Site description and sampling

Six arsenic contaminated groundwater samples (designated as AS1, AS2, AS3, AS9, AS40 and AS41) were collected from two most affected sites situated at Barasat and Chakdaha of North 24 Parganas and Nadia districts, respectively of West Bengal, during August, 2008 and January, 2009 (locations shown in Table 1). Samples were collected from household (domestic) tubewells. For sample collection no specific permission was required; however, the owners of the wells were informed about this work that is being carried out through our collaborators. The field study did not involve endangered or protected species. All the samples were collected in sterile 500 ml glass bottles after pumping the tubewells continuously for 15^_30 min until the temperature (°C), electrical conductivity (EC), dissolved oxygen (DO), oxidation reduction potential (ORP) and pH readings got stabilized. These parameters were measured on site with portable Orion multiparameter meter (Thermo electron corporation, Beverly, MA) which was calibrated before the use. For microbiological investigation, sample bottles were stored immediately at 4°C prior to their analysis. For chemical analysis, samples were collected following the standard procedure that include filtering them through 0.45 μm filter membrane and acidification with trace element grade HNO₃ (pH 2.0). Un-acidified filtered samples were collected for the analysis of major anions. The depth of the each tubewell (ranges 27–52 m) was noted from the record preserved by the well owners (Table 1). Arsenic containing orange sand used in this study was obtained from the inner part of the sediment core collected from Chakdaha block, Nadia district in West Bengal (location 23°01^′049.00^″N / 88°35^′07.05^″E) using a reverse circulatory drilling method done by Geological Survey of India, Kolkata. Samples were transported to the laboratory immediately after the collection and stored at 4°C prior to the use in our experiments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Geo-microbial properties of the groundwater samples.

https://doi.org/10.1371/journal.pone.0118735.t001

Analytical techniques

Major anions (SO₄²⁻, NO₃⁻, PO₄³⁻, HCO₃⁻, F⁻ and Cl⁻) and cations (Ca²⁺, Na⁺, K⁺ and NH₄⁺) in the water samples were estimated with an ion chromatograph (Dionex, USA). Total As, Fe and Mn were determined by atomic absorbance spectrometer (AAnalyst^TM 200, ParkinElmer, USA). Trace element concentration including As was measured with an inductively coupled plasma mass spectrometer (ICP-MS) (Varian 810 ICP-MS System, California). Total organic carbon, nitrogen, phosphorous and sulphur from the samples were assessed by following standard procedures as described APHA-AWWA-WPCF [34].

Mineralogy of the sediment sample (orange sand) was characterized by powder X-ray diffraction (XRD) using Panalytical high resolution XRD-I, PW 3040/60 diffractrometer (Almelo, Netherlands) with Cu-Kα1 radiation. Concentration of major and trace elements within the orange sand were determined by X-ray fluorescence spectrometry (PANalytical AXIOS, Almelo, Netherlands).

DNA extraction, amplification of 16S rRNA gene and clone library preparation

Community metagenomes were extracted from all the water samples using water master DNA extraction kit (Epicenter, USA) following manufacturer’s instruction. PCR amplifications of bacterial 16S rRNA genes were performed using eubacterial forward primer 27F and universal reverse primer 1492R maintaining the reaction and temperature cycling conditions as described by Reardon et al. [35]. Details of the PCR conditions and primers used are presented in S1 Table. Purity of the amplified products (~1.5 kb) was determined by electrophoresis of 5 μl samples in a 1% agarose tris-acetate-EDTA (TAE) gel. DNA was stained with ethidium bromide and viewed under short-wave UV light.

For 16S rRNA gene clone library construction, amplified 16S rRNA genes were gel purified with Qiagen kit (Hilden, Germany), ligated into pGEM-T Easy vector (Promega, Madison, WI) and cloned into Escherichia coli JM109. Clone libraries were constructed with 100^_200 randomly chosen white colonies per samples. Cloned 16S rRNA gene fragment from each positive colony was re-amplified using vector specific primers SP6 and T7. Amplified products were digested by restriction endonucleases (HaeIII and MspI) in separate reactions and were analyzed by 2.5% agarose gel electrophoresis. Amplified ribosomal DNA restriction analysis (ARDRA) patterns were grouped and each group was referred as an Operational Taxonomic Unit (OTU) or ribotype. At least one representative clone from each OTU was selected for sequencing of the 16S rRNA gene insert.

PCR-DGGE analysis

V3 regions of 16S rRNA gene were amplified from metagenomes using GC-clamp 341F and 518R primers [36]. A DCode Universal Mutation Detection System (Bio-Rad, USA) was used to separate the PCR amplified 16S rRNA gene fragments by DGGE. Details of the primers used and PCR conditions are presented in S1 Table. A 8% (w/v) acrylamide solution (40% acrylamide bisacrylamide, 37.5:1 stock solution) in 1× TAE buffer with a denaturing gradient from 35 to 65% [100% correspond to 7 M urea and 40% (w/v) formamide] was used. Following electrophoresis in 1× TAE at a constant temperature of 60°C for 12 h at 70 V, gel was stained with ethidium bromide and was visualized under UV illumination. Distinguished thirteen bands in DGGE profile were carefully excised and eluted by keeping in 20 μl DNase free PCR water at 4°C. Those eluted products were re-amplified by using 341F and 518R primers and subsequently were cloned into the pGEM-T vector for sequencing.

