OUP user menu

Arsenite modifies structure of soil microbial communities and arsenite oxidization potential

Raphaël Lami , L. Camille Jones , Matthew T. Cottrell , Brandon J. Lafferty , M. Ginder-Vogel , Donald L. Sparks , David L. Kirchman
DOI: http://dx.doi.org/10.1111/1574-6941.12061 270-279 First published online: 1 May 2013

Abstract

The influence of arsenite [As(III)] on natural microbial communities and the capacity of exposed communities to oxidize As(III) has not been well explored. In this study, we conducted soil column experiments with a natural microbial community exposed to different carbon conditions and a continuous flow of As(III). We measured the oxidation rates of As(III) to As(V), and the composition of the bacterial community was monitored by 454 pyrosequencing of 16S rRNA genes. The diversity of As(III)-oxidizing bacteria was examined with the aox gene, which encodes the enzyme involved in As(III) oxidation. Arsenite oxidation was high in the live soil regardless of the carbon source and below detection in sterilized soil. In columns amended with 200 μmol kg−1 of As (III), As(V) concentrations reached 158 μmol kg−1 in the column effluent, while As(III) decreased to unmeasurable levels. Although the number of bacterial taxa decreased by as much as twofold in treatments amended with As(III), some As(III)-oxidizing bacterial groups increased up to 20-fold. Collectively, the data show the large effect of As(III) on bacterial diversity, and the capacity of natural communities from a soil with low initial As contamination to oxidize large inputs of As(III).

Keywords
  • arsenite oxidation
  • soil bacterial communities
  • molecular approaches
  • 16S rRNA pyrosequencing

Introduction

Arsenic (As) contamination of soils and waters is a major public health issue in many parts of Southeast Asia and around the world (Nickson et al., 1998; Nordstrom, 2002; Smedley, 2003; Duker et al., 2005; Bhattacharya et al., 2007; Fendorf et al., 2010). This contamination is due to natural and anthropogenic release of various chemical forms of As. The environmental mobility and toxicity of inorganic As in soils is dependent on oxidation state, which is typically either arsenite [As(III)] or arsenate [As(V)] (Cullen & Reimer, 1989; Masscheleyn et al., 1991; Borch et al., 2010). At circumneutral pH, As(V), an oxyanion, can sorb to clay and mineral component of soil such as iron (Fe) oxides. Because arsenite is more mobile and more toxic than As(V) due to its neutral charge in soil, As(III) oxidation is a key reaction in As detoxification and immobilization (Oremland & Stolz, 2003). Because arsenite oxidation by oxygen (O2) is kinetically slow, soil minerals and bacteria are the relevant As(III) oxidants (Eary & Schramke, 1990). As a result of this complexity, the influence of As(III) on microbial communities and the role of microbes in As(III) oxidation in soil remain poorly characterized.

A few studies have examined the effects of As(III) contamination on bacterial composition and community structure (Jackson et al., 2001; Macur et al., 2004; Bruneel et al., 2006; Lorenz et al., 2006; Bouskill et al., 2010; Halter et al., 2011). However, because various other metals, such as cadmium or copper, also contaminated these environments examined by previous studies, the effects of specifically As(III) on microbial communities remain unclear. Additionally, most studies on As(III)-oxidizing bacteria have focused on cultivated strains, which are a small (< 1%) percentage of the total bacterial community of natural environments (Liao et al., 2011). A few studies examined uncultivated potential As(III) oxidizers as detected by 16S rRNA genes (Jackson et al., 2001). However, these genes are poor predictors of potential As(III) oxidation because this process is potentially carried out by many different bacteria (reviewed in Oremland & Stolz, 2003). Furthermore, few published studies have linked As(III) oxidation rates with As(III)-oxidizing gene diversity and abundance.

