The effect of human settlement on the abundance and community structure of ammonia oxidizers in tropical stream sediments

Ammonia-oxidizing archaea (AOA) and ammonia-oxidizing bacteria (AOB) are a diverse and functionally important group in the nitrogen cycle. Nevertheless, AOA and AOB communities driving this process remain uncharacterized in tropical freshwater sediment. Here, the effect of human settlement on the AOA and AOB diversity and abundance have been assessed by phylogenetic and quantitative PCR analyses, using archaeal and bacterial amoA and 16S rRNA genes. Overall, each environment contained specific clades of amoA and 16S rRNA genes sequences, suggesting that selective pressures lead to AOA and AOB inhabiting distinct ecological niches. Human settlement activities, as derived from increased metal and mineral nitrogen contents, appear to cause a response among the AOB community, with Nitrosomonas taking advantage over Nitrosospira in impacted environments. We also observed a dominance of AOB over AOA in mining-impacted sediments, suggesting that AOB might be the primary drivers of ammonia oxidation in these sediments. In addition, ammonia concentrations demonstrated to be the driver for the abundance of AOA, with an inversely proportional correlation between them. Our findings also revealed the presence of novel ecotypes of Thaumarchaeota, such as those related to the obligate acidophilic Nitrosotalea devanaterra at ammonia-rich places of circumneutral pH. These data add significant new information regarding AOA and AOB from tropical freshwater sediments, albeit future studies would be required to provide additional insights into the niche differentiation among these microorganisms.


Introduction
Nitrification, in which ammonia is converted to nitrite and subsequently to nitrate, is a key process in the global nitrogen cycle that is essential for the functioning of many ecosystems. Aerobic ammonia oxidation, the first step in nitrification, converts reduced inorganic nitrogen species to oxidized ones (Gruber and Galloway, 2008). This process is mediated by autotrophic ammonia-oxidizing bacteria (AOB) (Purkhold et al., 2000), and also by autotrophic ammonia-oxidizing archaea (AOA) of the Thaumarchaeota phylum (Wuchter et al., 2006).
Although efforts have been made to understand the AOB and AOA community structures and their ecological roles, the effect of human settlement associated with tropical metalmining on the diversity, abundance and distribution of these communities is still unknown. Herein, we hypothesize that such human settlement, which is also connected with mineral nitrogen pollution, would lead to different responses on the abundance and composition of AOA and AOB communities. From such different responses it might be able to detect some genera or groups as potential bioindicators of metal-mining pollution or of water quality. To address this hypothesis we used archaeal and bacterial amoA genes as a molecular marker together with Betaproteobacteria-and Thaumarchaeota-specific 16S rRNA gene-targeting primers, for tracking the distribution of AOA and AOB in tropical sediment streams with different metal concentrations due to historical pollution from metal and smelter activities. Moreover, we used the quantitative PCR approach to unveil the abundance of amoA genes among the prokaryotic community present in these freshwater sediment samples.

Study Area and Physicochemical Analysis
The Iron Quadrangle region (Minas Gerais, Brazil) is extremely rich in ores and has been historically explored. A total of six sites were sampled: three sites located in mining-impacted streams sediments, i.e., in the Mina stream (MS,19 • (Reis et al., 2014), with exception of the MTS, which is not subjected to the influence of human settlement and belongs to the environmental protection area of the Minas Gerais state spring sanitation company. Only the names of the Mina stream and the Mutuca stream are formal names.

DNA Extraction and Quantitative PCR (qPCR) Assay
Total DNA was extracted from the sediment samples (10 g wet weight) using the PowerMax soil DNA isolation kit (MoBio Laboratories, USA) according to the manufacturer's instructions. The DNA samples were stored at -20 • C until further processed. Table 3 gives details of all the primers and the PCR conditions used in this study. The copy numbers of archaeal and bacterial amoA genes were quantified using the primer set Arch-amoAF and Arch-amoAR (Francis et al., 2005), and amoA-1F and amoA-2R (Rotthauwe et al., 1997;Stephen et al., 1999), respectively. Each reaction was performed in a 20 μL volume containing 10 ng of DNA, 1 μL of 5 μM of each primer and 10 μL of SYBR Green using the Rotor-Gene SYBR Green PCR Kit (Qiagen, Hilden, Germany), specific for the Rotor-Gene Q2 plex HRM Platform (Qiagen, Hilden, Germany). In the qPCR assay clones of Escherichia coli JM109 containing the archaeal and bacterial amoA gene fragments were included as standard, generating standard curves from seven dilutions. A control reaction without template DNA was included in each qPCR assay. All DNA samples and the negative control were analyzed in triplicate to obtain an accurate value for the amoA gene abundance in each sediment. Aliquots of the qPCR products were run on an agarose gel to identify unspecific PCR products such as primer dimers or fragments with unexpected lengths (data not shown). The amplifications efficiencies of AOB and AOA amoA gene were 90 and 109%, with r 2 -values of 0.999 and 0.992, respectively.

