and N02- by Denitrification Bioassay and Mass Spectrometry

Transcription

1 APPLIED AND ENVIRONMENTAL MICROBIOLOGY, JUly 1994, p Vol. 6. No /94/$4. + Determination of 15N Abundance in Nanogram Pools of NO3- and N2- by Denitrification Bioassay and Mass Spectrometry OLE HJBERG,l* HENRIK SAABY JOHANSEN,2 AND JAN SRENSEN' Microbiology Section, Department of Ecology and Molecular Biology,1 and Physics Laboratory, Department of Mathematics and Physics,2 The Royal Veterinary and Agricultural University, Frederiksberg C, Denmark Received 9 March 1994/Accepted 6 May 1994 Suspensions of two strains of Pseudomonas aeruginosa (ON12 and ON12-1) were used to reduce N3- and N2-, respectively, to N2. The evolved N2 was quantified by gas chromatography with electron capture detection, and the 15N abundance was determined by mass spectrometry with a special inlet system and triple-collector detection. Sample gas containing unknown N2 pools as small as.5 ng of N was analyzed by use of a spike technique, in which a reference gas of N2 of natural '5N abundance was added to obtain enough total N for the mass spectrometer. In N3- or N2- pools, the 15N abundance could be determined in samples as small as approximately 3.5 ng of N. No cross-contamination took place between the N3- and N2- pools. The excellent separation of N3- and N2- pools, small sample size required, and low contamination risk during N2 analysis offer great advantages in isotope studies of inorganic N transformations by, e.g., nitrifying or denitrifying bacteria in the environment. The great interest in inorganic nitrogen cycling in both aquatic and terrestrial environments has resulted in development of numerous techniques to study the microbial processes involved. These include specific bioassay techniques to extract and quantify the pools of NO3- and NO,- (4, 6). In heterogeneous soil environments, it is important to be able to analyze very small subsamples of soil and small pools of NO3- and NO2-, which are involved in plant uptake and microbial nitrification and denitrification. NO,- is an intermediate in both processes and is therefore an important link between them. It has been reported that accumulation of NO- and NO2- can occur in the rhizosphere (4), but whether the NO,- accumulation results from nitrification or denitrification activity is not yet understood. To study soil N turnover in detail, it is necessary to use 15N isotope techniques. In the classical methods, NO3- is chemically reduced either to NO2- or N gases (15) or to NH4' (7, 8). The NH4+ may subsequently be collected by microdiffusion (5) and converted to N2 (8) before analysis. Common to these chemical methods, however, is that cross-contamination may occur during the separations (8, 12). Furthermore, the amount of N required is relatively large and the methods are not suitable for analysis of small amounts of NO- or NO2-. To facilitate work with small N3- pools, Christensen and Tiedje (6) used a denitrifying strain of Pseudomonas chlororaphis, ATCC 43928, to reduce '5N-labelled NO- (1 atom% 15N) to N2. The authors determined the isotope composition of NO3- pools with a precision of..2 atom% (two replicates). By comparison, Risgaard-Petersen et al. (18) determined the isotope composition of relatively small NO3 pools (containing 3 ng of N) by reduction to N2 using a denitrifying enrichment culture. The present study couples a denitrification bioassay technique (4, 6) to a sensitive mass-spectrometric (MS) analysis of '5N in the N2 being formed. The use of pure cultures of an NO3- (and NO-)-reducing Pseudomonas aeruginosa wild- * Corresponding author. Mailing address: Microbiology Section, Department of Ecology and Molecular Biology, The Royal Veterinary and Agricultural University, Rolighedsvej 21, DK-1958 Frederiksberg C, Denmark. Phone: Fax: type strain (ON12) and an NON2--reducing P. aeruginosa mutant strain (ON12-1) provides a tool to separate the NOand NOJ- pools in very small environmental samples (4). By reducing the pools to N2 rather than to N2, we avoid the risk of background contamination by atmospheric