Arsenite oxidation is mediated by arsenite oxidase, a heterodimeric periplasmic enzyme containing molydopterin (Ellis et al., 2001) encoded by aox genes (Silver & Phung, 2005). The aoxB genes have been found in cultivated bacteria from diverse prokaryotic groups, including Alphaproteobacteria, Gammaproteobacteria, and Archaea (Turner, 1954; Osborne & Enrlich, 1976; Oremland & Stolz, 2003; Quéméneur et al., 2008). These genes have been used successfully to monitor As(III) oxidizers in natural environments (Oremland & Stolz, 2003; Inskeep et al., 2007; Quéméneur et al., 2010; Heinrich-Salmeron et al., 2011). Culture-independent studies of aox genes have suggested that a few Alphaproteobacteria (mainly Bosea sp.), Betaproteobacteria (mainly Acidovorax), and Gammaproteobacteria (mainly Pseudomonas) are the major As(III) oxidizers in soil and freshwater environments (Oremland & Stolz, 2003). These reports suggest that the capacity for As(III) oxidation is widespread and could be important in the microbial world.

The goal of this study was to quantitatively evaluate the impact of As(III) on microbial communities in flow-through column experiments with and without an additional carbon source. We measured the time dependency of bacterial vs. abiotic driven As(III) oxidation by comparing live and sterile soil columns. We then examined the diversity of 16S rRNA genes by 454 pyrosequencing to monitor bacterial community structure before and after amendment with As(III) and organic material. The diversity and abundance of As(III)-oxidizing genes were examined with aox gene targeted clone libraries, and the abundance of these oxidizers was estimated using quantitative PCR (qPCR) analysis of aox genes. Our results show the importance of biological processes in As oxidation even though As(III) selected against many bacteria and reduced overall diversity of soil microbial communities.

Materials and methods

Design of soil columns experiments

Soil samples of the Matapeake series were collected from a depth of 5–10 cm at the University of Delaware Farm, Newark, DE (39° 39′57″ N; 75° 45′05″ W). Aliquots of soil samples were air-dried for soil analysis and sterilization. The Matapeake soil is a silt loam with 13.5% sand, 64.5% silt, and 22% clay. The cation-exchange capacity (CEC) was 16.34 cmol kg−1, pH was 6.2, bulk density of the column soil was 1.32 g cm−3, and organic matter content, estimated by loss on ignition, was 2.6%. Air-dried soil was acid-digested (nitric and hydrochloric acid) with microwave heating following the protocol for EPA 3051 soil analysis and analyzed by inductively coupled plasma optical emission spectrometry (ICP-OES) (US EPA, 1994). Total Fe was 15 476 mg kg−1, and total Mn was 400 mg kg−1, total Pb was 76.4 mg kg−1, and total As was 15.5 ± 4.9 mg kg−1. Air-dried samples of the Matapeake soil were sterilized by autoclaving (1 h, 121 °C) and γ-irradiation, which is considered to have minimal geochemical effect on soil (McNamara et al., 2003). Air-dried soil was irradiated with a dosage of 25 000 Gy using a Sheperd Mark 1 γ-irradiator.

Soil columns were packed immediately after sampling to preserve the activity of native bacteria and were prepared in triplicate. All columns were assembled in a sterile aerobic environment at ambient room temperature. Fresh soil was gently crushed at field moist conditions, large (> 200 μm) particles were removed, and remaining soil was mixed with sterile coarse quartz sand in a 30 : 70 soil to sand ratio by mass. Column soils were not dried to minimize changes that may affect microbial community. Quartz sand was added to increase the porosity and prevent development of preferential flow paths. The soil–sand mixture was packed into sterile 20-mL plastic syringes (8.5 cm long) and capped with sterile glass wool on bottom and rubber stoppers on top. Influent entered at the bottom of the columns to prevent preferential flow paths and to reduce the effect of gravity on flow velocity.

The soil columns were amended with As(III)-spiked media with and without additional organic material. The concentration of As(III) used in all columns was 200 ± 50 μM added as NaAsO2. One measurement of As concentration was carried out per column. Organic carbon-amended media (OM media) included 5 mg L−1 yeast extract, 10 mg L−1 tryptic soy broth, and 500 mg L−1 NaCl. Organic carbon-free media contained only 500 mg L−1 NaCl. All media had trace metals added at the following concentrations: 0.100 mg L−1 MnCl2-4H2O, 0.120 mg L−1 CoCl2-6H2O, 0.070 mg L−1 ZnCl2, 0.060 mg L−1 H3BO3, 0.025 mg L−1 NiCl2-6H2O, 0.015 mg L−1 CuCl2-2H2O, 0.025 mg L−1 Na2MoO4-2H2O, and 0.001 mg L−1 12 M HCl. The pH of all media was 6.5–7. Media solutions were sterilized by autoclaving.

