OUP user menu

Silver (Ag+) reduces denitrification and induces enrichment of novel nirK genotypes in soil

Ingela Noredal Throbäck, Mats Johansson, Magnus Rosenquist, Mikael Pell, Mikael Hansson, Sara Hallin
DOI: http://dx.doi.org/10.1111/j.1574-6968.2007.00632.x 189-194 First published online: 1 May 2007


The use of silver ions in industry to prevent microbial growth is increasing and silver is a new and an overlooked heavy-metal contaminant in sewage sludge-amended soil. The denitrifying community was the model used to assess the dose-dependent effects of silver ions on microorganisms overtime in soil microcosms. Silver caused a sigmoid dose-dependent reduction in denitrification activity, and no recovery was observed during 90 days. Dentrifiers with nirK, which encodes the copper nitrite reductase, were targeted to estimate abundance and community composition for some of the concentrations. The nirK copy number decreased by the highest addition (100 mg Ag kg−1 soil), but the nirK diversity increased. Treatment-specific sequences not clustering with any deposited nirK sequences were found, indicating that silver induces enrichment of novel nirK denitrifiers.

  • denitrifying bacteria
  • heavy metals
  • microbial diversity
  • resilience
  • resistance
  • nitrite reductase


Silver ions are becoming an increasingly popular antimicrobial agent and silver therefore constitutes a new, and often overlooked, heavy-metal contaminant in sewage sludge-amended soil (Eriksson, 2001; Brandt et al., 2005). It was earlier shown to be the most toxic metal among 12 tested with respect to soil CO2-respiration (Cornfield, 1977). Nevertheless, microbial communities in soil have an inherent ability to resist disturbances from heavy metals, but also a capacity to recover from these. Microbial diversity has been proposed as an important factor determining resistance and resilience (Pankhurst et al., 1997), but studies that focus on the total microbial communities often produce ambiguous results (e.g. Bååth, 1989), probably due to a high degree of redundancy in the populations present. It has therefore been suggested that functional communities would be more suitable as indicators of toxic effects than the total community (Griffiths et al., 2000). Kandeler et al., (1996) concluded that heavy metals influenced enzymatic processes in nitrogen cycling more negatively than those in carbon cycling. Accordingly, the anaerobic respiration process denitrification, in which nitrate is reduced to nitrogen gas, was shown to be more sensitive to heavy metals than aerobic soil respiration (Bardgett et al., 1994). Furthermore, denitrification has been identified as being highly sensitive to silver in soil (Johansson et al., 1998). Therefore, the denitrifying community was the model used to assess the dose-dependent effects of silver on soil microorganisms overtime. Soil microcosms with 16 AgNO3 concentrations were incubated for 90 days, and denitrification activity was monitored to evaluate the resistance and capacity to recover after disturbance. Dentrifiers with nirK, which encodes the copper nitrite reductase, were targeted to estimate abundance and community composition.

Materials and methods

Microcosm experiment and denitrification activity assay

Soil (39% clay, 4.95% organic C, and 0.49% total N) with pH 7.8 was sampled from the top 0–20 cm layer of an arable field in Alunda, Uppsala, Sweden. It contained 2.4 mg Ag per kg dry soil. The soil was sieved (<4 mm), mixed and stored at −20°C. For the incubation, 19 jars were filled with 500 g of soil adjusted to 60% water-holding capacity. AgNO3 was added to 16 microcosms in amounts ranging from 0.003 to 100 mg Ag kg−1 dry weight (dw) soil to obtain 16 different concentrations of silver, while three remained nonamended and functioned as controls. Additional NO3 was added to all jars as 10 mL of a concentrated KNO3 solution that was thoroughly mixed with the soil to give a final concentration of 75 mg NO3 g−1 dw. The soil was incubated aerobically with closed lids for 90 days at 25°C. Twice a week, the jars were aerated and soil moisture content was adjusted. Samples of 80 g soil were withdrawn after 1, 14, 30 and 90 days.

Potential denitrification activity was measured in all samples according to Pell et al., (1996), while the rest of the analyses were performed using only the nonamended control from days 1 (C1) and 90 (C90), and in 90-day samples from the treatments with addition of 0.8 (Ag0.8), 12 (Ag12) and 100 (Ag100) mg Ag kg−1 dw soil. In an air-dried fraction of these samples, the amount of soluble silver was determined according to Maiz et al., (2000). The extractions and ICP-MS analyses were performed by the Swedish National Testing and Research Institute (Örebro, Sweden).