PCR-Denaturing Gradient Gel Electrophoresis (DGGE) Analysis
In attempt to assess the diversity and distribution of ammonia oxidizers, denaturing gradient gel electrophoresis (DGGE) of the AOA 16S rRNA gene was performed by a nested PCR with the archaeal primer set 21F and 958R followed by the thaumarchaeotal primer set Parch519F and Arch915R-GC, as previously described by Vissers et al. (2009). To detect the presence of archaeal amoA genes in the community the primer set Arch-amoAF and Arch-amoAR was used as described by Wuchter et al. (2006) and Vissers et al. (2012), (Table 3).  Muyzer et al., 1993) was attached to the 5 end of the primers in bold to prevent complete melting of the PCR products during DGGE. * Forward primer designed by Rotthauwe et al. (1997), but with minor modification made by Stephen et al. (1999).
For the PCR-DGGE of the AOB 16S rRNA gene, a nested PCR was performed using the bacterial primer set βAMOf and βAMOr, which are selective but not completely specific for betaproteobacterial ammonia oxidizers followed by a second PCR using the betaproteobacterial primer set CTO189f-GC and CTO654r, as described by Laanbroek and Speksnijder (2008). To detect the bacterial amoA gene we used the primer set amoA-1F (Nicolaisen and Ramsing, 2002) and amoA-2R (Rotthauwe et al., 1997). The GC clamp described by Muyzer and Smalla (1998) was incorporated into the 5 end of the primers as indicated in Table 3.
The final PCR products were separated by DGGE in a 7% polyacrylamide gel with a vertical gradient of 35-60% of formamide and urea denaturants. The running conditions were 100 V at a constant temperature of 60 • C for 18 h.

DGGE-Band Sequencing and Phylogenetic Analysis
The phylogenetic assignment of ammonia-oxidizing prokaryotes sequences was determined by excising the prominent bands from the DGGE gels, eluting in 20 μl of sterile Milli-Q water at room temperature for 2 h. After that, the amplicons were reamplified under the conditions described above and submitted for sequencing at Macrogen Inc., Amsterdam. To identify the closest relatives the obtained sequences were compared with available database using the BLASTn search tool from GenBank 1 . The Bellerophon program (Huber et al., 2004) was used to detect and omit chimeric DNAs.
For the amoA gene, the retrieved nucleotide sequences were aligned to amoA gene sequences available on the GenBank 1 database. The evolutionary history was inferred using neighborjoining algorithm (Saitou and Nei, 1987) with Jukes-Cantor correction (Jukes and Cantor, 1969). Evolutionary analyses were conducted in MEGA 6 (Tamura et al., 2013). For both genes, the bootstrap consensus tree inferred from 1000 replicates (Felsenstein, 1985) was taken to represent the evolutionary history of the taxa analyzed. The nucleotide sequences generated were deposited into the GenBank database with accession numbers KR028198-KR028271.

Data Analysis
For the DGGE data, a dendrogram from the fingerprint was calculated using the Dice coefficient of similarity and the unweighted pair-group method with arithmetic averages (UPGMA), within the BioNumerics version 6.5 software package (Applied Maths, Sint-Martens-Latem, Belgium).
Manual selection of environmental parameters was performed after the generation of a Pearson correlation-based heatmap matrix and analysis of the clusters using the free available 1 http://www.ncbi.nlm.nih.gov/ software packages R 2 . The hierarchical agglomerate clustering was performed using the agglomeration method. Clustering was performed with complete linkage and Euclidean distances as the distance measure.
To determinate multivariate relationships between DGGE banding profiles and environmental parameters we performed a canonical correspondence analysis (CCA), using R software. Moreover, in attempt to correlate the AOA/AOB gene copy numbers ratio with the environmental parameters we performed a principal component analysis (PCA). After standardization of environmental data (by subtracting the mean from each observation and dividing by the corresponding SD), PCA was obtained using the rda function in the Vegan library program implemented in R software.

