS PubMed Google Scholar
Ercoli, M. F., Shigenaga, A. M., Teixeira De Araujo, A., Jain, R. & Ronald, P. C. Tyrosine-sulfated peptide hormone induces flavonol biosynthesis to control elongation and differentiation in Arabidopsis primary root. Preprint at https://doi.org/10.1101/2024.02.02.578681 (2024).
Wang, Y., Chen, W., Ou, Y., Zhu, Y. & Li, J. Arabidopsis ROOT ELONGATION RECEPTOR KINASES negatively regulate root growth putatively via altering cell wall remodeling gene expression. J. Integr. Plant Biol. 64, 1502–1513 (2022).
Article CAS PubMed Google Scholar
Pruitt, R. N. et al. A microbially derived tyrosine-sulfated peptide mimics a plant peptide hormone. N. Phytol. 215, 725–736 (2017).
Article CAS Google Scholar
Aulakh, M. S., Wassmann, R., Bueno, C. & Rennenberg, H. Impact of root exudates of different cultivars and plant development stages of rice (L.) on methane production in a paddy soil. Plant Soil 230, 77–86 (2001).
Article CAS Google Scholar
Kesawat, M. S. et al. Genome-wide survey of peptides containing tyrosine sulfation (PSY) gene family and potential PSY specific miRNA revealed their role in plant development and diverse stress conditions in rice (Oryza sativa L. Plant Stress 11, 100412 (2024).
Article CAS Google Scholar
Tost, A. S., Kristensen, A., Olsen, L. I., Axelsen, K. B. & Fuglsang, A. T. The PSY peptide family-expression, modification and physiological implications. Genes Basel 12, 218 (2021).
Wang, L. et al. A dynamic gene expression atlas covering the entire life cycle of rice. Plant J. 61, 752–766 (2010).
Article CAS PubMed Google Scholar
Huang, L. & Schiefelbein, J. Conserved gene expression programs in developing roots from diverse plants. Plant Cell 27, 2119–2132 (2015).
Article CAS PubMed PubMed Central Google Scholar
Chandran, A. K. N. et al. Rice Chalky Grain 5 regulates natural variation for grain quality under heat stress. Front. Plant Sci. 13, 1026472 (2022).
Article CAS PubMed PubMed Central Google Scholar
Sharon, I. et al. Accurate, multi-kb reads resolve complex populations and detect rare microorganisms. Genome Res. 25, 534–543 (2015).
Article CAS PubMed PubMed Central Google Scholar
Delgado-Baquerizo, M. et al. A global atlas of the dominant bacteria found in soil. Science 359, 320 (2018).
Article ADS CAS PubMed Google Scholar
Ding, L. J. et al. Microbiomes inhabiting rice roots and rhizosphere. FEMS Microbiol. Ecol. 95, fiz040 (2019).
Ma, B. et al. A genomic catalogue of soil microbiomes boosts mining of biodiversity and genetic resources. Nat. Commun. 14, 7318 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Cho, S. R. et al. A new approach to suppress methane emissions from rice cropping systems using ethephon. Sci. Total Environ. 804, 150159 (2022).
Zecchin, S. et al. Microbial communities in paddy soils: differences in abundance and functionality between rhizosphere and pore water, the influence of different soil organic carbon, sulfate fertilization and cultivation time, and contribution to arsenic mobility and speciation. FEMS Microbiol. Ecol. 99, fiad121 (2023).
Breidenbach, B., Pump, J. & Dumont, M. G. Microbial community structure in the rhizosphere of rice plants. Front. Microbiol. 6, 1537 (2016).
Grosse, S. et al. Purification and characterization of the soluble methane monooxygenase of the type II methanotrophic bacterium sp strain WI 14. Appl. Environ. Microbiol. 65, 3929–3935 (1999).
Article ADS CAS PubMed PubMed Central Google Scholar
Garcia, P. S., Gribaldo, S. & Borrel, G. Diversity and evolution of methane-related pathways in archaea. Annu. Rev. Microbiol. 76, 727–755 (2022).
Article CAS PubMed Google Scholar
Lubitz, W., Ogata, H., Rudiger, O. & Reijerse, E. Hydrogenases. Chem. Rev. 114, 4081–4148 (2014).
Article CAS PubMed Google Scholar
