Unelongated Stems are an Active Nitrogen-Fixing Site in Rice Stems Supported by Both Sugar and Methane Under Low Nitrogen Conditions

Abstract

Enhancing nitrogen (N) fixation in rice plants can reduce N fertilizer application and contribute to sustainable rice production, particularly under low-N conditions. However, detailed microbial and metabolic characterization of N fixation in rice stems, unlike in the well-studied roots, has not been investigated. Therefore, the aim of this study was to determine the active N-fixing sites, their diazotroph communities, and the usability of possible carbon sources in stems compared with roots. The N-fixing activity and copy number of the nitrogenase gene in the rice stem were high in the outer part of the unelongated stem (basal node), especially in the epidermis. N fixation, estimated using the acetylene reduction assay, was also higher in the leaf sheath and root than in the inner part of the unelongated stem and culm. Amplicon sequence variants (ASVs) close to sugar-utilizing heterotrophic diazotrophs belonging to Betaproteobacteria and type II methanotrophic diazotrophs belonging to Alphaproteobacteria were abundant in the outer part of the unelongated stems. Media containing crushed unelongated stems exhibited N-fixing activity when sucrose, glucose, and methane were added as the sole carbon sources. This suggested that N fixation in the unelongated stems was at least partly supported by sugars (sucrose and glucose) and methane as carbon sources. ASVs close to sugar-utilizing heterotrophs belonging to Actinobacteria were also highly abundant in the unelongated stem; however, their functions need to be further elucidated. The present finding that diazotrophs in rice stems can use sugars such as sucrose and glucose synthesized by rice plants provides new insights into enhancing N fixation in rice stems.

Background

Rice is a major crop that plays a critical role in feeding the growing global population growth. One key factor affecting rice productivity is the availability of nitrogen (N), an essential element for plant growth. Chemical N fertilizers are produced by fixing atmospheric N₂ through the Haber–Bosch process, supporting global crop production (Erisman et al. 2008). As the concept of planetary boundaries indicates that artificially fixed N disrupts the global N cycle (Rockström et al. 2009; Richardson et al. 2023), reducing the amount of N fertilizer is necessary for sustainable rice production.

One solution is to enhance the biological N fixation in rice cultivation systems. During the N-fixing process, diverse bacteria and archaea containing nitrogenase enzymes (diazotrophs) reduce molecular N to ammonia. Biological N fixation has been extensively studied, and an increase in N fixation in arable land contributes to the sustainability of agricultural systems, especially in marginal lands under low N conditions (Vance and Graham 1995; Ladha et al. 2022). Legume plants are well known for their symbiosis with rhizobia, with the host plant supplying a carbon source to the rhizobia and the rhizobia supplying the fixed N to the host plant (Udvardi and Poole 2013). In addition to legume plants, the amount of N fixation in non-legume croplands is comparable to symbiotic legume N fixation in agricultural land worldwide (Herridge et al. 2008; Ladha et al. 2016). N fixation in rice cultivation systems has also been suggested to contribute greatly to maintaining the productivity of paddy fields. N fixation per crop season ranges from approximately 3 to 63 N kg ha^− 1 in rice cultivation systems, including rice plants, paddy soil, and irrigation water, as estimated using the δ¹⁵N-dilution or acetylene reduction assay (Yoshida and Ancajas 1973). Since N fixation in rice is known to be higher without the application of N fertilizer (Shrestha and Maskey 1970; Siddikee et al. 2016), a comprehensive understanding of N fixation under low-N conditions is crucial for improving the productivity and sustainability of rice production in the future.

Active N fixation has been found in the lower portions of the shoots and roots in rice plants, and the amount of fixed N was estimated to be equal in the shoots and roots in Asian countries, suggesting that both shoots and roots are essential for N fixation in rice (Watanabe and Barraquio 1979; Ito et al. 1980; Yoshida and Yoneyama 1980; Eskew et al. 1981; Ma et al. 2019). The lower part of the leaf sheath and the unelongated stem (basal node) in rice shoots displayed active N fixation (Ito et al. 1980; Eskew et al. 1981). Several diazotrophs have been isolated from the stem and root of rice plant (Ladha et al. 1983; Elbeltagy et al. 2000; Ji et al. 2014). However, it has been assumed that the active N-fixing community associated with rice is mainly composed of uncultivated organisms that are specific to the rice ecosystem and have not yet been characterized (Elbeltagy and Ando 2008). Recent advances in sequencing and metaproteomic technology have revealed that N-fixing type II methanotrophs, such as Methylosinus sp., are abundant under low-N conditions (Bao et al. 2014b; Ikeda et al. 2014). Bradyrhizobium sp. and Paraburkholderia sp. were shown to be the dominant diazotrophs in a wide range of rice stems (10 cm from unelongated stem) using amplicon analysis (Okamoto et al. 2021).

As N fixation is a highly energy-demanding process, diazotrophs require abundant carbon sources (Olivares et al. 2013). Photosynthetic products synthesized in rice or methane and carbon dioxide gases generated in soil or rice plants are possible carbon sources for diazotrophs in rice. N fixation by type II methanotrophs in rice roots uses methane as the carbon source (Shinoda et al. 2019; Hara et al. 2022). Type II methanotrophs belong to the class Alphaproteobacteria and use the serine cycle for carbon assimilation (Knief 2015). Simultaneously, methane-independent N fixation occurred in the roots, as indicated by the use of the methane monooxygenase inhibitor difluoromethane (DFM) (Hara et al. 2022). Our previous study suggested that the levels of soluble sugars such as sucrose and glucose influence stem N fixation in rice stems (Okamoto et al. 2021). Starch, sucrose, glucose, and fructose accumulate in the rice stems as nonstructural carbohydrates (Arai-Sanoh et al. 2011). However, the sugars produced in the rice plants have not been investigated for utilization by diazotrophs. Organic acids and amino acids may also be possible carbon sources, as it has been reported that organic acids and amino acids could be used in diazotrophs isolated from rice (Tripathi et al. 2006; Xie and Yokota 2006). Therefore, it seems possible to elucidate the N fixation mechanism in rice stems by identifying the active N fixation sites in rice stems in more detail and by investigating diazotrophs and metabolites present at these sites and examining their carbon availability.