N2. In this study, we demonstrate the use of the bioassay combined with 15N analysis of the N2 pools formed. If the amount of N2 in the sample gas is too small (<1 ng of N) for direct determination of '5N abundance in the MS, the sample is spiked with a reference gas containing N,O of natural '5N abundance and the '5N content of the original sample is calculated (1). The sensitive detection of N2O by gas chromatography and subsequent 15N analysis in the MS will facilitate studies of N turnover in very small N3- and NO,- pools in soil and other environments. MATERIALS AND METHODS Standards of N3-, N2-, and N2. Standards (2 mm) of 98 atom% '5N-enriched NaNO3 and NaNO2 (Cambridge Isotope Laboratories, Woburn, Mass.) were made in distilled water. To obtain lower enrichments, these standards were mixed with 2 mm solutions of NO3 and NO,- of natural '5N abundance. Distilled water was added to obtain standard solutions of lower concentrations. Standard mixtures of N2O were made by growing a culture of P. chlororaphis ATCC under anaerobic conditions in tryptic soy broth (Bacto; Difco Laboratories, Detroit, Mich.) which contained 1 mm concentrations of '5N-enriched NO3 -' The NO3- is reduced to N,O because the organism is unable to reduce N,O to N2 (6). The '5N-labelled N2O was purified by purging the headspace gas through a column with trapping material for CO2 (Carbosorb, 12/2 mesh; Merck, Dorset, United Kingdom) and H2O (Drierite; W. A. Hammond, Xenia, Ohio). High-purity helium ( %) was used for the purge, and the N2 was subsequently collected in a cold trap (liquid N2). The cold trap was closed off and connected to a 12-ml, helium-flushed bottle. The final N2O concentration was approximately 2.5% (vol/vol). Dilution series of N,O in helium were made in 3.5-ml blood-collecting tubes (Venoject; Terumo Europe N. V., Leuven, Belgium). 2467

2 2468 HJBERG ET AL. Denitrification bioassay of N3- and N2- pools. The bioassay was performed essentially as described by Binnerup and Srensen (4), with wild-type (ON12) and mutant (ON12-1) strains of P. aeruginosa to reduce N3- and NO2-, respectively. The NO3- (and NO2-)-reducing ON12 strain was grown to stationary phase in citrate minimal medium with 1 mm N3- and was then washed and resuspended in the spent growth medium to a higher cell density (-1"' cells per ml). The NO2--reducing ON12-1 strain resembles the C13- resistant mutants reported by other workers (1, 2). The mutant strain is thus unable to synthesize an active nitrate reductase when grown in the presence of C13-. In the present work, the ON12-1 strain was grown in the presence of 1 mm rather than 1 mm C13- as used in the original assay. Stationary phase was thus reached after approximately 3 days rather than 7 days. The ON 12-1 cells were then resuspended in spent ON12 medium including 1 mm C13- to a higher cell density (-3 x 19 cells per ml). The spent ON12 medium was completely free of N3-, since N3- was reduced during the growth of the ON12 strain. Before resuspension, the spent medium was boiled for 5 min to strip accumulated N2 from the solution. Unknown samples (5 to 1 [LI) to be analyzed for N3- and N2- concentrations including their '5N content were placed in 1.8-ml screw-cap glass vials with polytetrafluoroethylene (Teflon)-coated silicone septa and quickly heated (95 C for 1 min) in a water bath to kill indigenous bacteria. The vials were then flushed with helium and acetylene (1-kPa C,H2) was added to inhibit N2 reduction in the two P. aenuginosa strains. This was because the ability of the two strains to accumulate N2 in the citrate minimal medium (4) was shown to depend on unknown compounds released from the butyl rubber stoppers (14). When samples contained both N3- and N2-, 1 Kl (-3 x 18 cells) of strain ON12-1 was first added to reduce the NO,- pool to N2. To quantify the N2, a gastight glass syringe (Dynatech or Hamilton) was flushed carefully with helium and 1 pl1 of helium was injected into the sample vial in replacement of 1 pi of the headspace gas. The sample gas was injected into a gas chromatograph (Chrompack model 