Environmental Characterization
A Pearson correlation-based heatmap was generated to arrange the environmental parameters into clusters and correlate them subsequently with the AOA and AOB communities from different sediment samples. The resulting heatmap showed three different clusters: As, NO 3 − , Cu, NO 2 − , Zn, TP and Cr (cluster 1); Ni, Mn, Fe, pH, and DO (cluster 2); PO 4 3− , Pb, NH 4 + , temperature, and conductivity (cluster 3; Figure 1). From these clusters, NO 3 − , TP, Fe, DO, and NH 4 + were chosen for PCA and CCA analyses.
Principal component analysis, which was based on these five environmental parameters, revealed that the first component (PC1) of the PCA biplot, mainly explained the positive correlation of the CS sample with DO and Fe, and a negative correlation with NH 4 + , whereas the opposite was observed for the TS sample. The MS sample was determined by the second component (PC2), being mainly positively correlated with NO 3 − . The non-impacted sites S1, S2, and MTS were all negatively correlated with NO 3 − . In addition, TP and NH 4 + were also negatively correlated, being the first with the MTS sample and the latter with the S1 and S2 samples (Figure 2).

Abundances of AOA and AOB and their Correlation with Environmental Parameters
The absolute abundances of amoA gene, i.e., the number of copies of this gene per gram of sediment, were assessed by qPCR. Archaeal amoA gene was detected in all samples, whereas for bacteria no amoA genes were found in the MTS and S2 samples (Table 4). Notably, the S2 sample presented the highest archaeal amoA gene copy number/g of sediment (1.4 × 10 5 ). Moreover, the copy numbers of bacterial amoA genes were higher than those of archaeal amoA genes in the CS, MS, and TS samples from mining-impacted sediments (from one to three orders of magnitude). It should be noted that the bacterial amoA gene abundance of the MS sample was three orders of magnitude larger than the archaeal.  From the five parameters chosen in the heatmap (Figure 1) NH 4 + was only correlated with the archaeal amoA gene abundance. As observed in Figure 3, the regression line exhibited a negative correlation between the NH 4 + concentrations and AOA amoA gene abundance (r 2 = 0.6935).

Analysis of DGGE Banding Patterns
The DGGE banding profiles demonstrated a higher number of bands in the AOA communities than in the AOB communities based on both 16S rRNA and amoA genes (Figure 4). Moreover, only one bacterial amoA gene band was detected in the nonimpacted S1 sample, whereas no bands were observed for the AOB community in the non-impacted S2 and MTS samples ( Figure 4D).
Based on a cut-off of 60% of similarity, the dendrograms based on 16S rRNA DGGE fingerprinting analysis showed rather similar AOA and AOB communities clustering with the miningimpacted sediment sample (CS) grouping with non-impacted sediment samples (Figures 4A,C). Moreover, TS and MS were the most dissimilar samples based on archaeal and bacterial 16S rRNA genes, respectively. A similar clustering among the archaeal 16S rRNA and amoA genes was also found for the S1, S2, and CS samples (Figures 4A,B).

Phylogenetic Assignment of the AOA Community
To identify the phylogenetic affiliations of members of the AOA and AOB communities selected bands from the DGGE gels were   excised and sequenced. No chimeras were detected among the sequences. For the archaeal 16S rRNA gene, 36 sequences were subjected to phylogenetic analysis (Figure 5). It should be noted that all the archaeal sequences from the TS and S2 samples were affiliated with the Thaumarchaeota, whereas the MTS sample, which is oligotrophic and is not influenced by human settlement, was comprised of Crenarchaeota-related sequences. Sequences recovered from the other stream sediments were scattered throughout the Thaumarchaeota and Crenarchaeota phyla. Most sequences from the S1 and S2 samples grouped together in the Group 1.1a-associated clade (Figure 5).
From the DGGE fingerprinting profiles of the functional amoA gene, 17 sequences were used in the construction of a phylogenetic tree (Supplementary Figure S1). Phylogenetic affiliation revealed that the sequences from the CS and MS mining-impacted streams formed a distinct clade (Thaumarchaea group 1.1a) from the non-impacted streams (Group 1.1a-associated). The TS18 sequence was affiliated with Nitrosopumilus maritimus, whereas the remaining sequences were affiliated with unclassified archaea. The majority of archaeal amoA gene sequences were associated with freshwater sediments sequences present in the GenBank database.

