Introduction

Nitrification is an important process in the global nitrogen (N) cycle, oxidizing ammonia to nitrate and providing substrate for reductive nitrogen processes (denitrification). The rate-limiting step in nitrification is the oxidation of ammonia to nitrite, with nitrite typically being rapidly oxidized to nitrate by nitrite-oxidizing organisms. Until recently, ammonia oxidation has been considered to be entirely mediated through activities of ammonia-oxidizing bacteria (AOB; Kowalchuk and Stephen, 2001). Cultivation-dependent (Head et al., 1993), and subsequently cultivation-independent studies (Stephen et al., 1996, 1998), had suggested that members of a distinct clade of β-proteobacteria were the primary biological agents of ammonia oxidation in soil. The narrow phylogenetic breadth of this group facilitated the design of primers to monitor AOB using both phylogenetic (rRNA) and functional gene (amoA encoding the α-subunit of the key enzyme ammonia monooxygenase) markers (Rotthauwe et al., 1997; Kowalchuk et al., 2000; Kowalchuk and Stephen, 2001). Using these nucleic acid-based tools, there have been extensive environmental and ecotoxicological studies exploring the distribution, diversity and response to environmental stresses of AOB in relationship with soil nitrification (for example, Koops et al., 2001; Kowalchuk and Stephen, 2001; Aoi et al., 2004).

Recently, Venter et al. (2004) and Treusch et al. (2005) suggested the involvement of archaea in the global nitrification process. This genetic potential for archaeal ammonia oxidation was soon supported by the cultivation of an ammonia-oxidizing archaea isolate (AOA; Francis et al., 2005; Könneke et al., 2005), and it became evident that the pool of amoA genes across a range of soils appeared archaea-dominated (1.5–230 times>AOB; Leininger et al., 2006). Analysis of amoA mRNA copy levels also suggested that Crenarchaeota are the most active ammonia-oxidizing organisms in several soils (Leininger et al., 2006). These recent findings have resulted in a fundamental shift in our understanding of the biology underpinning nitrification in environmental samples (Francis et al., 2007). In particular, the current ethos is for an archaeal driven process, questioning the wider role of bacteria in this key N-cycle process in soil (Leininger et al., 2006; Nicol and Schleper, 2006; He et al., 2007).

The biological oxidation of ammonia can be influenced by numerous factors and is recognized as being a model process for ecological studies (Kowalchuk and Stephen, 2001). Although numerous studies have explored the links between environmental factors, AOB and nitrification, the response of AOA to environmental perturbations remains unknown (Nicol and Schleper, 2006; Francis et al., 2007; Boyle-Yarwood et al., 2008; Chen et al., 2008; Shen et al., 2008).

Soil nitrification is highly sensitive to elevated zinc (Zn; Smolders et al., 2004), although ecotoxicological effects can be transient (Rusk et al., 2004). It has been hypothesized that recovery of nitrification following Zn exposure is due to the development of Zn-tolerant AOB populations (Mertens et al., 2006). However, the recent insights into the role of AOA in soil ammonia oxidation (Leininger et al., 2006) place the role of AOB in nitrification restoration in doubt.

Here, we report the relative responses of AOB and AOA during the recovery of the ammonia oxidation process in an Australian agricultural soil (Spalding, South Australia) after experimental Zn contamination. To examine whether the functional recovery was due to changes in the size, activity or composition of dominant AOB and/or AOA, we measured the numbers of amoA gene copies and gene transcripts and assessed the amoA community composition for both AOB and AOA populations, and compared the data to patterns of ammonia oxidation activity.