Enumeration, isolation and identification of bacteria

Aerobic heterotrophic microbial populations were enumerated by dilution plate count technique. For dilution plating, groundwater samples were serially diluted in normal sterile saline (0.9%) and 100 μl of suspension from each dilution was plated on R2A agar plate in triplicate and incubated at 26°C for 7 days. Number of bacterial colony forming units (CFUs) was counted at selected time intervals throughout the incubation period and morphologically distinct colonies were selected and purified by repeated subculturing in R2A medium. Each purified bacterial strain was stored at −80°C with 15% glycerol.

Taxonomic identification of isolated strains was done by analyzing the 16S rRNA gene sequences. Genomic DNA from each bacterial strain was extracted using genomic DNA isolation kit (Qiagen Ltd., Hilden, Germany) as per manufacturer’s instructions. PCR amplification of nearly complete 16S rRNA gene (1.5 kb) was carried out following the procedure described previously [35]. In brief, PCR amplification of 16S rRNA gene from the isolated genomic DNA was done using primer pair 27F and 1492R. Each 100 μl PCR reaction contained: 100 ng template DNA, 10 μl 10x reaction buffer, 5 μl MgCl₂ (25 mM), 2 μl dNTP mix (10 mM), 10 mole of each primer, and 2U of Taq polymerase (Promega, USA). The PCR amplification was performed with an initial denaturation step of 94°C for 5 min; followed by 30 cycles of denaturation at 94°C for 60 s, annealing at 58°C for 45 s, elongation at 68°C for 90 s and a final elongation step at 68°C for 7 min. Amplified fragments were separated by 1% agarose gel electrophoresis and PCR products were purified by gel extraction according to manufacturer’s protocol (Fermentas, USA) and grouped together in phylotypes based on the RFLP analysis where two different restriction enzymes MspI and HaeIII were used [35]. At least one isolate from each RFLP group was identified by 16S rRNA gene sequencing.

Arsenic sensitivity and metabolic characterization of the bacterial strains

Arsenic sensitivity and metabolic characteristics of bacterial strains isolated from six samples were studied using minimal salt medium (MSM) [37] with/without specific amendment(s) as required for the particular test. Arsenic sensitivity of the isolated strains was determined by growing the cells on MSM agar plates supplemented with 0.1^_20 mM As³⁺ (as NaAsO₂) or 1^_100 mM As⁵⁺ (as Na₂HAsO_4. 7H₂O) (Merck, Germany) diluted from 1 M respective stock solution under aerobic and anaerobic conditions. Mid log phase cells were inoculated onto the agar and the plates were incubated at 26°C for 7 days under aerobic condition. Plates incubated under anaerobic condition were amended with cystein hydrochloride (1g/l) and placed inside the anaerobic workstation for 14 days. As³⁺ oxidase and As⁵⁺ reductase activities of the isolates were tested using AgNO₃ method described by Drewniak et al. [38], where reaction between AgNO₃ and As³⁺ or As⁵⁺ formed coloured precipitate. A brownish precipitate was observed in presence of Ag₃AsO₄ (silver arsenate) and yellow precipitate in presence of Ag₃AsO₃ (silver arsenite). Siderophore formation was studied using Chrome Azurol S (CAS) agar media [39]. CAS agar plates were inoculated with bacterial strains and incubated at 30°C for 7 days. Colonies showing orange hollow zone following incubation were recognized as siderophore positive. Swimming motility was assessed following the procedure described previously by Lee et al., (2007) [40]. Where swim plates were made by adding 0.3% agar to R2A broth and inoculated with overnight grown bacterial culture. The plates were incubated at 30°C and bacterial migration was observed after 24 h. To check the carbon source utilization ability, MSM agar plate was amended with 10 mM of either of the following compounds (casein, glycerol, glucose, acetate, pyruvate, lactate, citrate, starch, sucrose and ascorbic acid). Metabolism of selected hydrocarbons (nonadecane, docosane, dodecane, pentadecane, cyclohexane, phenanthrene, naphthalene, pyrene, flurone and anthracene) was tested following bacterial growth in MSM broth where the test hydrocarbon (each 50 μg/ml concentration) was added as sole carbon source. All the hydrocarbons were obtained from Sigma (analytical grade). Cultures (plates and tubes) were incubated aerobically at 30°C to allow bacterial utilization of the test compounds and growth. Control plate or broth was made with MSM without any carbon source and incubated with bacterial strains under the same condition [41]. Utilization of alternate electron acceptor was assayed by allowing growth under anaerobic conditions in modified MSM agar plates supplemented with either of the test electron acceptors [As⁵⁺ (5 mM), Se⁶⁺ (5 mM), NO₃⁻ (5 mM), SO₄²⁻ (5 mM), Fe³⁺ (20 mM), S₂O₃²⁻ (5 mM) and SO₃²⁻ (5 mM)] [42]. L-cysteine-HCl was added as a reducing reagent to quench the dissolved oxygen within the medium. Control plates were kept without addition of any electron acceptor and incubated with test strains under the same anaerobic environment. All anaerobic experiments were conducted within an anaerobic workstation (Coy laboratory products Inc.) filled with 95% N₂:5% H₂:5% CO₂ mix gases. Chemolithoautotrophic property of the isolates was tested following their growth on M1 and M2 media [43] supplemented with 100 mM NaHCO₃, vitamin (Composition mg/l: folic acid 12.5; riboflavin 25; thiamine 25; pantothenic acid 25; vitamin B₁₂ 250; niacin 25; ascorbic acid 40) and trace elements (Composition g/l: EDTA 0.05; MgSO₄.7H₂O 3.00; MnSO₄.H₂O 0.5; NaCl 1.0; FeSO₄.H₂O 0.1; anhydrous CaCl₂ 0.1; Al₂(SO₄)₃ 0.01; H₃BO₃ 0.01; Na₂MoO₄ 0.01; Na₂SeO₃ 0.001), pH 7 to 7.2. One hundred mg/l of As³⁺ was added as electron donor in the test experiment to the earlier composition.