Phylogenetic Assignment of the AOB Community
As expected, all the AOB community sequences were affiliated with Betaproteobacteria class. The 14 retrieved betaproteobacterial 16S rRNA-based sequences were allocated phylogenetically in two clades: Nitrosospira and Nitrosomonas (Figure 6). Moreover, the 16S rRNA gene tree revealed an interesting correlation between phylogenetic clustering and trophic state of the environment, i.e., oligo-and ultraoligotrophic (Nitrosospira) or meso-and eutrophic (Nitrosomonas) sediments.
All the amoA gene sequences from the CS, TS, and MS samples did not group into the Nitrosospira or Nitrosomonas clades (Supplementary Figure S2). Furthermore, all amoA gene sequences obtained in this study were more closely related to themselves than to reference sequences. Interestingly, sequences from the TS and MS samples, which exhibited high NH 4 + concentrations, were most closely related to sequences recovered from activated sludge and a wastewater treatment plant, respectively. The CS sample sequences were closely related to sequences derived from an estuarine sediment and a freshwater lake. Figure 7 shows the correlation of the five selected environmental parameters with both AOA and AOB community compositions based on the banding profile of both the 16S rRNA and amoA genes. The CCA for the archaeal 16S rRNA gene community showed S2 and CS to be positively correlated with Fe, TP, and NO 3 − . Moreover, S1 was positively correlated with DO and negatively correlated with NH 4 + , whereas the opposite was observed for the MS community ( Figure 7A). The bacterial 16S rRNA gene communities from CS, S2, and MTS were positively correlated with Fe, TP, and NO 3 − . Again, MS was positively correlated with NH 4 + and negatively with DO, and S1 was positively correlated with DO ( Figure 7B).

Correlation of Ammonia Oxidizers Community Structure with Environmental Parameters
Additionally, using the amoA gene banding profile, the AOA communities recovered from the S1 and S2 samples were positively correlated with NO 3 − , and MTS was negatively correlated with Fe. Otherwise, the MS and TS communities were positively correlated with NH 4 + and negatively with DO, whereas the CS community was positively correlated with Fe ( Figure 7C).
Regarding to the bacterial amoA gene, the TS community was equally explained by both axes CCA1 and CCA2, being negatively correlated with all tested parameters. NH 4 + was the parameter which best explained the segregation of the samples in the plot, and was positively correlated with the MS community. The AOB CS and S1 communities were positively correlated with Fe and NO 3 − , respectively ( Figure 7D).

Discussion
Many studies have revealed that bacterial and archaeal ammoniaoxidizing communities are shaped by different environmental drivers in many ecosystems (Tourna et al., 2008(Tourna et al., , 2011Daebeler et al., 2012;Liu et al., 2014). Overall, our findings demonstrated that distinct environmental characteristics can affect the distribution of AOB communities, clustering them in species-specific niches with Nitrosomonas dominating in meso-and eutrophic sediments, and Nitrosospira prevailing in oligo-and ultraoligotrophic sediments. Ammonia oxidizers from sediments often experience oxygen depletion due to competition with facultative anaerobes, favoring microaerophilic or anaerobic respiration (Arp et al., 2007). The dominance of Nitrosomonas-related sequences from the MS sample could be explained by the inverse correlation between NH 4 + and DO concentrations. Although we were not able to assign the obtained sequences to more narrow defined Nitrosomonas and Nitrosospira clusters, it is interesting to note for example that the Nitrosomonas eutropha genome analysis (IMG, 2006a) revealed genes related to microaerophilic respiration, promoting its growth in environments with oxygen depletion in comparison with Nitrosospira multiformis, which lacks these genes in its genome (IMG, 2006b). In addition, it should be highlighted that the Nitrosospira and Nitrosomonas clades harbor clusters from different environments (Freitag and Prosser, 2003 and references therein). Among these clusters, cluster 7 encompasses the N. eutropha and N. europaea species, which are often recovered from particularly NH 4 + -rich environments, such as wastewater treatment plants (Hugenholtz et al., 1998;Herbert, 1999; , 1999). Another interesting characteristic that could have favored the occurrence of Nitrosomonas in the CS and MS samples is the capacity of N. eutropha to cope with exposure to toxic compounds (Arp et al., 2007). Therefore, human settlement activities could favor Nitrosomonas over Nitrosospira. These findings are in agreement with the study of Dang et al. (2010), who observed that eutrophic sediment with high concentrations of heavy metals favor the predominance of Nitrosomonas over Nitrosospira, likely due to exogenous input of microbes and nutrients via polluted rivers.
Regarding to AOA phylogeny, it was not possible to obtain a deeper affiliation of the Thaumarchaeota phylum, as all retrieved amoA gene sequences were related to unclassified taxa, corroborating with a study in another freshwater system (Vissers et al., 2012). Thus, these sequences could represent novel variants of the amoA gene or genes related to unknown ammonia oxidizers. In addition, our findings show that the diversity of amoA genes is larger than earlier described, once none amoA gene sequence were affiliated with known reference strains. The AOA communities were grouped according to the origin of each sampled sediment, as also reported by Pester et al. (2012) for geographically distinct soils. Nitrosotalea devanaterra was formerly described by Lehtovirta-Morley et al. (2011) as an obligate acidophilic ammonia oxidizer (pH from 4 to 5), growing at extremely low ammonia concentration (0.18 nM). In contrast to this report, N. devanaterra (Figure 5 and Supplementary Figure S1) was only detected in the nonimpacted S1 and S2 sediments characterized by pH 6.5 and NH 4 + concentration of 1,390 nM. Thus, our finding suggests the ability of members of this taxon to occupy NH 4 + -rich and circumneutral pH niches, likely due to the presence of novel ecotypes with different physiological characteristics within the Nitrosotalea group. N. devanaterra was also the most abundant phylotype among the Thaumarchaea encountered in a volcanic grassland soils on Iceland, which had a pH of 6.7 and an NH 4 + concentration of 0.5-1.1 μg per g dry soil (Daebeler et al., 2014). Some members of the soil Thaumarchaeota group 1.1b, such as Nitrososphaera found in this study (Figure 5), are able to hydrolyze urea leading to an ecological advantage in NH 4 + -poor environments (Lu et al., 2012). In the present study the sites are NH 4 + -rich, demonstrating the ability of Nitrososphaera to deal in environments with a wide range of NH 4 + concentration. Most of the studies showed that archaeal amoA genes are more abundant than bacterial amoA genes in many ecosystems (Adair and Schwartz, 2008;Dang et al., 2008;Prosser and Nicol, 2008;Herrmann et al., 2009), with a few reports showing contradictory results in environments such as an activated wastewater treatment bioreactor (Wells et al., 2009), a lake sediment (Jiang et al., 2009), an estuarine sediment (Mosier and Francis, 2008), and now also in mining-impacted freshwater sediments as observed in the present study. In our study, AOB were about one or three orders of magnitude more abundant than AOA in the MS, CS, and TS samples, but not in the S1 sample, suggesting that AOB might be the primary drivers of ammonia oxidation in the sediments impacted by human settlement. Indeed, we did not detect bacterial amoA genes in the S2 and MTS samples and rarely in the S1 sample. Auguet et al. (2012) also did not find these genes and reported that it was due to the NH 4 + concentrations in the water, which were below the AOB affinity threshold value, i.e., <1 μM NH 4 + (Bollmann et al., 2002). In our study, S1 and S2 showed a NH 4 + concentration of 1.39 μM, which is slightly higher than this threshold value, whereas the MTS sample had an NH 4 + concentration that was lower than the threshold value (i.e., 0.69 μM). A key finding in our study was that the NH 4 + concentration was only inversely proportional to the AOA amoA gene abundance and not to the AOB amoA gene abundance, likely indicating that lower NH 4 + concentrations lead to higher numbers of archaeal amoA gene copies, but did not affect the bacterial amoA gene copy numbers. Interestingly, Verhamme et al. (2011) reported that high NH 4 + concentrations were proportional to the abundance of bacterial amoA gene copies. These findings could suggest that AOA and AOB occupy separate ecological niches, with AOA contributing for the nitrification of ammonia released through mineralization, whereas AOB dominate under high NH 4 + concentrations. Our data provide insights into the influence of physicochemical parameters of the overlying bulk water on species or ecotypes differentiation of ammonia oxidizer communities in stream sediments. Taken together, our data reveal a different response within the AOB community with Nitrosomonas taking advantage over Nitrosospira in environments impacted by human settlement. The findings also reveal a high diversity of largely unknown species or ecotypes of the Thaumarchaeota, such as members related to N. devanaterra inhabiting ammonia-rich habitats of circumneutral pH values. Moreover, our results demonstrated that the bacterial amoA gene was more abundant than the archaeal amoA gene in the miningimpacted streams. Regarding that the understanding about the evolutionary history and metabolic repertoire of ammonia oxidizers is still in its beginning, future works would be required to provide additional insights into niche differentiation among these microorganisms.