Sondergaard, D., Pedersen, C. N. S. & Greening, C. HydDB: a web tool for hydrogenase classification and analysis. Sci. Rep. 6, 34212 (2016).
Article ADS PubMed PubMed Central Google Scholar
Greening, C. et al. Genomic and metagenomic surveys of hydrogenase distribution indicate H is a widely utilised energy source for microbial growth and survival. ISME J. 10, 761–777 (2016).
Article CAS PubMed Google Scholar
Zehr, J. P. & Capone, D. G. Changing perspectives in marine nitrogen fixation. Science 368, 729 (2020).
Article Google Scholar
Dong, X. Y. et al. Phylogenetically and catabolically diverse diazotrophs reside in deep-sea cold seep sediments. Nat. Commun. 13, 4885 (2022).
Gralka, M., Pollak, S. & Cordero, O. X. Genome content predicts the carbon catabolic preferences of heterotrophic bacteria. Nat. Microbiol. 8, 1799–1808 (2023).
Article CAS PubMed Google Scholar
OECD & Food and Agriculture Organization of the United Nations. OECD-FAO Agricultural Outlook 2021–2030. (OECD, https://doi.org/10.1787/19428846-en. 2021).
Wang, X. et al. Global and regional trends in greenhouse gas emissions from rice production, trade, and consumption. Environ. Impact Assess. Rev. 101, 107141 (2023).
Article Google Scholar
Conrad, R., Klose, M., Lu, Y. & Chidthaisong, A. Methanogenic pathway and archaeal communities in three different anoxic soils amended with rice straw and maize straw. Front. Microbiol. 3, 4 (2012).
Hu, J. et al. A low-methane rice with high-yield potential realized via optimized carbon partitioning. Sci. Total Environ. 920, 170980 (2024).
Article CAS PubMed Google Scholar
Chidthaisong, A. & Conrad, R. Turnover of glucose and acetate coupled to reduction of nitrate, ferric iron and sulfate and to methanogenesis in anoxic rice field soil. FEMS Microbiol. Ecol. 31, 73–86 (2000).
Article CAS PubMed Google Scholar
Krüger, M., Frenzel, P. & Conrad, R. Microbial processes influencing methane emission from rice fields. Glob. Change Biol. 7, 49–63 (2001).
Article ADS Google Scholar
Jetten, M. S. M., Stams, A. J. M. & Zehnder, A. J. B. Methanogenesis from acetate - a comparison of the acetate metabolism in methanothrix-Soehngenii and Methanosarcina Spp. FEMS Microbiol. Lett. 88, 181–197 (1992).
Article CAS Google Scholar
Chin, K. J., Lueders, T., Friedrich, M. W., Klose, M. & Conrad, R. Archaeal community structure and pathway of methane formation on rice roots. Microb. Ecol. 47, 59–67 (2004).
Article ADS CAS PubMed Google Scholar
Lovley, D. R. Minimum threshold for hydrogen metabolism in methanogenic bacteria. Appl. Environ. Microbiol. 49, 1530–1531 (1985).
Article ADS CAS PubMed PubMed Central Google Scholar
Feldewert, C., Lang, K. & Brune, A. The hydrogen threshold of obligately methyl-reducing methanogens. Fems Microbiol. Lett. 367, fnaa137 (2020).
Liu, F. & Conrad, R. Thermoanaerobacteriaceae oxidize acetate in methanogenic rice field soil at 50 °C. Environ. Microbiol. 12, 2341–2354 (2010).
Article CAS PubMed Google Scholar
Mosbæk, F. et al. Identification of syntrophic acetate-oxidizing bacteria in anaerobic digesters by combined protein-based stable isotope probing and metagenomics. ISME J. 10, 2405–2418 (2016).
Article PubMed PubMed Central Google Scholar
Pan, X. et al. Deep insights into the network of acetate metabolism in anaerobic digestion: focusing on syntrophic acetate oxidation and homoacetogenesis. Water Res. 190, 116774 (2021).
Article CAS PubMed Google Scholar
Lyu, Z. & Lu, Y. Metabolic shift at the class level sheds light on adaptation of methanogens to oxidative environments. ISME J. 12, 411–423 (2018).
Article PubMed Google Scholar
Angel, R., Claus, P. & Conrad, R. Methanogenic archaea are globally ubiquitous in aerated soils and become active under wet anoxic conditions. ISME J. 6, 847–862 (2012).
Article CAS PubMed Google Scholar