In this study, we first aimed to identify active N-fixation sites in rice stems by segmenting the stems and measuring the N-fixation activity or the amount of fixed ¹⁵N₂-labeled gas at each stem site in comparison with the roots. We also aimed to identify diazotroph community structures and compositions at active N-fixing sites using amplicon sequencing analysis and candidate carbon sources for diazotrophs at active N-fixing sites using metabolome analysis. Furthermore, we aimed to confirm the usability of the possible carbon sources by incubating them in media containing a single carbon source with high N-fixing rice tissue.

Materials and Methods

Field Cultivation

Field experiments were conducted in 2020, 2021, and 2022 using an experimental paddy field at the Togo Field Center for Research and Education of Nagoya University in Aichi, Japan (35°06’36"N, 137°05’00"E). The field received no N, phosphorus (P), or potassium (K) from 2014 to 2019, and only P and K were applied from 2020 to 2022. Rice (Oryza sativa L.) cv. Nipponbare was used. Sowing and transplanting were conducted on April 20 and May 13, 2020, April 20 and May 11, 2021, and April 18 and May 11, 2022. Seedlings were transplanted at a density of 22.2 hills m^− 2 (15 × 30 cm), with one plant per hill. Each rice plant was allocated a plot 1.5 m wide by 1.5 m long, with two replicate plots. Chemical fertilizers were applied as basal dressings at a rate of 9 g P₂O₅ m^− 2 and 9 g K₂O m^− 2 as basal dressing. The rice plants were grown under conventional culture practices, and the soil was flooded to a depth of approximately 5 cm from transplantation until harvest.

Plant Samplings

Sampling was conducted at the tillering (50 days after transplanting [DAT]), heading (90 DAT), and maturing (127 DAT) stages in 2020 to clarify the stem part with high acetylene reduction activity (ARA) through different growth stages. Two plants from each replicate plot were sampled, and the soil adhering to the stems of the rice plants was carefully washed with tap water to eliminate algal activity (Watanabe et al. 1978). The stems of each plant were separated by hand, and the two thickest stems were selected and cut at fixed-length intervals for ARA measurements. The data below 35 cm (plant height) were recorded during the tillering stage.

Sampling to identify detailed sites with high ARA in the unelongated stem was performed at 148 DAT in 2022. Two thick stems were selected from the rice plants sampled in 2020. One of the stems was laced in a 50 mL tube and surface-sterilized with 25 mL of 70% ethanol for 1 min and 25 mL of 0.3% sodium hypochlorite for 2 min, followed by careful rinsing with distilled water. Both stems were then divided with a razor into the outer and inner parts of the unelongated stem, with the root 2 cm from the stem base, as shown in Fig. 1. The outer part of the unelongated stem was further divided into the “epidermis part” and “outer part without epidermis” by hand cutting. The “epidermis part” is a thinly shaved outer part of the unelongated stem that includes the epidermis and some inner tissue inside the epidermis and crown root primordia.

Fig. 1

Photos and illustration of the base of the rice plant sampled at the heading season. (A) Photos from outside and (B) inside, and (C) an illustration of inside. The colors in the illustration indicate the roots, outer part of the unelongated stems, inner part of the unelongated stems, leaf sheaths, and culms

Sampling for the evaluation of ¹⁵N₂ fixation, N content, sugar components, and DNA extraction was conducted at 87 DAT in 2021. Two plants from each replicated plot were sampled, and the main stem was sampled and cut into 2 cm lengths from the stem base, including 2 cm of root for ¹⁵N₂ feeding (Fig. 1A, B). Another cut stem from each plant was snap-frozen in liquid N and stored at − 80 °C until freeze-drying. After freeze-drying, the samples were cut into five sections, as shown in Fig. 1C. The samples were subjected to metabolomic analysis and DNA extraction.

Sampling was performed at 112 DAT in 2021 to assess the carbon availability of diazotrophs inhabiting unelongated stems. Two plants were sampled for each replicate. After washing the soil with tap water, only the unelongated stems were cut. Equally sized unelongated stems from each plant were selected to evaluate carbon availability.

ARA Measurement as N-Fixing Activity

ARA was performed as described by (Burris 1972). The plant parts were enclosed in a 160 mL glass test tube with a silicone cap or a 20 mL vial (SVG-20; NICHIDEN-RIKA GLASS Co. LTD, Hyogo, Japan) with a butyl rubber stopper. The 10% (v/v) gas phase in the test tube or vial was then replaced with acetylene gas, and the plants were incubated for 24 h at 25 °C in the dark. The ethylene concentration was determined using a gas chromatograph (GC-4000; GL Sciences Inc., Tokyo, Japan) equipped with a Porapak N 80–100 mesh and an FID connected to a Chromato Logger (LC Science Corporation, Nara, Japan). The plant samples were dried at 80 °C for 48 h after ARA measurement, and the dry weight was measured.