Amplification of arsenic related genes

Genomic DNA of selected bacterial strains were used as template for PCR amplification of target As homeostasis genes (viz., arsC, arsB, acr3(2), arrA and aioB). PCR reactions targeting the arsC, arsB, acr3(2), arrA and aioB genes were carried out using primers and conditions as previously described [44–47]. Details of the PCR conditions and primers used are presented in S1 Table. The primer sets chosen here have been successfully applied in a variety of environmental samples. Selected amplicons of respective genes were individually gel-purified using the Qiagen kit (Hilden, Germany) (following the specified procedure) and quantified using nanodrop spectrophotometer (Thermo, USA). PCR products were ligated into pGEM-T Easy vector (Promega, USA), and cloned into Escherichia coli JM109.

DNA sequencing and phylogenetic analysis

A total of 169 partial sequences (first 500–600 bp) of 16S rRNA genes from clones or isolated bacteria were obtained using an automated 3100 DNA sequencer. For each of our sequences, most similar sequences was retrieved from those available in public database by using the BLAST (NCBI) program (http://blast.ncbi.nlm.nih.gov/Blast.cgi) followed by initial classification using a web-based classifier program in ribosomal database project (RDP released 10 and with 95% of similarity) (http://rdp.cme.msu.edu/classifier/classifier.jsp). All the 16S rRNA gene sequences (retrieved from database and obtained in this study) were aligned by using ClustalW (http://www.ebi.ac.uk/Tools/msa/clustalw2/). Resulting alignments were used to construct the distance matrix followed by phylogenetic tree construction by neighbour-joining method using MEGA 4 software package [48]. Nucleotide sequences of As-resistance and As transformation genes were translated using the ExPASy tools (http://www.expasy.org/tools/dna.html), and appropriate reading frame for each gene was selected. Protein homology of translated products was determined using BLASTP of NCBI database. Phylogenetic trees were constructed using MEGA 4 with neighbour-joining method [48]. Bootstrap percentages (1,000 bootstrap replications) were used to test the robustness of phylogenetic relationships within the trees.