428) equipped with an electron capture detector and a manual backflush system to prevent C2H2 from passing the detector. Running time was about 4 min per sample. After determination of the N2 concentration, the '5N abundance in the sample was analyzed by MS (see below). Before analysis of the N3- pool in the subsequent step, the vials were opened and heated (95 C for 5 min) in a water bath to remove the NO,---derived N2 pool. The vials were then closed again and flushed with helium. New C2H, (1 kp followed by a 1-Rl suspension (-19 cells) of strain ON12 was added, and the NO--derived N,O was finally analyzed for both concentration and '5N content as described above. Analysis of standards or unknown samples containing only NO- (or N2-) was carried out in one step, with 1 Rl (-1'3 cells) of strain ON12 in the bioassay. MS. The MS was a triple collector instrument (Tracermass model; Europa Scientific Ltd., Crewe, United Kingdom) providing a determination of mass-to-charge ratios (m/z) of 44, 45, and 46. The latter represented the most abundant molecular species, 44N2 (14N 4NO -,45N2 ( N N and N'416), and 4('NO (N5N-N'- ), respectively, and some combinations with the '7O and 1O isotopes (see below). Gas samples (1 to 3 pul) were injected manually into the MS with a gastight syringe after initiation of the software program (ANCA; Europa Scientific Ltd.). Behind the injection port, the sample passed a stainless steel column (4-mm inside APPL. ENVIRON. MICROBIOL. diameter and 15 cm long) filled with the mixture of trapping materials for C2 and H2. To separate N2 from traces of 2 (see below), the gas sample then passed a chromatographic column (Porapak Q;.2 mm by.4 m) in an oven held at 7 C. The retention time was about 1.5 min when the inlet pressure of the helium carrier gas was approximately 15 kpa. The running time was 3 min per sample. When analyzing small (nanogram) pools of N2, we included a spike addition of reference gas, which contained natural N2 (15N abundance taken to be.366%) to obtain adequate total N in the samples. Corrections were made to determine the 15N abundance in the original sample gas of the unknown N2 pool (see below). Three pools of N2 were thus defined: (i) the unknown pool, representing N2 from the original N3- and NO- sample reduced in the bioassay; (ii) the spike pool, representing N2 of natural '5N abundance added to obtain enough total N for the MS analysis; unless otherwise stated, the spiking gas was also used as reference gas to calibrate the instrument; and (iii) the mixed pool, representing a mixture of the unknown and spike pools. In practice, we ran a reference gas sample of natural N2 followed by 5 to 1 samples of mixed N2 pools. The first and the last of the mixed pool samples served as references, as they were prepared with total N contents and '5N abundances similar (within 1%) to the unknown ones. The injection procedure was initiated by a flush of the gastight syringe with helium, and 1 p.l of reference gas (natural N2 in helium;.1 to 1% [vol/vol]) was injected into the MS. The syringe was flushed again in helium, and 1 to 3.I1 of sample gas was withdrawn from a vial. The syringe was then quickly inserted into the bottle of reference gas, and 1 1A of this gas was carefully drawn into the syringe as well. The sample gas, containing both unknown and spike N2 pools, was finally injected into the MS. Calculations of '5N abundance. The molecular species of N2 contributing to the ion currents (1) at m/z of 44, 45, and 46 are the following: 441, 14N14N'1O; N5N 1o, '5N 4N'6, and 14N'4N 17; and 46j, l5n5n 16, 14N 5N 17, '5N '4N'7, and '4N '4N'8O. This means that determinations of the molecular fractions (x) of the 45 and 46 species, 4-5x and 46x as defined by 451/( i + 461) and 46"/(44I + 45I + 461), will give higher values than expected from a random distribution of the common isotopes l4n, '5N, and 16 only. There may further be a discrepancy between measured and expected values of the molecular fractions caused by mass discrimination or nonlinearity between the readings at the three MS collectors (19). To compensate for the presence of heavy oxygen isotopes and instrumental effects, we used the following three assumptions to obtain correct values of 45j and 46I. (i) In the reference gas, '4N and '5N were considered to be randomly, i.e., binomially, distributed. The '5N abundance was defined to be.366 atom%. (ii) It was assumed that no isotope exchange between the unknown and spike pools of N2 during the MS analysis took place, i.e., the mixed pool represented two separate pools of N2. This means that the unknown and spike pools both had random distributions of 14N and '5N isotopes, whereas the mixed N2 pool had a nonrandom isotope distribution. (iii) The 44j, which represented only one molecular species (14N'4N16), constituted >99% of the total beam area in the reference gas and was considered to be correctly determined in the MS. Under these three assumptions, the correct (44IC) and measured (44JIR) values of 44I in the reference gas are determined as 44i,,? = 44ic = (1-152 X XN2O (1)

3 VOL. 15ON 6, 1994 IN N3- AND N2 POOLS 2469 where 15a is the natural l5n abundance (.366 atom%) and XN,o is the total amount of N2 molecules without heavy oxygen isotopes. XN2O could then be calculated from equation 1. Because 4N and 15N were randomly distributed in the reference gas, the correct values of 45I, and 46I, could be determined as 45ic = 2 X (15 X (1-15 XXN2 (2) 46ic (152 = XXN2 (3) The differences between the correct values (45IC and the measured values (45Im and 46Im) may be expressed as A45I and A46I, respectively. The latter are in turn proportional to the total amount of N2 in the injected gas sample (see and 46IC) Results and Discussion) and were used to obtain the 45I and 46IC values from the 45Im and 461m values. The equations for determining isotope abundances in a mixed pool of '5N-labelled and natural (atmospheric) N2 have been presented by Hauck et al. (13), Hauck and Bouldin (11), Mulvaney (16), and Arah (1). In the present study, the 15N abundance in an unknown N2 pool, i.e., derived from NO3- or N2 pools in the bioassay, was calculated from the following equation originally developed for N2 analysis (1, 2): (46Xm - 15 a 15 a.) (5am - '5a The 15ap value is the 15N abundance of the unknown N2 pool. The 15aa value is the 15N abundance of the spike pool (;.366 atom%). The '5am value is the 15N abundance [(½/24 Ic + 46IC)/( IC + 45IC + 46Ic)] of the mixed pool. The 46Xm value is the molecular fraction of 46N2 [46Jc/(4 Ic + 45IC + 46IC)] in the mixed pool. The 15N abundance of the unknown N2 pool ('5ap) can thus be calculated from the '5N abundances of the spike pool (15a and the mixed pool ('5am) and the 46N2 molecular fraction (46Xm) of the mixed pool. RESULTS AND DISCUSSION Interferences by C2H2, C2, and 2 in MS analysis of N2. Acetylene somehow resulted in a tailing of the N2 peak appearing on the monitor of the MS. To investigate whether this interference affected determination of the isotope composition in N2, two sample series (five replicates each) of "5N-enriched N2 (-56 atom%), with or without 1-kPa C2H2, were analyzed. The measured l5n abundances (± standard deviations) were (±.28) atom% (with C2H2) and (±.42) atom% (without C2H2) for 2-ng N gas samples. The results showed that the altered shape of the N2 peak had no significant effect on the determination of its isotope composition. Carbon dioxide (m/z, 44) interferes with the MS analysis of N2 and must be removed. This was done by passing the sample gas (helium gas purge) through the CO2 trap in the MS inlet system; the trap removed CO2 completely from samples containing up to 5% (vol/vol) CO2 (data not shown). When the samples contained 2, small amounts of gas presumed to be NO2 (mlz, 46) affected the N2 analysis. A reaction between 2 and N2 at the ion source filament has been reported to produce NO (m/z, 3), which interferes with N2 analysis (3, 9, 19). One solution to remove the 2 may be to install a reduction furnace at the inlet before the sample enters the MS system (19). This cannot be done when analyzing N2, however, since the furnace will also reduce