¹⁵N₂ Feeding

Four stem bases comprising 2 cm of the stem and 2 cm of the roots were individually placed in 20 mL vials (SVG-20; NICHIDEN-RIKA GLASS Co. LTD) and sealed with a butyl rubber stopper. All air in the vials was replaced with helium gas. Subsequently, 40% of the gas phase (v/v) was replaced with ¹⁵N₂ gas (99.6 atom %), and 20% was replaced with O₂. The vials were then incubated in the dark at 25 °C. After 24 h, the plants were removed and snap-frozen in liquid N and stored at − 80 °C. The samples were freeze-dried and cut into 5 sections (Fig. 1C). The samples were ground and weighed using tin capsules. The ¹⁵N:¹⁴N ratio and total N concentration were determined using a Flash2000-DELTAplus Advantage ConFloIII System (Thermo Fisher Scientific, Waltham, MA, USA).

DNA Extraction

DNA was extracted using DNAs-ici!-W (RIZO Inc., Ibaraki, Japan). The ground samples were incubated in DNAs-ici!-W buffer at 65 °C for 30 min, followed by incubation with RNase at 37 °C for 30 min. DNA was extracted using one volume of chloroform, precipitated with 2-propanol, washed with 70% ethanol, and dissolved in DNase-free water.

Real-Time PCR and Digital PCR

Real-time PCR was performed as described by Liu et al. (2016) and Masuda et al. (2021), with slight modifications to target nifH and nifD in general diazotrophs (universal nifH and nifD), nifD in Anaeromyxobacter and Geobacter (A&G nifD), and mcrA, using universal and specific primer sets (Table S1). The primer set of A&G nifD was designed to target the nitrogenase gene of iron-reducing bacteria, which was recently proposed to be responsible for N fixation in rice paddy soils (Masuda et al. 2017, 2023). The primer set of the mcrA gene was designed to target the methyl coenzyme M reductase gene from five different methanogenic species (Luton et al. 2002). A StepOnePlus System (Life Technologies, Carlsbad, CA, USA) or Thermal Cycler Dice Real-Time System (Takara Bio Inc., Shiga, Japan) with TB Green Premix Ex Taq II (Takara Bio Inc.) was used. The PCR cycling parameters are listed in Table S1.

For digital PCR, the method described by Shinjo et al. (2023) was followed to quantify the copy number of pmoA gene (encoding the β-subunit of pMMO [particulate methane monooxygenase]) of type Ia, Ib, and IIa methanotroph using the QuantStudio 3D Digital PCR System (Thermo Fisher Scientific).

Amplicon Sequencing of the nifH and 16S rRNA Genes

The first PCR reaction of the nifH gene for amplicon sequencing was conducted using the PolF/PolR primers (Poly et al. 2001); the forward primer (5′-TGCGAYCCSAARGCBGACTC-3′) and the reverse primer (5′-ATSGCCATCATYTCRCCGGA-3′). The first PCR of the V5–V7 region of bacterial 16S rRNA for amplicon sequencing was conducted using the following blocking primers: the forward primer 799 F (5′-AACMGGATTAGATACCCKG-3′) and the reverse primer 1193R (5′- ACGTCATCCCCACCTTCC-3′) (Chelius and Triplett 2001). The 799 F/1193R primer set avoids amplification of chloroplast DNA sequences (Beckers et al. 2016; Wang et al. 2018).

The first PCR mixture of the nifH gene consisted of 5.0 µL of 10x reaction buffer, 4.0 µL of 2.5 mM deoxyribonucleotide triphosphates (dNTPs), 1.0 µL of 10 µM primers PolF/PolR (each), 5.0 µL template DNA (1.0 ng µL^− 1), 0.5 µL Ex Taq Hot Start Version (HS) enzyme (5 U µL^− 1) (Takara Bio Inc.), and 33.5 µL milli-Q water. The first PCR mixture of 16S rRNA gene consisted of 3.0 µL of 10x reaction buffer, 2.4 µL of 2.5 mM dNTPs, 1.5 µL of 10 µM primers 799 F/1193R (each), 3.0 µL of template DNA (1.0 ng µL^− 1), 0.3 µL of Ex Taq HS enzyme (5 U µL^− 1) (Takara Bio Inc.) and 18.3 µL milli-Q water. The thermal conditions for the first PCR for both primers were as follows: 94 °C for 2 min; 30 cycles of 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min; and 72 °C for 5 min. After the 1st PCR, only the PCR product of the nifH gene was electrophoresed, and the band was excised, purified using NucleoSpin Gel and PCR Clean-up (Takara Bio Inc.), and eluted with 25 µL of TE buffer.

Amplicon sequencing of the nifH and 16S rRNA genes was performed by the Bioengineering Lab. Co. (Kanagawa, Japan). Amplicons were purified using AMPure XP magnetic beads (Beckman Coulter, Brea, CA, USA). The second round of PCR was performed according to the manufacturer’s instructions. Sequencing was conducted on an Illumina MiSeq system using MiSeq Reagent Kit v3 (Illumina Inc., San Diego, CA, USA). MiSeq was performed in a 2 × 300 cycle configuration. Reads with matching primer sequences were extracted using the FASTX Toolkit (ver. 0.0.14). The raw tags were filtered to obtain high-quality clean tags using Sickle (ver. 1.33) (Joshi and Fass 2011). The paired-end read-binding script FLASH (ver. 1.2.11) (Magoč and Salzberg 2011) was used to bind the reads with a minimum overlap of 10 bases. After removing chimeric and noise sequences, representative sequences (amplicon sequence variants [ASVs]) were generated using the amplicon sequence variant method with the dada2 plugin of QIIME 2 (ver. 2021.11) (Bolyen et al. 2019). The ASVs of 16S rRNA were estimated using the feature-classifier plugin for QIIME 2 against the EzBioCloud 16S database to verify the taxonomic affiliation (Yoon et al. 2017). For the nifH gene amplicon data, ASV sequences were translated into amino acid sequences, and sequences without Cys 97 and Cys 132 amino acids (protein numbering for the NifH protein in Azotobacter vinelandii; and 4Fe-4 S iron-sulfur cluster ligating cysteines) (Howard et al. 1989) were removed. Amino acid sequences were estimated using BLASTp searches against the non-redundant protein sequence (nr) database, excluding uncultured bacteria, to verify the taxonomic affiliation of the ASVs of NifH. The ASV was classified as an uncultured bacterium if the closest sequence identity was less than 90%.