Materials and methods

Soil treatment and nitrification

Top soil was sampled from an ongoing field trial in Spalding, South Australia, within 1 week after ZnSO4 addition (T0) and annually for 3 years (T1–T3). The Spalding soil is classified as a chromosol (Isbell, 1996), and has a pH (0.01 M CaCl2) of 6.3, 1.9% organic carbon and 27% clay. The background Zn concentration is 0.9 mmol Zn per kg. Zinc treatments were duplicated, and metal salts were incorporated into the soil using a rotary hoe. At each sampling, four topsoil samples (0–10 cm) were collected per plot using a 5-cm-diameter auger, and samples were combined to yield one mixed sample. Details about the experimental design, the spiking procedure, soil sampling and soil physicochemical properties have been reported previously (Broos et al., 2007b). Nine Zn treatments were imposed; a control soil (‘unexposed soil’) and 8 soils which received increasing Zn doses (1.1–113 mmol Zn per kg). After sampling, soils were air dried, ground, sieved to <2 mm and stored in closed containers at room temperature. Before further functional and molecular analysis, soils were remoistened with distilled water to 50% of the maximum water-holding capacity and aerobically incubated at 20 °C for 14 days in the dark to minimize the effects of storage time and of temperature and moisture content at the time of sampling.

Substrate induced nitrification (SIN) was measured in all Zn treatments (that is, the unexposed soil sample and 8 soil samples with increasing Zn doses) at T0 to T3 in duplicate per soil sample, and SIN was expressed as the percentage NH4+ substrate used (%) 28 days after substrate addition (100 mg NH4+-N per kg soil; Broos et al., 2007b). Effects of increasing Zn concentrations on the nitrification process were related to Zn concentrations in a CaCl2 soil extract. Total Zn concentrations in the CaCl2 extract ranged from 0.01 to 109 mM Zn at T0 and 0.01 to 21 mM Zn at T2. The plot that received 28.3 mmol Zn per kg, corresponding to 22 mM Zn in a CaCl2 soil extract at T0 and 6.6 mM Zn in a CaCl2 soil extract at T2 (‘Zn-exposed soil’), represented the largest Zn dose at which soil nitrification fully recovered following initial inhibition.

Zn-tolerance testing

Zn tolerance was tested in the unexposed and the Zn-exposed soil samples at T0 to T3 using the spike-on-spike test of Mertens et al. (2006). In short, ZnCl2 was added to duplicate suspensions of soil samples (1:10 soil/CaCl2 0.01 M ratio) to final concentrations of 0, 0.77, 1.5 and 3.1 mM added Zn. Ammonium sulfate was added at 0.7 mM NH4-N, and the pH in suspension was adjusted immediately after ammonia addition and daily during the 3-day test to pH 5.8 using 0.1 M NaOH or HCl. We preferred to adjust pH to pH 5.8 to approach the in situ soil pH rather than the optimal pH for nitrification (that is, pH 7–8; Jiang and Bakken, 1999). The potential nitrification rate (PNR; Smolders et al., 2001) was measured, and is the nitrate production rate in the initial period (0–7 days) after ammonia addition (mg NO3−-N per kg per day) during which there is unlimited substrate present. The experimental setup for SIN and PNR is essentially identical, but whereas the SIN is defined as the amount of substrate used within a 28-day incubation experiment (during which the substrate may have already been exhausted—note that SIN never exceeds 100%), PNR is defined as the rate of substrate used over a shorter period of time. PNR is very sensitive to metal stress, and also to slight changes of other physicochemical soil characteristics (for example, pH; Broos et al., 2005). Therefore, PNR rather than SIN was used as a measure of the nitrifying activity for Zn-tolerance testing in the well-defined soil suspensions (that is, constant pH and ionic strength for all soils) whereas SIN was preferred for screening the nitrification process in field soil samples over time (Broos et al., 2007b). Effects of increasing Zn concentrations to the PNR were related to Zn concentrations in a CaCl2 soil extract.

