New Method of Denitrification Analysis of Bradyrhizobium Field Isolates by Gas Chromatographic Determination of ¹⁵N-Labeled N₂

Abstract

To evaluate the denitrification abilities of many Bradyrhizobium field isolates, we developed a new ¹⁵N-labeled N₂ detection methodology, which is free from interference from atmospheric N₂ contamination. ³⁰N₂ (¹⁵N¹⁵N) and ²⁹N₂ (¹⁵N¹⁴N) were detected as an apparent peak by a gas chromatograph equipped with a thermal conductivity detector with N₂ gas having natural abundance of ¹⁵N (0.366 atom%) as a carrier gas. The detection limit was 0.04% ³⁰N₂, and the linearity extended at least to 40% ³⁰N₂. When Bradyrhizobium japonicum USDA110 was grown in cultures anaerobically with ¹⁵NO₃⁻, denitrification product (³⁰N₂) was detected stoichiometrically. A total of 65 isolates of soybean bradyrhizobia from two field sites in Japan were assayed by this method. The denitrification abilities were partly correlated with filed sites, Bradyrhizobium species, and the hup genotype.

Denitrification is anaerobic respiration using nitrate (NO₃⁻), nitrite (NO₂⁻), nitric oxide (NO), and nitrous oxide (N₂O) as terminal electron acceptors. Denitrifying species are distributed over a broad variety of bacteria, including Proteobacteria spp., gram-positive bacteria, and archaea (31). Most denitrifying bacteria reduce the electron acceptors sequentially as NO₃⁻ → NO₂⁻ → NO → N₂O → N₂, and each step of these reactions is catalyzed by specific respective reductases (31). To evaluate denitrification capability, the acetylene inhibition assay (30) has been widely used in pure cultures (13) as well as for environmental samples such as soils (8, 14, 27) and water (24) because of sensitivity and atmospheric N₂ contamination (13). In this assay, the excess of N₂O emission in the presence of acetylene over the level of N₂O emission in the absence of acetylene is regarded as the N₂ evolution (because acetylene inhibits N₂O reductase). N₂O can be detected by a gas chromatograph (GC) equipped with a ⁶³Ni electron-capture detector possessing high-level sensitivity for N₂O. The amount of N₂O is calculated on the basis of the measured headspace concentration and corrected for dissolved gas by using the Bunsen coefficient (29).

However, the acetylene inhibition assay has some drawbacks for evaluating bacterial denitrification. The most serious disadvantage is that it is impossible to measure N₂O reductase activity directly. For this reason, the activities and kinetic parameters of N₂O reductase have been studied directly by measurement of the N₂O decrease (4) or by spectrophotometric assay of reduced benzyl viologen for nitrous oxide reductase activity in vitro (12, 23). Moreover, the underestimation of denitrification capability may occur because of incomplete blockage of N₂O reductase at a low nitrate concentration or in the presence of sulfide (24). In practice, two determinations of N₂O concentration per sample—in the presence and absence of acetylene—are required for estimation of the amount of N₂ evolved by N₂O reductase.

Acetylene inhibition assays have revealed the diversity of the denitrification activities of Bradyrhizobium species (1, 2, 28). Bradyrhizobium japonicum USDA110, which possesses a full set of denitrification genes on the chromosome (11), consistently shows denitrification from nitrate to N₂ (2, 28). However, soybean bradyrhizobia often show diverse denitrification patterns. Their denitrification end products are N₂O and NO₂⁻ as well as N₂ when nitrate is supplied to media under anaerobic conditions, while a few strains are unable to denitrify at all (1, 2, 28). This polymorphism is probably due to truncation of the processes of anaerobic nitrate respiration (31). It has been shown that indigenous populations of soybean bradyrhizobia show significant field site variation in terms of DNA fingerprints according to insertion sequences, species level, and hydrogen uptake phenotype and genotype (7, 16, 17, 21). Similarly, we wanted to determine whether the denitrification patterns of indigenous populations of soybean bradyrhizobia differed in accordance with their field sites.