Bacterial community similarity structure analysis was performed using non-metric multidimensional scaling (NMDS) with the Bray–Curtis similarity algorithm. NMDS was calculated from the weighted UniFrac base matrix using the R package vegan 2.5-7 (Oksanen et al. 2020) as described by Sasaki et al. (2013). The Shannon index was calculated using the vegan package in R.

The phylogenetic tree was constructed with 14 ASVs that presented ≥ 1.5% relative abundance in at least one part of the NifH sequence data to examine the taxonomy of predominating diazotrophs. Reference sequences close to the ASVs were collected using BLASTp analysis and the ARB program (Ludwig et al. 2004) with nifH_database_2012 (Gaby and Buckley 2014). The 14 ASVs and reference sequences were aligned using the CLUSTAL W program (Larkin et al. 2007), with the default settings of MEGAX (Kumar et al. 2018). Phylogenetic trees of NifH were constructed using the maximum likelihood method with the LG model (Le and Gascuel 2008). The robustness of the tree topologies was tested using 1,000 bootstrap replicates. Roseiflexus castenholzii DSM 13,941 (GCA 000017805.1) was used as the outgroup. The ASVs were classified into NifH clusters such as Clusters I and III, molybdenum nitrogenase (Mo-nitrogenase, Nif), and Cluster II, which is an alternative nitrogenase (Fe- or V-only nitrogenase, Anf or Vnf), according to the definition of Zehr et al. (2003). For reference strains used in the phylogenetic trees of NifH, the N-fixing ability and carbon (sucrose, glucose, fructose, and methane) assimilation ability were estimated at the species or genus level with available information. N-fixing ability was evaluated as the ability to fix N, show ARA, or have the key gene for N fixation (nifHDKENB) (Dos Santos et al. 2012).

Metabolomic Analysis

For neutral sugar analysis by using LC-MS, the ground samples were extracted by adding 90% MeOH containing 100 µM ribitol as a reference compound in 10 times volume relative to dry weight and vortexed for 3 min. After centrifugation at 20,400 × g for 3 min, the supernatant was filtered through Millex LG PTFE 0.20 μm (SLLG025SS; Merck Millipore, Darmstadt, Germany) and used as the analysis sample. Samples were then analyzed using an Agilent 1290 Infinity analytical HPLC instrument (Agilent Technologies, Santa Clara, CA, USA) with linearly-connected Unison UK-Amino 3 μm 2.0 × 250 mm and 2.0 × 100 mm columns (Imtack, Kyoto, Japan), coupled to an Agilent 6120 Single Quadrupole mass detector (Agilent Technologies). The analytical conditions were as previously described by Kawade et al. (2024). The neutral sugar concentrations were calculated as the sum of the sugars detected using LC-MS.

For ionic metabolite analysis by using CE-MS, 10 mg freeze-dried samples were added to 500 µL methanol containing 8 µM of two reference compounds (methionine sulfone and camphor 10-sulfonic acid for cation and anion analyses, respectively) and extracted using Mixer Mill MM310 (Verder Scientific, Haan, Germany) at a frequency of 20 Hz for 1 min with φ5 mm zirconia beads. The extracts were centrifuged at 20,400 × g for 3 min at 4 °C. The supernatant (500 µL) was transferred into a new tube and subjected to CE-MS analysis, as described previously by Oikawa et al. (2011).

Incubation of Crushed Unelongated Stem with Candidate Carbon Sources

As candidate carbon sources for diazotrophs in the unelongated stem, sucrose, glucose, fructose, and methane were selected. Sucrose, glucose, and fructose were selected because of their high concentrations among the sugars in the metabolomic analysis. Methane was selected because methane is a carbon source for methanotrophs, which were abundant in the amplicon analysis. The LGI-P medium (Reis et al. 1994) was prepared without crystallized cane sugar, and sucrose, glucose, and fructose were added to the medium at a carbon concentration of 0.5%. No carbon source was added to the medium for the control or methane treatments. After each solution was adjusted to pH 5.5, 0.3% gellan gum was added and heated with stirring to dissolve in the solution and produce a semi-solid medium. Gellan gum is more suitable than agar for measuring ARA from diazotrophs in soil (Hara et al. 2009). Then, 5 mL of the medium was dispensed into 20 mL vials, stoppered, sealed with butyl rubber stoppers, and autoclaved (121 °C for 20 min).

The sampled unelongated stems were surface-sterilized with 70% ethanol for 1 min and 0.3% sodium hypochlorite for 2 min and then rinsed carefully in distilled water. The remaining roots on the sterilized unelongated stems were cut using a razor. The unelongated stems were then added to a mortar containing an equal weight of distilled water and ground manually. Then, 100 µL of milled solution was added to the medium, and the vials were tightly closed again with rubber stoppers. In the methane-supplemented treatments, 90% (v/v) of the gas in the vials was replaced with methane. Vials were incubated in an incubator for 3 days at 25 °C in the dark. Then, 10% (v/v) of the gas phase was replaced with acetylene, and the vials were incubated in an incubator for 1 day at 25 °C in the dark. After incubation, gases were measured using a gas chromatograph (GC-4000; GL Sciences Inc.) with an FID detector equipped with a methanizer (MT-221; GL Sciences Inc.) to determine the average ethylene production rate for one day and the total carbon dioxide production for four days.