N2. We therefore chose to purge the sample vials and syringes with helium, CL zo 2 - uz M Total beam area (nc) FIG. 1. Differences between measured (Im) and correct (I') values of ion currents at mlz of 45 and 46 plotted against total beam area. The differences are designated A41I () and A46I (), respectively. Correlation coefficients (r2) are.9997 and 1., respectively. which removed the 2 and reduced the presumed NO2 peak to an absolute minimum (data not shown). Traces of 2 appearing in the MS were separated from N2 by the chromatographic column inserted behind the injection port. If accidental formation of NO2 still took place in the MS, this could be observed immediately on the monitor and the sample was rejected. MS analysis of '5N abundance in N2 pools. It has been claimed that MS analysis of N2 should be preferred to analysis of N2 because the latter is influenced by CO2 interference (17), N2 dismutation to NO and N2 (8), and lack of sensitivity (17). Some workers have thus chosen to reduce N2 to N2 (by Cu catalysis) prior to determination of the isotope composition (17). However, recent developments in the determination of the isotope abundance directly in N2 pools have been made, and combined gas chromatography and MS analysis of N2 has become a routine (3, 19). The latter studies have dealt with unspiked samples with "5N enrichments of 1 atom% or less, which is the optimal range of 15N abundances for most instruments (3). The 15N abundances in the spiked samples of the present study never exceeded 5 atom% and were also well within the optimal range (3). The spiking technique is therefore well suited for MS analysis of samples with a small N content but high "5N enrichment. It is common practice in MS analysis to adjust the total N content in reference samples according to that of the unknown samples (19). Using the spiking technique, we found that it was necessary to correct for even small differences in total N (<5%) between reference and sample gases in order to obtain reproducible results. As shown in Fig. 1, the differences between correct and measured ion currents in the reference gas were linearly related with the total beam areas (working range of 1-9 to 1.5 x 18 C) with correlation coefficients (?) of.9997 and 1., respectively. The Al values could thus be adjusted according to the difference in total N between reference and sample gases. The AI values were subsequently used A46I A451

4 247 HJBERG ET AL. - E A C) Q. z v U) Un Expected 15N abundance (atom%) FIG. 2. Measured '5N abundances in N2 standards containing expected '5N abundances between 1 and 98 atom%. The amount of N in sample gas was approximately 8 ng, while the spiking was 575 rig. The regression line isy = 1.16 x x, and the correlation coefficient (r2) is.9996 (n = 5). to correct the measured ion currents, 45I,, and 46m, of the sample gas. The importance of this correction procedure was illustrated by results from a sample series (1 replicates of -4 ng of N), in which the '5N abundance was determined to be 3.96 (+.32) atom% with correction of the AI values and (+2.572) atom% without correction. The precision was much poorer without correction of the AI values. The difference in total N content between reference and sample gases did not exceed 5% in the sample series described above, but the correction procedure was in fact reliable over a much wider range of total N contents. In a different sample series (three replicates of -85 ng of N), the 15N abundance was thus determined to be (±.79) atom% when reference and sample gases differed less than 5% in total N content and (+.75) atom% when the difference was approximately 75%. The important point here is that the precision levels, as judged by the standard deviation, are similar in the two series. The spiking technique may therefore be used even if the total N contents differ significantly, provided that the correction procedure is followed. However, since the spike pool usually constituted >95% of the total N in the mixed pool, our analysis routine always resulted in samples that differed less than 