DNA extraction, PCR and denaturing gradient gel electrophoresis

DNA was extracted from all soil samples at T0 and T2 in duplicate per soil sample (Mertens et al., 2006). For denaturing gradient gel electrophoresis (DGGE) community profiling, PCR targeting the AOB amoA gene used primers amoA-1F* and amoA-2R (Rotthauwe et al., 1997; Stephen et al., 1999), PCR targeting the AOA amoA gene used primers CrenamoA23f and CrenamoA616r (Tourna et al., 2008). For amplification of AOB amoA genes, 1 μl of purified DNA was added to a PCR mixture containing 1.25 U of Taq polymerase (Qiagen, Venlo, the Netherlands), 20 pmol amoA-1F*, 20 pmol amoA-2R, 10 nmol of each dNTP, 1 × PCR buffer (Qiagen), 5 μl bovine serum albumin (1%) and PCR-grade water to a total volume of 50 μl. PCR was conducted in an Eppendorf Mastercycler (Eppendorf, Hamburg, Germany) and PCR conditions were 10 min at 94 °C, followed by 35 cycles of 60 s at 94 °C, 60 s at 60 °C and 60 s at 72 °C, and a final elongation step for 10 min at 72 °C. A GC-rich clamp was attached to the 5′ end of the AOB forward primer for DGGE fingerprinting. Analysis of PCR products size and integrity, and DGGE profiling of amplified bacterial amoA gene fragments were performed as described (Mertens et al., 2006). Amplification of AOA amoA genes used Qiagen HotStar Taq PCR chemistry. DGGE separation of the AOA amoA genes was conducted without requirement for a GC-clamp in an Ingeny PhorU DGGE system (Tourna et al., 2008). DGGE profiling was performed as previously described (Nicol et al., 2005; Tourna et al., 2008).

amoA sequence and phylogenetic analysis

Band excision, cloning and sequencing were performed on the unexposed and Zn-exposed soil samples at T2 as described (Mertens et al., 2006). Sequences for dominant bands ‘a’ to ‘k′ (Figure 3c) have been submitted to the GenBank database under accession numbers EU515192 through to EU515202. AOB amoA DNA sequences ‘a’ to ‘k’ had open reading frames with translation products matching known AmoA protein sequences. The sequence data for each clone was compared with sequences on GenBank using the BlastN tool and similar and dissimilar sequences spanning the amoA region of interest were recovered. Sequences were aligned in ClustalX; the alignment was manually checked. Phylogenetic analyses were conducted using MEGA version 4 (Tamura et al., 2004) using the UPGMA method (Sneath and Sokal, 1973) with evolutionary distances computed using the maximum composite likelihood model (Tamura et al., 2007). Phylogeny was tested using bootstrapping (Felsenstein, 1985) with 500 replicates (Supplementary Figure 2).

amoA gene copy and gene transcript quantification

Copy numbers of AOB and AOA amoA gene fragments were quantified in all soil samples at T2 and, additionally, in the unexposed and Zn-exposed soil samples at T0, T1 and T3 by real-time quantitative PCR (qPCR) on a Rotor-Gene 3000 Real-time PCR Cycler (Corbett Research, St Neots, UK). Real-time PCR of amoA gene fragments used the primers amoA-1F* and amoA-2R primers as previously described (Rotthauwe et al., 1997; Stephen et al., 1999), and primers Amo196F and Amo277R (Treusch et al., 2005) for AOA. The protocols for quantitative amplification were based on methods described elsewhere (Okano et al., 2004; Treusch et al., 2005). In short, qPCR was performed in 25 μl reaction mixtures using 5 μl purified DNA as template and 12.5 μl of SYBR Green master mix (Qiagen). Primer concentrations were 2.25 pmol for primers amoA-1F* and amoA-2R, 1.5 pmol Amo169F and 3 pmol Amo277R for the AOA amoA amplification. Reaction mixtures were made up to a final volume of 25 μl using PCR-grade water. The protocol for amplification of AOB amoA fragments was 2 min at 50 °C, 15 min at 95 °C, 45 cycles of 45 s at 95 °C, 60 s at 60 °C and 45 s at 72 °C (based on Okano et al., 2004), and the protocol for amplification of AOA amoA fragments was 15 min at 95 °C, 45 cycles of 15 s at 95 °C, 40 s at 55 °C and 45 s at 72 °C (based on Treusch et al., 2005). Both protocols were followed by a final elongation step of 5 min at 72 °C. Dissociation curve analysis and agarose gel electrophoresis were performed to confirm amplification specificity. Dissociation curves for AOB and AOA amoA quantification were conducted by stepwise increasing the temperature with 1 °C from 60 to 95 °C or 55 to 95 °C (45 s at the lowest temperature, and 5 s for each subsequent temperature step). Dilution series of purified PCR products of plasmids (amoA gene of Nitrosomonas europaea ATCC 19718) and fosmid vectors (clone 54d9) were used as qPCR standards to allow quantification of each target number per sample.

Total soil RNA was extracted from the unexposed and Zn-exposed soil samples at T0 and T2 using the MoBio Soil RNA extraction kit and was treated with DNAse (Promega, Leiden, the Netherlands). cDNA generation and quantification of AOA and AOB amoA transcript numbers were performed as described by Leininger et al. (2006).

Statistical analyses

Dose–response curves were fitted by log-logistic modeling (Doelman and Haanstra, 1989) using the Marquardt method (proc NLIN, SAS 9.1; NC, USA). Significant (P<0.05) inhibitory effects on amoA gene copy and transcript numbers were detected using Student's t-tests.

Results

Zinc contamination of the arable field decreased nitrification with a characteristic log-logistic dose–response relationship (Figure 1). The EC50 for SIN (effective concentration with 50% reduction of SIN compared to the unexposed soil sample) was 3.6 mM Zn one week after exposure (T0; Broos et al., 2007b). The EC50 increased significantly to 13.1 mM Zn after 2 years of exposure (T2). Zinc concentrations in the Spalding soil samples were expressed as 0.01 M CaCl2 soil extracts, which mimic in situ Zn concentrations in porewater (Degryse et al., 2003) and, hence, bioavailable Zn concentrations in soil (Mertens et al., 2007). A Zn contamination level of 28.3 mmol Zn per kg soil, corresponding to 22 mM Zn at T0 and 6.6 mM Zn at T2, was the highest contamination level for which ammonia oxidation was restored within 2 years after initial inhibition (Figure 1), and this soil sample was selected for further analysis (‘Zn-exposed soil’).

Figure 1
figure 1

Substrate induced nitrification (SIN) in the Spalding field site over time. SIN, expressed as percentage substrate used (%), is plotted versus Zn concentrations in a 0.01 M CaCl2 soil extract at T0 (open symbols) and T2 (closed symbols). The arrow indicates the shift in EC50 (Effect Concentration with 50% decrease compared to the unexposed soil sample) values between T0 and T2.

Full size image

Zinc-tolerance testing of the unexposed soil sample at T0 and T2 revealed EC50 values of 0.09 and 0.22 mM Zn, respectively (Figure 2). No nitrifying activity was detected for the Zn-exposed soil sample at T0. Two years after Zn addition, the PNR of the Zn-exposed soil sample was comparable to the PNR of the unexposed soil sample, and application of increasing Zn doses to the Zn-exposed soil sample at T2 did not affect the PNR (Figure 2).

Figure 2
figure 2

Zn-tolerance testing at T0 and T2. Potential nitrification rate (PNR) (±s.e.), expressed as mmol NO3−-N per kg per day, of the unexposed soil sample (circle) and a Zn-exposed soil sample (triangle) at T0 (open symbols) and T2 (closed symbols), is plotted versus Zn concentrations in a 0.01 M CaCl2 soil extract (±s.e.).

Full size image

The AOB and AOA amoA gene copy numbers in the unexposed soil samples did not differ significantly from each other, and were not significantly different between T0 and T2 (Figure 3a). AOA amoA gene copy numbers in the Zn-exposed soil sample were lower than AOB amoA gene copy numbers at both T0 and T2, with a significant difference at T2 (Figure 3a). Also, in the examination of T2 samples exposed to a range of Zn doses, severe decreases of AOA amoA gene copy numbers were observed at a soluble Zn concentration two orders of magnitude lower than for AOB amoA gene copy numbers (Supplementary Figure 1a).

Figure 3
figure 3

amoA gene copy numbers, amoA gene transcript numbers and amoA community profile of AOB and AOA in the unexposed and the Zn-exposed soil sample over time. Average number of amoA gene copies (±s.e.; a) and number of amoA gene transcripts (±s.e.; b) of two field replicates and the amoA community profile (c) of AOB and AOA in the unexposed soil sample and the Zn-exposed soil sample at T0 and T2. Bands marked with a letter (‘a’ to ‘k’) where cloned and sequenced and belong to the Nitrosospira lineage.

Full size image

At T0, the numbers of AOB and AOA amoA gene transcripts in the unexposed soil samples were 3.7 × 106 and 1.2 × 106 per g of soil, respectively, and similar values were found at T2 (Figure 3b). In contrast, the number of AOB and AOA amoA gene transcripts in the Zn-exposed soil samples was more than 3 orders of magnitude reduced at T0, compared to the values in the unexposed soil sample (Figure 3b). After two years of exposure, the number of AOB amoA gene transcripts recovered to the level of the unexposed soil sample at T0, but remained a bit lower than the level of the unexposed soil sample at T2. In contrast, the number of AOA amoA gene transcripts remained 4 orders of magnitude below that in the unexposed soil sample (Figure 3b).

Community profiling of amoA genes (DGGE) showed a distinct shift in the AOB community structure in the Zn-exposed soil samples at T2, yet no such shift in amoA genotypes was observed for the AOA community (Supplementary Figures 1b and c). In the AOB community profile of the unexposed soil, the amoA genotypes corresponding to bands labeled ‘b’, ‘c’, ‘d’ and ‘e’ in the DGGE fingerprint were most prevalent (Figure 3c). However, at T2, bands ‘h’ and ‘i′ were most prominent in the AOB amoA community profile (Figure 3c). Sequence analysis revealed that all detected AOB amoA bands were affiliated with several sequence clusters within the Nitrosospira lineage (Avrahami et al., 2002), and no clustering of sequences was observed with respect to the Zn tolerance of the community (Supplementary Figure 2).

Discussion

The increased EC50 value over time, expressed as Zn concentrations in 0.01 M CaCl2 soil extracts (Figure 1), was not attributable to slow immobilization reactions of Zn in soil, but rather to adaptation of the ammonia-oxidizing community, as observed for other long-term Zn contaminated soils (Rusk et al., 2004; Mertens et al. 2006). This was verified by testing for Zn tolerance in the Zn-exposed soil sample. In accordance to the SIN data, no nitrifying activity was observed in the Zn-exposed soil sample at T0, whereas nitrification levels recovered to the level of the unexposed soil sample at T2. At the same time, the recovery of the nitrification process in this soil sample was accompanied with increased Zn tolerance within the nitrifying community at T2 (Figure 2).

The AOB and AOA amoA gene copy numbers in the unexposed soil sample were high, but comparable to those found in previous studies (Leininger et al., 2006; He et al., 2007). In several recent studies, the number of AOA amoA gene copies in soil exceeded that of AOB (for example, Adair and Schwartz, 2008; He et al., 2007; Shen et al., 2008), whereas numbers in the unexposed Spalding soil were roughly equal. The reduction of the AOA amoA gene copy numbers on contamination is very rapid (Figure 3a; Supplementary Figure 1a). All evidence suggests the death of AOA cells on contamination, as storage and rewetting itself does not significantly affect levels of AOA, AOB or total bacteria in these soils (not shown). It should be noted that the coverage of amoA-based primers, especially for AOA, is not yet well established, and the assays used in this study may therefore underestimate actual amoA gene copy and gene transcript numbers. At T2, the nitrification activity in the Zn-exposed soil sample recovered to the value in the unexposed soil sample (Figures 1 and 2), and AOB amoA gene copy numbers were appreciable (Figure 3a). In contrast, the AOA amoA gene copy number in the Zn-exposed soil sample was fivefold lower compared to the unexposed soil sample (Figure 3a), suggesting that recovery of SIN in the Zn-exposed soil sample was exclusively attributable to the AOB community in this soil. This finding is supported by a tight association between AOB amoA gene copy numbers and SIN across Zn application rates at T2 (50% reduction of AOB amoA gene copy numbers and SIN at 8.1 mM and 13.1 mM Zn, respectively, and at 0.07 mM Zn for the AOA amoA gene copy numbers; Figure 1; Supplementary Figure 1a). SIN, level of Zn tolerance and AOB and AOA amoA gene copy numbers of the unexposed and the Zn-exposed soil samples at T1 (that is, after 1 year of exposure) were similar to those at T0 and corresponding data at T3 (that is, after 3 years of exposure) were similar to those at T2 (details not shown).

