Reduction in denitrification activity in field soils exposed to long term contamination by 2,4,6-trinitrotoluene (TNT)


*Corresponding author. Tel.: +1 (514) 496-6182; Fax: +1 (514) 496-6265


Terrestrial sites contaminated with 2,4,6-trinitrotoluene (TNT) are a widespread and persistent problem and often contain non-vegetated areas with TNT concentrations well in excess of 1000 mg kg−1. In this study, we examined the effect of TNT on denitrification activity in field soils, and compared the sensitivity of denitrifying enzymes to TNT. DNA probes assessed the prevalence of nirS, nirK and nosZ (encoding cd1 or copper nitrite reductase and nitrous oxide reductase, respectively), denitrifying genotypes in the culturable and total microbial community. The nitrate (NaR), nitrite (NiR) and nitrous oxide (N2OR) reductase activities in field soil and in isolates were assessed by gas chromatography. The relative occurrence of the nirK, nirS or nosZ genotypes increased in the cultured community and in total uncultured community DNA as nitroaromatic concentrations increased. However, denitrifying activity decreased in response to increasing TNT concentrations, with an IC50 for NaR+NiR+nitric oxide reductase (NOR) of 400 mg TNT kg−1 soil and for N2OR of 26 mg TNT kg−1 soil. The denitrifying activity of four soil isolates also decreased in response to TNT, with N2OR activity being three times more sensitive to TNT than NaR+NiR+NOR activity. Interestingly, there were 118 times more nirK isolates than nirS isolates in uncontaminated soil but only 1.5 times more in soil containing 17 400 mg kg−1 TNT. The results from this study indicated that TNT reduced denitrification activity in field soils, and N2OR was much more sensitive to TNT than NaR+NiR+NOR.


Terrestrial sites contaminated with 2,4,6-trinitrotoluene (TNT) are a widespread and persistent problem [1]. These sites often contain non-vegetated areas with TNT concentrations in excess of 1000 mg kg−1[2,3]. Understanding the microbial ecology of ‘hot-spots’ such as these can aid in the design of bioremediation systems and natural attenuation monitoring procedures. Typically, TNT is reductively transformed in aerobic soils [1] and the amino-metabolites that consequently arise are less toxic and less bioavailable than the parent compound [4,5]. Despite the ease with which TNT is transformed in soil under laboratory conditions, many contaminated sites still contain high TNT concentration decades after introduction of the pollutant was stopped.

Fuller and Manning [3] investigated sub-populations of microbial communities in soils chronically exposed to TNT. Using phospholipid fatty acid biomarkers and selective plating, they found that TNT was more toxic to Gram-positive than Gram-negative bacteria. These findings confirmed earlier in vitro studies that used cultured isolates to assess TNT toxicity to bacteria [6,7]. In addition to this differential toxicity, Fuller and Manning [3] found that levels of organic nitrogen, ammonium (NH4+) and nitrate (NO3) in soil were abnormally high and correlated with TNT. High levels of ammonium and nitrate in TNT-contaminated soil may be due to the denitration of TNT and deamination of TNT metabolites in soil [8]. Fuller and Manning [3] found that ammonium oxidation potential was weakly inversely correlated with the TNT concentrations and hypothesized that TNT disrupted the denitrification portion of the nitrogen cycle.

Denitrification is the respiratory reduction of nitrate or nitrite to dinitrogen or nitrous oxide by some aerobic bacteria under anoxic conditions [9,10]. During denitrification, NO3 is sequentially reduced to N2 via a set of four enzymes: nitrate reductase (NaR), nitrite reductase (NiR), nitric oxide reductase (NOR) and nitrous oxide reductase (N2OR). These enzymes convert nitrate to nitrite (NaR), nitrite to NO (NiR), NO to N2O (NOR) and N2O to N2 (N2OR), respectively. Sequences of the corresponding structural genes for these four enzymes are known for at least three different denitrifiers. These genes are narGH, nirS/K, norCB and nosZ. There are two types of NiR enzyme: a cd1 cytochrome encoded by the nirS gene and a copper reactive center encoded by the nirK or nirU gene. Most soil denitrifiers exhibit cd1 NiR but the copper NiR is widely distributed taxonomically. No known organism contains both forms of the NiR enzyme [11].