Microcosm experiment

Arsenic bearing orange sand obtained from aquifer sediment was incubated anaerobically in artificial groundwater with selected strains (Acinetobacter BAS123i, Arthrobacter CAS4101i, Brevundimonas CAS4005i, Pseudomonas BAS323i, Pseudomonas CAS907i, Phyllobacterium BAS224i, Staphylococcus BAS108i and Rhodococcus CAS922i). Orange sand was thoroughly homogenised and sterilized by three rounds of autoclaving at 120°C, 15 psi for 40 min to remove the viable cells present within it. A total of nine microcosms, eight bioaugmented individually by eight individual bacterial strains and one control (without bioaugmentation by any bacteria), were prepared in triplicate using glass serum bottles (Sigma-Aldrich, St. Louis, USA). In each microcosm, 10 g of sterilized sand was added in 20 ml sterile artificial groundwater [15] amended with glucose and Na-acetate (10 mM each) as carbon source. In all biotic microcosms the initial bacterial cell density (added as inoculum) was maintained as 10⁶–10⁷ CFU/ml. All microcosm bottles were purged with ultrapure N₂ gas for 2 hrs, sealed with butyl rubber stopper and incubated in the dark at 26°C over the entire period. Water samples (3 ml) were removed periodically from the bottles, centrifuged at 14,000 ×g for 5 min, and passed through a 0.45 μm membrane filter. The filtered but unacidified supernatant was used to determine concentrations of As⁵⁺ and As³⁺ by HPLC-HG-AAF (ParkinElmer, USA). Concentration of Fe²⁺ was measured spectrophotometrically using the ferrozine method [49]. Prior to analysis, the sample was completely solubilised by adding 1 ml of 0.5 N HCl to 50 μl of the desired sample and digested at 25°C for 24 h. Then Fe²⁺ was determined by adding 200 μl of sample digest to 1.5 ml of the ferrozone solution (1 g of ferrozine to 1 l of 50 mM HEPES buffer) and measuring absorbance at 562 nm. Total Fe was quantified using atomic absorbance spectrophotometer (AAS; AAnalyst 200, PerkinElmer, USA). Change in liquid phase arsenic and other element concentrations was measured by ICP-MS (Varian 810 ICP-MS System, California), as well as by AAS (ParkinElmer, USA) and/or Flame photometer (52A Flame Photometer Perkin-Elmer, USA) as appropriate. Organic acid concentrations present in the aqueous phase of microcosms were estimated by ion chromatography (Dionex, USA) using the procedure as describe by Frey et al. [50]. Change in sediment mineralogy before and after incubation with bacteria was tracked using XRD (Panalytical high resolution XRD-I, Almelo, Netherlands) and XRF (PANalytical AXIOS, Almelo, Netherlands) analyses. Colony forming units were determined at various time points using R2A agar medium and following the procedure as describe by Miles and Misra [51].

Diversity indices and statistical analysis

For 16S rRNA gene clone libraries calculation of diversity indices and analysis of rarefaction were performed based on the number and frequency of ribotypes identified in each library. Shannon diversity indices (H and E) and reciprocal of Simpson’s index of dominance (1/D) were calculated following the standard procedure [35]. Percentage coverage of individual ribotypes in respective clone libraries was calculated using the equation C = 1-(n/N) x100, where n is the number of unique ribotypes and N is the number of sequences analyzed in the library [52]. Rarefaction curve was generated to compare the relative diversity and coverage of each library using the Analytical Rarefaction software version 1.3 (http://www.uga.edu/strata/software/index.html). Experimental observations were recorded with three replications (n = 3) and data were expressed as mean ± SD. The means were compared using analysis of variance (ANOVA) at 5% significant level (p≤0.05). One way ANOVA followed by Duncan’s multiple range tests was performed to determine the significant variations. Associations between variables were calculated by Pearson’s correlation. Relationship among the samples with respect to their geochemical properties was ascertained using Principle Component Analysis (PCA). Interrelationship between constituent bacterial populations and geochemical factors, in each sample was determined by Canonical Correspondence Analysis (CCA) using the XLSTAT (Addinsoft, New York, NY) package. Relationship among the samples with respect to cumulative metabolic properties of isolated bacteria was ascertained by Unweighted Pair Group Method with Arithmetic Mean analysis (UPGMA). Cumulative metabolic profiles of isolated bacterial strains were determined by calculating percentage value of positive response with respect to a particular metabolic parameter obtained within the isolates from individual samples (e.g., if all the isolates obtained from AS40 sample showed positive result for utilization of glucose as C (carbon) source, the value for AS40, glucose was computed as 100%). Interrelationship among the strains (selected) and their metabolic/genetic characters viz., ability to utilize different C sources and inorganic electron acceptors during anaerobic growth, chemolithiotropic growth, siderophore production and presence As transformation (arsC, arrA and aioB) and transporter [arsB and acr3(2)] genes were further determined by multivariate analysis using two component PCA. Data used in this PCA analysis were presented in S2 Table. In order to identify the relationship within bacterial As-resistance, As⁵⁺ reductase and As³⁺ oxidase activities and presence of As related genes UPGMA was done. MVSP 3.1 software was used for PCA and UPGMA analyses.

Sequence data accession number

The sequences obtained in this study have been deposited in the GenBank database under accession no. KF379586^_KF379706, KF442753^_KF442789 and KF793264^_KF793274.