DNA extraction, nirK quantification and construction of nirK clone libraries

DNA was extracted in duplicate using the FastDNA® Spin Kit for soil (Qbiogene), according to the manufacturer's instructions, after mixing 300 mg soil 300 mg with 978 µL phosphate buffer and 122 µL MT buffer from the kit, using a blender. The nirK copy number, which is equivalent to the number of nirK genotypes (Philippot, 2002), in each extract was quantified in triplicate with real-time PCR according to Henry et al., (2004) in an ABI PRISM® 7000 thermal cycler. A standard curve was created using three 10-fold-dilution series of a linearized plasmid with a nirK insert, kindly provided by L. Philippot (INRA). One nirK clone library was constructed for each of the five samples by cloning PCR products amplified in triplicate during 28 cycles, previously described by Throbäck et al., (2004). Prior to cloning, the six amplicons for each sample were pooled and purified with the MiniElute Gel Extraction kit (Qiagen). Plasmids in transformed clones with the correct insert size were isolated using the QIAprep Spin Miniprep Kit (Qiagen).

Sequencing and sequence analysis

From each library, 100 clones were sequenced with the DYEnamic Et Terminator Cycle Sequencing kit (GE Healthcare) using the plasmid-specific primer M13F, and an ABI PRISM 377 automated DNA sequencer. Clones were grouped into operational taxonomic units (OTUs) at a level of >97% amino acid sequence identity. Library coverage (C) was estimated as C=1−nN−1, where n is the number of different OTU types encountered only once, and N is the total number of clones analyzed. The diversity of nirK genotypes was analyzed at a 95% confidence interval level using the analytic rarefaction software (http://www.uga.edu/~strata/software/Software.html). Derived amino acid sequences of nirK were aligned with translated amino acid sequences from GenBank using the clustal w software (http://www.ebi.ac.uk/clustalw/). Distance matrix analyses were performed with the Poisson correction and a Neighbor-Joining phylogram was constructed, with bootstrap values based on 100 replicates. Topology was corroborated by construction of cladistic trees (Maximum likelihood and Parsimony), based on both nucleotide and translated amino acid sequence alignments, using the paup4b10 software (Sinauer Associates Inc.). The 316 unique nirK clones of the 500 sequenced are deposited in GenBank with accession numbers DQ304119DQ304434.

Statistical analyses

The dose–response curves were calculated using a three-parameter logistic equation: Embedded Image where y is the denitrification rate (percent of nonamended control) at silver concentration c, max the highest rate, logEC50 the concentration of silver required to reduce the max by half and Hillslope the slope of the curve. Nonlinear regression analysis and influential outlier identifications, identified using standardized residuals and leverage, were carried out using sigmaplot software. Real-time PCR results were compared using an unpaired t-test (P<0.05). To elucidate the effect of silver on activity, nirK numbers and diversity, an integrated analysis of the data set from the four samples at day 90 (supplemental material) was performed by principal component analysis (PCA) using the unscrambler software. All variables were standardized to N(0,1) distribution.

Results and discussion

The immediate inhibitory effect of silver on denitrification activity was described by a sigmoidal dose–response curve, and <20% of the activity in the control remained in samples with the highest silver concentration (Fig. 1). A similar response has been described previously (Johansson et al., 1998). After 14, 30 and 90 days of incubation, the inhibition patterns were similar, and no recovery was observed (Fig. 1). The concentrations of silver having an effect on denitrification were within the range that occurs in soil after sewage sludge applications at the maximum recommended dose in Sweden (4000 kg dry matter ha−1 every fourth year; Swedish EPA). To compensate for the decreased activity in the nonamended soil caused by the incubation, the rates were expressed as percent of the control on each sampling day. The EC50 values showed no trend, and were 7.7, 10.4, 7.4 and 6.2 mg Ag kg−1 soil for days 1, 14, 30 and 90, respectively. The indigenous soil silver content (2.4 mg kg−1 soil) had possibly made the denitrifiers adapted to silver, which could partly explain a twice as high EC50 value day 1 as that reported in an earlier study (Johansson et al., 1998). However, only 0.0055 mg of the ions initially present originated from soluble, exchangeable and chelated silver, which is thought to be bioavailable according to Maiz et al., (2000). Even at the highest application rate, <0.05 mg Ag kg−1 soil was available after 90 days (supplemental material). Accordingly, Murata et al., (2005) demonstrated that although the exchangeable fraction was small compared with the added amount of silver, the effect on soil dehydrogenase activity was severe and bacterial colony growth was inhibited at levels between 0.1 and 0.5 mg Ag kg−1 soil. Early studies in freshwater indicated that microbial growth may be inhibited by silver at concentrations below those of other heavy metals (Albright & Wilson, 1974).

Figure 1

Potential denitrification activity in soil with different amounts of added AgNO3 after incubation for 1, 14, 30 and 90 days. The control was soil without addition of silver (mean±SD, n=3). The 95% confidence interval for the regression is indicated by the dotted lines. Identified outliers not included in the regression are shown as open circles.