A peristaltic pump was used to flow media through the columns at a rate of 10 mL h−1, resulting in a pore water velocity of 4.25 cm h−1. Aerobic conditions in the columns were verified by monitoring Fe(II)/Fe(III) in the effluent solution using the ferrozine spectrophotometric method (Stookey, 1970). Samples of the effluent were taken daily, filtered with a 0.22-μm nylon filter, and refrigerated prior to As speciation analysis. Arsenic speciation [As(III) and As(V)] was determined by high-performance liquid chromatography with inductively coupled plasma mass spectrometry (HPLC-ICP/MS) (HPLC: Agilent 1200; ICP/MS: Agilent 7500cx) following the protocol described in Parikh et al. (2010). Arsenite concentration over time was measured in triplicate columns, and all As chemical data errors are the standard deviations of these replicates.

The experiment consisted of four treatments: (1) sterilized soil and As(III) in the flow-through media; (2) live soil and As(III) and without the OM media; (3) live soil with the OM media; and (4) live soil with As(III) and organic material in the flow-through media. The experiment was run for 22 days after which the columns were sacrificed for chemical and molecular analyses of bacterial communities. All samples for microbial analyses were preserved at −80 °C until analyses. The columns were run in triplicate. The molecular biology analyses were conducted on one column per treatment, but the qPCR assays were run in triplicate. The As(III) and As(V) kinetics were very similar in the triplicate columns, suggesting that the biological variation among the three columns was minimal. Also, our conclusion about the effect of As on bacterial diversity was based on two As-treated columns vs. two samples without added As (the initial soil and the column receiving only organic material; see Results).

Nucleic acids extraction and identification of aoxB genes

DNA was extracted using an UltraClean Soil DNA Isolation Kit (MoBIO) and quantified in triplicate via a standard picogreen assay (Invitrogen) on a POLARstar Optima fluorometer (BMG Labtech). The aoxB genes were recovered using universal aoxB genes primers (Table ) described previously (Quéméneur et al., 2008). A detailed review and list of phyla targeted by these primers are discussed by Quéméneur et al. (2008). Ten nanograms of DNA was used in PCR amplifications, with the following conditions: one cycle of 94 °C, 2 min, followed by 30 cycles of 94 °C, 30 s; 63 °C, 30 s; and 72 °C, 30 s. Final concentrations of primers and MgCl2 were 1.87 mM and 2 mM, respectively. The PCR products were subjected to electrophoresis (1.5% agarose), and the products of the expected size were excised from the gel and eluted with a Gene Clean gel extraction kit (Q-Biogene). The purified fragments were cloned with a TOPO TA cloning kit for sequencing (Invitrogen) according to the manufacturer's instructions. Colonies were screened via PCR to determine the insert size, and 120 clones of the expected size were sequenced by standard Sanger sequencing. Sequences were trimmed of the vector sequence and aligned using the clustal program in Mega version 3.1 (Tamura et al., 2007). Sequences were added to an alignment containing most of the known aoxB sequences. Within-group mean distances were calculated in Mega 3.1. Phylogenetic trees of the translated sequences were constructed in Mega 3.1 with the default parameters, based on both neighbor-joining and minimum evolutionary algorithms with 500 bootstrap replicates. Rarefaction analysis, coverage estimations, and Shannon diversity index calculations were performed with DOTUR (Schloss & Handelsman, 2005). The libraries were compared using libshuff (Singleton et al., 2001).

View this table:
1

List of PCR and qPCR primers used in this study

NameTargetSequence (5′ -> 3′)Annealing (T °C)Reference
w49 dirAll bacterial 16S rRNA genesCGG TCC AGA CTC CTA CGG61°Delbès et al. (2001)
w34 revTTA CCG CGG CTG CTG GCA CLee et al. (1996)
aoxBM1-2FAll aox genes carrying bacteriaCCA CTT CTG CAT CGT GGG NTG YGG NTA52°Quéméneur et al. (2008)
aoxBM”-2RTGT CGT TGC CCC AGA TGA DNC CYT TYT CQuéméneur et al. (2008)
G1_27FAox genes from bacterial group 1GTC GTC CAG GAA CAA GGT69°This study
G1_209RTCG CCT ACT CCC ATA TCAThis study