It is not clear which of the four denitrification enzymes is the most sensitive to pollutants because the majority of studies have only investigated toxicity to the NaR+NiR+NOR system. There is some evidence that N2OR may be more sensitive to pollutants than the NaR+NiR+NoR system. Richards and Knowles [9] found that a component of Hamilton Harbour sediment, contaminated with polycyclic aromatic hydrocarbons and heavy metals, inhibited N2OR activity more than NiR activity. Pesticides and their metabolites also inhibit N2OR activity [12]. As noted by Richards and Knowles [9], denitrification is commonly measured using the acetylene blockage technique, which inhibits the conversion of nitrous oxide to dinitrogen and would mask any effects of pollutants on N2OR activity.

This study investigated the interactions of TNT with the denitrifying community present at two field sites contaminated for more than 20 years by nitroaromatics. The prevalence, activity and TNT sensitivity of denitrifying enzymes were determined, at both the community and isolate level. The objective of this study was to evaluate the hypothesis that TNT was adversely affecting denitrification in field soils. Furthermore, we wished to determine if N2OR was more sensitive than NaR or NiR to TNT.

2Materials and methods

2.1Field sites

The two study sites (A and B) were within 200 m of each other, at a TNT manufacturing facility, in use between 1939 and 1972. At each site, three samples (0–15 cm depth), located along a transect running from a central monitoring well in the non-vegetated ‘hot-spot’ to the vegetated area, were taken on June 17, July 14 and August 26 1998. The soil was a clay-loam with a pH ranging between 6.8 and 7.8. Reference soil was a clay-loam soil taken from a nearby (300 m) uncontaminated area. New transects were drawn at each sampling date. One sample was collected from the non-vegetated area (area 1), the second sample (ca. 9 m from the first) was at the edge of the vegetated area (area 4) and the third sample was 6 m further into the vegetated area from the second (area 6) [22], and two additional samples were collected from the vegetated area. Soil was collected, placed in plastic bags, stored in coolers and frozen at −20°C. Because the samples were frozen, samples collected at different time points in the summer are additional replicate samples and not an indicator of changes occurring over the course of the summer. Sample collection tools were washed with warm water, rinsed with hexane, then acetone and dried between each sample collection.

2.2Bacterial strains and plasmids

Pseudomonas stutzeri ATCC 14405 and Achromobacter cycloclastes ATCC 21921 were purchased from ATCC. Plasmids pnir9 (nirS) and pMW12 (nosZ) were obtained from Dr. Y.K. Chan (Agriculture Agri-Food Canada) with permission from Dr. W.G. Zumft. Plasmid pRTC19 (nirK) was also obtained from Dr. Y.K. Chan with permission from Dr. J.M. Tiedje.

2.3Amplification, purification, sequencing and labeling of gene probes

Three sets of primers were used to amplify genes encoding the cd1 NiR (nirS), the copper NiR (nirK) and the N2OR (nosZ), respectively. The primers for the three reductases were designed following alignment of the sequences from four different denitrifying bacteria. Conserved regions were used for primer design. The primer sets (Hukabel Scientific, Montreal, Que., Canada) were as follows: for nirS (forward) 5′-CGGCTACGCGGTGCATATCTCGCGTCTGTC-3′ and (reverse) 5′-GATGGACGCCACGCGCGGCTCGGGGTGGTA-3′; for nirK (forward) 5′-GGGCATGAACGGCGCGCTCATGGTGCTGCC-3′ and (reverse) 5′-CGGGTTGGCGAACTTGCCGGTGGTCCAGAC-3′; and for nosZ (forward) 5′-CTGGGTCTCGGGCCGCTGCACACCACCTTC-3′ and (reverse) 5′-GATCAGCTGCTCGTTCTCCGGATGCAGCGG-3′[21]. The primers amplified a 316-bp fragment from the cd1 NiR (nirS), a 378-bp fragment from the Cu NiR (nirK) and a 299-bp fragment from the N2OR (nosZ).