Statistical Analyses

The Tukey–Kramer HSD test was used to examine the differences between the rice parts and carbon sources. Student’s t-test was used to assess the effect of sterilization. One-way analysis of variance (ANOVA) was used for metabolomic analysis. All statistical analyses were conducted using JMP Pro v.15.1.0 (SAS Institute Inc., Cary, NC, USA).

Results

N-fixing Activity, Fixed ¹⁵N, Total N Concentrations in Different Parts of the Rice Plant

Different sections were sampled and used for measurements to identify the parts of rice plants with high ARA (Fig. 1). The unelongated stems had the highest ARA among the rice stems at different heights throughout the tillering, heading, maturing stages (Fig. 2A). Stems (0–5 cm) showed the second-highest ARA at the heading stage (Fig. 2A). Among the detailed parts of the unelongated stems, ARA was the highest in the epidermis, followed by the outer and inner parts (Fig. 2B). The epidermis of the unelongated stem had approximately half of the ARA after sterilization, but the other parts were not affected by sterilization (Fig. 2B). Regardless of sterilization, ARA showed little value in the roots, as well as in the inner part of the unelongated stem (Fig. 2B).

Fig. 2

Evaluation of nitrogen-fixing ability at detailed parts of rice plant. (A) Changes in acetylene reduction activity (ARA) of different parts of rice stems throughout 3 growth stages. (B) ARA in 4 parts of the base of rice plant with/without sterilization. (C) Fixed ¹⁵N and (D) total nitrogen concentration in five parts of rice. Data are presented as mean ± SE (n = 3–4). Different lowercase letters indicate significant differences based on Tukey’s HSD test (P < 0.05) for each stage. The ** indicates significance at P = 0.01 level assessed with/without sterilization in each part using Student’s t-test

Exposure of the stem base to ¹⁵N₂ resulted in the highest N fixation in the root, followed by the outer part of the unelongated stem, leaf sheath, inner part of the unelongated stem, and culm (Fig. 2C). Total N concentrations were the highest in the leaf sheath and lowest in the root; no correlation was found between fixed ¹⁵N and total N concentrations (Fig. 2C, D).

Bacterial Gene Quantification and Community Structure

Quantification of nitrogenase genes showed that the copy number of the universal nifH gene was the highest in the root and lowest in the inner part of the unelongated stem and culm (Fig. 3A). The copy number of universal nifD was not detected in the inner part of the unelongated stem and culm, and the number in other parts was approximately half that of the universal nifH (Fig. 3A, B). The copy number of the A&G nifD gene was lower than that of the universal nifH and universal nifD genes (Fig. 3A–C). Quantification of genes related to methane production or oxidation showed that the copy numbers of mcrA and pmoA were highest in the roots (Fig. 3D, E). Among the pmoA genes, type IIa had the highest (Fig. 3E).

Fig. 3

Quantification of genes involved in nitrogen fixation and methane production and oxidation. (A) Universal nifH gene, (B) universal nifD gene, (C) A&G (Anaeromyxobacter and Geobacter) nifD gene, (D) mcrA gene, and (E) pmoA gene. Data are presented as mean ± SE (n = 3–4). Different lowercase letters above the boxplots indicate significant differences based on Tukey’s HSD test (P < 0.05). “n.d.” and “n.s.” indicates not detected and not significant, respectively

By amplifying the nifH gene, 21,654–46,075 valid reads and 172–1014 ASVs were obtained from the 20 samples. By amplifying the 16S rRNA gene, 1,007–26,907 valid reads and 82–476 ASVs were obtained from 20 samples. The raw reads, valid reads, ASVs, and Shannon index data are presented in Table S2. The ratio of non-valid reads to raw reads was high in 16S rRNA, especially in the inner part of the unelongated stem and culm, and these non-valid sequences were identified as rice mitochondrial gene sequences (Table S2). The Shannon index was highest in the roots and lowest in the inner part of the unelongated stem for both nifH and 16S rRNA genes (Table S2).

The NMDS plots for the nifH and 16S rRNA genes were mostly located close to each other within the same rice parts based on the ASV composition (Fig. S1A).

Based on the sequence analysis, Proteobacteria was the most dominant bacterial phylum in the nifH gene (60.1–69.3% for the total reads) and 16S rRNA (43.7–67.1% for the total reads) (Fig. S1B). In the phylum Proteobacteria, members of the class Alphaproteobacteria and Betaproteobacteria were approximately equally abundant in nifH and 16S rRNA (Fig. S1C). The second most abundant phylum after Proteobacteria was Actinobacteria in all rice parts for both nifH and 16S rRNA (Fig. S1B). In particular, the unelongated stems had a greater abundance of Actinobacteria than other parts in both nifH and 16S rRNA (Fig. S1B).

Community Analysis of Potential Diazotroph and Their Carbon Assimilation Capacity

The ASVs ≥ 1% in the relative abundance of at least one sample of nifH gene and 16S rRNA are listed in Table S3 and S4, respectively. The ASVs with ≥ 1.5% relative abundance (NifH-ASVs) used in the phylogenetic tree accounted for 16.1% in roots, 25.4% in the outer part of unelongated stems, 27.8% in the inner part of unelongated stems, 22.2% in leaf sheaths, and 13.2% in culms. The NifH-ASVs were classified into two NifH clusters (clusters I and II) (Fig. 4).

