Development and application of a PCR-denaturing gradient gel electrophoresis tool to study the diversity of Nitrobacter-like nxrA sequences in soil


  • Sophie Wertz,

    1. Laboratoire d'Ecologie Microbienne, Université de Lyon, Université Lyon 1, Villeurbanne, France
    2. CNRS, UMR 5557 Laboratoire d'Ecologie Microbienne, Villeurbanne, France
    3. INRA, USC 1193, Laboratoire d'Ecologie Microbienne, Villeurbanne, France
    Search for more papers by this author
  • Franck Poly,

    1. Laboratoire d'Ecologie Microbienne, Université de Lyon, Université Lyon 1, Villeurbanne, France
    2. CNRS, UMR 5557 Laboratoire d'Ecologie Microbienne, Villeurbanne, France
    3. INRA, USC 1193, Laboratoire d'Ecologie Microbienne, Villeurbanne, France
    Search for more papers by this author
  • Xavier Le Roux,

    1. Laboratoire d'Ecologie Microbienne, Université de Lyon, Université Lyon 1, Villeurbanne, France
    2. CNRS, UMR 5557 Laboratoire d'Ecologie Microbienne, Villeurbanne, France
    3. INRA, USC 1193, Laboratoire d'Ecologie Microbienne, Villeurbanne, France
    Search for more papers by this author
  • Valérie Degrange

    1. Laboratoire d'Ecologie Microbienne, Université de Lyon, Université Lyon 1, Villeurbanne, France
    2. CNRS, UMR 5557 Laboratoire d'Ecologie Microbienne, Villeurbanne, France
    3. INRA, USC 1193, Laboratoire d'Ecologie Microbienne, Villeurbanne, France
    Search for more papers by this author

  • Present address: Sophie Wertz, Department of Biology, University of New Brunswick, Bag Service 45111, Fredericton, New Brunswick, E3B 6E1, Canada.

  • Editor: Riks Laanbroek

Correspondence: Franck Poly, 16 rue Dubois, Villeurbanne 69100, France. Tel.: +33 4 72 43 13 79; fax: +33 4 26 23 44 68; e-mail:


A new PCR–denaturing gel gradient electrophoresis (DGGE) tool based on the functional gene nxrA encoding the catalytic subunit of the nitrite oxidoreductase in nitrite-oxidizing bacteria (NOB) has been developed. The first aim was to determine if the primers could target representatives of NOB genera: Nitrococcus and Nitrospira. The primers successfully amplified nxrA gene sequences from Nitrococcus mobilis, but not from Nitrospira marina. The second aim was to develop a PCR-DGGE tool to characterize NOB community structure on the basis of Nitrobacter-like partial nrxA gene sequences (Nb-nxrA). We tested (1) the ability of this tool to discriminate between Nitrobacter strains, and (2) its ability to reveal changes in the community structure of NOB harbouring Nb-nrxA sequences induced by light grazing or intensive grazing in grassland soils. The DGGE profiles clearly differed between the four Nitrobacter strains tested. Differences in the structure of NOB community were revealed between grazing regimes. Phylogenetic analysis of the sequences corresponding to different DGGE bands showed that Nb-nxrA sequences did not group in management-specific clusters. Most of the nxrA sequences obtained from soils differed from nxrA sequences of NOB strains. Along with existing tools for characterizing the community structure of nitrifiers, this new approach is a significant step forward to performing comprehensive studies on nitrification.


Nitrification, the conversion of NH4+ to NO3 via NO2, is a key step for nitrogen cycling, which largely determines soil fertility (including the balance between ammonium and nitrate) and influences N oxide emission from soils. This process is carried out by two different and specialized bacterial groups: ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB).

In recent years, most of the ecological studies dealing with the impacts of environmental factors and anthropogenic disturbances on the activity and diversity of nitrifiers have focused on AOB (Phillips et al., 2000; Kowalchuk & Stephen, 2001; Avrahami et al., 2003; Patra et al., 2005; Webster et al., 2005). Indeed, the first step of nitrification is often assumed to be rate-limiting (Prosser, 1989). Furthermore, molecular tools for investigating the diversity of AOB based either on the 16S rRNA gene or the functional gene amoA have long existed (Kowalchuk et al., 1997; Rotthauwe et al., 1997; Nicolaisen & Ramsing, 2002), whereas no equivalent tool was available to study the diversity of NOB until recently.