Denitrification DNA probes were derived from plasmid pnir9 for nirS[13], plasmid pRTC19 for nirK[14] and plasmid pMW12 for nosZ[15]. Genomic DNAs from P. stutzeri ATCC 14405 (nirS) and A. cycloclastes ATCC 21921(nirK) were also used as positive controls for the colony hybridization experiments. The probes derived from the cloned genes did not cross hybridize.

PCR reactions were performed in a 50-μl final volume with 200 μM each of dATP, dCTP, dGTP and dTTP, 2 mM MgCl2, 0.2 μM of each primer, 2 U of Taq polymerase (Pharmacia Biotech, Baie d'Urfé, Que., Canada) and 40–50 ng of DNA template. PCR was performed in a DNA thermal cycler (Perkin Elmer Cetus, Montreal, Que., Canada) for 30 cycles using the following conditions: 1 min at 94°C, 1 min at 70°C, 1 min at 72°C and a final extension of 3 min at 72°C. PCR fragments were analyzed by agarose (1.6%) gel electrophoresis and visualized by ethidium bromide staining. The PCR fragments were purified with QIAquick PCR Purification kit (Qiagen, Mississauga, Ont., Canada) and 32P internally labeled (22–56 ng of purified PCR fragment; approximately 400 000 disintegrations per min per ml) using the Amersham multiprime kit (Amersham, Oakville, Ont., Canada). Unincorporated nucleotides were removed with ProbeQuant TM G-50 Micro Columns (Pharmacia Biotech, Baie d'Urfé, Que., Canada).

2.4Community DNA extraction and DNA dot-blotting

Microbial community DNA was extracted from soil by chemical lysis [16] and purified on polyvinylpolypyrrolidone spin columns [17]. Purified DNA was dot-blotted in triplicate on Zeta-Probe (Bio-Rad membranes) and hybridized at high stringency (i.e. 65°C) with 32P-labeled oligonucleotide probes according to the manufacturer's instructions. Scintillation counting of cut dot-blot membranes determined the amount of bound probe. Standard curves compiled from six concentrations of genomic DNA from the nirS, nirK or 23S rDNA-positive organisms bound to the membrane were constructed for each gene probe. Only those membranes whose standard curves were in excess of r2>0.90 were used. Levels of nirS and nirK are expressed as a percentage of community DNA in samples as determined by 23S rDNA.

2.5Bacterial enumeration and colony hybridization

Five g of sieved (5 mm) soil was placed in a tube (25 mm×150 mm) containing approximately 2.5 g of glass beads (3 mm diameter). The soil was diluted with 15 ml of 0.1% (w/v) tetrasodium pyrophosphate (pH 7.0) and vortexed for 2 min. The water–soil mixture was serially diluted and 0.1 ml of the 10−1, 10−2 and 10−3 dilutions spread-plated onto triplicate plates of YTSN containing per l of tap water (250 mg yeast extract (Becton Dickinson, Cockeysville, MD, USA), 250 mg tryptone (Difco Laboratories, Detroit, MI, USA), 250 mg soluble starch (Anachemia, Montreal, Que., Canada), 0.5 g NaNO3, 15 g granulated agar (Becton Dickinson)). Plates were incubated anaerobically (BBL GasPak Anaerobic System, Becton Dickinson, Cockeysville, MD, USA) at room temperature (21–24°C) for 2 weeks, the bacterial colonies counted and lifted onto nylon membranes. Cells adhering to the membranes were lysed, the DNA was denatured, fixed and cross-linked to a membrane and the DNA hybridized to DNA probes [13]. The proportion of Gram-positive colonies was determined by the heat shock method [3] and confirmed by conventional Gram staining of selected putative Gram-positives.