Quantification of aoxB and 16S rRNA gene abundance

One group of aoxB genes (Group 1), related to Alphaproteobacteria, was targeted by qPCR. This group was selected because of its relative abundance in aoxB clone libraries (see Results). Primers for this aoxB groups were designed using ARB (Ludwig et al., 2004) with an alignment containing a large set of environmental and strain-derived aoxB sequences. One aoxB-containing plasmid was used as a positive control and for standard curves in the qPCR assays. The aoxB-containing plasmid related to the targeted Alphaproteobacteria Group 1 was isolated from environmental clones (Qiaprep kit, Qiagen) and linearized with PstI (Invitrogen). Products were quantified in triplicate using the picogreen assay. Standard reactions with linearized plasmid contained approximately 101–106 copies per reaction. All standard curves were linear within the ranges tested. All qPCR reactions were performed in triplicate or quadruplicate with 1 μL of diluted DNA (0.1 ng mL−1) in a final volume of 12.5 μL using the Stratagene SYBR green mix with an ABI 7500 (Applied Biosystems). The qPCR conditions were an initial denaturation step at 95 °C (10 min), followed by amplification at 95 °C for 15 s, the indicated annealing temperature for 45 s (Table ), and 72 °C for 45s, with a final dissociation step. Final primer concentrations were 0.2 mM. Only single peaks were observed in the dissociation curves for both the standards and samples (data not shown), indicating specific amplification with each set of primers. Amplification efficiencies were between 95% and 105%, and no inhibition was detected when a known quantity of standard was added to each sample (data not shown). The detection limit of qPCR assays was about 10 copies. To verify the specificity of the aoxB primers, the qPCR primer pair was used to amplify aoxB genes from the column experiments, and the resulting products were cloned. The PCR conditions were as outlined above and in Table . PCR products were cloned and sequenced as described above. Of 10 clones that were analyzed, all contained the expected aoxB sequences related to the targeted group. We evaluated the coverage of our new primer set targeting Group 1. These tests indicated that our primers match 60% of sequences relative to this group in GenBank allowing for one mismatch for our probes. 16SrRNA genes were quantified following a similar protocol, using primers indicated in Table .

Sequencing of 16S rRNA genes

All 16S rRNA genes were pyrosequenced on a Roche 454 FLX instrument (Research and Testing Laboratory, Lubbock, TX) using a mixture of Hot Start and HotStar high-fidelity taq polymerases, which generated amplicons of 250–550 bp starting from position 27 (Escherichia coli 16S rRNA number). Using QIIME (Caporaso et al., 2010), the 16S rRNA sequences were demultiplexed and quality filtered, including homopolymer removal. Sequences were denoised using PyroNoise (Quince et al., 2009), and chimeras were removed using ChimeraSlayer (http://sourceforge.net/projects/microbiomeutil). The data were randomly resampled to check whether the observed patterns were similar to those observed when each sample had the same number of sequences. Diversity indices were also calculated using QIIME (Quince et al., 2009). All 16S rRNA genes sequences were deposited in GenBank (JN388250JN388300).

Results

As(III) oxidation

We examined the rates of biotic and abiotic As(III) oxidation and the effect of As(III) on microbial communities over 22 days in soil columns. Arsenite oxidation was not detected in the sterilized soil, as As(III) was near the added concentration in the effluent over the entire experiment, while As(V) remained below 18 μmol kg−1 (Fig.a). In contrast, oxidation was apparent in live columns where As(III) was added (Fig.b and c). After the first day, the concentration of As(III) in the effluent from the live soil column amended with only As(III) was 140 μmol kg−1 or about 60% of the As(III) concentration in the influent (Fig.b). Arsenite oxidation remained steady in the live soil column for the first 10 days of incubation. After approximately 12 days, As(III) oxidation increased substantially, and after 16 days As(III) oxidation was high enough to reduce the concentration of As(III) to only 1% of the added concentration in the influent (Fig.b). Microbial As(III) oxidation resulted in a concentration of As(V) equal to 180 μmol kg−1 in the effluent, which accounted for 80% of the decrease of As(III) within the column. High rates of As(III) oxidation continued for the remaining 8 days of incubation (Fig.b).