The molecular approaches developed and used to assess NOB diversity to date were based on 16S rRNA gene sequences, targeting members of the genera Nitrobacter or Nitrospira (Navarro et al., 1992; Daims et al., 2000; Grundmann & Normand, 2000; Freitag et al., 2005; Siripong & Rittmann, 2007). In contrast to AOB, which are restricted almost exclusively to a monophyletic group, NOB are distributed more widely, among the Alphaproteobacteria, Gammaproteobacteria and Deltaproteobacteria and phylum Nitrospira for Nitrobacter, Nitrococcus, Nitrospina and Nitrospira, respectively (Orso et al., 1994; Teske et al., 1994; Ehrich et al., 1995). Due to this polyphyletic distribution, the use of a different set of 16S primers for each NOB genus would be necessary. Moreover, characterization of NOB diversity among the genus Nitrobacter is difficult with a set of 16S primers owing to the close similarity of 16S sequences among these organisms (Orso et al., 1994). Recently, Vanparys et al. (2007) designed primers targeting 16S and nxr genes for phylogenetic analysis of Nitrobacter strains. In addition, to explore Nitrobacter diversity in soils, Poly et al. (2008) developed a PCR cloning/sequencing approach targeting the functional gene nxrA encoding the catalytic subunit of nitrite oxidoreductase (NXR), the key enzyme in nitrite oxidation (Bock et al., 1986). No fingerprint-like tool targeting this functional gene has been developed thus far to investigate the diversity of NOB belonging to the genus Nitrobacter and perhaps also to other NOB genera, although such a tool would be more suitable for comparing community structure between a high number of environmental samples or treatments than would cloning/sequencing approaches.

The four objectives of the present study were:

  • 1To test the ability of the nrxA primers to target different phylogenetic groups of NOB.
  • 2To develop and test a fingerprint-type approach, here denaturing gradient gel electrophoresis (DGGE), targeting the functional gene nxrA.
  • 3To test the capacity of the PCR-DGGE approach to detect modifications in the diversity of nxrA gene sequences induced by different management regimes (intensive vs. light grazing by sheep) in a grassland ecosystem. Above-ground grazing regime is likely to influence soil NOB community structure, because grazers strongly modify key environmental parameters for NOB. In particular, grazers increase the availability of labile forms of nitrogen (by input of urine and dung) and labile organic substrates (by increasing root/shoot allocation and root exudation) (Patra et al., 2005, 2006).
  • 4To provide a phylogenetic analysis of partial nrxA gene sequences corresponding to different bands observed in DGGE profiles of strains and soils.

Materials and methods

Nitrite-oxidizing strains

Nitrobacter winogradskyi strain AG (ATCC 14123), Nitrobacter alkalicus strain AN4 (Sorokin et al., 1998), Nitrospira marina (ATCC 43039) and Nitrococcus mobilis Nb 231 (ATCC 25380) were cultivated in an autotrophic medium with 2 g NaNO2 L−1 (Schmidt et al., 1973). Nitrobacter hamburgensis strain X14 (Bock et al., 1983) and Nitrobacter vulgaris strain Z (Bock et al., 1990) were cultured in a mixotrophic medium with 2 g NaNO2 L−1, 1.5 g peptone L−1, 1.5 g yeast extract L−1 and 0.55 g Na pyruvate L−1 (Bock et al., 1983).

Grassland sites and soil sampling

Soil was sampled in a seminatural grassland at Theix (45°43′N, 3°1′E), France (for soil characteristics, see Le Roux et al., 2003). Before establishment of the study sites in 1989, the grassland had experienced moderate grazing pressure by sheep for more than 35 years. In 1989, two sites (500 m2 each) were fenced at an upslope location and two sites (500 m2 each) at a downslope location along a topographical transect. Two grazing regimes have been prescribed for 13 years on the sites (Patra et al., 2005): ewes were allowed to graze once (light grazing, LG) or four times (intensive grazing, IG) per year. For each grazing event, eight ewes per plot were allowed to graze until the mean height of the sward reached 6 cm in the IG plot (typically a few days). Ewes were kept permanently in the enclosures, allowing redistribution of N to the soil as labile forms in urine and dung. In addition, the IG plots experienced one mowing event each June, which mimicked some export by herbivores without redistribution to the soil. No fertilizer was used in any treatment. The plant species composition on each plot after 13 years is presented by Le Roux et al. (2003). Nine soil samples (0–8 cm layer, 8 cm diameter) were taken randomly on each site, after removing the litter layer. The 36 fresh soil samples were then sieved (2 mm mesh) and stored at −20 °C.