Fig. 4

Maximum likelihood phylogenetic tree of NifH amplicon sequence variants (ASVs) obtained from rice plant. Bootstrap values of ≥ 50% are indicated at their respective nodes. The red bar graphs on the right hand of the ASV-NifH name indicate the mean relative abundance of the ASVs in rice parts (n = 4). The black-and-white shading in the columns on the right hand of the the reference strains are colored according to nitrogen-fixing ability and carbon (sucrose, glucose, fructose, and methane) assimilation ability of those bacteria based on the data from (1), Helene et al. (2019); (2), Martínez-Pérez et al. (2018); (3), Satari et al. (2022); (4), Wasai-Hara et al. (2020); (5), Helene et al. (2020); (6), Hungria et al. (1998); (7), Rivas et al. (2004); (8), Murrell and Dalton (1983); (9), Chu and Alvarez-Cohen (1996); (10), Chen et al. (2006); (11), Oliveira-Filho et al. (2021); (12), Coenye et al. (2001); (13), Reinhold-Hurek and Hurek (2000); (14), Chen et al. (2013); (15), Maszenan et al. (2002); (16), Weon et al. (2008); (17), Hwang et al. (2018); (18), Bae et al. (2006); (19), Akita et al. (2024)

In the outer part of the unelongated stem, where N fixation was active, ASV-NifH-I 3–8 and 10–11 were present in high relative abundance in NifH (Fig. 4). NifH-I 3 and 4 were 99.1–100% identical to that of Methylosinus trichosporium SM6, known for N fixation and methane assimilation (Chu and Alvarez-Cohen 1996). It was high in the root and outer part of the unelongated stem but low in the leaf sheath (Fig. 4, Table S3). ASV-NifH-I 5–8, classified as Betaproteobacteria, are known for their N fixation and sugar assimilation abilities (Reinhold-Hurek and Hurek 2000; Chen et al. 2006, 2013; Oliveira-Filho et al. 2021), and were abundant in the leaf sheath (Fig. 4). ASV-NifH-I5 and 6 were 97.2% identical to that of Paraburkholderia mimosarum STM 3726 (Table S3). ASV-NifH-I7 and 8 were 100% identical to that of Azonexus fungiphilus and Azoarcus olearius, respectively (Table S3). ASV-NifH-I 10 and 11 were 100% identical to Uliginosibacterium gangwonense, which is known for sugar utilization but not for N fixation (Weon et al. 2008; Hwang et al. 2018) (Fig. 4, Table S3). The relative abundances of ASVs classified as Alphaproteobacteria, such as ASV-NifH-I 1, which showed 96% identity with Sagittula sp. and ASV-NifH-I 2, which showed 100% identity with Bradyrhizobium cosmicum, were high in the leaf sheath (Fig. 4, Table S3). Regarding iron-reducing bacteria, only two ASVs among the ASVs whose relative abundance was greater than 1% in at least one part of nifH were close to that of Geobacter sp., and their total relative abundance was highest in the outer part of the unelongated stem (2.5%) and lowest in the culm (0.4%) (Table S3).

For the AnfH sequences (Nif Cluster II), ASV-NifH-II 1–3 were 93.4% identical to Brooklawnia cerclae, all classified as Actinobacteria and had a high relative abundance in all parts, especially in the unelongated stems of nifH (Fig. 4).

Metabolite Concentrations and Their Availability for N Fixation in the Unelongated Stem

Neutral sugar concentrations were the highest in the inner part of the unelongated stem, followed by the outer part of the unelongated stems and roots (Fig. 5A). Among the sugars, the sucrose concentration was the highest (82.8–86.7%) (Fig. 5A). Glucose (4.9–6.9%) and fructose (5.3–8.9%) concentrations were comparable, and other sugar concentrations were low (Fig. 5A). Data for the anions and cations are listed in Table S5. For anions, malate was higher in roots than in unelongated stems, whereas, for cations, lysine and threonine were lower in the roots than in unelongated stems (Table S5).

Fig. 5

Accumulated carbon sources in unelongated stem and their availability. (A) The mean amount of neutral sugar concentration in three parts of rice. (B) Acetylene reduction activity (ARA) and (C) CO₂ production of sterilized crushed unelongated stems cultivated in semi-solid medium with different carbon sources. Distilled water was added to control. Data are presented as mean ± SE (n = 3). Different lowercase letters above bars indicate significant differences based on Tukey’s HSD test (P < 0.05)

The medium containing sucrose as the sole carbon source incubated with crushed unelongated stems showed the highest ARA (Fig. 5B). Medium with glucose increased ARA, but fructose had almost no effect on ARA (Fig. 5B). ARA also increased in the medium with methane, a candidate carbon source other than metabolites (Fig. 5B). Total CO₂ production from vials containing sucrose was the highest, followed by those containing glucose (Fig. 5C). The addition of fructose and methane produced the same amount of CO₂ as the control (Fig. 5C).

Discussion

N fixation in the rice stem grown under low-N conditions was active in the outer part of the unelongated stem, especially in the epidermis, associated with sugar-utilizing heterotrophs (Class Betaproteobacteria) and methanotrophs (Figs. 2, 3 and 4). Incubation experiments with sole carbon sources showed that N fixation in unelongated stems was supported by sugars (sucrose and glucose) and methane (Fig. 5). Active N fixation was also observed in the leaf sheaths and roots (Fig. 2), consistent with Ito et al. (1980) and Eskew et al. (1981). Methanotrophs were abundant in the roots; however, sugar-utilizing heterotrophs were more abundant than methanotrophs in the leaf sheaths (Figs. 3 and 4). A hypothetical model of N fixation in the basal part of rice is shown in Fig. 6.

