N2O Emission from Degraded Soybean Nodules Depends on Denitrification by Bradyrhizobium japonicum and Other Microbes in the Rhizosphere

A model system developed to produce N2O emissions from degrading soybean nodules in the laboratory was used to clarify the mechanism of N2O emission from soybean fields. Soybean plants inoculated with nosZ-defective strains of Bradyrhizobium japonicum USDA110 (ΔnosZ, lacking N2O reductase) were grown in aseptic jars. After 30 days, shoot decapitation (D, to promote nodule degradation), soil addition (S, to supply soil microbes), or both (DS) were applied. N2O was emitted only with DS treatment. Thus, both soil microbes and nodule degradation are required for the emission of N2O from the soybean rhizosphere. The N2O flux peaked 15 days after DS treatment. Nitrate addition markedly enhanced N2O emission. A 15N tracer experiment indicated that N2O was derived from N fixed in the nodules. To evaluate the contribution of bradyrhizobia, N2O emission was compared between a nirK mutant (ΔnirKΔnosZ, lacking nitrite reductase) and ΔnosZ. The N2O flux from the ΔnirKΔnosZ rhizosphere was significantly lower than that from ΔnosZ, but was still 40% to 60% of that of ΔnosZ, suggesting that N2O emission is due to both B. japonicum and other soil microorganisms. Only nosZ-competent B. japonicum (nosZ+ strain) could take up N2O. Therefore, during nodule degradation, both B. japonicum and other soil microorganisms release N2O from nodule N via their denitrification processes (N2O source), whereas nosZ-competent B. japonicum exclusively takes up N2O (N2O sink). Net N2O flux from soybean rhizosphere is likely determined by the balance of N2O source and sink.

Nitrous oxide (N 2 O) is a key atmospheric greenhouse gas that contributes to global warming and the destruction of stratospheric ozone (14,46,47). Agricultural land is a major source of N 2 O through the microbial transformation of nitrogen in the soil (13,24,58), and contributes significantly to the net increase in atmospheric N 2 O (46). Legume crops emit more N2O than non-legumes (10,15,32). Yang and Cai (55) reported that the emission of N2O from a soybean field greatly increased in the late growth period, suggesting that senescence and the decomposition of roots and nodules contributed to emissions. Ciampitti et al. (7) also reported marked N 2 O emissions from a soybean field in the late growth period regardless of N fertilization. N2O emission from a field with nodulating soybeans was several times higher than that from a field with non-nodulating soybeans (27). N2O was emitted directly from degraded nodules of field-grown soybeans in the late growth period (20). Thus, soybean nodules emit N 2 O under field conditions, although the mechanism remains unresolved.
Microorganisms might be involved, as N2O can be generated by several microbial processes (4,13). Using microbial community analysis, Inaba et al. (20) nominated potential N2O producers that increased in abundance in degraded nodules. Among them, Bradyrhizobium japonicum was one of the dominant microbes as endosymbionts of soybean nodules and rhizosphere soil bacteria (9,29,30,33,35,39). It reduces nitrogen oxides during denitrification as where each step is catalyzed by specific reductases. These reductases are encoded, respectively, by napA (encoding periplasmic nitrate reductase), nirK (Cu-containing nitrite reductase), norCB (nitric oxide reductase), and nosZ (nitrous oxide reductase) (5). The aim of this study was to clarify the involvement of B. japonicum in the emission of N 2 O from the soybean rhizosphere. The N2O flux from denitrification mutants of B. japonicum was compared in the laboratory.

Preparation of soil suspension
Soil was collected from an experimental field at Tohoku University (Kashimadai, Miyagi, Japan). This gray lowland soil had pH[H2O] 5.6, pH[KCl] 4.2, total C 1.37%, total N 0.132%, and Truog P 48 mg P2O5 kg −1 . Fresh soil (10 g) was extracted twice with 30 mL distilled water to remove nitrate and nitrite. The suspension was shaken for 10 min in centrifuge tubes and then centrifuged at 5,555×g for 15 min (Himac CR20E; Hitachi, Tokyo, Japan). The pellet was resuspended in 30 mL distilled water.