DNA extraction

DNA was extracted from pure cultures using a DNeasy tissue kit (Qiagen). For the soil samples, DNA was extracted from 0.5 g (equivalent dry mass) sieved and frozen soil using the FastDNA spin kit for soil (BIO 101, Qbiogene, Carlsbad, CA).

Primers and PCR conditions

The nxrA primers from Poly et al. (2008) that have been shown to allow amplification of a 322-bp fragment of the nxrA gene in Nitrobacter were modified by eliminating the degeneracy of the reverse primer to avoid DGGE multiple band patterns while not affecting the amplified targeted sequences: F1370 F1 nxrA (5′-CAGACCGACGTGTGCGAAAG-3′) and F2843 R2 nxrA (5′-TCCACAAGGAACGGAAGGTC-3′). Because of the low abundance of NOB in soils and the difficulty of annealing using primers with GC clamp, amplification of nxrA gene fragments was achieved in two PCR steps for the PCR-DGGE approach. Such a procedure has sometimes been required in DGGE or restriction fragment length polymorphism approaches for PCR amplification of sequences corresponding to other functional groups with low abundances in the environment, for example for some studies on ammonia-oxidizing and nitrogen-fixing bacteria via the amplification of fragments of the functional genes amoA (Nicolaisen & Ramsing, 2002) and nifH (Bürgmann et al., 2005), and for ammonia-oxidizing, nitrite-oxidizing and methanotrophic bacteria via the amplification of 16S rRNA gene fragments (Freitag & Prosser, 2003; DeJournett et al., 2007). The first amplification was done using 20 ng of DNA with nxrA primers. To avoid nonspecific amplification in the second PCR step, products of the first PCR were loaded on 2% agarose gels, and gel slices containing the nxrA PCR products were excised, crushed in 200 μL of sterile water and incubated at 4 °C overnight to elute the DNA. The second PCR was then carried out from the eluted material with the same primers containing a GC clamp (Muyzer et al., 1993) added at the 5′ end of the forward primer (F2842 F1 nxrAgc, CGCCCGCCGCGCGCGGCGGGCGGGGCGGGGGCACGGGGGGCAGACCGACGTGTGCGAAAG). The final reagent concentrations for the two PCR steps were 0.4 μM primers, 200 μM of each dNTP, 1.75 U of Taq (Qbiogene, Carlsbad) in 50 μL of 10 mM Tris-HCl, 50 mM KCl, 0.1% Triton X-100, 1.5 mM MgCl2, pH 9. Thermocycling conditions for the two PCR steps were: 3 min at 94 °C followed by 35 cycles of 94 °C for 30 s, annealing at 55 °C for 45 s and elongation at 72 °C for 45 s with terminal elongation at 72 °C for 5 min.

DGGE conditions

DGGE analysis of PCR products was carried out using the D-Code Universal Mutation Detection System (Bio-Rad, Marnes la Coquette, France). PCR products were loaded on 5.5% polyacrylamide gels containing a gradient of 39–61% denaturant, 100% denaturing solution being defined as 7 M urea and 40% formamide. Gels were run for 16 h at 75 V in 1 × TAE buffer at 60 °C. Gels were stained with SYBR Green (Roche, Meylan, France) and then photographed with a UV source system (Fischer Biolock Scientific, Illkirch, France) using the software biocaptmw (Vilber-Lourmat, France).

Cloning strategy

To characterize amplified nxrA sequences corresponding to major DGGE bands observed for soil samples, clone libraries were generated from nxrA PCR products obtained from two soil samples from the upslope location: one sample for LG treatment and one sample for IG treatment. Cloning was carried out using the pGEM T-Easy vector system (Promega Ltd, Southampton, UK) and JM109 supercompetent Escherichia coli cells (Stratagene Inc., Maidstone, UK). We randomly chose 100 clones for the LG soil sample and 100 clones for the IG soil sample. The clones were subjected to PCR, and the PCR products were screened by DGGE (as described above). Clones were grouped on the basis of their migration distance on the DGGE gels. One to eight representatives of each clone group were selected for sequencing (Genome Express, Meylan, France). Sequences showing ≥99% similarity were not considered as different due to methodological uncertainties linked to the PCR cloning/sequencing procedure (Acinas et al., 2005).