2.6NirS and nirK probe assessment

Bacterial colonies that were positive for the nirS or nirK gene probe were isolated and purified on YTS plates. Pure cultures were inoculated into tubes of YTS or YTSN broth, and incubated anaerobically for 144 h. Cultures that could not grow on YTS, but converted 80% of the nitrate to nitrous oxide, were considered as denitrifying bacteria [18]. Dr. J.J. Germida at the University of Saskatchewan, Dept. of Soil Science, identified 15 isolates by fatty acid methyl ester analysis using the MIDI system. Strains identified with a similarity index <0.3 were considered tentatively identified.

2.7Response of denitrifying isolates to TNT

The ability of eight denitrifying isolates, four nirS-positives and four nirK-positives, to metabolize TNT was determined. Isolates were grown in YTS overnight and 0.5 ml of the overnight culture (optical density=1) was inoculated into 9 ml of four media: YTSN, YTSN containing 50 mg TNT l−1, YTS or YTS containing 50 mg TNT l−1. TNT dissolved in acetone was added to the media and initial TNT concentrations were verified using US EPA Method 8330. All media were incubated aerobically as well as anaerobically with abiotic controls for each media type. The ability of these isolates to transform TNT was assessed by incubating the isolates in YTS or YTSN amended with 2.5 mg TNT l−1. The optical density (590 nm) was determined daily, and after 7 days of incubation, TNT and metabolites were determined using US EPA Method 8330.

2.8Denitrifying activity of soil

To assess the combined NaR, NiR and NOR activity in field soil, soil (20 g fresh weight) was placed in glass serum bottles (120 cm3), 800 μl of 1 M NaNO3 was added. The microcosms were purged with dinitrogen and sealed with teflon seals and aluminum crimps. Conversion of nitrous oxide to dinitrogen was blocked by the addition of 10 kPa acetylene [19,20]. Nitrous oxide concentrations were measured by gas chromatography (GC) of headspace samples. N2OR activity was assessed by measuring the disappearance of N2O (1000 ppmv) added to sealed microcosms rendered anoxic by evacuating and backfilling with N2[20]. These microcosms were previously incubated anaerobically for 6 weeks before assessing N2OR activity. No conversion of nitrate to nitrous oxide was detected during the N2OR assay. Preliminary experiments found that TNT added to sterile sand did result in the consumption or production of N2O. Regression analysis and inspection of residuals (Minitab, State College, PA, USA) determined the relationship between soil TNT concentrations and denitrification activity.

2.9Toxicity of TNT to NaR+NiR+NOR, NiR+NOR and N2OR activity

The toxicity of TNT to the NaR+NiR+NOR, NiR+NOR and N2OR activities was determined. Four isolates, two nirS-positive and two nirK-positive, were incubated for 3 days, washed and inoculated into YTSN with 10 kPa acetylene (NaR+NiR+NOR), or YTS containing 0.41 g NaNO2 with 10 kPa acetylene (NiR+NOR), or YTS containing 1000 ppmv N2O (N2OR). Nitrous oxide concentrations were determined by GC daily. After 3 days, TNT concentrations were determined.

2.10Analytical methods

US EPA Method 8330 was used to analyze nitroaromatic concentrations in soil [4]. Gases (N2O, CO2) were measured by GC on a SRI 8610C gas chromatograph (SRI, Torrance, CA, USA) with a thermal conductivity detector (TCD), and a flame ionization detector (FID) in parallel, and on a separate injection port, an electron capture detector (ECD). The SRI 8610C GC had the following configuration: the TCD was set at 100°C, the FID at 150°C and the ECD at 330°C; the oven was set at 60°C, each detector was connected to a separate column (2 m×3.1 mm) stainless steel packed with Porapak Q (Supelco, Missisauga, Ont., Canada). Helium (ultra high purity grade, Prodair, Montreal, Que., Canada) was used as carrier gas with flow rates of 23 ml min−1 for the TCD, 20 ml min−1 for the FID and 30 ml min−1 for the ECD. The detection limits were: N2O, 1 ppmv (ECD) or 200 ppmv (TCD); CH4, 20 ppmv (FID) or 800 ppmv (TCD); and CO2, 300 ppmv (TCD). The TCD response was linear for N2O, CH4 and CO2 (up to 200 000 ppmv), the FID response was linear for CH4 up to 15 000 ppmv and the ECD response was linear for N2O up to 300 ppmv [21].