5% in total N content. Sensitivity of denitrilication bioassay and MS analysis. A series of N2 standards enriched with 1 to 98 atom% '5N was prepared from '5N-labelled N3- by use of the ON12 strain. The 15N abundances were determined in samples of N2 standards containing approximately 8 ng of N plus 575 ng of N in reference gas (natural 15N abundance) for spiking. Figure 2 shows a good agreement between expected and measured means (slope = 1.16 and r2.9996) for = the whole range of '5N abundances between 1 and 98 atom%. The standard deviations (five replicates) varied from.387 for the 1 atom% series to.59 for the 98 atom% series; the standard deviation values are too low to be seen in Fig. 2. The results demonstrate that determination of isotope compositions by the spike technique becomes less precise as the amount of '5N decreases in the sample. To estimate the minimum sample size required at a fixed '5N abundance, a sample series of N2 standards (-44 atom% '5N), representing variable amounts of total N, were analyzed with and without addition of a spike pool; spiking was with 575 ng of reference gas (natural '5N abundance). When no spike pool was added, a highly enriched batch of N2 standard (98 atom%) was used as the reference gas to equilibrate the instrument. The data in Fig. 3 (upper part) show that the precision of analysis is poor in unspiked samples when the total amount of N in the sample gas becomes lower than approximately 1 ng of N. However, when the spike pool was added the measured mean of approximately 44 atom% was obtained in as little as 1 ng of N in sample gas. It was clear, however, that standard deviations became larger as the N content decreased below approximately 1 ng of N in the sample gas. Thus, in the whole range of N contents (1 to 1 ng of N), the spiked samples gave much better determination of the isotope abundance than the unspiked ones, as judged from both the measured means and the standard deviations. To confirm that good sensitivity could be obtained also when the denitrification bioassay was coupled to the MS analysis, three sample series (1 replicates in each) of N3- solutions containing 3.5, 7, and 7 ng of N and a fixed 15N abundance of approximately 3 atom% were reduced to N2 by the ON12 strain. The total N contents of sample gas injected into the MS were.8, 4, and 4 ng of N in the three series, and spiking was with 288, 575, and 575 ng of N in reference gas (natural 15N abundance), respectively. The results shown in Fig. 3 (lower part) show that the measured mean was well determined even at the lowest sample size of approximately 1 ng of N. The standard deviations were also comparable to the ones determined on N2 standards (Fig. 3, upper part). This indicates that the denitrification bioassay combined with MS analysis operates under the same detection limits for sample size and range of suitable '5N abundances as the MS analysis alone. Discrimination of '5N abundances in N3- and N2- pools. To examine if there was any contamination of '5N isotope from the N2- pool to the NO3 APPL. ENVIRON. MICRO131L. pool, we performed an MS analysis of 1 replicates of an N3- standard (7 ng of N, -3 atom% '5N) mixed with an NO2 - standard (7 ng of N, -5 atom% 15N). After the N2- was reduced by the ON12-1 strain, the accumulated N2O was removed by opening the vials before heating (95 C for 5 min) and flushing with helium. Gas samples of the headspace were analyzed on the gas chromatograph and showed no detectable N2 (data not shown). The N3- was then reduced to N2 by the ON12 strain, and the '5N abundance was determined in 4 ng of N in sample gas spiked with 575 ng of N in reference gas (natural '5N abundance). The results in Table 1 (rows 1 and 2) show that the isotope composition in the N3- standard amended with NO,- could be determined with a low standard deviation similar to that of the NO- standard without NO2-. This confirms that the N2- -derived N2 was effectively removed from the vials. It may be noticed that the measured mean was somewhat higher in the N2--amended series. This was not an error of the assay, however; according to the isotope supplier (Cambridge Laboratories), the '5N-enriched N2- added to the N3- standards may actually contain 5 to 1% (wt/wt) of '5N-labelled NO3-. In a parallel series, we tested whether isotope contamination from the NO3- pool to the NO2- pool could take place. Five replicate samples of an NO2- standard (7 ng of N, -55 atom% '5N) was mixed with an NO- standard (7,ug of N, natural '5N abundance); a parallel series of '5N-labelled N2-