In order to characterize amplified nxrA sequences of Nitrococcus mobilis and compare them with known sequences from the genome database, a clone library was generated from nxrA amplicons obtained from this organism. Cloning was performed as described above and eight clones were sequenced. The nxrA sequences corresponding to the different bands in the DGGE profiles of the Nitrobacter strains X14, Z, AN4 and AG were determined by cloning of nxrA PCR products of these strains and then loading them on DGGE gels.

Phylogenetic sequence analysis

All sequences were aligned with partial nxrA sequences of cultured NOB available in Genbank: Nitrobacter strains X14, Z, AG, AN4, Nb-255 and Nb-311A and Nitrococcus mobilis Nb-231 (Starkenburg et al., 2006; Poly et al., 2008; and with partial sequences of narG genes of nitrate-reducing strains (encoding the catalytic subunit of the nitrate reductase) that are known to be close to nxrA gene sequences (Kirstein & Bock, 1993; Poly et al., 2008) (Table 1) using clustal x 1.81 (EMBL, Heidelberg, Germany). The length of the sequences compared was 282 bp, i.e. 322 bp (amplicons) minus 40 bp (primers). The nxrA and narG nucleotide sequences were translated to amino acid sequences using BioEdit ( Phylogenetic analysis was performed based on three treeing methods: (1) parsimony, (2) neighbour joining with HKY model of substitution (using 1000 bootstraps) for nucleotide sequences and with PAM distance for amino acid sequences (100 bootstraps), and (3) maximum likelihood, using four categories of substitutions with an estimated γ distribution parameter and HKY model of substitution for nucleotide sequences and WAG model for amino acid sequences. The first two methods were performed with phylowin software ( (Galtier et al., 1996), while the third method was performed with phyml software ( (Guindon & Gascuel, 2003). A consensus tree was then computed with Consense in the phylip package 3.66 ( (Felsenstein, 1993). The consensus tree was visualized using nj plot ( (Perriere & Gouy, 1996). Sequences determined in this study have been deposited in the GenBank database (Table 1).

Table 1.   GenBank accession numbers of partial nxrA sequences from NOB strains and soils and partial narG sequences of nitrate-reducing strains represented in Figs 2 and 5
Partial sequencesAccession numbers
 Nitrobacter hamburgensis X14: clones 1–3AF344872, DQ421368 and DQ421367
 Nitrobacter vulgaris Z: clones 1 and 2AF344875 and DQ421377
 Nitrobacter alkalicus AN4: clones 1 and 2AF344873 and DQ421372
 Nitrobacter winogradskyi AG: clones 1–3AF344874, DQ421373 and DQ421374
 Nitrococcus mobilis Nb 231: clone 1DQ514250
From genome sequences
 Nitrobacter sp. Nb.-311AAAMY01000022.1
 Nitrobacter winogradskyi Nb-255CP000115.1
 Nitrobacter hamburgensis X14NC_007964.1 and CP000319.1
 Nitrococcus mobilis Nb 231AA0F01000001 and AA0F01000011
 Soil under light grazing regime (LG)DQ514257DQ514274
 Soil under intensive grazing regime (IG)DQ514275DQ514293
 Geobacter metallireducens GS15CP000148
 Geobacter lovleyi SZAAVG01000003 and AAVG01000020
 Thermus thermophilus HB8Y10124
 Anaeromyxobacter sp. FW109-5AAWM01000006
 Pseudomonas aeruginosa PA01AE004091.2
 Mycobacterium tuberculosis CDC1551AE000516.2
 Staphylococcus carnosus TM300AF029224.3
 Aeropyrum pernix K1BA000002.3

DGGE profile analysis

DGGE banding profiles from all soil samples were analysed as follows. Two types of data matrices were constructed using gel comparii software (Applied Maths, Kortrijk, Belgium) and primer-e ldt software (Plymouth, UK): a binary matrix consisting of the presence/absence and relative position of each DNA band for all soil samples, and a matrix consisting of the relative intensity and position of each DNA band, total band intensity being normalized among samples. Matrices of similarity (Bray–Curtis coefficient) were calculated using primer-e ldt software. Then, similarity ranking of community structure between samples was represented by multi-dimensional scaling (MDS) (Kruskal & Wish, 1978). Distortion between rank similarity matrices and MDS representation was assessed by a stress factor. Three-dimensional MDSs were chosen so that the stress factor was sufficiently low.