For gas determinations, 0.5 ml of the gas samples was injected onto the GC system with simultaneous integration of peaks using the PeakSimple II software (SRI, Torrance, CA, USA). Gas standards were injected at the beginning and at the end of each day of analysis. A gas standard that contained 7092 ppmv of each of the following gases: CH4, CO2, N2O and C2H2 was prepared at the beginning of each day of analysis. Calibrated gas standards (990 ppmv N2O in N2) (Prodair) were also used.

2.11Accession numbers

The GenBank accession numbers of the organisms used for the construction of primers were for nirS: P. stutzeri#M80653, Pseudomonas aeruginosa#X16452, Paracoccus denitrificans Pd1222 #U05002 and Ralstonia eutropha#X91394. For nirK: A. cycloclastes#Z48635, Pseudomonas aureofaciens#Z21945 and P. aeruginosa G-179 #M97294. For nosZ: A. cycloclastes#X94977, R. eutropha#X65278, P. stutzeri#M22628 and P. denitrificans#X74792.


3.1Prevalence of denitrifying genotypes

The concentration of nitroaromatics varied between the three sampling areas (Table 1), increasing from area #6 to #1, with the exception of the amino-nitrotoluenes, which peaked in area #4 and were undetectable in area #1. The nitrotoluenes, 2-nitrotoluene and 2,3-dinitrotoluene, were rarely detected and therefore not included in the results. Soil nitrate concentrations did not vary significantly between areas.

Table 1.  Selected properties and concentrations (mg kg−1) of nitroaromatics in soils used in this study
  1. aTNT, 2,4,6-trinitrotoluene; 2,4-DNT, 2,4-dinitrotoluene; 2,6-DNT, 2,6-dinitrotoluene; 1,3,5-TNB, 1,3,5-trinitrobenzene; 2-ADNT, 2-amino-dinitrotoluene; 4-ADNT, 4-amino-dinitrotoluene.

  2. bnd, not determined.

AreaTimeSiteMoisture (%)pHNO3-NTNTa2,4-DNT2,6-DNT1,3,5-TNB2-ADNT4-ADNT
11A137.05113 384109155000
11B107.22420 76900000
12A156.89027 6103494862400
12B117.51210 0276603600
13A107.23515 162271309000
13B167.01917 1552001400
41B176.2276 1950162000
Reference  137.1ndbndndndndndnd

The prevalence of the nirK, nirS or nosZ genotypes increased in total community DNA as the nitroaromatic concentration increased (Fig. 1A). For example, levels of nirK doubled between areas #6 (6.56 mg TNT kg−1) and #4 (1150 mg TNT kg−1), and then doubled again between areas #4 and #1 (17 400 mg TNT kg−1). Prevalence of nirK and nirS in the culturable community also increased as the TNT concentration in the soil increased, tripling between areas #4 and #1 (Fig. 1B). The increase in nirK prevalence in total community DNA was weakly correlated (r=0.648, n=18) with the exponential of the TNT concentration in the soil. The increase of positive nirK in the culturable community was also correlated (r=0.768, n=18) with TNT concentrations in the soil. NirS and nosZ followed similar correlation patterns. Interestingly, more of the nirK genotype was detected in soils with low TNT concentrations but as TNT levels in soil increased, the relative occurrence of the nirS genotype increased. For example, in soils with TNT concentrations below 10 mg kg−1, nirK was 118 times more prevalent than nirS but only 1.5 times more prevalent in soils contaminated with high concentrations of TNT. In contrast to the denitrifying community, heterotrophs decreased in response to elevated TNT concentrations in soil (Fig. 1C). Thus, although the relative occurrence of denitrifier genotypes increased as TNT concentrations increased, the absolute numbers of nirS and nirK organisms did not change, remaining constant at approximately 104 colony forming units (cfu) per g of soil. Gram-positive organisms decreased more than heterotrophs or anaerobes, decreasing from 106 cfu g−1 in uncontaminated soil to 103 cfu g−1 in contaminated soil.

Figure 1.

Composition of microbial communities in soils from areas contaminated with three concentrations of TNT (given in parentheses). (A) Prevalence of denitrifying genotypes in total community DNA, (B) prevalence of denitrifying genotypes in culturable anaerobes, and (C) enumeration of culturable aerobes, anaerobes and aerobic Gram-positive bacteria. Error bars are S.E.M.

3.2Diversity of denitrifiers

Twenty putative denitrifying isolates, as suggested by gene probes (10 nirK and 10 nirS), were tested for denitrifying activity. Eighteen of the isolates converted greater than 80% of the added nitrate to N2O but did not grow in the absence of nitrate under anaerobic conditions. Only 15 isolates, (six nirK and nine nirS) grew on the media required for MIDI analysis. All six nirK-positive strains were identified as Pseudomonas sp. (one tentatively). In contrast, out of nine nirS-positives analyzed, six were identified as Pseudomonas sp. (three tentatively), one as a Photobacterium sp. and two as Aquaspirillum sp. Three non-pseudomonad species and five pseudomonad species were isolated from areas with low TNT concentrations, whereas all seven isolates from areas with high TNT concentrations were pseudomonads.

The presence of 50 mg TNT l−1 decreased the aerobic growth rate of nirK isolates by 55% but had no effect on the growth rate of nirS isolates (data not shown). Aerobic transformation of TNT varied between nirK and nirS isolates (Fig. 2). Under aerobic conditions, only 2.5% of the TNT remained with nirK isolates but 43% remained with nirS isolates. Concentrations of TNT remaining in the medium were negatively correlated (r=−0.93) with the stationary phase optical density (data not shown). In contrast, anaerobic growth rate and TNT metabolism did not differ between nirK or nirS organisms. One isolate, nirS-E (identified as Aquaspirillum autotrophicum with a SIM=0.438), rapidly transformed TNT to its amino-metabolites under both aerobic and anaerobic conditions and completely consumed TNT in the growth medium.

Figure 2.

Nitroaromatic concentrations remaining in broth cultures (YTS aerobic or YTSN anaerobic) after 7 days aerobic or anaerobic incubation in the presence of 2.5 mg TNT l−1 with either four nirK or four nirS-positive isolates. Error bars are S.E.M.


Increasing soil TNT concentrations reduced NaR+NiR+NOR (r2=0.94; P<0.001; n=18) as well as N2OR (r2=0.90; P<0.001; n=18) activity in soil (Fig. 3). N2OR activity was more sensitive to TNT than NaR+NiR+NOR activity. N2OR activity decreased 32% for every 10-fold increase in TNT concentration in soil, whereas, NaR+NiR+NOR activity only decreased 18% for every 10-fold increase in TNT concentration. Thus, the IC50 for N2OR was 26 mg TNT kg−1 soil with a 90% confidence limit of 13–97 mg TNT kg−1 soil. NaR+NiR+NOR activity had an IC50 of 400 mg TNT kg−1 soil with a 90% confidence limit of 170–1280 mg TNT kg−1 soil. There was always a basal N2OR activity even in soils containing the highest TNT concentrations. Whereas NaR+NiR+NOR activity continued to decrease until there was no observable activity in the five samples contaminated with greater than 13 000 mg kg−1 TNT. For example, in soils contaminated with 17 400 mg TNT kg−1, N2OR activity was still apparent although greatly reduced compared to soils containing only 7 mg TNT kg−1 (Fig. 4). Despite having similar nitrate concentrations, the soil contaminated with 17 400 mg TNT kg−1 produced 25 μmol g−1 soil of N2O compared to 1100 μmol g−1 soil produced by a soil contaminated with 7 mg TNT kg−1. In contrast to this greatly reduced NaR+NiR+NOR activity, N2OR in the highly contaminated soil was still active, reducing N2O by approximately 55% in 28 days compared to a reduction in slightly contaminated soils of 70% in 21 days.

Figure 3.

Correlation of NaR+NiR+NOR and N2OR activity with concentrations of TNT in soil. Data are expressed as a percentage of the activity of a reference soil which was not contaminated with TNT. Consumption of N2O did not change at soil TNT concentrations greater than 800 mg kg−1 and were not used in the correlation analysis.

Figure 4.

Nitrous oxide production and consumption (μmol flask−1) in the presence of different concentrations of TNT in soil. Each point is the mean of six replicates sub-sampled three times. Error bars are S.E.M.

Similar results were obtained with pure cultures (Fig. 5). N2OR activity decreased exponentially (r2=0.94; P<0.001; n=8) compared to the relatively minor and linear decreases in NiR+NOR (r2=0.93; P<0.001; n=8) and NaR+NiR+NOR (r2=0.97; P<0.001; n=8) activity. Interestingly, nirS isolates had three times greater (P<0.001) NiR+NOR activity than their nirK counterparts and twice (P<0.001) the NaR+NiR+NOR activity (data not shown).

Figure 5.

Sensitivity of NaR, NiR and N2OR activity to TNT concentration in broth culture. Each point is the mean of four isolates, two nirK and two nirS, sub-sampled twice, and is expressed as percentage of medium containing isolates but no TNT. Error bars are S.E.M.


TNT inhibited denitrification activity in field soils as well as in soil isolates. This inhibition was dependent on and proportional to the level of TNT present in the media or soil. The results support the assertion of Fuller and Manning [3] that TNT affects the denitrification portion of the nitrogen cycle in addition to a more general decrease in the heterotrophic microbial community. Typically, it is assumed that ammonia-oxidizing bacteria are the most susceptible to pollution [22], yet the correlation between TNT levels in soil and ammonia-oxidizing ability can be quite low, e.g. r2=0.393 [3] and r2=0.49 [23]. In contrast, in this study, denitrification activity was strongly negatively correlated, r2≥0.90, with TNT concentrations in soil. The extensive manipulation required for the assessment of nitrification activity and prolonged incubation under slurry conditions may be responsible for the differences seen between these two assays [24]. Despite these differences, the IC50 values for nitrification and denitrification at this field site were similar with an IC50 of 51 mg TNT kg−1 soil for nitrification [23] and 26 mg TNT kg−1 soil for N2OR activity.

NaR+NiR+NOR activity was only 55% as sensitive to TNT as N2OR activity and this is reflected in NaR+NiR+NOR's higher IC50 of 400 mg TNT kg−1. In pure culture work, N2OR activity was three times more sensitive to TNT than NaR+NiR+NOR activity. We did not directly measure the product of NaR activity, i.e. NO2, instead, the combined products of NaR, NiR and NOR were measured. Hence, our assay cannot differentiate if NaR is less sensitive to TNT than NiR because of the sequential nature of denitrification. Similarly, it is possible that either NaR or NiR or both their activities are less sensitive to TNT than NOR and that the NOR step is the rate limiting step in response to TNT toxicity. Other authors have also found that N2OR activity is more sensitive to toxicants than NaR+NiR+NOR activity. Fifty mg of 4-chloro-o-toluidine kg−1 soil inhibited NaR+NiR+NOR activity by only 5%, but inhibited N2OR activity by 11%[12]. An unknown component of Hamilton Harbour sediment was more toxic to N2OR activity than to NaR+NiR+NOR activity [9]. These authors caution that the use of the acetylene blockage technique to assess denitrification activity may mask significant toxicity to the N2OR system. For example, in a study of the effects of trichloroethylene and toluene on the nitrogen cycle in soil, Fuller and Scow suggest that nitrite oxidation potential is more sensitive than denitrification activity [25]. The authors used the acetylene blockage technique to measure denitrification activity and this technique only measures NaR+NiR+NOR activity. Comparing our results to those of Gong et al. [22] indicates that nitrification is more sensitive than the NaR+NiR+NOR portion of the denitrification pathway. However, the final step in denitrification, i.e. N2OR activity, is more sensitive than nitrification potential. Hence, from Fuller and Scow's data, it is not clear what the comparative toxicity of toluene and trichloroethylene is between N2OR and nitrification. The different components of the denitrification pathway require differentiation when investigating the impacts of toxicants on nitrogen cycling. Otherwise, it is likely that the most sensitive enzyme, N2OR, will not be assessed.