Physiochemical characteristics of groundwater

Physicochemical properties of the six groundwater samples were presented in Table 1. The samples were characterized by nearly neutral pH (6.5 to 6.9), low dissolved oxygen (0.1 to 0.9 mg/l) and negative ORP (^_18.0 to ^_86.9 mV). Among the anions and cations tested, concentrations of As, HCO₃⁻, and Ca were found significantly (p≤0.05) higher. Arsenic concentration varied widely within the samples with lowest value (15 μg/l) obtained in AS2 and highest (1364 μg/l) in AS3. All the groundwater samples showed As level beyond the permissible limit for drinking water, i.e., 10 μg/l as set by World Health Organization [53]. Nevertheless, based on the concentration regime of test samples we categorized three of them (AS2, AS40 and AS41) as relatively low As (<50 μg/l) and remaining three (AS1, AS3 and AS9) as high As (>50 μg/l) containing samples. Concentrations of Na, K and Ca were nearly consistent within the samples while Fe and Mn showed considerable variation (0.67^_12 mg/l Fe; 0.073^_0.49 mg/l Mn). Among the trace elements Al, Ba, W and Zn were detected at relatively higher concentration ranges, e.g., 9^_63 μg/l, 42^_286 μg/l, 10^_76 μg/l and 31^_318 μg/l, respectively (S3 Table). Dissolve organic content was within the range of 3.6 to 4.7 mg/l. Culturable cell counts as enumerated using R2A agar medium showed 2×10^2_3.6×10² cells/ml of the samples. No significant correlation was found between the concentration of As and Fe in all the samples (R² = 0.029, p≤0.05). Presence of total organic carbon (TOC) was found positively correlated (R² = 0.4, p≤0.05) with As, while no significant correlation was observed between CFU and As or TOC (S4 Table).

16S rRNA gene clone library and community analysis

Six clone libraries were constructed using the 16S rRNA gene amplified from groundwater community DNA and the clones were classified by ARDRA. Numbers of OTUs obtained from most of the libraries were within the range of 17^_21 with number of clones ranging from 136^_210. Goods’ coverage (>90%) and rarefaction data indicated high degree of clone coverage for the libraries (Table 1, S1 Fig.). Diversity index values (Table 1) showed considerably low diversity (H<3.0) with uneven community composition (E<1.0). Composition of bacterial communities was delineated by analysing 16S rRNA gene sequences of representative OTUs obtained from each library. It was noted that members of β- and γ-Proteobacteria were ubiquitously present within the samples followed by Bacteroidetes and α-Proteobacteria (Fig. 1A). Members of δ-Proteobacteria, Armatimonadetes, Actinobacteria, Chlorobi and Cyanobacteria were detected as relatively minor populations in a fewer samples. Presence of high GC containing bacterial groups Bacteroidetes and Armatimonadetes as well as Cyanobacteria and a few unclassified groups were observed in both high and low As samples.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Distribution of major phylogenetic groups of bacteria as detected in clone libraries.

Abundance of bacterial groups is plotted with respect to their affiliation at (a) phylum level and (b) genus level. The horizontal bar indicates gradient of As concentrations as detected in six samples.

https://doi.org/10.1371/journal.pone.0118735.g001

Identification of bacterial groups at lower taxonomic level revealed presence of a number of genera (Fig. 1B). Genera Pseudomonas, Rheinheimera, Polaromonas, Methyloversatilis, Methylotenera, Hydrogenophaga of γ- and β-Proteobacteria and Flavobacterium (of Bacteroidetes) represented the major populations within the communities. Based on their ubiquity, bacterial genera detected here were grouped into two categories; one comprising of those detected in two or more samples (designated as frequently detected bacterial genera), while the other constituting genera found in only one of the samples (designated as less frequently detected bacterial genera).

Frequently detected bacterial genera

Genus Pseudomonas represented the most abundant population (29^_68%) in all three low As samples as well as a minor population (5.6%) in the highest As containing sample (AS3). Phylogenetic analysis with often similar sequences showed that Pseudomonad related sequences segregated into two clades (Fig. 2). Sequences obtained from the low As samples were showed lineages to various cultured and uncultured Pseudomonas species reported earlier from rhizosphere and other subsurface environments. Absence of similarity was found between these sequences with other Pseudomonas sequences retrieved previously from As contaminated sites or aquifers. In contrast to this sequences retrieved from AS3 showed similarity and close lineage with cultured P. putida, P. stutzeri and uncultured Fe reducing Pseudomonadace bacteria previously obtained from As contaminated aquifers of Bangladesh and other As rich sites.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Neighbor-joining phylogenetic tree of γ-Proteobacteria.

The tree is constructed based on 16S rRNA gene sequences by using Jukes-Cantor distances. 1000 bootstraps analyses are conducted and more than 50% are denoted in nodes. Sequences retrieved from As contaminated environment, groundwater and chromium contaminated sites are suffixed here as ‘A’, ‘G’, and ‘Cr’, respectively. Sequences represented in bold italic font are derived from clone libraries and bold italic font with a suffix ‘i’ are derived from isolated strains.