We therefore developed a new methodology for determination of ³⁰N₂ levels by using a GC equipped with a thermal conductivity detector (TCD). This method, which we refer to as the ¹⁵N/TCD method, enables ¹⁵N tracer analysis without the need for isotope ratio determination by mass spectrometry and will overcome some of the problems of the acetylene inhibition method when large numbers of field isolates are analyzed.

MATERIALS AND METHODS

Bacterial strains.

Strain USDA110 was used as a standard strain of B. japonicum (11). A total of 65 field isolates of Bradyrhizobium were isolated from the soils of the Nakazawa fields at the Niigata Agricultural Experiment Station (Nagaoka, Niigata, Japan) and the Tokachi field at the Tokachi Agricultural Station (Memuro, Tokachi, Hokkaido, Japan) as described previously (9, 15, 18).

Standard gases.

A gas mixture composed of 40% (vol/vol) ³⁰N₂ (¹⁵N¹⁵N) (¹⁵N, 99.7 atom%), 40% (vol/vol) Ar, and 20% (vol/vol) O₂ (Shoko Co., Ltd., Tokyo, Japan) was used as a ³⁰N₂ original standard gas. A standard gas of ²⁹N₂ (¹⁵N¹⁴N) was prepared by using codenitrification of Fusarium solani IFO9425 anaerobically grown in cultures under He with Na¹⁵NO₂ according to the method of Shoun et al. (22) and Tanimoto et al. (26). The amount of ²⁹N₂ was determined by using a GC-mass spectrometer (GC-MS QP5050; Shimadzu, Kyoto, Japan).

Growth media and growth conditions.

Bradyrhizobium cells were aerobically grown to full growth at 30°C in 15-ml HM salts medium (5) supplemented with 0.1% arabinose and 0.025% (wt/vol) yeast extract (Difco, Detroit, Mich.). For anaerobic growth for denitrification, HM medium was further supplemented with 0.55 μM Na₂MoO₄2̇H₂O, 1 μM FeCl₂, and 1 μM CuSO₄2̇H₂O, which were required for nitrate, nitrite, and nitrous oxide reductase activities, respectively (31). Filter-sterilized K¹⁵NO₃ (¹⁵N, 99.6 atom%) (Shoko Co., Ltd.) or K¹⁴NO₃ was added to the medium at a final concentration of 2 mM.

We used two bottle systems for denitrification experiments: 70 ml of medium in a 123-ml airtight specimen vial with a butyl rubber stopper and an aluminum seal (Maruemu, Osaka, Japan), and 15 ml of medium in a 27-ml test tube with a double butyl rubber stopper. Full-grown cells of bradyrhizobia were inoculated to the media to achieve an initial optical density of 0.05 at 660 nm. Just after the vials and test tubes were sealed, the overhead space gas was replaced several times with N₂ or He in a vacuum line. Gas samples were collected 10 days after incubation at 30°C. For the acetylene inhibition assay, ¹⁴NO₃⁻ was used as a substrate of denitrification in the presence and absence of 10% (vol/vol) acetylene.

Gas chromatography.

For ²⁸N₂, ²⁹N₂, and ³⁰N₂ analyses, gas samples (1 ml or 0.5 ml) were injected into a GC (GC-7A; Shimadzu) equipped with a TCD and a Molecular Sieve 5A column (80/100 mesh, 0.3-mm diameter by 2-m length). The carrier gas used was He or commercially available N₂ gas that has a natural abundance of ¹⁵N (0.366 atom%). The flow rate of the carrier gas was 30 ml/min. Throughout the TCD GC analyses, the temperatures of injection, column, and detector were 100, 50, and 100°C, respectively. Determination of N₂O levels was carried out by the TCD GC system described above except for use of a PorapakQ column (80/100 mesh, 0.3-mm diameter by 2-m length) and He carrier gas (flow rate, 65 ml/min). The amount of N₂O emission was calculated on the basis of the measured headspace concentration and corrected for the presence of dissolved gas by using the Bunsen coefficient (29). Peak areas were calculated from chromatograms by an integrator (ChromatoPack C-R18; Shimadzu).