1

As(V) and As(III) in effluent from column experiments. (a) Sterile soil, (b) arsenite-amended soil, and (c) arsenite and OM media-amended soil. Error bars represent the standard deviation of As concentrations over time in triplicate columns.

Based on concentrations of As(III) in the effluent, As(III) oxidation appeared to be similar in soils with and without the organic material (OM) addition, at least for the first 8 days of the incubation (Fig.B and C). By the end of the experiment, the concentration of As(III) in the effluent from the soils with and without OM averaged 158 ± 36 μmol kg−1 and 128 ± 10 μmol kg−1, respectively. The OM addition resulted in more As(V) being eluted after 10 days compared with the column without OM added (Fig.B and C). After 10 days, the concentration of As(III) in the effluent from columns with and without the OM addition averaged 10 ± 18 μmol kg−1 and 44 ± 40 μmol kg−1, respectively (Fig.B and C), and the majority of As recovered was As(V).

Total As content was measured in the initial soil, and after the column experiments were completed to examine As retention. The initial soil contained approximately 0.21 ± 0.0013 mmol kg−1 of As, which originated from historical pesticide use at the soil collection site. The column treated with only As(III) retained the most As (2.05 ± 0.26 mmol kg−1) followed by the column treated with As(III) and OM (1.78 ± 0.20 mmol kg−1). The sterile soil had about half the amount of residual As(III) (0.98 ± 0.07 mmol kg−1) compared with the live soils. Retention of As indicates that As(III) oxidation was accompanied by sorption of As(V) to mineral or organic phases in the soil. Sorption of As was enhanced in columns with live soil and significant but lower in sterile soils.

Bacterial community structure

A total of 1300 operational taxonomic units (OTUs, grouped at ≥ 97% similarity) were identified in the soil used for the column experiments. The coverage index indicated that only 60% of the total 16S rRNA gene diversity was sampled (Fig.). The rarefaction curves and diversity indices indicated that diversity was lowest in the columns with added As(III) (Fig.). The Shannon index was highest in the initial soil (7.1 ± 0.32) and lower in all treatments (< 7). When As(III) was added, the Shannon index was 5.4 ± 0.09, lower than the value (6.1 ± 0.21) found in the columns with As(III) and OM. In the OM treatment without added As(III), this index was 6.9 ± 0.23. Similar trends were found using the Simpson index, and after sample size was equalized among all samples by random resampling (data not shown). This decrease of bacterial diversity was also evident in Chao1 values, which were as much as 50% lower in soils with added As(III) than in the initial soil (data not shown).

2

Rarefaction analysis of 16S rRNA genes in the various treatments and in the initial soil. Operational taxonomic units (OTU) were defined as those sequences that were ≥ 97% similar. Coverage is given from a Chao1 analysis.

We analyzed the degree of similarity among the different communities by UPGMA clustering of the 16S rRNA gene sequences (Fig.). The communities were similar in the top, middle, and bottom section of the columns (similarity index > 70%), and any differences were small compared with the variations between the different columns (similarity index ≤ 55%). The sterilized soil differed the most from the initial soil. The communities in OM amended and As(III)+OM amended columns were the most similar (55%), while the columns with only added As(III) differed the most (Fig.). An analysis based on OTU presence/absence rather than abundance was similar (data not shown).

3

UPGMA dendrogram of 16S rRNA genes in the various treatments and in the initial soil, based on OTUs sharing 97% similarity (Bray–Curtis distance).

We used rank-abundance curves to examine the evenness of the soil microbial communities exposed to As(III). In the As(III)-amended columns, three OTUs dominated the community at the end of the experiment, while the abundance of all other OTUs was low (Fig.). By contrast, the microbial communities not exposed to As(III) were more even, and the rank-abundance curves were below those for the As(III)-exposed communities. The rank-abundance curves of microbial communities not exposed to As(III) were very similar to the rank-abundance curves for the initial soil community (Fig.). In the treatments where only As(III) was added, an OTU affiliated with the genus Acidovorax dominated the bacterial community, accounting for 9.6% of total 16S rRNA gene sequences. Similarly, in the treatment where As(III) and OM were added, Chitinophaga dominated the community, with 12% of all 16S rRNA gene sequences. By contrast, no single OTU dominated the community in the other treatments or in the initial soil. In those soils, the relative abundances of six OTUs were 1–2% of all 16S rRNA gene sequences, while all other sequences were < 1% of the total community (data not shown).