In the present study, a pseudoreplicated experimental design was used to test the grazing treatment effect on NOB community structure. Such a design is commonly used in ecology (e.g. Oksanen, 2004) provided that the limits of applicability of the results are recognized, i.e. we do not test here general patterns about grazing effect sensu stricto, but rather test patterns observed at particular sites and the ability of the new DGGE tool to discriminate community structure between treatments at these sites. Grazing effect on nitrification and the structure of the ammonia-oxidizing community was tested for the same sites in previous studies (Patra et al., 2005, 2006). We tested the effect of grazing treatments on NOB community structure using one-way analysis of similarity (anosim).


Analysis of partial nxrA sequences of cultured NOB

Amplification of 322-bp nxrA fragments was obtained for five of the six NOB strains tested, belonging to the four different species of Nitrobacter and to Nitrococcus mobilis, whereas no amplification was obtained for Nitrospira marina. However, comparison of amplified sequences from Nitrococcus mobilis with nxrA sequences known from genome data showed that only one of the two copies of the nxrA gene was successfully amplified. The four Nitrobacter strains for which PCR amplifications were successful were characterized by distinct DGGE profiles of nxrA fragments (Fig. 1). With the exception of Nitrobacter alkalicus AN4, DGGE profiles of each strain were composed of several bands: three bands for Nitrobacter hamburgensis X14, and two bands for Nitrobacter vulgaris Z and for Nitrobacter winogradskyi AG (Fig. 1).

Figure 1.

 DGGE analysis of nxrA sequences amplified by PCR from the NOB strains Nitrobacter hamburgensis X14 (Nb X14), Nitrobacter vulgaris Z (Nb Z), Nitrobacter alkalicus AN4 (Nb AN4) and Nitrobacter winogradskyi AG (Nb AG). The ladder consisted of reference sequences for ammonia-oxidizers (Mahmood & Prosser, 2006). The clone numbers correspond to the partial nxrA sequences presented in Fig. 2.

Phylogenetic analysis based on either nucleotide sequences or amino acid sequences showed that partial nxrA sequences corresponding to the different bands in the DGGE profiles of Nitrobacter strains were grouped in a cluster separated from narG sequences (Fig. 2 and supplementary Fig. S1). The sequence retrieved from Nitrococcus was distant from the other nxrA sequences (Fig. 2 and supplementary Fig. S1): this sequence showed 77% similarity to the sequence of Nitrobacter sp. Nb 311A and with the two sequences derived from Nitrobacter alkalicus, the three most similar nxrA gene sequences. It was closer to narG sequences (similarity of 72%, 76% and 73%, respectively, to narG sequences of Geobacter metallireducens) than to the other nxrA sequences. For each Nitrobacter species, the different copies of partial nxrA nucleotide sequences were grouped together in the phylogenetic tree (Fig. 2). However, phylogenetic analysis based on amino acid sequences showed no strain-specific clustering of nxrA sequences (supplementary Fig. S1).

Figure 2.

 Consensus phylogenetic tree based on nucleotide sequences of nxrA sequences from Nitrobacter hamburgensis X14, Nitrobacter vulgaris Z, Nitrobacter alkalicus AN4, Nitrobacter winogradskyi AG, Nitrobacter sp. Nb-311A, Nitrobacter winogradskyi Nb-255 and Nitrococcus mobilis Nb 231 in relation to narG sequences of nitrate-reducers. Sequences corresponding to DGGE bands are in bold. Because the consensus tree was consistent with the tree obtained with the neighbour-joining method, bootstrap values >50% (1000 replicates) from the neighbour-joining tree are indicated.

Identical DGGE migration distances but differences in partial nxrA sequences were found respectively between clones 1 and 2 for Nitrobacter alkalicus AN4 and between clones 1 and 2 for Nitrobacter winogradskyi AG (Figs 1 and 2). For each sequence pair, the two sequences showed 98.1% similarity.