5 V51N IN NO3 AND NO, POOLS 2471 Voi-. 6, 19) E o 42-- i' 41 - c cu Z 32- ') C',, 31 - cn E c Total N in sample gas (ng) FIG. 3. (Upper part) Measured '5N abundances in N,O standards (-44 atom% '5N) containing different amounts of N in sample gas. Samples with () and without () spiking (575 ng of N) are shown. (Lower part) Measured '5N abundances in NO3 standards (-3 atom% '5N) containing different amounts of N in sample gas. Spiking was 288 ng of N for the sample with smallest amount of N and 575 ng of N for the others. standard, but without NO3 was also made. The NO2 was then reduced to N2O by the ON12-1 strain, and the '5N abundance was determined in 4 ng of N in sample gas spiked with 575 ng of N in reference gas (natural '5N abundance). The results in Table I (rows 3 and 4) show that a high background of nonenriched NO-, about 1 times higher than the NO2- standard, had no effect on determination of the isotope abundance in the NO< pool. This confirms that the ONl2-1 strain does not reduce NO3 under the assay conditions (4). TABLE 1. Analysis of '5N abundance in NO3 and/or NO, -containing standards" Sourcc I < + spike NO3- standard Measured '5N abundancc" (atomrn,;) NO NO3 (+NO )' NO , NO (+NO. )" " The total amount of N in NO3 or NO, standard solutions used in the dcnitritication bioassay was 7t) ng, the total amount of N in the sample gas (unknown N2 pool) analyzed in the MS was 4 ng, and the total amount of N in the reference gas (spike N2O pool) analyzed in the MS was 575 ng. " Measured means and stalndard deviations (n = 1 for NO3 samples; it = 5 for NO, samples). N3N standard was supplemented with NO2 (-5 atom%4 '5N) at a molar ratio of 1:1. The labelled NO, contained -3.5'7r labelled NO3 (see the text). "NO2 standatrd was supplemented with NO3 (.366 atomk '5N) at a molar ratio of 1: I()(). p The denitrification bioassay thus had an excellent specificity for both NO3- and N2- analyses. To our knowledge, the present study is the first to demonstrate determination of '5N content in both NO- and NO, pools in the same sample. Furthermore, the combination of the denitrification bioassay and MS analysis is a new approach to investigate small, nanogram pools of NO3-- and NO,, including their '5N contents, in natural samples. The assay determines the '5N content in much smaller pools of NO3- and/or NO2- than has previously been reported by Christensen and Tiedje (6) and Risgaard-Petersen et al. (18). From the results presented above, the denitrification bioassay in combination with MS analysis might be successfully optimized to a lower sample size limit than 1 to 1 ng of N and a lower '5N abundance than approximately 1 atom% in the unknown N2O pool. Further improvement of the sensitivity of the present method may be obtained by reducing the headspace volume in the vials, i.e., increasing the N2O content available as sample gas. Also, the spike-to-sample gas ratio may be adjusted for particular combinations of sample size and '5N abundance. In this study, we have developed a spiking technique for examination of isotope composition in small samples of relatively high '5N enrichment (>1 atom%) of the N2. Risgaard-Petersen et al. (18) used a denitrifying enrichment culture to reduce NO3- to No for analysis of '5N abundance in the MS. However, these authors reported a standard deviation of -.2 atom% (three replicates) for the measured mean abundance in NO- samples, representing a range of 1 to 99 atom%. Their precision was thus poorer than that obtained in