In the integrated analysis, principal component 1 explained 87% of the total variation in the data obtained from the samples day 90 (Fig. 2; supplemental material). Samples with increasing silver addition followed the trajectory from left to right, explained by decreasing denitrification rates, nirK copy numbers and activity per nirK copy, and increased amounts of soluble Ag and number of OTUs. The trajectory curved downwards due to the discrepancy between nirK genotypes per ng DNA and per g dw soil, although the low information content in PC2 (8%) indicates that the difference is less significant. The specific activity, calculated from absolute denitrification rates and nirK copy numbers, ranged from 2 to 40 fg N2O-N cell−1 min−1 and corroborated with the values reported in other studies (Martin et al., 1988; Mahne & Tiedje, 1995; Etchebehere et al., 2001). They showed a tendency to be negatively affected by silver. This is supported by the similar specific activity determined in the control day 1 and day 90, despite that both activity and nirK abundance decreased in the control as an effect of the incubation itself (supplemental material). This indicates intact physiological properties of the nirK community in the controls during incubation in contrast to the communities in the soils with silver addition.

Figure 2

PCA of dentrification rate, nirK copy number, both per dry weight and per ng DNA, rate per nirK copy number, amount of soluble Ag and total number of OTUs in the nonamended soil (C90) and soil with different amounts of silver added (Ag0.8, Ag12 and Ag100) sampled on day 90. The loadings (of the variables) are superimposed on the scores (of the samples) to show the influence of the variables on the samples. Axes are scaled to display the relative information content.

The rarefaction curves for each library did not reach a plateau, indicating that denitrifier diversity was high in all soil samples (Fig. 3). As calculated, 63%, 81%, 67%, 76% and 49% of the diversity was captured in C1, C90, Ag0.8, Ag12 and Ag100, respectively. Therefore, only differences in dominant nirK genotypes were detected among the 500 clones, which all had an insert showing homology with nirK. Elevated concentrations of silver resulted in increased nirK diversity (Figs 2 and 3). Although the most commonly reported effect of heavy metals is decreased genetic diversity (Kozdroj, 2001; Muller et al., 2001; Moffett et al., 2003), others have observed similar results as found in this study in soil (Sandaa et al., 2001; Ranjard et al., 2006). Giller et al., (1998) explained increased diversity in heavy metal-contaminated soils with the intermediate disturbance hypothesis. It postulates that environments with high numbers of competitive species have increased diversity because metal stress reduces the innate competitive exclusion between bacterial populations and induces enrichment of others. In the phylogram, a unique cluster containing 26% of the clones specific for the Ag-libraries, and some common for all libraries, was observed (Fig. 4). These clones did not relate to any previously reported nirK sequences, suggesting that unknown silver-tolerant denitrifier populations had started to establish in the microcosms. In addition to growth of tolerant denitrifiers, the observed increased nirK gene diversity can also originate from PCR amplification of DNA from nonviable cells still present due to toxification of microorganisms involved in its degradation. Nevertheless, the novel nirK denitrifiers showing up after silver treatment warrants further exploration as heavy-metal tolerance of soil denitrifiers is connected to N2O dynamics (Holtan-Hartwig et al., 2002).

Figure 3

Rarefaction curves with 95% confidence intervals showing numbers of partial nirK sequences in the clone libraries from the unamended soil (C90), and soil with different amounts of silver added (Ag0.8, Ag12 and Ag100) sampled on day 90. The solid line shows the maximum possible diversity that could be observed in a clone library (i.e. each clone being a different OTU).

Figure 4

Neighbor-Joining phylogram of nirK genes translated into amino acid sequences (158 aa). The scale bar indicates 10% nucleotide substitution. Bootstrap values greater than 70 from 100 replicate trees are reported at the nodes. The clones from this study are framed and shaded in gray, and sequences specific for a library are marked with an asterisk (*). The early branching Ag-cluster is shaded in dark gray. The sequences of aniA from Neisseria meningitides Z2491 (Accession no. AL162757), panI from Neisseria gonorrhoea (Accession no. M97926), omp from Aspergillus fumigatus Af293 (Accession no. XM_754129), nir1 from Ajellomyces capsulatus (Accession no. AY816318), sufI from Mannheimia succiniciproducens MBEL55E (Accession no. AE016827) and omp from Bdellovibrio bacteriovorus HD100 (Accessions no. BX842653) served as the out-group to root the phylogram.

Supplementary material

The following supplementary material is available for this article:

Table S1. Data set for soil samples analyzed day 90 and the non-amended control from day 1.


The authors thank M. Tångring for helping with the clone libraries. Grants were provided by The Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (Formas).


  • Editor: Elizabeth Baggs


View Abstract