Inoculation and plant cultivation
Surface-sterilized soybean seeds (Glycine max cv. Enrei) were germinated in sterile vermiculite for 2 days at 25°C. The seedling was then transplanted into a Leonard jar pot (one plant per pot) (28,53,56), which contained sterile vermiculite and nitrogen-free nutrient solution (31, 34) (Fig. S1). The seedlings were then inoculated with B. japonicum cells at 1×10 7 cells per seedling. Plants were grown in a phytotron (Koito Industries, Tokyo, Japan) providing 270 μmol photons m −2 s −1 of photosynthetically active radiation (PAR, 400-700 nm) for 30 days at 25/20°C with a 16-h light/8-h dark cycle. A nitrogen-free sterilized nutrient solution (34) was periodically supplied to the pots. Thirty days after inoculation, a soil suspension (10 g in 30 mL) was added to the vermiculite in the pot (soil addition, S), or the aboveground parts of plants were excised (decapitation, D), or both treatments were performed (DS) (Fig. 2). The aim of the S treatment was to introduce soil microbes into the aseptic pot. That of the D treatment was to stop photosynthate supply to the soybean roots; because field N2O emission occurred more than 100 days after sowing (55), shoot decapitation was used to promote nodule senescence and degradation. The pots were left in the phytotron until N2O determination for 15 days except otherwise indicated.
N2O determination N2O flux was determined with a gas chromatograph (GC-14BpsE; Shimadzu, Kyoto, Japan) equipped with a 63 Ni electron capture  Model system for N2O emission from degraded nodules USDA110 (nosZ+), USDA110ΔnosZ (nosZ−), and T9 (nosZ−) were used as inoculants. Thirty days after inoculation, treatments were applied (Fig. 2). Ten days later, nodules were collected from soybean roots, washed with sterilized water, and weighed. The nodules were introduced into a 19-mL airtight vial. Gas in the vial was sampled 1, 2, and 3 h after the vials were sealed to determine the N2O concentration. This was the "excised nodule method" (Fig.  2).

Long-term N2O monitoring
T9 was used as inoculum. Thirty days after inoculation, the D or DS treatment was applied. The N2O flux from the pot was intermittently monitored during 2 months. On each measurement day, the pot was sealed with a lid with a gas sampling port (Fig.  2). After 5 h, the gas was sampled to determine N2O concentration. After the gas sampling, the pot was returned to the phytotron. This was the "sealed jar method" (Fig. 2 N2 was supplied to the root zone of soybeans inoculated with USDA110ΔnosZ in seven pots for 8 h (Fig. 2 and S1). The nodules from three plants were separately collected and dried at 80°C for 3 days. The 15 N concentrations of the powdered nodules were determined by mass spectrometer (EA 1110 DeltaPlus Advantage ConFlo III; Thermo Fisher Scientific, Bremen, Germany). The other four pots received the DS treatment. Fifteen days later, the gas phase was sampled by the sealed jar method (Fig. 2). The 15 N concentrations were determined by gas chromatography/mass spectrometry (GC/MS-QP2010 Plus; Shimadzu) (21,22).
N2O emission from degraded nodules with denitrification mutants USDA110, USDA110ΔnosZ, USDA110ΔnapAΔnosZ, and USDA110ΔnirKΔnosZ were used as inoculants. Thirty days after inoculation, D or DS treatment was applied (Fig. 2). Fifteen days later, the N2O flux from the nodules was determined by the excised nodule method.
N2O flux from soybean rhizosphere with denitrification mutants USDA110 and its ΔnosZ, ΔnirK, and ΔnirKΔnosZ mutants were used as inoculants to evaluate the effect of the nirK and nosZ genes on N2O emission from the rhizosphere. The nirK mutation was selected as a nitrate-to-N2O denitrification mutation, because the nirK mutant is not able to denitrify both nitrate and nitrite that exist in the rhizosphere (4). Thirty days after inoculation, DS treatment was applied. Fifteen days later, the N2O flux from each pot was determined by the sealed jar method, 3 h after the pot was sealed. In addition, 50 mL of 5 mM KNO3 solution was applied to each pot, the pots were immediately sealed, and the N2O flux was determined as above.