Cabrol, L. et al. Microbial ecology of fermentative hydrogen producing bioprocesses: useful insights for driving the ecosystem function. FEMS Microbiol. Rev. 41, 158–181 (2017).
Article CAS PubMed Google Scholar
Kerdchoechuen, O. Methane emission in four rice varieties as related to sugars and organic acids of roots and root exudates and biomass yield. Agric. Ecosyst. Environ. 108, 155–163 (2005).
Article CAS Google Scholar
Qian, P. et al. The CLE9/10 secretory peptide regulates stomatal and vascular development through distinct receptors. Nat. Plants 4, 1071–1081 (2018).
Article CAS PubMed Google Scholar
Zampieri, M., Hörl, M., Hotz, F., Müller, N. F. & Sauer, U. Regulatory mechanisms underlying coordination of amino acid and glucose catabolism in Escherichia coli. Nat. Commun. 10, 3354 (2019).
Article ADS PubMed PubMed Central Google Scholar
Chandel, N. S. Amino acid metabolism. Cold Spring Harb. Perspect. Biol. 13, a040584 (2021).
Article CAS PubMed PubMed Central Google Scholar
Moe, L. A. Amino acids in the rhizosphere: from plants to microbes. Am. J. Bot. 100, 1692–1705 (2013).
Article CAS PubMed Google Scholar
Aulakh, M. S., Wassmann, R., Bueno, C., Kreuzwieser, J. & Rennenberg, H. Characterization of Root Exudates at Different Growth Stages of Ten Rice (Oryza sativa L.) Cultivars. Plant Biol. 3, 139–148 (2008).
Article Google Scholar
Zhalnina, K. et al. Dynamic root exudate chemistry and microbial substrate preferences drive patterns in rhizosphere microbial community assembly. Nat. Microbiol. 3, 470–480 (2018).
Article CAS PubMed Google Scholar
Heuermann, D. et al. Distinct metabolite classes in root exudates are indicative for field- or hydroponically-grown cover crops. Front. Plant Sci. 14, 1122285 (2023).
Article CAS PubMed PubMed Central Google Scholar
Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012).
Article CAS PubMed PubMed Central Google Scholar
Crooks, G. E., Hon, G., Chandonia, J.-M. & Brenner, S. E. WebLogo: a sequence logo generator: Fig. 1. Genome Res 14, 1188–1190 (2004).
Article CAS PubMed PubMed Central Google Scholar
Karimi, M., Depicker, A. & Hilson, P. Recombinational cloning with plant gateway vectors. Plant Physiol. 145, 1144–1154 (2007).
Article CAS PubMed PubMed Central Google Scholar
Debernardi, J. M. et al. A GRF–GIF chimeric protein improves the regeneration efficiency of transgenic plants. Nat. Biotechnol. 38, 1274–1279 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Hellemans, J., Mortier, G., De Paepe, A., Speleman, F. & Vandesompele, J. qBase relative quantification framework and software for management and automated analysis of real-time quantitative PCR data. Genome Biol. 8, R19 (2007).
Article PubMed PubMed Central Google Scholar
Taylor, I. et al. Mechanism and function of root circumnutation. Proc. Natl. Acad. Sci. 118, e2018940118 (2021).
Article CAS PubMed PubMed Central Google Scholar
Ursache, R., Andersen, T. G., Marhavý, P. & Geldner, N. A protocol for combining fluorescent proteins with histological stains for diverse cell wall components. Plant J. 93, 399–412 (2018).
Article ADS CAS PubMed Google Scholar
Fonseca-García, C. et al. Defined synthetic microbial communities colonize and benefit field-grown sorghum. ISME J. 18, wrae126 (2024).
Article PubMed PubMed Central Google Scholar
Suzuki, S. et al. High-throughput determination of thioglycolic acid lignin from rice. Plant Biotechnol. 26, 337–340 (2009).
Article CAS Google Scholar
Nurk, S., Meleshko, D., Korobeynikov, A. & Pevzner, P. A. metaSPAdes: a new versatile metagenomic assembler. Genome Res. 27, 824–834 (2017).
Article CAS PubMed PubMed Central Google Scholar
Kang, D. W. D. et al. MetaBAT 2: an adaptive binning algorithm for robust and efficient genome reconstruction from metagenome assemblies. Peerj 7, e7359 (2019).
Article PubMed PubMed Central Google Scholar