https://doi.org/10.1371/journal.pone.0118735.g002

Members of the genus Polaromonas were present in two high as well as two low As samples with maximum abundance of 50% in the high As sample AS40. Polaromonas sequences were closely related to P. aquatica and similar organisms reported from As contaminated water samples. Sequences retrieved from high As samples AS3 and AS9 showed similarity (98%) with P. aquatica and uncultured β-Proteobacteria from As-Fe nodules and formed a separate clade indicating their phylogenetic distinctness (Fig. 3). Methane (or C1 compound) utilizing genera Methyloversatilis, Methylotenera and Methylophylaceae represented the most abundant groups in both high (AS3 and AS9) as well as low As sample (AS41) (Fig. 1B). Sequences affiliated to these groups showed lineage with mostly uncultured bacteria previously retrieved from As contaminated aquifers, As-Fe co-precipitates, Kalahari shield and alkenes’ contaminated river sediment (Fig. 3). Genus Flavobacterium was detected as a major population (23%) in low As sample AS2 and as minor group in two other samples (including high As sample AS9) (Fig. 4). Sequences affiliated to this genus showed strong relation with Flavobacterium strains isolated earlier from water bodies and As contaminated sediment. Genus Hydrogenophaga was detected as moderately abundant population (16%) in both high and low As samples (AS1 and AS41) (Fig. 1B). Sequences of this group showed similarity and lineage with uncultured β-Proteobacteria and H. defluvii reported earlier from As contaminated environment (Fig. 3) [53].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Neighbor-joining phylogenetic tree of β-Proteobacteria.

The tree is constructed based on 16S rRNA gene sequences by using Jukes-Cantor distances. 1000 bootstraps analyses are conducted and more than 50% are denoted in nodes. Sequences retrieved from As contaminated environment, groundwater and mercury contaminated sites are suffixed here as ‘A’, ‘G’, and ‘Hg’, respectively. Sequences represented in bold italic font are derived from clone libraries and bold italic font with a suffix ‘i’ are derived from isolated strains.

https://doi.org/10.1371/journal.pone.0118735.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Neighbor-joining phylogenetic tree of δ-Proteobacteria, Actinobacteria, Cyanobacteria and other groups.

The tree is based on 16S rRNA gene sequences using by Jukes-Cantor distances. 1000 bootstraps analyses are conducted and more than 50% are denoted in nodes. Sequences retrieved from As contaminated environment and groundwater are suffixed here as ‘A’, and ‘G’, respectively. Sequences represented in bold italic font are derived from clone libraries and bold italic font with a suffix ‘i’ are derived from isolated strains.

https://doi.org/10.1371/journal.pone.0118735.g004

Less frequently detected bacterial genera

Members of the genera Aquincola, Brevundimonas, Parvibaculum, uncultured Actinobacteria, etc. of high As samples and genera Aquabacterium, Cyanobacteria along with unclassified bacteria of low As samples represented the less-frequent and -abundant (<5%) populations. Genera Rheinheimera and Curvibacter found in high As sample AS1 as the predominant populations (≥30%) could not be detected elsewhere (Fig. 1B). Genera Rheinheimera (γ-Proteobacteria) showed close lineage with Rheinheimera strains previously reported from various freshwater as well as As contaminated environments (Fig. 2). Sequences related to Curvibacter and Brevundimonas showed lineage to similar organisms previously obtained from As contaminated sites including the groundwater from Brahmaputra river basin (Figs. 3, 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Neighbor-joining phylogenetic tree of α-Proteobacteria.

The tree is based on 16S rRNA gene sequences by using Jukes-Cantor distances. 1000 bootstraps analyses are conducted and more than 50% are denoted in nodes. Sequences retrieved from As contaminated environment are suffixed as ‘A’. Sequences represented in bold italic font are derived from clone libraries and bold italic font with a suffix ‘i’ are derived from isolated strains.

https://doi.org/10.1371/journal.pone.0118735.g005

PCR-DGGE analysis

Bacterial community structures within the samples were further elucidated using DGGE fingerprinting of PCR amplified 16S rRNA gene. Separation of expected 170–220 bp PCR fragments by DGGE from the samples produced distinct banding patterns (S2 Fig.). Presence of a few common and/or prominent bands across the samples was observed. All together 13 bands from DGGE gel were sequenced and identified (Table 2). Sequences of bands AS3-1, AS2-3, AS41-7 and AS40-12 showed identity with uncultured and cultured Pseudomonas sp. retrieved from various environments. Affiliation with uncultured γ- and δ-Proteobacteria was shown by AS2-4 and AS9-10, respectively. Bands AS1-5, AS41-8 and AS9-9 showed close lineage with Acinetobacter sp., Acidovorax sp. and uncultured Methylophilaceae bacterium, respectively. Sequences of a few bands, e.g., AS3-2, AS1-6, AS9-11 and AS40-13 showed identity with uncultured unidentified bacteria. It was noted that overall data corresponded well with the clone library based observations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Taxonomic affiliation of 16S rRNA gene clones retrieved from major DGGE bands.

https://doi.org/10.1371/journal.pone.0118735.t002

Identification of cultured bacterial isolates

As many as 173 bacterial strains isolated in the present study were identified and belonged to 16 genera under four phyla (Fig. 6). Isolates affiliated to the genus Pseudomonas were most frequently detected (five out of six samples) with moderate to high abundance (7^_61%). Genera Brevundimonas, Rhodococcus and Bacillus were present in several (≥3) samples, representing major parts of culturable populations. Bacterial strains affiliated to the genera Rhizobium, Rheinheimera, Phyllobacterium, Herbaspirillum, Stenotrophomonas and Arthrobacter though detected less frequently (≤2 samples), they often showed higher abundance (10^_86%). A few strains were identified as Hydrogenophaga, Microbacterium, and Mucilaginibacter representing minor populations (3^_6%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Distribution of major phylogenetic groups of isolated bacteria.