Soils with high TNT concentrations, e.g. 17 400 mg TNT kg−1, had a higher prevalence (although similar absolute numbers) of denitrifying bacteria than soils with lower, e.g. 1150 mg TNT kg−1, concentrations. This trend was similar for all three genotypes, nirK, nirS and nosZ, in the total uncultured and cultured communities. The reasons for this increase are not immediately evident. Denitrifying bacteria could not use TNT as a sole carbon source, nor were denitrifying bacteria more resistant to TNT under anaerobic, i.e. actively denitrifying, compared to aerobic, i.e. normal heterotrophic metabolism, conditions. Thus, it does not appear that denitrifying activity imparts a physiological capability to resist TNT toxicity. The decrease in anaerobic heterotrophs and Gram-positive bacteria partially explains the increase in denitrifier prevalence. As mentioned in Section 3.1, numbers of denitrifiers did not increase, rather their prevalence did, which is a function of their proportion within the community. Hence, if numbers of bacteria capable of anaerobic growth decreased dramatically but denitrifiers did not decrease, the proportion of denitrifiers relative to the total number of bacteria would increase. In this study, levels of anaerobic bacteria decreased by more than one order of magnitude in soils with 17 400 mg TNT kg−1. However, this does not explain why denitrifiers are resistant to TNT, while the remainder of the facultative anaerobic community was sensitive to TNT. All the denitrifiers isolated in this study were Gram-negative organisms, and TNT is more toxic to Gram-positive organisms, for reasons that are not yet clear [3,6,7]. In our study, Gram-positive organisms decreased by one order of magnitude in soils containing 17 400 mg TNT kg−1. Hence, the apparent resistance of denitrifiers is likely due to the preponderance of Gram-negative denitrifying organisms and not due to physiological benefits arising from denitrification.

The nirK genotype predominated over the nirS genotype in samples containing low TNT concentrations. This is in contrast to previous reports that nirS numerically predominates over nirK in soil [11,26]. Denitrifiers are typically isolated after some sort of enrichment process whereas in this study, denitrifiers were isolated based on hybridization to the nirS or nirK genes without any prior selective enrichment. Also, nirS isolates grew faster in broth culture and produced twice as much nitrous oxide as did nirK isolates, suggesting that nirS isolates are likely more competitive in enrichment cultures. What ecological advantage allows nirK isolates to dominate in the soil environment compared to nirS? There was little difference between the two genotypes in terms of TNT transformation but TNT had a greater effect on nirK isolate growth compared to that of nirS isolates. Perhaps the presence of TNT offsets whatever ecological advantage the slow growing nirK isolates normally enjoy over nirS organisms. For example, one organism, a nirS-positive, was highly efficient in TNT transformations but it is not known if the increase in nirS prevalence is due to this organism.

Our results clearly underline the importance of assessing the NaR+NiR+NOR as well as the N2OR portions of the denitrification pathway in the presence of toxicants. As noted by Richards and Knowles [9], using only the acetylene blockage technique masks effects of toxicants on the complete denitrification pathway. The results from this study have confirmed that TNT decreased the denitrification activity in field soils [3]. Furthermore, this effect occurred at very low TNT concentrations, i.e. an IC50 of only 26 mg kg−1. The importance of this inhibition to natural attenuation is not yet clear but indicates that anaerobic, non-fermentative, metabolism may be altered in the presence of TNT.


The authors would like to thank Dr. P. Gong and L. Paquet for providing the TNT concentration data and Dr. R. Knowles for reviewing this manuscript. Contribution 13307, National Research Council Canada.