Fig. 6

Hypothetical model of nitrogen (N) fixation in the basal part of rice. N fixation occurs in roots and stems. In the stem, both the outer part of the unelongated stem and the leaf sheath are assumed to have higher N fixation. In the unelongated stem, N fixation would be higher closer to the epidermis. Methanotrophs in root and sugar-utilizing heterotrophs in leaf sheath would be responsible for N fixation, respectively. Both sugar-utilizing heterotrophs and methanotrophs are probably responsible for N fixation in the outer part of the unelongated stem

Detailed measurements of the N-fixing ability of rice stems revealed the importance of the outer part of the unelongated stem (Fig. 6). Phylogenetic analysis and NMDS plots revealed that the composition and structure of diazotrophs differed among the outer parts of the unelongated stem, leaf sheath, and root (Fig. 4, S1A). Such differences in N-fixing ability and diazotroph structure among the different parts observed in this study may be related to the accessibility of diazotrophs to the rice tissue. Since sterilization of the epidermis of the unelongated stem greatly reduced ARA in the outer part of the stem (Fig. 2B), N fixation in the outer part would be particularly active around the surface. Previous studies have shown that bacteria colonize only the surface of the roots in most instances, and N-fixing endophytes in rice enter from the base of the emerging lateral roots between epidermal cells (Cocking 2003). Hence, diazotrophs in unelongated stems and roots would have inhabited the surface, epidermis, and cracks of emerging crown roots. At the same time, the inner parts of the unelongated stem fixed ¹⁵N and showed ARA even after surface sterilization, indicating that the endophytes living inside the unelongated stem also fixed N. The inner part of the unelongated stem had a higher abundance of Methylosinus sp. than the outer part. Bao et al. (2014a) reported that Methylosinus sp. colonized the area around the epidermal cells and vascular cylinder in wild rice roots, suggesting that Methylosinus sp. may have the ability to enter the rice tissue. In addition, Rhizobium sp. and Azoarcus olearius, detected in this study (Fig. 4, Table S4, S5), were reported to densely colonize the epidermis and spaces close to emerging lateral roots, indicating crack entry (Mitra et al. 2016; Faoro et al. 2017). This accessibility to bacterial invasion probably contributed to the structure of diazotrophs at each site. Nif-ASVs with abundance ≥ 1% in this study were mostly classified in Nif Cluster I, with no ASVs classified in Cluster III (Fig. 4, Table S4). Cluster I contains primarily aerobic and facultatively anaerobic bacteria, and Cluster III contains exclusively obligate anaerobes (Gaby and Buckley 2014). The rice plants were grown under waterlogged conditions, which are known to enable oxygen to reach the roots owing to the presence of a radial oxygen loss barrier (Nishiuchi et al. 2012), which may have enabled aerobic and facultative anaerobic bacteria to live in the plants. Meanwhile, ASV-NifH-I1 (96% identical to that of Sagittula sp. P11) were more abundant in the leaf sheaths than in the unelongated stems and roots (Fig. 4). Because Sagittula sp. is strictly aerobic (Gonzalez et al. 1997), it is likely that the unelongated stems and roots were not supplied with enough oxygen to support strictly aerobic bacteria, and the oxygen concentration in plant tissues can also affect the diazotroph structure.

This study revealed that sugars such as sucrose and glucose are the potentially important carbon sources of N fixation in unelongated stems (Fig. 6) because most ASVs have been reported to be able to utilize sugars, such as P. mimosarum (Fig. 4) and that incubation of crushed unelongated stems has shown ARA from a medium containing sucrose or glucose (Fig. 5B). The unelongated stem and leaf sheath, but not the root, showed higher ARA, even in the presence of acetylene (Fig. 2A, B). This also supported the presence of heterotrophic diazotrophs other than methanotrophs, as acetylene is a strong competitive inhibitor of methane oxidation for methanotroph (Dalton and Whittenbury 1976). This is consistent with the results of our previous study, which showed that sucrose and glucose concentrations influenced ARA in rice stems (Okamoto et al. 2021). Neutral sugars in the outer part of the unelongated stem were highest in sucrose, followed by glucose and fructose, with sucrose accounting for as much as 85% (Fig. 5A), which is consistent with the composition of sugars accumulated in the stem (Arai-Sanoh et al. 2011). Using abundantly accumulated sugars as carbon sources is reasonable for heterotrophic diazotrophs. In addition, sugar concentrations were higher in the inner part of the unelongated stem than in the outer part, where N fixation was higher (Fig. 5A). This indicated that the amount of sugar accumulation, as well as accessibility to the invasion of diazotrophs, is important, as mentioned above. Fructose was not a primary carbon source for the bacteria in the unelongated stem since neither ARA nor CO₂ production was low in the fructose treatment (Fig. 5B, C). Further investigation is needed to explore the causes of the unavailability of fructose. Among substances such as amino acids and organic acids, malate, whose concentration was higher in the roots than in the unelongated stems (Table S2), was reported to improve the cell concentration of M. trichosporium OB3b in the presence of methane (Xing et al. 2006). However, sufficient data to explain the relationship between most of the detected substances and N fixation are not available.