N2O emission from degraded nodules
When B. japonicum USDA110 (nosZ+) was used as the inoculum, N2O was not emitted in any treatment (Fig. 3A). When USDA110ΔnosZ or T9 (each nosZ−) was used, the DS treatment induced marked N2O emission, whereas the D and S treatments alone did not induce N2O emission ( Fig. 3B and  C). Indeed, the nodules in the DS treatment were clearly degraded (Fig. S2), similar to those of field-grown soybean in the late growth period (20). On the other hand, the nodules in the S treatment stayed intact, and those in the D treatment looked slightly degraded (Fig. S2). These results indicate that both soil microbes and nodule degradation are required for N 2 O emission. In addition, N 2 O was emitted only from DS-treated nodules with nosZ− strains, suggesting that the B. japonicum nosZ gene is critical in the emission of N2O from degraded nodules.
Long-term monitoring of N2O flux from the soybean rhizosphere Substantial N2O was emitted from the rhizosphere of soybeans inoculated with T9 (nosZ−) in DS treatment, but none was emitted in D treatment throughout the experimental period (5-63 days) (Fig. 4). This result is similar to the results in the excised nodule method (Fig. 3B and C). As the N2O flux in the DS treatment peaked 15 days after the treatment was applied and then gradually decreased (Fig. 4), we measured N 2 O flux at 15 days in later experiments.

Origin of N2O-N
The profile of N2O flux (Fig. 4) suggests that the source of N2O was limited. Thus, we examined whether N2O is derived from N fixed in the nodules by using 15 N-labeled dinitrogen. The supply of 15 N2 to the root zone of USDA110ΔnosZ plants just before DS treatment produced 15 N concentration in N 2 O emitted 15 days later of 1.32±0.42 atom% excess (mean ± SD), similar to the concentration of nodule N (1.13±0.08 atom% excess). This result clearly indicates that the N2O-N emitted from the soybean rhizosphere was derived from N fixed symbiotically in the nodules.

N2O emission from degraded nodules with denitrification mutants
N2O emissions from the nodules formed with USDA110 and its mutants were determined by the excised nodule method to reveal the involvement of bradyrhizobial denitrification (Fig. 5). Nodules inoculated with ΔnosZ, ΔnapAΔ nosZ, and ΔnirKΔnosZ emitted marked amounts of N2O in DS treatment. Nodules inoculated with USDA110 emitted negligible N 2 O even in DS treatment (Fig. 5A).
Because the nosZ gene is responsible for the reduction of N2O to N2 (18,43), and no N2O was emitted from nosZ+ nodules (Figs. 3A and 5A), N 2 O reductase encoded by nosZ is likely a sink for N2O in the soybean rhizosphere. In the absence of nosZ, N2O emission from nodules inoculated with double mutants (ΔnapAΔnosZ and ΔnirKΔnosZ) was lower than that from nodules with ΔnosZ, although there was no significant difference (Fig. 5B, C, and D, t-test [P<0.05]).

N 2 O flux from the soybean rhizosphere with denitrification mutants
When soybean plants were inoculated with USDA110 and ΔnirK, a small quantity of N2O was released (1.9-2.6 nmol h −1 per pot; Fig. 6A). When plants were inoculated with ΔnosZ and ΔnirKΔnosZ, N 2 O emission was significantly higher (16.7 and 9.9 nmol h −1 per pot, respectively). These results strongly suggest that the nosZ gene of B. japonicum is involved in the uptake of N2O that is released from degraded nodules. In Fig. 6A, the relative contribution of the   nosZ gene to N2O flux is shown as "CZ1". In the absence of nosZ, there was a significant difference in N 2 O flux between ΔnosZ and ΔnirKΔnosZ (CK1 in Fig. 6A). This difference is due to the loss of nitrite reductase in the denitrifying pathway of B. japonicum. Therefore, the N 2 O flux from soybeans inoculated with ΔnosZ could have had two distinct sources; denitrification up to N2O by B. japonicum (CK1 [41%] in Fig. 6A), and other soil microbes (CS1 [59%] in Fig. 6A).
KNO3 was added to the rhizosphere to clarify whether NO 3 − is a precursor of N 2 O. When KNO 3 was supplied before N2O determination, the N2O flux from the pots with each inoculant was markedly enhanced, particularly from pots with ΔnosZ (78.1 nmol h −1 per pot) and ΔnirKΔnosZ (31.3 nmol h −1 per pot; Fig. 6B). This result confirms that N2O was produced from NO3 − through microbial denitrification. KNO3 application also enhanced the contribution of B. japonicum to N2O flux (60% [CK2, Fig. 6B] cf. 41% [CK1, Fig. 6A]). These results suggest that B. japonicum prefers nitrate as a substrate for N 2 O production.