6 2472 HJBERG ET AL. the present study, even if approximately 4 times more NO3- in total sample N content was used. By reducing or omitting the spiking, we believe it will be possible to analyze small pools of NO3- and N2 at low enrichment (<1 atom%). We are presently using the methods to investigate microbial transformation of small, '5N-labelled NO3 and N2 pools in soil microenvironments which support a close coupling between nitrification and denitrification processes. In the heterogenous soil environment, the application of 15N isotope techniques in small-scale studies will be useful to obtain information about N turnover and simultaneous mineralization, nitrification, and denitrification processes. Organic aggregates, plant residues, and the root surface environment (rhizosphere) are important niches for rapid N3 and N2 turnover by nitrification and denitrification. The NO3 and N2- are obligatory intermediates in the processes and are important to analyze simultaneously when the 15N technique is used. ACKNOWLEDGMENTS We thank Svend Jrgen Binnerup for helpful assistance and Elfinn Larsen for critical reading of the manuscript. This work was supported by the Danish Center for Microbial Ecology. REFERENCES 1. Arah, J. R. M New formulae for mass spectrometric analysis of nitrous oxide and dinitrogen emissions. Soil Sci. Soc. Am. J. 56: Arah, J. R. M., I. J. Crichton, and K. A. Smith Denitrification measured directly using a single-inlet mass spectrometer and by acetylene inhibition. Soil Biol. Biochem. 25: Atkins, J. R. M., A. Barrie, S. J. Prosser, P. D. Brooks, and D. J. Herman Rapid isotopic methods to study the fluxes of radiatively active trace gases, p In Applications of isotopes and radiation in conservation of the environment. Proc. Int. Symp., Karlsruhe, Germany, 9-13 Mar International Atomic Energy Agency, Vienna. 4. Binnerup, S. J., and J. Srensen Nitrate and nitrite microgradients in barley rhizosphere as detected by a highly sensitive denitrification bioassay. Appl. Environ. Microbiol. 58: Blackburn, T. H Method for measuring rates of NH4' APPL. ENVIRON. MICROBIOL. turnover in anoxic marine sediments, using a '5N-NH4' dilution technique. Appl. Environ. Microbiol. 37: Christensen, S., and J. M. Tiedje Sub-parts-per-billion nitrate method: use of an N2-producing denitrifier to convert N3- or 15NO3- to N2. Appl. Environ. Microbiol. 54: Crumpton, W. G., T. M. Isenhart, and C. M. Hersh Determination of nitrate in water using ammonia probes and reduction by titanium (III). J. Water Pollut. Control Fed. 59: Fiedler, R., and G. Proksch The determination of nitrogen-15 by emission and mass spectrometry in biochemical analysis: a review. Anal. Chim. Acta 78: Giese, C. F The reaction O+ + N2 -- NO' + N. Adv. Chem. Ser. 58: Grifliths, L., and J. A. Cole Lack of redox control of the anaerobically-induced nirb+ gene of Escherichia coli K-12. Arch. Microbiol. 147: Hauck, R. D., and D. R. Bouldin Distribution of isotopic nitrogen in nitrogen gas during denitrification. Nature (London) 191: Hauck, R. D., and J. M. Bremner Use of tracer for soil and fertilizer nitrogen research. Adv. Agron. 28: Hauck, R. D., S. W. Melsted, and P. E. Yankwich Use of N-isotope distribution in nitrogen gas in the study of denitrification. Soil Sci. 86: Hjberg, O., and S. J. Binnerup. Unpublished data. 15. Koike, I., and A. Hattori Simultaneous determinations of nitrification and nitrate reduction in coastal sediments by a '5N dilution technique. Appl. Environ. Microbiol. 35: Mulvaney, R. L Determination of '5N-labeled dinitrogen and nitrous oxide with triple-collector mass spectrometers. Soil Sci. Soc. Am. J. 48: Mulvaney, R. L., and L. T. Kurtz A new method for determination of 15N-labeled nitrous oxide. Soil Sci. Soc. Am. J. 46: Risgaard-Petersen, N., S. Rysgaard, and N. P. Revsbech A sensitive assay for determination of 14N/15N isotope distribution in N3-. J. Microbiol. Methods 17: Stevens, R. J., R. J. Laughlin, G. J. Atkins, and S. J. Prosser Automated determination of nitrogen-15-labeled dinitrogen and nitrous oxide by mass spectrometry. Soil Sci. Soc. Am. J. 57: Stouthamer, A. H Biochemistry and genetics of nitrate reductase in bacteria. Adv. Microb. Physiol. 14: