INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Chemolithotrophic oxidation of ammonia to nitrate via nitrate (autotrophic nitrification) is of major importance in the global cycling of nitrogen in terrestrial, aquatic, and marine ecosystems (22). The first, and rate-determining, step of nitrification, ammonia oxidation, is carried out by the autotrophic ammonia-oxidizing bacteria, whose growth in liquid batch culture rarely occurs at pH values below 6.5. Nevertheless, autotrophic nitrification has been reported in acid soils at pH values as low as 3.5 (5). Nitrification in acid soils may be explained to some extent by growth on surfaces (2) or in aggregates (6), by ureolytic activity (1, 4), and by heterotrophic nitrifiers (14). An additional explanation is the existence of strains adapted to low-pH environments (7, 26), but although an acidophilic nitrite oxidizer has been isolated (9), acidophilic ammonia oxidizers have proved difficult to isolate in pure culture. Low growth rates, low biomass yield, and the limited number of distinguishing phenotypic characters for ammonia oxidizers have prevented the analysis of natural communities, in particular those in acid soils. The application of 16S ribosomal DNA (rDNA)-based techniques, however, enables the study of community structure in environmental samples, without the requirement for laboratory cultivation (8). Such studies have been particularly productive when applied to ammonia-oxidizing bacteria. With the exception of a small number of cultured marine strains belonging to the -proteobacteria, the phylogenetic analysis of rDNA sequences places all ammonia oxidizers in a monophyletic group within the -proteobacteria. This group consists of two distinct monophyletic genera, Nitrosomonas and Nitrosospira (10, 28, 30, 31). An analysis of natural ammonia oxidizer populations from a variety of environments has revealed further subdivision (26), based on phylogenetic analysis of a 300-bp fragment of the 16S rDNA molecule spanning the V2 and V3 regions (20). An analysis of 1.1-kb regions (incorporating this 300-bp fragment) of representative environmental sequences has shown that Nitrosospira can be further subdivided into at least four clusters, designated 1 to 4 (26). The Nitrosomonas genus may also be subdivided, and Stephen et al. (26) recognized three clusters within this genus, designated 5 to 7 (Fig. 1). Of particular relevance to this study is a cluster containing three soil clones, here referred to as cluster 6a. These and other studies (11, 15) indicate that sequences from members of the genus Nitrosospira were more abundant than Nitrosomonas in a range of environments and suggest that the available pure cultures represent only a limited selection of the phylogenetic and, by inference, physiological diversity of natural ammonia oxidizer populations. These studies are based on the hypothesis that environmental sequences falling within the Nitrosomonas and Nitrosospira clades originate from autotrophic ammonia oxidizers (26). Evidence for this hypothesis is based on their phylogenetic positions relative to cultured taxa which uniformly possess this phenotype, including recently isolated cultures of Nitrosospira (28). In addition, no sequence from a nonautotrophic ammonia oxidizer falls within this clade.

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Schematic tree of the -subgroup ammonia-oxidizing bacteria based on 303 bp of 16S rDNA sequence spanning E. coli positions 198 to 500 (3) and target groups for probes described in Table 1. Ambiguous sites were removed by using the GDE 2.2 "mask" function (16) before transfer of the data to the ARB sequence analysis program (25). The tree was generated by neighbor joining (24) with the Jukes and Cantor (13) correction in ARB. Clusters of sequences are scaled vertically to represent the number of sequences and horizontally to represent the extent of variation within each cluster. Cluster designations are as described in Stephen et al. (26) with the exception of cluster 6a, which contains three soil clones (see text), and cluster 6b, which contains the remainder of cluster 6 sequences previously described (26). Soil clone pH4.2/28 and marine enrichment sequence AEM3 cannot be placed unequivocally in this scheme. The scale bar represents 0.1 estimated changes per nucleotide position. Sequence data were compiled from references 10, 15, 17, 21, 26, and 28).

The available 16S rDNA sequence data for -subgroup ammonia oxidizers can be interpreted to suggest that particular phylogenetic clusters may be associated with specific habitats (26). For example, sequences from Nitrosospira cluster 2 and Nitrosospira cluster 3 were apparently more common in acid and neutral agricultural soils, respectively (26), while sequences from Nitrosomonas cluster 6 and Nitrosospira cluster 4 were detected at similar frequencies in both soils. Biogeographic or temporal studies of community structure by determination of sequence abundance in gene libraries are severely limited by the time required for sequence acquisition and analysis, and the approach is not necessarily quantitative. Denaturing gradient gel electrophoresis (DGGE) provides a complementary tool for the analysis of complex microbial communities (19, 25). It involves the separation of DNA fragments of identical length on the basis of differences in denaturant sensitivity in an acrylamide gel matrix, resulting from differences in the primary sequence. PCR products generated from different samples can thereby be compared directly, based on their mobility, without the need for cloning or DNA sequence analysis. Specific amplification and DGGE analysis of 16S rDNA for -subgroup ammonia-oxidizing bacteria successfully demonstrated differences between communities in dune soil samples (15). However, differences in mobility between the various sequence clusters and overlapping banding patterns prevented precise identification of the community members, which necessitated band excision and sequence determination.

The recent expansion of the 16S rDNA sequence database for -subgroup ammonia oxidizers provides the potential for the design of improved oligonucleotide probes for the analysis of natural populations (10, 15, 21, 26, 28). Published -subgroup ammonia oxidizer probes have been based on sequence information from a limited number of cultured organisms (11, 12, 29) and would not detect much of the diversity determined by Stephen et al. (26), restricting their value for the analysis of natural communities.

The first aim of this study was to design and test oligonucleotide probes capable of distinguishing the subgroups within the different clusters of -subgroup ammonia oxidizers from each other. The second aim was then to use these probes to identify bands separated by DGGE analysis of PCR products generated from soil with ammonia oxidizer-specific primers. In particular, we aimed to investigate further the hypotheses (26) that Nitrosospira may be of significance for ammonia oxidation in acid soils and that the relative abundances of different sequence clusters are related to soil pH.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Design of oligonucleotide probes. Probes were designed with the assistance of the ARB "probe design" function (27) from a manually aligned data set of all 16S rDNA sequences related to the -subgroup ammonia oxidizers derived from pure cultures, enrichment cultures, and environmental clones. Clustering of sequences (Fig. 1) was based on an analysis of 303 bases spanning the homologous Escherichia coli positions 198 to 500 (Fig. 1). Potential specific probe sequences showed at least one highly destabilizing base pairing (purine/purine) or two less-destabilizing mismatches (purine/pyrimidine or pyrimidine/pyrimidine) between the probe sequence and the most similar nontarget sequences towards the center of the target site. Table 1 shows the oligonucleotides synthesized for use in this study (Isogen Bioscience BV, Maarssen, The Netherlands). Potential specificities of probes were assessed by using the Ribosomal Database Project CHECK-PROBE facility (16) and FastA searches of the EMBL sequence databases to ensure that the target site did not occur in published sequences of non-ammonia-oxidizer organisms.

Probe hybridization analysis. Elution profiles were established for each oligonucleotide probe and its membrane-bound target sequence, and the temperature at which 50% of the probe was eluted was determined. Profiles were also established for closely related heterologous sequences to optimize hybridization conditions for analysis of environmental samples. Plasmid or genomic DNA was amplified by using 20 pmol each of the ammonia oxidizer-specific primers CTO189f and CTO654r (15) and 2 U of Tbr polymerase (Dynazyme; Finnzymes, Espoo, Finland) under published amplification conditions. Each product (100 ng) was spotted in triplicate into a dot blot apparatus (Bio-Rad Laboratories, Surrey, United Kingdom) containing Hybond N+ hybridization support membrane (Amersham International plc, Bucks, United Kingdom).

1

DGGE, membrane transfer, and hybridization. PCR, DGGE conditions, and the CTO primers were as described previously (15). Briefly, fragments of representative environmental library clones and genomic DNA from cultured representatives of each sequence cluster were amplified with PCR primer pair CTO189f-GC and CTO654r by using 1.25 U of Expand High Fidelity polymerase (Boehringer GmbH, Mannheim, Germany) in a 25-µl reaction volume according to the manufacturer's instructions. A fraction (10%) of each amplification reaction mixture was run on a 2% agarose-TAE (0.04 M Tris base, 0.02 M acetic acid, 1.0 mM EDTA, pH 7.5) gel, and DNA was visualized by fluorescence following staining with ethidium bromide. The concentrations of products were estimated by visual reference to molecular weight standards. Approximately 200 ng of each product (4 to 12 µl of crude PCR reaction mixture) was removed from the reaction tube for analysis by DGGE. DGGE gels (0.5× TAE, 8% acrylamide) were cast according to the protocol of Muyzer et al. (19) by using 38 to 50% denaturant (100% denaturant was 7 M urea with 40% [vol/vol] formamide [Boehringer GmbH]) and run at a constant temperature of 60°C. Gels were stained with ethidium bromide (0.5 mg liter¹) in MilliQ water (Millipore B.V., Etten-Leur, The Netherlands) and destained in 0.5× TAE buffer. Images were captured by use of The Imager system (Ampligene, Illkirch, France). DNA was transferred to nylon hybridization membranes (Hybond N+; Amersham International plc) by use of a Semi-Dry electroblotter (model SD; Bio-Rad) according to the manufacturer's instructions, with transfer medium consisting of 0.5× TAE buffer, and run at 40 mA for 1 h. Following transfer, DNA was simultaneously denatured and covalently cross-linked to the hybridization membrane by incubation on a pad of 3MM paper (Whatman International Ltd., Kent, United Kingdom) soaked in 0.4 M NaOH for 30 min, followed by neutralization on two pads of 1 M Tris-HCl, pH 7.0, for 2 min each. Filters were then rinsed in distilled water and air-dried prior to prehybridization.

Analysis of environmental samples. The study site consisted of an agricultural field (Craibstone, Scotland) divided into seven pH-controlled plots held at approximately pH 6.6, 6.0, 5.5, 5.0, 4.5, and 4.2 since 1961 by the addition of lime. The unamended pH of this site is approximately 3.9. All plots were sampled in February 1997, having supported a crop of potatoes in the previous year. Soil samples (100 g) were collected from the surface 2 cm of each pH-controlled plot and homogenized by passing them through a 4-mm-mesh-size sieve. Soil (0.5 g) and glass beads (0.5 g, 0.1 mm in diameter; BioSpec Products, Techno Lab, Alkmaar, The Netherlands) were suspended in 0.5 ml of water-saturated phenol, pH 8.0, in a 2-ml screw-cap polypropylene tube. Cells were lysed mechanically by bead beating soil suspensions for three periods of 30 s in a minibeadbeater (BioSpec Products) set to 5,000 rpm. Suspensions were chilled on ice between shaking periods, and DNA was extracted in phenol followed by ethanol precipitation and passage through a two-stage agarose gel containing 1% polyvinylpolypyrillodone (Sigma, St. Louis, Mo.) (15). PCR amplification of 16S rDNA fragments related to the -subgroup ammonia oxidizers was by 35 cycles of 92 (30 s), 57 (30 s), and 68°C (60 s) with the primer pair CTO189f-GC and CTO654r and 2.5 U of Expand High Fidelity polymerase (Boehringer). PCR products were subjected to DGGE analysis alongside products from each cluster group, transferred to positively charged hybridization membranes, and probed with genus- and cluster-specific ³²P-labelled oligonucleotides.

Quantification of relative abundances of sequence types. Gels were first probed with -AO233 to quantify the total -proteobacterial ammonia oxidizer signal in PCR products. Bands for clusters 4 and 6a are well separated, and the relative abundances of PCR products represented in these clusters in each environmental sample were calculated as the percentages of the total hybridization signal to -AO233 within each lane represented by the cluster 4 or cluster 6a bands. The combined relative abundance of clusters 2 and 3 was calculated by subtraction from the total hybridization signal. Relative abundances of clusters 2 and 3 were determined by probing gels with NspCL2_458 and NspCL3_454, specific for clusters 2 and 3, respectively. Gels probed with -AO233 showed that controls for clusters 2 and 3 were loaded at equal levels (within 0.5% of each other). Differences in binding efficiency and other factors between cluster 2 and 3 probes were determined by comparison of the hybridization signals for the respective controls, consisting of representative clones. Resultant correction factors were applied to hybridization signals for environmental samples to calculate relative proportions of cluster 2 and 3 signals. Finally, these values were expressed as proportions of the combined cluster 2 and 3 signal to give relative abundances of each cluster as a percentage of the total ammonia oxidizer signal. All analyses were carried out on two independent sets of soil samples, and results are expressed as means of duplicates.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Probe design. Probe specificities determined both empirically and by database searching are summarized in Table 1. Natural communities were analyzed by probing DGGE gels of PCR products obtained with primers specific to -proteobacterial ammonia oxidizers (15). Probes were therefore chosen for their ability to distinguish between ammonia oxidizers belonging to different clusters. Some also have the potential to distinguish ammonia-oxidizing from non-ammonia-oxidizing bacteria, although finding such probes was not the main aim of the study. These carry at least one or two mismatches to nontarget sequences currently in the database (Table 1), although this is likely to represent a small percentage of natural diversity. No single probe target site which was specific to all -subgroup ammonia oxidizers and which could differentiate these organisms from all other -proteobacteria was found within the amplified region. The -AO233 probe did not show complete specificity for -subgroup ammonia oxidizers in an unconstrained search of the available data banks. However, all nontarget strains with similar sequences bore strongly destabilizing mismatches to the CTO PCR primer pair used to produce PCR products for DGGE. The -AO233 probe was used to standardize radioactive probing of all products resulting from amplification with the CTO primer pair in quantifying the relative proportions of each band separated by DGGE.

Genus Nitrosospira. Probe Nsp436 is specific to currently known sequences related to the Nitrosospira genus and showed at least two mismatches to all other proteobacterial sequences available except Thiobacillus thioparus (NCIMB 8370), to which it showed only a single weakly destabilizing mismatch, and Comamonas testosteroni, with which it is identical. The closest sequence which could be amplified with the CTO primer pair was that from Nitrosomonas europaea, from which target 95% of probe had eluted at the T_d value (45°C). Nitrosospira cluster 1 was specifically differentiated from related groups with probe NspCL1_249. Specific hybridization of probe NspCL2_458 to cluster 2 sequences was achieved at 44°C, retaining approximately 40% of maximum probe binding (Fig. 2a). Several sequences from Nitrosospira cluster 2 contained only weakly destabilizing mismatches to the probe NspCL3_454, which was designed to Nitrosospira cluster 3 (Fig. 2b). Elimination of cross-reaction to these heterologous sequences therefore required hybridization under very stringent conditions, at 47°C, where binding to cognate sequences was reduced to approximately 20% of the maximum. Elimination of binding of the probe NspCL4_446, designed to Nitrosospira cluster 4, against its closest nontarget sequence, EnvB1-7 (Nitrosospira cluster 1; derived from marine sediment), required high stringency with washing at 44°C, although dissociation from the closest target associated with the soil environment occurred at 42°C (Fig. 2c).

View larger version (36K): [in this window] [in a new window]   FIG. 2.   Dissociation (DNA-DNA) kinetics for probes to clusters of Nitrosospira (a to c) and Nitrosomonas (d) detected on DGGE gels: NspCL2_458 (a), NspCL3_454 (b), NspCL4_446 (c), and NmoCL6a_205 (d). The rDNA targets used for Nitrosospira were derived from Nitrosospira briensis (accession no., M96396) and clones pH4.2A/K2 (accession no., Z69167), pH4.2A/C (accession no., Z69159), pH4.2A/J2 (accession no., Z69166), pH7C/30 (accession no., Z69191), pH4.2A/19 (accession no., Z69150), and EnvB1-7 (accession no., Z69102). Nitrosospira sp. strains B6, T7, KI, and GM41 are described by Utåaker et al. (28). The rDNA targets for Nitrosomonas were derived from N. europaea (accession no., M96399) and clones pH7C/54 (accession no., Z69196), pH4.2A/23 (accession no., Z69151), pH7C/56 (accession no., Z69197), DT1.9 (accession no., U62884), pH4.2A/3E (accession no., Z69155), and pH4.2/17 (accession no., Z69149). + denotes the cognate target sequence, and  and  represent the closest available nontarget sequences. Asterisks denote that the sequence is identical to the top sequence in that position. Differences are denoted by the replacement nucleotides shown below the target sequence. Probes were named to agree as closely as possible with the scheme used by Mobarry et al. (18), except that the position of the 5' nucleotide complementary to the 3' probe nucleotide is given as a suffix (E. coli numbering [3]).

Genus Nitrosomonas. Probe Nmo254 was designed to recognize all Nitrosomonas-like sequences to the exclusion of all Nitrosospira-like sequences. Its specificity was improved, during the study, as new sequences entered the database, and probe Nmo254a provides fewer mismatches with the full range of target sequences. Nevertheless, two recently published Nitrosomonas sequences, those of N. ureae and N. communis (21), and that of an environmental clone (pH7C56) have a single mismatch, and further probe design is required for their detection. It was generally more difficult to design probes for the Nitrosomonas genus than for Nitrosospira, as might be predicted from the large degree of 16S rDNA sequence variation observed within Nitrosomonas. A single probe specific to Nitrosomonas cluster 6 could not be designed because of lack of sequence similarity between three library soil clones (cluster 6a) recovered in a previous study (26) and all other sequences in this cluster. Because of their potential importance in these soils, a highly specific probe, Nmo6a_205, was developed for the detection of these sequences; the probe provided clear differentiation between target and nontarget standards at 44°C Fig. 2d). Probe NmoCL6b_376 was designed to detect all remaining Nitrosomonas cluster 6 sequences to the exclusion of all other -subgroup ammonia oxidizer sequences at a hybridization temperature of 46°C. Nitrosomonas cluster 7 (which is represented by N. europaea, N. eutropha, and Nitrosomonas sp. strain NM57) and a single marine enrichment sequence, B2aW2 (accession no., Z69138), were differentiated from all other Nitrosomonas-related sequences by the use of highly specific probe NmoCL7_439 at a hybridization temperature of 44°C. This probe carries at least four mismatches to other ammonia oxidizer sequences and three mismatches to the closest non-ammonia-oxidizer sequence, i.e., that from Cytophaga uliginosa NCIMB 1863. The only sequence cluster for which no specific probes could be designed within the amplified 16S rDNA region was Nitrosomonas cluster 5, previously detected only in highly polluted marine samples.

Analysis of environmental samples by DGGE and probing. Samples of soil from Craibstone sites maintained at a range of pH values were analyzed by DGGE followed by hybridization with probes designed against the different clusters of ammonia oxidizers. Very similar banding patterns were recovered for all samples, and amplified products from all Craibstone soil plots, irrespective of pH, could be separated into three groups of bands at 43.4 to 43.7, 44.0 to 45.3, and 45.5 to 46.4% denaturant (Fig. 3A). These bands comigrated with products from cloned standards representing sequence clusters Nitrosomonas cluster 6a, Nitrosomonas clusters 6b and 5 and Nitrosospira clusters 2 and 3, and Nitrosospira cluster 4, respectively (Fig. 3A). Bands comigrating with standards from clusters 2 to 6 were present in all samples, supporting studies of the community structure of ammonia oxidizers based on 16S rRNA gene libraries from the same soil (26). No bands comigrated with the Nitrosospira cluster 1 or Nitrosomonas cluster 7 standards used. None of the bands recovered could be ascribed to a sequence cluster on the basis of migration alone because of comigration of certain bands from different clusters.

View larger version (74K): [in this window] [in a new window]   FIG. 3.   Analysis of -subgroup ammonia oxidizers by DGGE, membrane transfer, and hierarchical probing. Soil DNA was amplified by 35 cycles of PCR with the CTO primer pair, and the products were separated by electrophoresis on a 38 to 50% denaturant gradient gel. Phylogenetically defined amplification products from each sequence group were included as migration and hybridization standards (left eight lanes). The gel was stained with ethidium bromide (A) before transfer of the DNA to a nylon hybridization support membrane. The transfer membrane was subsequently hybridized with 32P-labelled oligonucleotides -AO233 (B), NspCL2_458 (C), NspCL3_454 (D), NspCL4_446 (E), and NmoCL6a_205 (F). The membrane was stripped of bound probe between experiments by being washed in 2× SSC-0.1% sodium dodecyl sulfate at 65°C. Hybridization conditions are given in Table 1.

Presence or absence of sequence clusters. All of the bands visible by ethidium bromide fluorescence were detected by hybridization with the general -subgroup ammonia oxidizer probe -AO233 following membrane transfer (Fig. 3B). Subsequent hybridization with cluster-specific probes identified the four sequence clusters previously shown by cloning and sequence analysis to be present in this soil: Nitrosospira clusters 2, 3, and 4 and Nitrosomonas cluster 6a (Fig. 3C to F). Hybridization with probe Nsp436, the general probe for the genus Nitrosospira, targeted all bands visible in Fig. 3A and B, except for the Nitrosomonas cluster 6a band (data not shown). This band hybridized strongly to general Nitrosomonas probe Nmo254. The central set of four bands comigrated with Nitrosospira cluster 2 and cluster 3 standards. These bands hybridized with probes designed for these groups, NspCL2_458 and NspCL3_454 (Fig. 3C and D). The lowest group of three bands comigrated with products from the Nitrosospira cluster 4 standard and hybridized strongly to NspCL4_446, the cognate probe for this cluster (Fig. 3E).