Discussion
The term "rhizosphere" was first coined in 1904 by Lorenz Hiltner in Germany, who had a special interest in complicated N transformations around leguminous nodules with higher N contents in fields (16). In a sense, the present study advances such historical work on leguminous rhizospheres.
The results show that N2O emission from degraded nodules in the soybean rhizosphere is due to B. japonicum and other soil microbes. When plants were inoculated with B. japonicum nosZ− strains and treated with shoot decapitation and soil addition (DS), N2O was markedly produced (Figs. 3, 4, 5, and 6). On the other hand, when plants were inoculated with a nosZ+ strain, almost no N2O was emitted, even in DS treatment. These results suggest that N2O emission from degrading nodules formed with nosZ− strains was due to denitrification by both B. japonicum (nosZ−) and other soil microbes (Fig. 7). It is likely that N2O produced by soil microbes was offset by nosZ-competent B. japonicum with its N2O reductase. In other words, both B. japonicum and other soil microorganisms release N 2 O during nodule degradation (N2O source), and nosZ-competent B. japonicum (nosZ+ strains) takes up N2O (N2O sink) (Fig. 7).
The decline in N2O emission after the peak (Fig. 4) indicates that the source of N in the rhizosphere is limited. Indeed, the 15 N tracer experiment showed that nodule N is a major source of N2O emission from the soybean rhizosphere. Thus, complicated N transformation in the soybean rhizosphere would involve ammonification, nitrification, and denitrification.   7. Schematic representation of N2O metabolism in the soybean rhizosphere induced from the present study. Bradyrhizobium japonicum and other soil microorganisms generate N2O during nodule degradation. nosZ+ strains of B. japonicum are exclusively able to take up N2O via N2O reductase. The relative contributions of N2O emission (CK1 and CS1 in Fig. 6 and text) are shown as percentages at arrows of B. japonicum and soil microorganisms. Net N2O flux is determined by the balance between source and sink. NAP, NO3 − reductase; NIR, NO2 − reductase; NOR, NO reductase; NOS, N2O reductase.
KNO3 addition enhanced N2O emission (Fig. 6), supporting the idea that NO3 − is a precursor of N2O. When NH4Cl was preliminarily added to the rhizosphere, the addition did not change N2O emission (Inaba et al., unpublished data), suggesting that it is unlikely to be due to nitrification. KNO3 addition also enhanced the contribution of B. japonicum to N2O emission in relation to the other soil microbes (Fig. 6). Nitrate might be more available to B. japonicum, whereas other microbes might prefer other substrates. In fact, nitrite is a better substrate for denitrifying fungi to produce N2O (45). New approaches are needed to understand soil N2Oproducing microorganisms and N transformation from fixed nitrogen in the rhizosphere (4).
In soybean fields, it is likely that soybean roots are infected with multiple strains that differ in denitrifying activity. nosZ− strains of B. japonicum that produce N2O as the denitrification end product often dominate in agricultural fields (3,6,11,41,42,54). Both N2-and N2O-producing strains occurred in paddy-upland rotation fields (3). Similarly, both nosZ+ and nosZ− strains of B. japonicum were isolated from soybean fields (41,42). Thus, it is easily conceivable that both N2and N2O-producing strains of B. japonicum coexist in soybean fields. Consequently, the flux of N 2 O from soybean fields during the late growth period may be partly determined by biotic factors, namely the balance between N2O emission due to soil microbes and B. japonicum (nosZ−) and N 2 O uptake by B. japonicum (nosZ+) (Fig. 7).
The use of nosZ+ strains of B. japonicum as inoculants has been expected to reduce N 2 O emissions from soybean fields (42,43). Indeed, nosZ+ strains produced no N2O and were able to take up N2O from degraded nodules (Fig. 7). Recently, N 2 O reduction by nosZ-carrying inoculants was shown in a soil-filled pot planted with soybeans (17). Thus, B. japonicum mutants with increased N2OR activity (23) might be more effective to reduce net N 2 O flux from soybean rhizosphere.