Alneberg, J. et al. Binning metagenomic contigs by coverage and composition. Nat. Methods 11, 1144–1146 (2014).
Article CAS PubMed Google Scholar
Wu, Y. W., Simmons, B. A. & Singer, S. W. MaxBin 2.0: an automated binning algorithm to recover genomes from multiple metagenomic datasets. Bioinformatics 32, 605–607 (2016).
Article CAS PubMed Google Scholar
Nissen, J. N. et al. Improved metagenome binning and assembly using deep variational autoencoders. Nat. Biotechnol. 39, 555–560 (2021).
Article CAS PubMed Google Scholar
Sieber, C. M. K. et al. Recovery of genomes from metagenomes via a dereplication, aggregation and scoring strategy. Nat. Microbiol. 3, 836 (2018).
Article CAS PubMed PubMed Central Google Scholar
Olm, M. R., Brown, C. T., Brooks, B. & Banfield, J. F. dRep: a tool for fast and accurate genomic comparisons that enables improved genome recovery from metagenomes through de-replication. ISME J. 11, 2864–2868 (2017).
Article CAS PubMed PubMed Central Google Scholar
Chklovski, A., Parks, D. H., Woodcroft, B. J. & Tyson, G. W. CheckM2: a rapid, scalable and accurate tool for assessing microbial genome quality using machine learning. Nat. Methods 20, 1203–1212 (2023).
Article CAS PubMed Google Scholar
Chaumeil, P. A., Mussig, A. J., Hugenholtz, P. & Parks, D. H. GTDB-Tk v2: memory friendly classification with the genome taxonomy database. Bioinformatics 38, 5315–5316 (2022).
Article CAS PubMed PubMed Central Google Scholar
Parks, D. H. et al. A complete domain-to-species taxonomy for Bacteria and Archaea. Nat. Biotechnol. 38, 1079–1086 (2020).
Article CAS PubMed Google Scholar
Kopylova, E., Noe, L. & Touzet, H. SortMeRNA: fast and accurate filtering of ribosomal RNAs in metatranscriptomic data. Bioinformatics 28, 3211–3217 (2012).
Article CAS PubMed Google Scholar
Bushmanova, E., Antipov, D., Lapidus, A. & Prjibelski, A. D. rnaSPAdes: a de novo transcriptome assembler and its application to RNA-Seq data. Gigascience 8, giz100 (2019).
Hyatt, D. et al. Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinformatics 11, 119 (2010).
Steinegger, M. & Soding, J. MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nat. Biotechnol. 35, 1026–1028 (2017).
Article CAS PubMed Google Scholar
Steinegger, M. & Soding, J. Clustering huge protein sequence sets in linear time. Nat. Commun. 9, 2542 (2018).
Article ADS PubMed PubMed Central Google Scholar
Kanehisa, M., Furumichi, M., Sato, Y., Kawashima, M. & Ishiguro-Watanabe, M. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. https://doi.org/10.1093/nar/gkac963. (2022).
Bateman, A. et al. UniProt: the Universal Protein Knowledgebase in 2023. Nucleic Acids Res. https://doi.org/10.1093/nar/gkac1052 (2022).
Edgar, R. C. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461 (2010).
Article CAS PubMed Google Scholar
Drula, E. et al. The carbohydrate-active enzyme database: functions and literature. Nucleic Acids Res. 50, D571–D577 (2022).
Article CAS PubMed PubMed Central Google Scholar
Zheng, J. et al. dbCAN3: automated carbohydrate-active enzyme and substrate annotation. Nucleic Acids Res. 51, W115–W121 (2023).
Article CAS PubMed PubMed Central Google Scholar
Finn, R. D. et al. The Pfam protein families database: towards a more sustainable future. Nucleic Acids Res. 44, D279–D285 (2016).
Article CAS PubMed Google Scholar
Eddy, S. R. Accelerated profile HMM searches. Plos Comput. Biol. 7, e1002195 (2011).
Article ADS MathSciNet CAS PubMed PubMed Central Google Scholar
Olm, M. R. et al. Consistent metagenome-derived metrics verify and delineate bacterial species boundaries. Msystems 5, e00731–19 (2020).
Article CAS PubMed PubMed Central Google Scholar
Rognes, T., Flouri, T., Nichols, B., Quince, C. & Mahe, F. VSEARCH: a versatile open source tool for metagenomics. Peerj 4, e2584 (2016).
Boratyn, G. M. et al. BLAST: a more efficient report with usability improvements. Nucleic Acids Res. 41, W29–W33 (2013).