4

Abundance-rank curves for the different treatments. For clarity, the figure gives only the 25 most abundant OTUs, which accounted for 40% of all sequences. The OTUs are defined at 97% similarity. The curves from the two As(III)-amended treatments were significantly higher than the OM-only amended treatment and the initial soil (Student's t-test, P < 0.005).

Diversity of arsenite-oxidizing genes

Most of the aoxB gene sequences found in this study were difficult to assign to known As oxidizers because they were very different from aoxB genes in cultivated strains (Fig.). In all treatments, the As(III)-oxidizing genes were most similar to those from Alphaproteobacteria (Group 1 and 8, for example) and Betaproteobacteria-related taxa (Group 8 and 9). No aoxB genes related to Gammaproteobacteria or Chloroflexi were detected (Fig.). Overall, the diversity and composition of aoxB genes did not differ significantly among the different treatments and from the initial soil (P > 0.05; libshuff analysis).

5

Phylogenetic tree of aoxB amino acid sequences based on a neighbor-joining analysis of translated nucleotide sequences. Numbers on branches represent bootstrap proportions from 500 replicates. The phylogenetic identification was based on cultured representatives containing aoxB genes.

Abundance of Group I aoxB genes and 16S rRNA genes

The aoxB sequence data were used to design primers to target alphaproteobacterial Group 1 (defined in Fig. ) and to estimate its abundances using qPCR. Bacterial abundance differed significantly among the different treatments. We detected 2.4 × 106 ± 1.7 × 106 copies of Group1 aoxB related genes per g of soil in the initial sampled soil (Fig.a). This abundance sharply increased in the As(III)-only column (up to 1.7 × 107 ± 1.3 × 106 copies per g of soil) and in the As(III)+OM column (up to 8.07 ± 1.3 × 106 copies per g of soil). In contrast, the abundance of Group 1 aoxB genes did not differ significantly between initial soil and the OM-only treatment (Fig.a).

6

(a) Abundance of aoxB genes related to Group 1, as defined in Figure . Top, middle, and bottom refer to the different sections of the columns. (b) Abundance of total 16SrRNA genes. Top, middle, and bottom refer to the different sections of the columns. Errors bars indicate the standard deviations of triplicate measurements.

Bacterial abundance in the different treatments was assessed by qPCR assays of 16S rRNA genes. The initial soil had 5.2 × 107 ± 8.9 × 106 copies of 16S rRNA genes per g of soil (Fig.b). Abundance of 16S rRNA genes increased in all treatments and in all sections of the columns, reaching about 108 copies of 16S rRNA genes per g of soil. Abundances were highest for the middle section of the AsIII + OM treatment, with 1.1 × 109 ± 1.1 × 108 copies of 16S rRNA genes per g of soil.

Discussion

In this study, our objective was to quantitatively evaluate the impact of As(III) on natural soil microbial communities by measuring microbiologically driven As(III) oxidation, the effect of As(III) toxicity on microbial diversity, and the growth of potential As(III) oxidizers over time in soil columns. We found that As(III) oxidation was primarily driven by biological or coupled biogeochemical processes, as oxidation was not measurable in sterile controls. Because of microbially driven oxidation, As(III) decreased to unmeasurable levels, while As(V) increased to nearly the maximum possible concentrations. With added As(III), bacterial diversity decreased by nearly one-half, and the dominant bacterial species shifted, resulting in a bacterial community enriched in As(III)-oxidizing genes. The abundance of some types of arsenite-oxidizing genes increased nearly 20-fold over initial levels.

Arsenite oxidation in the two treatments with live soil was greater than in the two controls, suggesting that purely chemically driven oxidation of As(III) in these experiments was limited. Because soil minerals such as manganese (Mn) oxides can oxidize As(III) rapidly (Lafferty et al., 2010), we expected that the abiotic component would have a greater role in our column experiments. However, the soil used in these experiments had relatively low Mn content (Jones, 2011), and oxidization by Mn oxides may have been low due to age, crystallinity, and surface passivation of the oxides (Post, 1999; Lafferty et al., 2010). As(III) oxidation by O2 was likely low because it is a very slow process (Cherry et al., 1979; Langner et al., 2001; Eary & Schramke, 1990). The rapid biological As(III) oxidation observed in this study is similar to rates observed previously with As(III)-oxidizing bacteria (Langner et al., 2001; Kulp et al., 2004; Macur et al., 2004; Jones et al., 2012). Our observation that adding OM enhanced oxidation of As(III) is consistent with a previous study using isolated strains (Santini et al., 2002). Also, the OM addition probably led to the appearance of copiotrophic bacteria, including Gammaproteobacteria, as previous shown in other studies (Cleveland et al., 2007).