Community structure of NOB in soils on the basis of Nitrobacter-like nrxA sequences

NxrA DGGE profiles for soil under LG and IG were successfully obtained. An example of DGGE profiles obtained for some soil samples is presented in Fig. 3. The number of bands in the DGGE profiles ranged from 16 to 25 per sample and did not differ significantly between LG and IG samples (data not shown).

Figure 3.

 Example of DGGE profiles of nxrA sequences amplified from soil DNA for some samples under LG or IG, at downslope (D) and upslope (U) locations. The ladder is as in Fig. 1.

Because nxrA DGGE profiles of cultured Nitrobacter strains were generally composed of several bands, we tested the occurrence of possible autocorrelation between bands within DGGE profiles obtained from soils. Simple regression analyses between the most abundant and discriminating bands (i.e. 27 bands among the 56 different bands detected) of the DGGE profiles of soils under LG and IG treatments showed that these bands were either not significantly correlated or weakly correlated to each other (R2 values of 0.001–0.37, except for one pair of bands for which R2=0.49, data not shown).

Comparison of community structures between IG and LG soils from nxrA DGGE profiles was carried out by taking into account (1) in a first analysis the relative intensities of the different bands (Fig. 4), and (2) in a second analysis the presence/absence of bands (data not shown). Both analyses revealed that above-ground grazing regime significantly modified the genetic structure of the soil Nitrobacter-like NOB community at both the downslope location (P=0.003, R=0.4, and P=0.005, R=0.3 for analysis with relative band intensities and with presence/absence of bands, respectively) and upslope location (P=0.01, R=0.2 and P=0.013, R=0.2 for analysis with relative band intensities and with presence/absence of bands, respectively).

Figure 4.

 MDS ordination of similarity ranks between the genetic structures of the NOB community of soils under (▵) LG treatment, and (▴) IG treatment at two topographical locations (a, downslope; b, upslope). The MDS is represented in three dimensions. The more similar the genetic structures, the closer the points. The stress (distortion factor between actual similarity rankings and the corresponding distance rankings in the ordination plot) is indicated.

Phylogeny of partial Nitrobacter-like nxrA gene sequences from soils

When considering either nucleotide sequences or amino acid sequences, the partial nxrA sequences retrieved from the two soil samples tested (one under LG and one under IG treatments) clustered with nxrA sequences of cultured NOB apart from narG sequences (Fig. 5 and Supplementary Fig. S2). Both analyses based on nucleotide and amino acid sequences showed that, inside the cluster of nxrA sequences, the majority of partial nxrA sequences obtained from soils grouped together, and did not cluster with the sequences of cultured nitrite-oxidizers (Fig. 5, supplementary Fig. S2). When several clones showing identical DGGE migration distance were sequenced, a unique sequence was obtained for 25% of the clone groups tested, whereas two and three different sequences were obtained for 60% and 15% of the clone groups tested, respectively (Fig. 5). No management-specific clusters of partial nxrA gene sequences belonging to either IG or LG treatments were observed (Fig. 5, supplementary Fig. S2).

Figure 5.

 Consensus phylogenetic tree based on nucleotide sequences of soil nxrA sequences in relation to nxrA sequences of cultivable NOB and narG sequences of nitrate-reducers. The names of NOB strain sequences are as in Fig. 2. Sequence names are prefixed ‘LG’ and ‘IG’ when they were retrived from soil samples under LG and IG, respectively. Different letters following LG or IG indicate clones with different DGGE migration patterns. Numbers are added after the letters for clones with identical DGGE migration distance but with different sequences. Because the consensus tree was consistent with the tree obtained with the neighbour-joining method, bootstrap values >50% (1000 replicates) from the neighbour-joining tree are indicated.


Primer specificity and ability to target different phylogenetic groups of NOB

High levels of similarity have been reported between nxrA sequences of Nitrobacter strains and narG sequences of nitrite-reducing bacteria (Kirstein & Bock, 1993; Poly et al., 2008). Here, we have shown that nxrA sequences retrieved from NOB strains and soils were grouped together apart from narG sequences, which highlights the good specificity of our PCR-DGGE tool. Furthermore, our primers were recently also tested on 10 other Nitrobacter strains obtained from environmental samples (Navarro et al., 1992), and amplification was obtained for each strain (unpublished results), which suggests good sensitivity of the primers to Nitrobacter.