Article PubMed PubMed Central Google Scholar
Katoh, K., Misawa, K., Kuma, K. & Miyata, T. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30, 3059–3066 (2002).
Article CAS PubMed PubMed Central Google Scholar
Capella-Gutierrez, S., Silla-Martinez, J. M. & Gabaldon, T. trimAl: a tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics 25, 1972–1973 (2009).
Article CAS PubMed PubMed Central Google Scholar
Nguyen, L. T., Schmidt, H. A., von Haeseler, A. & Minh, B. Q. IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol. Biol. Evol. 32, 268–274 (2015).
Article CAS PubMed Google Scholar
Letunic, I. & Bork, P. Interactive Tree Of Life (iTOL) v5: an online tool for phylogenetic tree display and annotation. Nucleic Acids Res. 49, W293–W296 (2021).
Article CAS PubMed PubMed Central Google Scholar
Neukirchen, S. & Sousa, F. L. DiSCo: a sequence-based type-specific predictor of Dsr-dependent dissimilatory sulphur metabolism in microbial data. Microb. Genomics 7, 000603 (2021).
Aroney, S. T. N. et al. CoverM: read alignment statistics for metagenomics. Bioinformatics 41, btaf147 (2025).
Article CAS PubMed PubMed Central Google Scholar
Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–U54 (2012).
Article CAS PubMed PubMed Central Google Scholar
Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930 (2014).
Article CAS PubMed Google Scholar
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15,550 (2014).
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate - a practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B-Stat. Methodol. 57, 289–300 (1995).
Article MathSciNet Google Scholar
Nothias, L.-F. et al. Feature-based molecular networking in the GNPS analysis environment. Nat. Methods 17, 905–908 (2020).
Article CAS PubMed PubMed Central Google Scholar
Schmid, R. et al. Integrative analysis of multimodal mass spectrometry data in MZmine 3. Nat. Biotechnol. 41, 447–449 (2023).
Article CAS PubMed PubMed Central Google Scholar
Wang, M. et al. Sharing and community curation of mass spectrometry data with Global Natural Products Social Molecular Networking. Nat. Biotechnol. 34, 828–837 (2016).
Article CAS PubMed PubMed Central Google Scholar
Kim, H. W. et al. NPClassifier: a deep neural network-based structural classification tool for natural products. J. Nat. Prod. 84, 2795–2807 (2021).
Article CAS PubMed PubMed Central Google Scholar
Sumner, L. W. et al. Proposed minimum reporting standards for chemical analysis: chemical analysis working group (CAWG) Metabolomics Standards Initiative (MSI). Metabolomics 3, 211–221 (2007).
Article CAS PubMed PubMed Central Google Scholar
Arkin, A. P. et al. KBase: the United States department of energy systems biology knowledgebase. Nat. Biotechnol. 36, 566–569 (2018).
Article CAS PubMed PubMed Central Google Scholar
Dukovski, I. et al. A metabolic modeling platform for the computation of microbial ecosystems in time and space (COMETS). Nat. Protoc. 16, 5030–5082 (2021).
Article CAS PubMed PubMed Central Google Scholar
Bacilio-Jiménez, M. et al. Chemical characterization of root exudates from rice (Oryza sativa) and their effects on the chemotactic response of endophytic bacteria. Plant Soil 249, 271–277 (2003).
Article Google Scholar
Tawaraya, K., Horie, R., Wagatsuma, T., Saito, K. & Oikawa, A. Metabolite profiling of shoot extract, root extract, and root exudate of rice under nitrogen and phosphorus deficiency. Soil Sci. Plant Nutr. 64, 312–322 (2018).
Article CAS Google Scholar
R. Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ (2022).
Leys, C., Ley, C., Klein, O., Bernard, P. & Licata, L. Detecting outliers: do not use standard deviation around the mean, use absolute deviation around the median. J. Exp. Soc. Psychol. 49, 764–766 (2013).
Article Google Scholar

Download references

Acknowledgements

We would like to thank Ryan C. Packer, Christina Fossum, and Rory M. Greenhalgh for their help setting up the greenhouse experiment. We thank Bethany C. Kolody and Jackie Zorz for assistance in using R, and Daniel A. Gittins for help with hydrogenase HMMs. We thank Leylen Miloslavich for help with soil incubation experiments. This project has been made possible in part by grant number CZIF2022-007203 from the Chan Zuckerberg Initiative Foundation (to J.F.B., P.C.R., and J.P.R.), and by support from the Emerson Collective (J.F.B.). This material is based upon work supported by the Joint BioEnergy Institute, U.S. Department of Energy, Office of Science, Biological and Environmental Research Program under Award Number DE-AC02-05CH11231 with Lawrence Berkeley National Laboratory (P.C.R. and H.V.S.). Funding was also provided by the Bill and Melinda Gates Foundation (grant number INV-037174 to J.F.B.). The findings and conclusions are those of the authors and do not necessarily reflect positions or policies of the Bill and Melinda Gates Foundation. The metabolomic analysis was conducted by the U.S. Department of Energy Joint Genome Institute (https://ror.org/04xm1d337), a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy, operated under Contract No. DE-AC02-05CH11231 (T.R.N.). A.M.S. is supported by a U.S. Department of Agriculture (USDA) National Institute for Food and Agriculture (NIFA) Postdoctoral Fellowship (2023-67012-39889). Work conducted by J.P.R. and K.E.M. was supported under the auspices of the DOE, LLNL Contract DE-AC52-07NA27344, and DOE BER award SCW1632. This material by m-CAFEs Microbial Community Analysis & Functional Evaluation in Soils (https://mcafes.lbl.gov), a Science Focus Area led by Lawrence Berkeley National Laboratory, is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Biological & Environmental Research under contract number DE-AC02-05CH11231(S.M.K., D.S. and T.R.N.).

Author information

These authors contributed equally: Ling-Dong Shi, Maria Florencia Ercoli.

Authors and Affiliations

Innovative Genomics Institute, University of California, Berkeley, CA, USA
Ling-Dong Shi, Katerina Estera-Molina, Rohan Sachdeva, Jennifer Pett-Ridge, Pamela C. Ronald & Jillian F. Banfield
Department of Plant Pathology and the Genome Center, University of California, Davis, CA, USA
Maria Florencia Ercoli, Artur Teixeira de Araujo Junior, Subah Soni, Tracy Satomi Weitz, Alexandra M. Shigenaga & Pamela C. Ronald
Environmental Science, Policy and Management, University of California, Berkeley, CA, USA
Junhyeong Kim & Jillian F. Banfield
Physical and Life Sciences Directorate, Lawrence Livermore National Laboratory, Livermore, CA, USA
Katerina Estera-Molina & Jennifer Pett-Ridge
Bioinformatics Program, Faculty of Computing and Data Sciences, Boston University, Boston, MA, USA
Ilija Dukovski & Daniel Segrè
Biological Design Center, Boston University, Boston, MA, USA
Ilija Dukovski & Daniel Segrè
Center for Advanced Interdisciplinary Research, Ss. Cyril and Methodius University, Skopje, North Macedonia
Ilija Dukovski
Feedstocks Division, Joint BioEnergy Institute, Emeryville, CA, USA
Halbay Turumtay, Henrik V. Scheller & Pamela C. Ronald
Environmental Genomics and Systems Biology Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA
Halbay Turumtay, Benjamin P. Bowen, Suzanne M. Kosina, Henrik V. Scheller & Trent R. Northen
Department of Energy System Engineering, Karadeniz Technical University, Trabzon, Turkey
Halbay Turumtay
Joint Genome Institute, Lawrence Berkeley National Laboratory, Berkeley, CA, USA
Katherine B. Louie, Benjamin P. Bowen & Trent R. Northen
Department of Plant and Microbial Biology, University of California, Berkeley, Berkeley, CA, USA
Henrik V. Scheller & Jillian F. Banfield
Life and Environmental Sciences Department, University of California Merced, Merced, CA, USA
Jennifer Pett-Ridge
Department of Physics, Boston University, Boston, MA, USA
Daniel Segrè
Department of Biomedical Engineering, Boston University, Boston, MA, USA
Daniel Segrè
Department of Biology, Boston University, Boston, MA, USA
Daniel Segrè
Earth and Planetary Science, University of California, Berkeley, CA, USA
Jillian F. Banfield