Nitrate assay.

The concentration of nitrate in the culture was measured colorimetrically (3).

Statistics.

Regression analysis was carried out with Microsoft Excel X for Mac software (Microsoft Co., Redmond, Wash.).

RESULTS

Detection of ¹⁵N-N₂ by TCD gas chromatography.

We have often used a gas mixture composed of 40% (vol/vol) ³⁰N₂ (¹⁵N 99.7 atom%), 40% (vol/vol) Ar, and 20% O₂ (vol/vol) to determine nitrogenase activity of rhizobia and diazotrophic endophytes associated with plants. When we analyzed, by chance, the gas mixture with a TCD GC using a Molecular Sieve 5A column and N₂ carrier gas, an obvious peak appeared at a retention time after Ar and O₂ had already eluted (data not shown). Although TCD senses any gaseous materials that have thermal conductivity different from that of carrier gas (6, 10, 20), it is surprising that chemical species identical to the carrier gas might be able to form a peak on the gas chromatogram. Thus, we wanted to examine whether the peak is really derived from ¹⁵N-labeled N₂ (¹⁵N-N₂) gas and whether this detection could be utilized for denitrification studies.

Three kinds of 4% (vol/vol) N₂ gas, ²⁸N₂ (¹⁴N¹⁴N), ²⁹N₂ (¹⁵N¹⁴N), and ³⁰N₂ (¹⁵N¹⁵N), and their balance gases (a mixture of Ar and O₂ gas for ²⁸N₂ and ³⁰N₂ and He for ²⁹N₂) were measured by using 0.5-ml injection and He or N₂ as a carrier gas (Fig. 1). When He was used as a carrier, ²⁸N₂, ²⁹N₂, and ³⁰N₂ were detected at similar retention times of 4.1 min with similar peak sizes. When N₂ gas was used as a carrier, on the other hand, ²⁸N₂ was not detected at all (Fig. 1B); however, the use of ³⁰N₂ gave rise to an apparent peak at a retention time of 3.4 min, as expected (Fig. 1C). Moreover, ²⁹N₂ gas injection showed a peak at the same retention time as ³⁰N₂ gas injection. The size of the ²⁹N₂ peak was about half (43%) of that of the ³⁰N₂ peak on a molar basis (Fig. 1D). Since the retention times of ²⁸N₂, ²⁹N₂, and ³⁰N₂ molecules were identical when He and N₂ carriers were used (Fig. 1), a Molecular Sieve 5A column did not separate these molecules. The thermal conductivity values of ¹⁵N-N₂ molecules are likely different from those of normal N₂ with a natural abundance of ¹⁵N, however, which enabled us to observe a peak (3.4 min) specific for ¹⁵N-N₂ in the chromatogram. Interestingly, ¹⁵N-N₂ behaved as if the chemical species of ¹⁵N-N₂ differed from that of the N₂ carrier gas. These results demonstrated that the system of TCD gas chromatography is able to detect ¹⁵N-N₂ gases including ³⁰N₂ and ²⁹N₂. In this work, we refer to this ¹⁵N-N₂ detection system as the “¹⁵N/TCD method.”

Fig. 1.

Gas chromatograms of ¹⁵N-N₂ standards and reference gases with He or N₂ as a carrier gas. (A) 80% Ar and 20% O₂; (B) 4.0% ²⁸N₂, 77% Ar, and 19% O₂; (C) 4.0% ³⁰N₂, 76% Ar, and 20% O₂; (D) 4.3%²⁹N₂ in He; (E) He. Gas samples (0.5 ml) were injected into a GC (GC-7A, Shimadzu) equipped with a TCD and a Molecular Sieve 5A column (80/100 mesh, 0.3-mm diameter by 2-m length). The carrier gas used was He or commercially available N₂ gas that has a natural abundance of ¹⁵N of 0.366 atom%. The flow rate of the carrier gases was 30 ml/min. ²⁹N₂ gas was generated by codenitrification culture of F. solani IFO9425 under He atmosphere conditions (22, 26), and the amount of evolved ²⁹N₂ was determined by gas chromatography-mass spectrometry (GC-MS QP5050; Shimadzu). RT, retention time of the corresponding peak on gas chromatograms. The intensity of the ²⁹N₂ peak was about half (43%) of that of the ³⁰N₂ peak on a molar basis when the gas was eluted with N₂ carrier.