To date, four phylogenetic groups of NOB, each corresponding to a genus, have been described, i.e. Nitrobacter, Nitrospira, Nitrococcus and Nitrospina. Nitrobacter is to date the only genus isolated from soils (Pan, 1971; Bock et al., 1983), where Nitrobacter 16S rRNA gene sequences have been commonly found (Degrange & Bardin, 1995; Freitag et al., 2005). However, according to recent studies, Nitrospira-like 16S rRNA gene sequences can been detected in soils. (Freitag et al., 2005), and therefore genera other than Nitrobacter are likely to be present in soils.

Poly et al. (2008) designed primers based on the functional gene nxrA encoding the catalytic subunit of the NXR of nitrite-oxidizers that allow amplification of nxrA sequences from cultured Nitrobacter strains. Several nxrA sequences of Nitrobacter species and two nxrA sequences retrieved from the genome of Nitrococcus mobilis currently sequenced are available (Starkenburg et al., 2006; Poly et al., 2008). Sequencing of the genomes of Nitrospira moscoviensis and Nitrospina defluvii (H. Daims, pers. commun.) is in progress but no sequences are available to date. Here, we show that the primers we used allow amplification of nxrA sequences not only from Nitrobacter strains but also from one nxrA copy of Nitrococcus mobilis, demonstrating that our tool may target NOB genera other than Nitrobacter when applied to environmental samples. However, no PCR amplification could be obtained for Nitrospira marina. This is in accordance with priliminary results from the sequencing of the genomes of Nitrospira moscoviensis and Nitrospira defluvii, showing important differences between nxrA sequences of Nitrobacter strains and Nitrospira (H. Daims, pers. commun.). This suggests that primers defined from nrxA sequences of Nitrobacter strains would not be suitable to target nxrA from Nitrospira-like NOB. In that case, the use of a different set of primers would be required for characterization of Nitrospira-like nxrA sequences.

Discrimination of cultivated Nitrobacter strains with the DGGE tool

The DGGE profiles allowed discrimination of strains belonging to the four different species of Nitrobacter. With the exception of Nitrobacter alkalicus AN4, the DGGE profiles of the strains were composed of several bands. Furthermore, two different sequences corresponded to the DGGE band of Nitrobacter alkalicus AN4, and to one of the DGGE bands of Nitrobacter winogradskyi AG. Our results thus show that Nitrobacter strains harbour two to three different nxrA sequences, which strongly suggests that the presence of several nxr operons per cell is a common feature in NOB. This is consistent with genome sequencing data for Nitrobacter strains (three and two nxr operons were identified from the genomes of Nitrobacter hamburgensis X14 and Nitrobacter winogradskyi Nb-255, respectively) (, Starkenburg et al., 2006). The presence of several copies of functional genes per cell has been revealed in other microbial functional groups, e.g. amoA and amoB genes of AOB (Norton et al., 1996; Klots et al., 1997; Hommes et al., 1998) and nifH genes of N2-fixing bacteria (Raymond et al., 2004).

Methodological issues when assessing NOB community structure in soils with the DGGE tool targeting Nitrobacter-like nrxA

The results show that our primers are probably suitable to characterize the structure of one important group of NOB, namely Nitrobacter-like NOB. The first results from sequencing of the genomes of Nitrospira moscoviensis and Nitrospira defluvii suggest important differences between nxrA sequences of Nitrobacter and Nitrospira (H. Daims, pers. commun.). Thus, the use of a second set of primers targeting nxrA of the other important group of NOB assumed to be important in soil, namely the phylum Nitrospira, will be required rather than modification of the current primers. Similarly, at least two sets of primers are commonly used to characterize the genetic structure of the nitrite-reducing bacteria community, owing to the low similarity of nirK and nirS genes (Hallin et al., 2006). Hence, the present work is a significant step forward in characterizing NOB diversity in natural environments that will need to be completed when the first nxrA sequences retrieved from organisms of the phylum Nitrospira become available.