Along with sugars, methane is also an important carbon source for certain bacteria in rice stems (Fig. 6). This is indicated by the results that ASV-NifH-I3 and 4, which were 99–100% identical to that of type IIa methanotroph M. trichosporium SM6, were abundant in outer part of unelongated stem (Fig. 4) and that incubation of crushed unelongated stems has shown ARA from a medium with methane (Fig. 5B). This N fixation by methanotrophs is supported by the fact that the roots showed a higher relative abundance of ASV-NifH-I3 and 4 and little ARA, whereas ¹⁵N fixation was higher (Figs. 2 and 4). This is consistent with results from Bao et al. (2014a) and Shinoda et al. (2019) that showed that N fixation in the roots is primarily due to type IIa methanotrophs. The particularly high abundance of methanotrophs in this study may be due to the rice growing under low-N conditions (Bao et al. 2014a). Thus, diazotrophs in the unelongated stem seemed to fix N using sugar and methane as carbon and energy sources. Further investigation is needed to determine whether methanotrophs can fix N using methane monooxygenase inhibitors (Hara et al. 2022). ARA was detected during the incubation of crushed unelongated stems with methane as the carbon source (Fig. 5B). Methanotrophs were assumed to use the methanol produced during the 3-day incubation as a carbon source before acetylene exposure. Many strains of M. trichosporium utilize methanol (Whittenbury et al. 1970; Bowman et al. 1993), and the type I methanotroph Methylococcus capsulatusis Bath utilize methanol even in the presence of acetylene (Dalton and Whittenbury 1976). In addition, because acetylene is a competitive inhibitor of methane oxidation, it is possible that some methanotrophs were able to obtain energy from methane oxidation to reduce acetylene, as most of the gas phase in this incubation consisted of methane substrate.

The leaf sheaths showed ARA and ¹⁵N₂ fixation, and the relative abundance of methanotrophs was low (Figs. 2 and 4), suggesting that heterotrophic bacteria were primarily responsible for N fixation (Fig. 6). The high relative abundances of Alphaproteobacteria and Betaproteobacteria, which are assumed to fix N and utilize sugars (Fig. 4), are consistent with our previous report on diazotroph structure analysis in rice stem (Okamoto et al. 2021). At the same time, the copy number of methanotrophs in the leaf sheath was roughly equal to that in the outer part of the unelongated stem, and the possibility of N fixation by methanotrophs in the leaf sheath cannot be excluded. The copy number of mcrA, which encodes a key enzyme in methanogenesis, was much lower in the leaf sheaths than in the roots (Fig. 3D). Since the main pathway for methane emission from paddy soil is reported to be the leaf sheath at the lower leaf position (Nouchi and Mariko 1993), if methanotrophs fix N in the leaf sheath, they would use methane produced in the soil and roots, rather than methane produced inside the leaf sheath.

Another important finding from the bacterial community composition was the high relative abundance of AnfH (an alternative nitrogenase) sequences such as ASV-NifH-II1–3 (93% identical to that of B. cerclage (Actinobacteria) in the unelongated stem as compared with other parts (Fig. 4). Since Anf have less efficiently in N₂ reductions compared to Nif (Eady 1996), the contribution of these bacteria to Anf would be small, although further clarification is needed. N-fixation by Actinobacteria is well known in Frankia, although the Anf sequence found in this study did not belong to Frankia and its function remains unknown. The 16S rRNA results also showed that the relative abundance of Actinobacteria was higher in unelongated stems (Table S5), suggesting that unelongated stems may be a suitable environment for Actinobacteria.

Rice plants grown in no N plots were used in this study. Although N fixation in soil is known to be generally inhibited by the application of N fertilizer, and the low N conditions are considered suitable for rice to perform high N fixation (Shrestha and Maskey 1970; Siddikee et al. 2016), further investigation is needed to clarify the effect of N conditions on endophytic N fixation in rice.

This study revealed detailed sites of N fixation in the stem and the diazotroph community structure at these sites. Although the fixed ¹⁵N per unit dry matter weight was higher in roots than in stems, the ratio of stems to roots in rice was generally large, suggesting that stems may have a comparable N-fixing potential to roots, as indicated by previous studies (Ito et al. 1980; Ma et al. 2019). Further research is required to clarify how sugars and methane can be effectively supplied to diazotrophs without restricting the growth of rice plants. Since sugar is produced through photosynthesis in rice, genetic improvement of rice may be one of the solutions to promote its N-fixing activity. Also, searching for metabolites other than sugars that effectively enhance N fixation would be useful. Methane production from paddy soils varies depending on the growth stage of rice and water management. It is important to clarify whether changes in methane supply from the soil affect N fixation by methanotrophs inhabiting the surface and inside rice plants. By integrating these factors, strategies for enhancing N fixation in rice plants can be developed.

Conclusion

The outer part of the unelongated stem, particularly the epidermis, was found to be an active N-fixing site in the rice stem. Amplicon sequencing analysis and incubation experiments with sole carbon sources suggested that N fixation in unelongated stems was at least partly supported by both sugar-utilizing heterotrophic bacteria belonging to Betaproteobacteria and type II methanotrophs belonging to Alphaproteobacteria. The leaf sheath also showed high N-fixing activity, and sugar-utilizing bacteria were considered the primary diazotrophs. Roots that exhibited high N-fixation were suggested to be responsible for N-fixation by methanotrophs, as reported in previous studies. The uniqueness of these active N-fixing sites could be influenced by the accessibility of diazotrophs to rice tissue and the oxygen concentration. Among the sugars, sucrose and glucose have been suggested to be key sugars for N fixation, but the reasons for the unavailability of fructose and the impact of other metabolites on N fixation need to be clarified. Regarding methane oxidation, it is necessary to clarify the relationship with the amount of methane supplied from paddy fields. Surprisingly, ASVs close to Actinobacteria were found in high abundance in unelongated stems, and their function needs to be further investigated. Combining these findings will clarify strategies for enhancing the N-fixation ability of rice plants.