Quantification of ³⁰N₂ by the ¹⁵N/TCD method.

We further investigated the usefulness of this ¹⁵N/TCD method for quantification. A standard gas of ³⁰N₂ was diluted with air, N₂, or He to make concentrations of ³⁰N₂ from 0.01 to 40% (vol/vol). A total of 1 ml of each gas sample was injected. Fig. 2 shows a standard curve between the amounts of ³⁰N₂ and the areas of peaks. The limit of detection was 0.04% (32.7 nmol of ¹⁵N). A best-fit curve without an intercept was calculated by linear regression. The equation of the resultant curve was given by y = 1,031.9x (r² = 0.9985). From the best-fit curve, the quantitative range was 0.1% (81.8 nmol of ¹⁵N) to 40% (32.7 μmol of ¹⁵N). Changes in the dilution medium (air, atmospheric N₂, or He) had no effect on the measurements.

Fig. 2.

Standard curve for ³⁰N₂ quantification by the ¹⁵N/TCD method. Gas samples (1 ml) were injected into the TCD GC with a Molecular Sieve 5A column and N₂ as a carrier gas. Areas were calculated from a peak specific for ¹⁵N-N₂ (retention time, 3.4 min) as described for Fig. 1. Standard gas (solid circles) contains 40% ³⁰N₂, 20% O₂, and 40% Ar. Dilution series were prepared with air (open squares), N₂ gas with a natural abundance of ¹⁵N (solid triangles), or He (open circles). The best-fit curve was found by linear regression analysis (y = 1,031.9x, r² = 0.9985). The interior figure shows an enlarged area representing the results seen with 0 to 5 μmol of ¹⁵N.

Detection of ¹⁵N-N₂ generated from B. japonicum cells grown in cultures under denitrification conditions.

After anaerobic cultivation of B. japonicum USDA110 for 10 days with 2 mM ¹⁵NO₃⁻ or 2 mM ¹⁴NO₃⁻ in a 123-ml airtight specimen vial (70 ml of culture and 53 ml of headspace), 1 ml of headspace gas was taken and introduced to the GC. When the headspace gas of the vial with ¹⁵NO₃⁻ was measured, the peak of ¹⁵N-N₂ was observed (Fig. 3B) in the same retention time (3.8 min) as for ³⁰N₂ standard gas (Fig. 3A). The slight delay of retention time compared to the result shown in Fig. 1 (3.4 min) was possibly derived from a decrease in the flow rate of carrier gas by manual adjustments. In contrast, no peak was detected at this position when 2 mM ¹⁴NO₃⁻ was supplied (Fig. 1C). Therefore, it is apparent that ¹⁵N-N₂ was generated from ¹⁵NO₃⁻ by cells of B. japonicum USDA110. Recently, the process of anaerobic ammonia oxidation, which converts nitrogen from NH₄⁺ and NO₂⁻ to N₂ by the activity of a microorganism, was discovered to be another route of anaerobic N₂ generation (25). We thus examined whether anaerobic N₂ generation by B. japonicum occurs exclusively through classical denitrification. Because NH₄Cl is the sole nitrogen source of the HM medium, we replaced 6 mM ¹⁴NH₄Cl with ¹⁵NH₄Cl (99.7 atomic percentage) (Shoko Co., Ltd.) and added 2 mM ¹⁴NO₃⁻ as a denitrification substrate. No ¹⁵N-N₂ peak value was detected from the gas phase that was sampled from the anaerobic culture with ¹⁵NH₄Cl and ¹⁴NO₃⁻ (Fig. 3D). Therefore, the production of ¹⁵N-N₂ observed in the culture of B. japonicum USDA110 was exclusively a consequence of classical denitrification.

Fig. 3.

Gas chromatograms of ¹⁵N-N₂ produced in the B. japonicum USDA110 culture cultivated under denitrifying conditions in the presence of 2 mM ¹⁵NO₃⁻ (B), 2 mM ¹⁴NO₃⁻ (C), or 2 mM ¹⁴NO₃⁻ and 6 mM ¹⁵NH₄Cl (D). (A) To identify the position of ³⁰N₂ elution, a gas mixture of 4% ³⁰N₂, 21% O₂, and 5% Ar was analyzed. The arrowhead indicates the position of a ¹⁵N-N₂ peak. The retention time of the ³⁰N₂ peak was slightly longer than that seen in a previous analysis of TCD gas chromatography (Fig. 1), probably because the flow rate of N₂ gas was slightly reduced compared to the value of 30 ml/min obtained earlier. O₂ was partially coeluted with Ar and showed a peak in the minus direction at a retention time of about 1.8 min, because the gas contained as much as 21% of O₂. Gas samples (1 ml) were injected into the TCD GC with a Molecular Sieve 5A column and N₂ as a carrier gas.

Stoichiometric conversion of NO₃⁻ to N₂ observed using the ¹⁵N/TCD method.

To examine whether the conversion of ¹⁵NO₃⁻ occurs stoichiometrically or not, the nitrogen amounts of NO₃⁻, N₂O, and ¹⁵N-N₂ were monitored in an anaerobic culture of B. japonicum USDA110 supplemented with 2 mM ¹⁵NO₃⁻ in a 123-ml airtight specimen vial (70 ml of culture and 53 ml of headspace). We periodically sampled 0.1 ml of liquid culture for NO₃⁻, 0.2 ml of gas for N₂O, and 0.5 ml of gas for ¹⁵N-N₂ determinations. Fig. 4A shows that the final amount of ¹⁵N-N₂ was 140 μmol of N/vial, which was equal to the amount of ¹⁵NO₃⁻ initially supplied. This result clearly indicates that all nitrogen supplied as ¹⁵NO₃⁻ was transformed into ³⁰N₂.

Fig. 4.

Comparison of the ¹⁵N/TCD method (A) and the acetylene inhibition assay (B and C) in the detection of NO₃⁻ conversion via denitrification by B. japonicum USDA110. Concentrations of ¹⁵N-N₂ (solid circles), N₂O (open triangles), and NO₃⁻ (open squares) in culture were determined periodically as described in the text. B. japonicum USDA110 was cultivated anaerobically in the presence of 2 mM ¹⁵NO₃⁻ (A) or 2 mM ¹⁴NO₃⁻ with (C) or without (B) 10% (vol/vol) acetylene. Results are shown as means of triplicate determinations. Bars indicate standard deviations.

Comparison with the acetylene inhibition method.

To compare the ¹⁵N/TCD method with an acetylene inhibition assay, conversion of ¹⁴NO₃⁻ to N₂O-N and N₂-N were observed using two anaerobic cultures of B. japonicum USDA110 in the presence and absence of C₂H₂. Almost no evolution of N₂O-N was detected in the culture without C₂H₂ (Fig. 4B), whereas the evolution of N₂O-N in the culture with C₂H₂ (Fig. 4C) corresponded to that of ³⁰N₂-N in the culture without C₂H₂ (Fig. 4A). Moreover, the pattern of ³⁰N₂ evolution (Fig. 4A) was in accordance with that of N₂O evolution (Fig. 4C). These results indicate that the ¹⁵N/TCD method is compatible with the conventional denitrification assay by acetylene inhibition. In addition, ¹⁵N/TCD method is a useful assay for direct and simple analysis for bacterial denitrification.

Denitrification abilities of indigenous soybean bradyrhizobia.