Authors

Ling-Dong Shi
View author publications
Search author on:PubMed Google Scholar
Maria Florencia Ercoli
View author publications
Search author on:PubMed Google Scholar
Junhyeong Kim
View author publications
Search author on:PubMed Google Scholar
Artur Teixeira de Araujo Junior
View author publications
Search author on:PubMed Google Scholar
Katerina Estera-Molina
View author publications
Search author on:PubMed Google Scholar
Subah Soni
View author publications
Search author on:PubMed Google Scholar
Tracy Satomi Weitz
View author publications
Search author on:PubMed Google Scholar
Alexandra M. Shigenaga
View author publications
Search author on:PubMed Google Scholar
Ilija Dukovski
View author publications
Search author on:PubMed Google Scholar
Rohan Sachdeva
View author publications
Search author on:PubMed Google Scholar
Halbay Turumtay
View author publications
Search author on:PubMed Google Scholar
Katherine B. Louie
View author publications
Search author on:PubMed Google Scholar
Benjamin P. Bowen
View author publications
Search author on:PubMed Google Scholar
Suzanne M. Kosina
View author publications
Search author on:PubMed Google Scholar
Henrik V. Scheller
View author publications
Search author on:PubMed Google Scholar
Jennifer Pett-Ridge
View author publications
Search author on:PubMed Google Scholar
Daniel Segrè
View author publications
Search author on:PubMed Google Scholar
Trent R. Northen
View author publications
Search author on:PubMed Google Scholar
Pamela C. Ronald
View author publications
Search author on:PubMed Google Scholar
Jillian F. Banfield
View author publications
Search author on:PubMed Google Scholar

Contributions

The study was conceived by L.D.S., M.F.E., P.C.R., and J.F.B. M.F.E., S.S., A.T.A.J., T.S.W., and A.M.S. processed the soil from rice paddies at UC Davis, planted rice in the greenhouse, and performed phenotypic analyses. L.D.S. and J.K. measured CH₄ flux and collected rhizosphere soil samples with the help of M.F.E., A.T.A.J., A.M.S., S.S., and T.S.W. L.D.S. extracted DNA and RNA from collected soil samples and sequenced them and performed all metagenomic and metatranscriptomic analyses. M.F.E., T.S.W. and A.T.A.J. collected rice root tissue and exudate samples for metabolomic analysis. T.R.N., K.B.L. and B.P.B. designed the metabolomics study, generated and processed the metabolomic data, and M.F.E. analyzed the metabolomic data. S.M.K. prepared the NPOC measurements. H.T. performed the lignin analysis in young and adult rice plants. H.V.S. supervised cell wall analysis. L.D.S. constructed microbial metabolic models and performed dynamic flux balance analysis with the help of I.D. and D.S. R.S. provided technical support in processing sequencing data. L.D.S., M.F.E., and K.E.M. performed soil incubation experiments with the help of A.T.A.J. and supervised by J.P.R. L.D.S., M.F.E., P.C.R. and J.F.B. wrote the manuscript with input from all the authors.

Corresponding authors

Correspondence to Pamela C. Ronald or Jillian F. Banfield.

Ethics declarations

Competing interests

J.F.B. is a co-founder of Metagenomi. T.R.N. is an advisor to Brightseed Bio. The remaining authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Pil Joo Kim, and Jens Sundström, who co-reviewed with Yunkai Jin, for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information (download PDF )

Description of Additional Supplementary Information (download PDF )

Supplementary Data 1 (download XLSX )

Supplementary Data 2 (download XLSX )

Supplementary Data 3 (download XLSX )

Supplementary Data 4 (download XLSX )

Supplementary Data 5 (download XLSX )

Supplementary Data 6 (download XLSX )

Supplementary Data 7 (download XLSX )

Supplementary Data 8 (download XLSX )

Supplementary Data 9 (download XLSX )

Supplementary Data 10 (download XLSX )

Reporting Summary (download PDF )

Transparent Peer Review file (download PDF )

Source data

Source Data (download XLSX )

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Shi, LD., Ercoli, M.F., Kim, J. et al. Reduced methane emissions in transgenic rice genotypes are associated with altered rhizosphere microbial hydrogen cycling. Nat Commun 17, 2028 (2026). https://doi.org/10.1038/s41467-026-68640-9

Download citation

Received: 27 November 2024
Accepted: 13 January 2026
Published: 26 January 2026
Version of record: 26 February 2026
DOI: https://doi.org/10.1038/s41467-026-68640-9