Abundance of bacterial groups is plotted with respect to their affiliation at (a) phylum level and (b) genus level. The horizontal bar indicates gradient of As concentrations as detected in six samples.

https://doi.org/10.1371/journal.pone.0118735.g006

Pseudomonas strains retrieved from both high and low As samples showed lineage to several cultured Pseudomonas spp. as well as with uncultured γ-Proteobacteria previously reported from various subsurface environments (Fig. 2). Genus Rheinheimera represented the predominant culturable population (53%) in As rich sample AS1 and its sequences showed relatedness to other Rheinheimera sp. and uncultured γ-Proteobacteria bacteria retrieved earlier from subsurface or Cr contaminated environment (Fig. 2). Genus Brevundimonas was detected as a predominant (24^_32%) group in three low As samples and as a minor (4%) population in one high As sample (AS1) (Fig. 6B). It was noted that Brevundimonas sequences bear strong lineage with uncultured α-Proteobacteria bacterium, B. intermedia and other Brevundimonas reported from As contaminated groundwater of Assam (India) and other places (Fig. 5). Genus Rhodococcus was found only within the samples from Chakdaha area, of which its abundance was maximum (64%) in As rich sample AS9 (Fig. 6B). Phylogenetic analysis revealed their close lineage with R. qingshengi and a Rhodococcus strain retrieved recently from As contaminated groundwater of Assam (India) (Fig. 4). Genus Bacillus was present in both high and low As samples and its sequences showed strong phylogenetic similarity with B. arsenicus and B. Nanhaiensis strains recovered earlier from As contaminated environment or sea coast of West India, respectively. Bacterial genera Staphylococcus as well as Arthrobacter and Herbaspirillum were found as abundant groups (≥10%) in AS1 or AS41 sample (Fig. 6B). Sequences of these genera showed close lineages with members of similar groups reported earlier from various As contaminated environments. Genus Rhizobium was present in both low and high As samples. Its abundance was higher in As rich sample AS3. Sequences related to this genus showed close lineage to unclassified Rhizobiaceae members previously reported from a As contaminated sample of West Bengal [26] (Figs. 5, 6B).

Similar to the clone library based observation, culturable populations also indicated frequent presence of a number of genera (e.g., Pseudomonas, Brevundimonas and Rhodococcus) across the samples. Members of the genera Phyllobacterium, Herbaspirillum, Stenotrophomonas, Arthrobacter and Hydrogenophaga, however, could be isolated from only one of the samples. Our results on community composition indicated the complementary nature of culture-independent and -dependent approaches, although, some degrees of overlap with respect to a few taxa were also found. Genera Pseudomonas, Rheinheimera, Brevundimonas and Hydrogenophaga were detected by both the approaches, while Stenotrophomonas, Enterobacter, Acidovorax, Herbaspirillum, Phyllobacterium, Rhizobium, Rhodococcus, Microbacterium, Bacillus and Staphylococcus could only be isolated, but not found within the 16S rRNA clone library. Genera Methyloversatilis, Methylotenera, Polaromonas, Curvibaculum and Methylobacterium were only detected through culture-independent approach.

Arsenic resistance and metabolic characterization of bacterial isolates

Maximum tolerable concentrations (MTC) of As³⁺ and As⁵⁺ for all the 173 strains was determined following growth with graded concentration of respective As species under aerobic and anaerobic conditions. Prior to the As resistance study, ability of all the test bacteria to grow under anaerobic condition was tested. Nearly 60% of the strains, mostly affiliated to the genera Acinetobacter, Brevundimonas, Pseudomonas, Rheinheimera, Rhizobium, Rhodococcus, Staphylococcus, Stenotrophomonas, etc. were able to grow anaerobically (Table 3). Resistance to As (either As³⁺ or As⁵⁺ or both) under aerobic as well as anaeorobic condition with considerably high MTC values was found to be an omnipresent phenotype of the bacterial strains. Actinobacteria, Rhizobium and Rhodococcus strains from As rich AS1, AS3, and AS9 samples, respectively showed relatively higher MTC values [≥100 mM] for As⁵⁺ under both conditions (Table 3). Members of the genera Pseudomonas, Rheinheimera, Brevundimonas and Herbaspirillum although showed higher As⁵⁺ resistance under aerobic condition, their inability to withstand the same under anaerobic condition was noted. Resistance to As³⁺ (MTC ≥5 mM) under aerobic condition was found widely distributed among Rhizobium (AS3), Rhodococcus and Pseudomonas (AS9), and Bacillus (AS1 and AS2) as well as in Acinetobacter and Rheinheimera strains. Under anaerobic condition comparable As³⁺ resistance was mainly found within Rhizobium (AS3) and Rhodococcus (AS9) strains and within a few Pseudomonas strains from AS9 sample.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Percentile distribution of positive phenotypic test results among the bacterial strains isolated from arsenic contaminated groundwater.