We used 65 Bradyrhizobium isolates from soybeans inoculated with the soils from the Nakazawa and Tokachi field sites. The species, hup genotype, and insertion sequence accumulation of the isolates had been previously characterized (9, 15, 18). The production of ³⁰N₂ from anaerobic culture of the isolates with ¹⁵NO₃⁻ was examined by the ¹⁵N/TCD method. A total of 28 isolates from the Nakazawa field emitted an amount of ³⁰N₂ equivalent to the amount of ¹⁵NO₃⁻ supplied in the medium, whereas no evolution of ³⁰N₂ was detected from the other isolates (Table 1). To examine whether these were truncated variants lacking N₂O reductase (2, 28, 31), we determined the N₂O concentration 7 days after inoculation in the medium supplemented with ¹⁴NO₃⁻ (2 mM). N₂O accumulated from 3 isolates from the Nakazawa field and from 18 isolates from the Tokachi field (Table 1). The amount of N₂O-N was approximately equivalent to that of the added NO₃⁻-N (even in the test tube assay).

TABLE 1.

Denitrification capability of soybean bradyrhizobia isolated from two field sites in Japan by the ¹⁵N/GC method^a

Field site (no. of isolates)	Denitri- fication end productb		Species	hupc	HRSd	No. of isolates
	N2	N2O
Nakazaka (n = 42)	+	−	B. japonicum	+	−	22e
	+	−	B. japonicum	−	−	6f
	−	+	B. japonicum	−	−	3g
	−	−	B. japonicum	+	+	2h
	−	−	B. elkanii	−	−	9i
Tokachi (n = 23)	−	+	B. japonicum	−	−	17j
	−	+	B. japonicum	−	+	1k
	−	−	B. japonicum	−	−	1l
	−	−	B. japonicum	−	+	4m

^aSoybean bradyrhizobia previously isolated from Nakazawa and Tokachi field soils were characterized for species, uptake hydrogenase, and repeated sequence fingerprints (9, 18).

^bThe N₂ column indicates completion of the full process of denitrification from ¹⁵NO₃⁻ (2 mM) to ³⁰N₂ with from 98 to 110% recovery after 7 days (+) except for a low recovery (62%) of isolate NC34a. −, the isolate evolved no ³⁰N₂ (<0.03% [vol/vol]). The N₂O column indicates positive (+) and negative (−) N₂O accumulation in the presence of ¹⁴NO₃⁻ (2 mM) after 7 days. ³⁰N₂ and N₂O were determined at least in duplicate.

^cGenotype and phenotype of uptake hydrogenase (16, 18).

^dHRS, highly reiterated sequence-possessing strains carrying high copy numbers of insertion sequences (9, 15, 18, 21).

^eIsolates NC4a, NC5a, NC6a, NC10a, NC11a, NC12a, NC13a, NC14a, NC15a, NC16a, NC17a, NC18a, NC20a, NC21a, NC22a, NC24a, NC26a, NC29a, NC34a, NC35a, NC37a, and NC41a.

^fIsolates NC2a, NC19b, NC27a, NC28a, NC38a, and NC39a.

^gIsolates NC8a, NC33b, and NC36a.

^hIsolates NC3a and NC32a.

ⁱIsolates NC7a, NC23a, NC25a, NC31b, NC42a, NC43a, NC44a, NC45a, and NC46a.

^jIsolates T1, T3, T4, T5, T6, T7, T8a, T9, T10a, T11, T12, T13, T20, T27, T29, T37, and T39.

^kIsolate T25.

^lIsolate T40.

^mIsolates T2, T15, T22, and T31.

DISCUSSION

We developed a new method of GC measurement of ³⁰N₂ generated through denitrification by bacterial pure cultures. As far as we know, this is a first report of ¹⁵N-N₂-specific determination by TCD GCs (Fig. 1). The sensitivity of TCD is directly proportional to the differences in thermal conductivity between the sample and the carrier gases (20). The elute profiles of ²⁸N₂, ²⁹N₂, and ³⁰N₂ (Fig. 1) showed that a ¹⁵N-N₂ peak appeared in the direction opposite from that of the peak of He with a higher thermal conductivity in the chromatogram eluted with N₂ carrier (20). This suggests that the thermal conductivities of ³⁰N₂ and ²⁹N₂ are slightly lower than that of ²⁸N₂. Recently, it has been shown that the absolute response of TCD to the eluting component cannot be interpreted simply but is involved in a number of heat-transfer factors such as flow rate, molar heat capacity, convection, and radiation (6, 10). Thus, it has so far been difficult to predict theoretically molar response factors of ³⁰N₂ and ²⁹N₂ for TCD.