Soil minerals still probably have a role in As redox reactions, and they were likely involved in the retention of As in our soils (Macur et al., 2004). Soil minerals including Mn and Fe oxides and hydroxides are known to sorb As, and those minerals were likely responsible for some of the sorbed As found after digesting the soil at the end of our experiment (Fendorf et al., 1997; Tournassat et al., 2002). The live soils retained about twofold more As than the sterile soil, suggesting that removal of As from solution in the columns was a coupled biogeochemical process.

We found that As had a large impact on bacterial communities in these soil columns. Among the few relevant previous studies, one also found major variations in bacterial community structure due to As(III) mixed with other metals in riverine waters (Bouskill et al., 2010). The toxicity of As(III) probably explains the large decrease in bacterial diversity in our soil columns. However, some of the bacteria in our experiments seem to have had enough time, given a generation time of days (Demoling et al., 2008), to adapt to the added As(III). Longer experiments are needed to determine whether diversity recovers to levels seen in the absence of added As(III).

Addition of As(III) led to a microbial community dominated by a few types of bacteria, based on analyses of 16S rRNA genes. By contrast, no phylotype dominated the community in the treatments without As(III). Interestingly, the dominant bacteria in the As(III) and As(III)+OM treatments were affiliated with the genera Acidovorax and Chitinophaga, respectively, which have previously been reported in As(III)-contaminated environments (Santini et al., 2002; Shivaji et al., 2011). These results suggest that some phylotypes are able to grow in the presence of As(III), at least at the concentration of 200 μM provided in the study. Our qPCR measurements confirmed that some bacterial groups were able to grow in the treatments, as 16S rRNA gene abundance increased substantially during the experiment, up to 20-fold. The phylotypes that dominated the bacterial community in As(III) treatments were probably resistant to this metalloid (Paez-Espino et al., 2009). Our analysis revealed that in all treatments, some Alphaproteobacteria and Betaproteobacteria phylotypes were aox gene carriers. Some of these phylotypes may be chemolithotrophs able to oxidize As(III). Interestingly, the diversity of aoxB genes did not vary substantially among the different treatments and in comparison with the initial soil. Previous studies demonstrated that As(III) oxidizers are able to grow over a large range of environmental conditions, including a variety of carbon and energy sources (Santini et al., 2002).

Our data on the abundance of aoxB genes show that As(III)-oxidizing population increased when As(III) was added. This result is not surprising because As(III) oxidizers in As(III) contaminated environments have been easily cultivated (Santini et al., 2000; Liao et al., 2011). Our observation is also consistent with Quéméneur et al. (2008), which found that the total abundance of aoxB genes as measured by qPCR correlated with As(III) concentrations in French rivers. These results demonstrate that high fluxes of As(III) lead to bacterial communities enriched in As(III)-oxidizing genes.

In this study, we show that the oxidation of As(III) in soil were driven by microbially mediated processes. Also, we found that As(III) shifted the dominant bacterial species and was probably toxic to some species because bacterial diversity decreased substantially. These effects led to the selection of adapted As(III) oxidizers, which increased in the whole community and maintained rapid As(III) oxidation, especially in the presence of an additional source of organic carbon. Our study suggests that microbially driven oxidation and detoxification of As(III) in the natural environment can be driven at high rates by natural microbial populations.

Acknowledgements

We thank Liying Yu for help in soil DNA extraction and Dr. Cameron Koch of the University of Pennsylvania Perelman School of Medicine for performing soil gamma irradiation. This work was supported by grants from the National Science Foundation (EPSCoR EPS-0814251 and EAR 0544246), the State of Delaware, and the Partners for Universities Fund and the Environmental Protection Agency (EPA) Science to Achieve Results (STAR) graduate fellowship.

References

View Abstract