Distribution of clusters as a function of pH. Our earlier observations of clone libraries recovered from soil plots at pHs 4.2 and 7.0 suggested that the abundances in libraries of Nitrosospira cluster 2 and 3 sequences from soil were dependent on soil pH (26). Cluster 2 was more common in soil at pH 4.2, while Nitrosospira cluster 3 was more prevalent at pH 7. These data provided no evidence for pH-associated effects on the abundance of Nitrosospira cluster 4, while only three clones were obtained for Nitrosomonas cluster 6a. Hybridization of blotted DGGE gels loaded with DNA amplified from the full range of pH-controlled plots at the Craibstone site with probes NspCL2_458 and NspCL3_454 demonstrated that PCR products from neutral soil hybridized strongly with probe NspCL3_454 and poorly with probe NspCL2_458 (Fig. 3C and D). PCR products from acidic soil samples generated the opposite hybridization pattern, with a reduction of Nitrosospira cluster 3 PCR products and an increase in Nitrosospira cluster 2. Nitrosospira cluster 4 varied with pH (Fig. 3E) but no significant effect of soil pH on Nitrosomonas cluster 6a could be detected (Fig. 3F).

5

5

View larger version (33K): [in this window] [in a new window]   FIG. 4.   Quantification of hybridization signals from DGGE gels of soil samples for Fig. 3C to F. Quantification was achieved by phosphorimaging and profile analysis of each lane of the membranes used. The percent total signal is the hybridization signal for each sample expressed as a percentage of the signal for the cluster standard, normalized as described in the text, and averaged over duplicate sets of lanes from independent DNA extractions from each soil. Least significant difference values for clusters 2, 3, 4, and 6a were 2.13, 2.36, 1.50, and 1.15, respectively.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The recent expansion of 16S rDNA sequence information from the -subgroup ammonia oxidizers has increased the potential value of molecular techniques in the study of the ecology of this environmentally and commercially important group of organisms. Phylogenetic analysis has demonstrated a division of -subgroup ammonia oxidizers into a number of clusters or related sequences. Analysis of the relative abundances of sequences obtained by PCR amplification from environmental samples with ammonia oxidizer-specific primers further suggests that the distribution of different clusters is linked to environmental factors. An alternative approach to the study of distribution of groups within the environment is the use of ammonia oxidizer-specific oligonucleotide probes, but previous studies have been limited by the use of sequence information from pure cultures to design and test probes. The initial aim of this study was to use the expanded database to improve and extend phylogenetic probes for autotrophic ammonia oxidizers, in particular probes specific for the clusters defined by phylogenetic analysis of rDNA sequences.

On the basis of current rDNA sequence information, the CTO primer pair (15) is specific to all -subgroup ammonia oxidizers. These primers, however, only provide specificity when used in combination, as in PCR amplification, and individual primers are not reliable as ammonia oxidizer-specific probes. Further analysis of the target site used in this study, 303 bases spanning the homologous E. coli positions 198 to 500, did not provide a unique sequence specific to all ammonia oxidizers. It was possible, however, to design a probe which appears to be specific for the Nitrosospira group. Nitrosomonas probe Nmo254 was specific for all Nitrosomonas sequences obtained from our study site but will not hybridize with N. ureae and N. communis, recently sequenced by Pommerening-Röser et al. (21). It is possible that the analysis of new environments will identify additional sequences which do not hybridize with Nmo254 or with the Nitrosospira probe. This stresses the need for the generation of clone libraries when initiating studies of new environments and the use of low-specificity PCR primers to detect related but uncharacterized groups (17).