The inhibition of SIN in the Zn-exposed soil sample at T0 coincided with a sharp decrease in the AOB and AOA amoA gene transcript numbers, indicating that both groups were severely affected by Zn stress (Figures 1 and 3b). The restoration of the nitrification process in the Zn-exposed soil sample at T2 was associated with a recovery of the amoA gene transcript number within the AOB community, but not within the AOA community (Figure 3b), providing evidence for a dominant role of AOB in the restoration of the ammonium oxidation activity. At the same time, the AOB amoA gene community profile in the Zn-exposed soil sample had changed, yet no corresponding change could be observed within the AOA amoA gene community profile (Figure 3c). As sequence variation in the AOB amoA gene is closely linked to 16S rRNA phylogeny (Purkhold et al., 2000, 2003; Aakra et al., 2001), this observation suggested a shift in AOB community structure that was associated with restoration of nitrification. Interestingly, we did not observe any clustering of recovered sequences with respect to Zn tolerance of the communities (Supplementary Figure 2). Thus, it appears that the ability to acquire Zn tolerance is widespread throughout the Nitrosospira lineage, and specific to particular ecotypes not discerned using this functional marker gene. The selection of genotypes of Zn-tolerant AOB species occurred from low background levels of these AOB, or from adaptation and selection of Zn-tolerant AOB over time. Our failure to detect a shift in the AOA community suggested the AOA community did not have the capacity to respond to Zn-induced stress over a 2-year time period. Similar to our results, changes of the numbers of AOB and AOA amoA gene copies, gene transcripts and/or amoA gene profiles suggested AOB dominated ammonia oxidation in an alkaline sandy loam soil under different long-term fertilization regimes (Shen et al., 2008), in grassland under different grazing managements (Le Roux et al. 2008) and in long-term acidified soil samples (Nicol et al., 2008). In contrast, data on the numbers of AOB and AOA amoA gene transcripts in soils incubated at different temperatures (Tourna et al., 2008) and in nitrogen fertilized paddy rhizosphere soils (Chen et al., 2008) suggest an AOA-dominated ammonia-oxidation process. It must be recognized that the nitrification process, amoA gene community profiles, amoA gene copy numbers and amoA gene transcript numbers were analyzed in dried, stored and reconditioned soils. Detected target numbers may therefore not be an accurate portrayal of in situ conditions. However, we have observed that soil drying and storage, which mimic the conditions present for extended periods in these fields (Broos et al., 2007a), had no detectable effects on determined process rates, as well as on bacterial numbers (data not shown). Furthermore, our molecular analyses were geared toward determining the populations responsible for activities in the reconditioned soil samples, not necessarily in situ field conditions at the time of sampling.

Thus, contrary to the general view of archaea being more tolerant to chronic stress conditions than bacteria (Schleper et al., 2005; Valentine, 2007), we demonstrated that AOA were more sensitive to Zn contamination than AOB in this soil. It is unlikely that the effects on the AOA community were caused by other physicochemical or biological soil processes besides Zn, as the numbers of AOA and AOB amoA gene copies and gene transcripts and AOA and AOB amoA gene profiles of the unexposed soil samples were similar across all replicates and sampling times (that is, T0 to T3). Similar consistent results were observed by Wheatley et al. (2003) in three uncontaminated fields. Interestingly, significant numbers of AOA amoA genes still remained in the Zn-exposed soil samples at T2, even though virtually no expression was observed. Thus, the prospect remains that some AOA populations may be able to survive such stress, in dormant states or as spores (Onyenwoke et al., 2004). Alternatively, these organisms may obtain energy through processes other than ammonia oxidation. Further research will be necessary to determine if AOA have long-term adaptive capabilities in response to heavy metal-induced stress.