The advantages of this ¹⁵N/TCD method are (i) use of a more commonly available piece of equipment (a TCD GC) rather than a mass spectrometer, (ii) no interference with ³⁰N₂ determination by atmospheric N₂ contamination, (iii) simultaneous analysis of many cultures, and (iv) presumptive application to N₂O reductase activity assay by product measurement. Although field isolates of soybean bradyrhizobia were analyzed in this work, the ¹⁵N/TCD method is equally applicable to other bacteria in their corresponding media by using compounds highly enriched with ¹⁵N such as NO₃⁻, NO₂⁻, and N₂O.

Use of the ¹⁵N/TCD method rapidly demonstrated that the Bradyrhizobium isolates can apparently be categorized into three denitrification types: (a) full denitrifier (up to N₂), (b) truncated denitrifier (up to N₂O), and (c) nondenitrifier. No full denitrifiers was found among the Tokachi isolates, interestingly, whereas the majority of Nakazawa isolates (67%) showed full denitrification from nitrate to N₂ (Table 1). In contrast, truncated denitrifiers (up to N₂O) accounted for 78% of the Tokachi isolates (Table 1). If soybean bradyrhizobia contribute to N₂O emission from field soils, then an indigenous population dominated by Bradyrhizobium full denitrifiers would reduce N₂O emission from agricultural soils. When the three types of denitrification were compared with previously characterized traits, we found some relationships among the tested isolates (Table 1). All of the isolates with the ability to take up hydrogen (hydrogen uptake-positive isolates) fell into the full-denitrifier category. All of the B. elkanii isolates were found to be nondenitrifiers by this assay. Highly reiterated sequence-possessing isolates carrying high copy numbers of insertion sequences tended to belong to the nondenitrifiers. To confirm these relationships, examination of additional field isolates and appropriate standard strains of Bradyrhizobium spp. (together with analyses of their denitrification genes) would be required.

To apply this method to environmental samples such as soil and water we must consider ²⁹N₂ (¹⁵N¹⁴N) evolution, because denitrification produces ³⁰N₂ (¹⁵N¹⁵N), ²⁹N₂(¹⁵N¹⁴N), and ²⁸N₂ (¹⁴N¹⁴N) through random isotope pairing (19) and because anaerobic ammonia oxidation and codenitrification contribute to ²⁹N₂(¹⁵N¹⁴N) production (22, 25, 26). Since ²⁹N₂ gave rise to a level of signal intensity about half of that seen with ³⁰N₂ (Fig. 1C and D), ¹⁵N incorporation into N₂ molecule can be monitored by the ¹⁵N/TCD method as a first screening assay for environmental samples. Afterwards, mass spectrometry analysis would be required to specify ³⁰N₂ (¹⁵N¹⁵N) and ²⁹N₂(¹⁵N¹⁴N) molecules or their ratios for the environmental samples.

Finally, we summarize the recommended procedure of the ¹⁵N/TCD method for denitrification assays of microbial pure cultures. A gas mixture of 4% (vol/vol) ³⁰N₂ (¹⁵N, more than 99 atom%) and 96% (vol/vol) Ar was recommended as an ideal standard gas, because the Ar peak was oriented in the same direction as the ³⁰N₂ peak and because the Ar peak is smaller than the He peak (Fig. 1). Commercially available N₂ gas that has a natural abundance of ¹⁵N (0.366 atom%) should be used as a carrier gas. The recommended conditions of TCD GC are as follows: flow rate of carrier gas, 30 ml/min; column, Molecular Sieve 5A (80/100 mesh, 0.3-mm diameter by 2-m length); detector temperature, 100°C: column temperature, 50°C; injection volume, 0.5 to 1.0 ml.