The approach of using DGGE and probing adopted in this study involved initial amplification using the ammonia oxidizer-specific CTO primers prior to the use of probes. A hybridization analysis of amplification products from our agricultural soil samples demonstrated that all amplification products could be assigned to previously described cluster groups within the -subgroup ammonia oxidizers. This confirms the specificity of the PCR primers and the described reaction conditions in diverse environments. Under these conditions, cluster-specific probes applied to the analysis of PCR products generated with the CTO primer pair need only be faithful within the -subgroup ammonia oxidizer clade. None of the bacterial species which show closest sequence similarity to the PCR primer pair show close similarity to any of the probes presented. Additionally, some of these probes do show sufficient specificity in sequence similarity searches to hold potential for use in more complex analysis such as in situ probing (Table 1). It is also clear that further probes will need to be designed for the analysis of DGGE patterns described here, particularly for the genus Nitrosomonas, which cannot be exhaustively analyzed with the probe set presented. This probe set allows detection of, and discrimination between, each of the ammonia oxidizer clusters described by Stephen et al. (26), except for cluster 5, which has so far only been detected in polluted marine samples.

This study also provided information on the hybridization conditions required for use of each cluster probe. Hybridization at T_d values calculated from elution profiles of some probe and target sequences was capable of distinguishing target and known nontarget ammonia oxidizer sequences. This was not the case, however, for all probes to the clusters present in the soil samples used in this study, necessitating the use of more-stringent conditions for hybridization.

An earlier study (26) suggested the existence of two "specialist" clusters, Nitrosospira clusters 2 and 3, favored by acid (pH 4.2) and neutral (pH 7) conditions, respectively. DGGE analysis of PCR products followed by probing for the specialist clusters confirmed these findings. Signals from cluster 2 probes showed a continuous increase as soil pH values decreased from 6.6 to 3.9, while those from cluster 3 probes decreased over this range. Both studies also demonstrated similar levels of cluster 6 at all soil pH values investigated and indicated that Nitrosospira cluster 1 and Nitrosomonas clusters 5, 6b, and 7 were either absent from this environment or below the detection limits of this approach at the time of sampling. Sequences from the earlier study were obtained following PCR amplification with the AMOf primers of McCaig et al. (17), while the present study involved initial amplification with a completely different primer set, CTO, targeting a different region of 16S rDNA. Soil samples were obtained from the same study sites. Despite the use of different primers, both experimental approaches indicated similar effects of soil pH on the relative abundances of ammonia oxidizer clusters 2 and 3, although the earlier study was based on a relatively small number of sequences. The similarity in results from the two sets of primers indicates that primer bias is unlikely to be selecting for particular sequences, thereby greatly increasing confidence that the findings obtained represent in situ abundances. Other potential biases include lysis bias between clusters, although this is likely to be minimized by the use of bead beating. The suggestion that Nitrosospira cluster 4 is a "generalist" cluster was based on similar sequence abundances at only two pH values, 4.2 and 7, while the present study involved finer-scale analysis of soil pH. DGGE and probing analyses provided further evidence for similar levels of cluster 4 at low and neutral pH values but indicated that levels were greatest at pH 5.5, decreasing at lower and higher pH values. The effect of pH on the relative abundance of cluster 4 was less marked than on those of clusters 2 and 3 and would not have been detected in the earlier study, for which only two samples, one acid and one neutral, were analyzed.

Confirmation of the existence of generalist and specialist ammonia oxidizer strains highlights the need for physiological studies of these groups, in particular, representatives of Nitrosospira clusters 2, 3, and possibly 4, to identify factors leading to pH-related selection. Several cluster 3 strains are available in pure culture, and two pure cultures of Nitrosospira cluster 2 and Nitrosospira sp. strains B6 and T7 have recently been obtained (28). Currently, nothing is known of the physiological differences which may lead to pH-based selection, but our findings suggest that Nitrosospira cluster 2 may be an important genotype in acid soils and that physiological studies of representatives of this group are required.

The design of genus- and cluster-specific probes and their use in combination with DGGE analysis of PCR amplification products from directly extracted soil DNA provide a powerful, rapid, and quantitative technique for the analysis of the structure of ammonia oxidizer populations in natural environments. The technique has demonstrated clear links between genotype and ecotype in soils maintained at a range of pH values for 36 years, identifying candidate generalist and specialist strains. The findings provide the basis for further studies on the physiological characteristics of these strains leading to differences in their distribution and on their significance for nitrification rates and the persistence of ammonia oxidizers in natural environments.