Because NOB strains can harbour several nxrA sequences giving different DGGE bands, the relative intensities of some bands observed on DGGE profiles of nxrA amplicons from soils could be correlated if those bands are mainly explained by the presence of one major population. This could lead, for instance, to misevaluation of a treatment effect. In our study, the lack of (or weak) correlations observed between the relative intensities of the different DGGE bands for soil samples suggests that this was not a major problem when testing the management effect on the genetic structure of a soil NOB community with the Nitrobacter-like nrxA PCR-DGGE tool.

As observed with other DGGE tools targeting sequences of other organisms (Kowalchuk et al., 1998; Nicolaisen & Ramsing, 2002), an identical DGGE migration distance did not always imply sequence identity. Here, two or three different sequences were obtained for 75% of the tested clone groups derived from soil. This indicates a limit for DGGE tools in assessing the diversity of the targeted community in terms of richness or evenness. Furthermore, owing to the low abundance of NOB in soil, two PCR amplifications had to be used, which could also restrict the ability of the DGGE tool to evaluate community diversity per se. However, our tool, as with other DGGE tools, will be very useful in comparing the genetic structure of the targeted community between a large number of environmental samples (see Discussion in Kowalchuk et al., 1998).

Evaluation of community structure of NOB harbouring Nitrobacter-like nxrA in grassland soils and effect of grazing regime

Analysis of nxrA DGGE profiles between LG and IG plots showed that the genetic structure of the Nitrobacter-like NOB community differed between grazing regime at both upslope and downslope locations. Freitag et al. (2005) also demonstrated, using a molecular approach based on 16S rRNA genes of NOB, that N fertilizer management practices influenced the NOB community structure in agricultural grassland soils. On our study sites, it has been shown previously that grazing regime increased ammonium concentration and enhanced nitrification (Patra et al., 2005, 2006). Although nitrite concentrations in the study soils were below the detection limit, the increase of ammonium concentration and overall nitrification are likely to imply enhanced nitrite availability for nitrite-oxidizers. This could explain changes in Nitrobacter-like NOB community structure because the sensitivity to nitrite availability differs among NOB (Maixner et al., 2006). In addition, grazing can modify the availability of carbon substrates, which can also influence community structure because some NOB are mixotrophs.

Intensive grazing has been shown to promote nitrification activity and, concurrently, to induce modifications in the genetic structure of the ammonia-oxidizing community at the same study sites (Patra et al., 2005). Our results show that anthropogenic disturbances such as agricultural practices inducing changes in nitrification lead to changes not only in ammonia-oxidizing community structure but also in NOB community structure. Thus, characterizing concurrently both community structures is important for a comprehensive understanding of nitrification in soil.

Most nxrA sequences retrieved from soils differed from the sequences of cultured NOB strains. This was expected because very few NOB strains have been isolated to date, so that few nxrA sequences of NOB strains are available. Furthermore, these strains were isolated from a few particular environments: for instance Nitrobacter hamburgensis X14 was isolated from one soil in Hamburg (Bock et al., 1983), Nitrobacter alkalicus AN4 from a soda lake (Sorokin et al., 1998) and Nitrococcus mobilis from seawater (Watson & Waterbury, 1971). The number of different Nitrobacter-like nxrA sequences retrieved for the two soil samples analysed and the fact that the majority of these sequences were different from the sequences of cultured NOB strains suggest that the diversity of NOB in soils may be higher than often assumed.

In conclusion, we have successfully developed a PCR-DGGE tool to study the diversity of Nitrobacter-like nxrA partial sequences from environmental samples. Using this new PCR-DGGE tool, we showed that different grazing regimes induced changes in Nitrobacter-like NOB community structure at the study sites. Along with molecular tools avalaible for AOB, nrxA DGGE will be useful when performing comprehensive studies of nitrification, which will require assessment of the structures of both ammonia-oxidizing and nitrite-oxidizing communities.


We thank the ‘Bureau des Ressources Génétiques’ BRG, the ‘Institut National de la Recherche Agronomique’ INRA, and the ‘Institut Fédératif de Recherche’ IFR41 for financial support. S.W. acknowledges funding of a postgraduate studentship by the French Ministry of Research. We thank T. Freitag for training with the DGGE technique, J. Prosser and H. Daims for helpful comments and suggestions, E. Spieck and D. Sorokin for providing the Nitrobacter alkalicus strain, and F. Louault and P. Loiseau for management of field sites and help during soil sampling.