https://doi.org/10.1371/journal.pone.0118735.t003

Metabolic characteristics of bacterial populations were ascertained by a number of tests including As³⁺ oxidase and As⁵⁺ reductase activities, utilization potential of several carbohydrates and different inorganic electron acceptors (Fig. 7). A consolidated profile of all the metabolic properties tested along with percentage of positive response obtained for each trait was presented in Fig. 7A. Arsenate reductase activity was found within nearly two third of the bacterial strains mostly affiliated to genera Brevundimonas (from samples AS2, AS40 and AS41), Pseudomonas (AS2, AS9, AS40 and AS41), Rhodococcus (AS9), Rhizobium (AS3), Phyllobacterium (AS2), Acinetobacter (AS1) and Arthrobacter (AS41) (Table 3). In comparison to this, As³⁺ oxidase activity was less abundant (16%), present within the strains mostly affiliated to genera Rhodococcus and Pseudomonas from AS9 sample. With respect to utilization of different carbon/electron sources, more than 80% of the isolates were able to metabolize all the compounds tested, except citrate and ascorbate. The latter two compounds were metabolized by 72% and 13% of the isolates, respectively (Fig. 7B). Interestingly, members of a few genera like Bacillus, Phyllobacterium, Staphylococcus and Stenotrophomonas showed their inability to utilize many of the test compounds (e.g., glucose, glycerol, acetate, pyruvate, etc.). With respect to utilization of different inorganic elements as terminal electron acceptors (TEAs) during anaerobic growth, 50% of the strains showed their ability to metabolize As⁵⁺ followed by Se⁶⁺ (40%), S₂O₃²⁻ (30%), Fe³⁺ and HSO₄⁻ (24%), NO₃⁻ (13%) and SO₄²⁻ (12%) (Fig. 7B). Ability to grow anaerobically using As⁵⁺ as TEA was found to be relatively frequent within a few genera isolated from specific samples (e.g., Rhizobium strains from AS3, Rhodococcus strains from AS9 and Bacillus strains from AS2) (Table 3). Ability to use Fe³⁺ as TEA was mainly present in Rhodococcus, Pseudomonas and Brevundimonas strains from AS9, AS40 and AS41 along with a few Rhizobium (AS3), Staphylococcus (AS1) and Phyllobacterium (AS2) strains. Interrelation among the test metabolic properties indicated presence of several distinct groups across the taxonomic domains (Fig. 7C). The UPGMA dendrogram clearly showed that bacterial strains belonged to taxonomically distinct groups share metabolic traits at high similarity (> 80%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Consolidated profiles of all the metabolic properties tested within the bacterial strains isolated from six As contaminated samples.

(a) Positive (red box) and negative (yellow box) responses obtained for various properties tested; (b) Percentage of positive response obtained for each property and (c) UPGMA based on metabolic properties of bacterial strains isolated from six As contaminated samples. Metabolic characters tested are (I) enzymes involved in As transformation: 1: arsenite oxidase; 2: arsenate reductase; (II) ability to utilize different carbon sources: 3: cacsin; 4: glycerol; 5: glucose; 6: acetate; 7: pyruvate; 8: lactate; 9: citrate; 10: starch; 11: sucrose; 12: ascorbic acid and (III) utilization of different terminal electron acceptors: 13: As⁵⁺; 14: Se⁶⁺; 15: SO₄²⁻; 16: S₂O₃²⁻; 17: SO₃²⁻; 18: NO₃⁻; 19: Fe³⁺; 20: O_2. Taxonomic affiliations of bacterial isolates are indicated using coloured bars presented right side of consolidated metabolic profiles.

https://doi.org/10.1371/journal.pone.0118735.g007

Along with these properties, presence of genes responsible for As transformation (arsC, arrA and aioB) and efflux [arsB and acr3(2)], utilization of As³⁺ as electron donor and HCO₃⁻ or long chain aliphatic and polyaromatic hydrocarbons as carbon source, production of siderophore and motility were also studied within selected 44 strains (Table 4). These strains were selected based on their taxonomic abundance and members of nearly all genera were included. Nearly 50% of the test strains belonged to the genera Pseudomonas, Rhizobium, Brevundimonas, Rhodococcus, etc., showed presence of cytosolic As⁵⁺ reductase arsC gene. Sequences of arsC gene obtained from Acinetobacter BAS123i, Arthrobacter CAS410i, Pseudomonas CAS907i, Rhodococcus CAS922i and Staphylococcus CAS108i shared high identity and lineages with As⁵⁺ reductase of several Proteobacteria (Vibrio sp., Escherichia coli, Acinetobacter sp., Ochrobactrum sp., Pseudomonas, etc.) (Fig. 8). Arsenate transporter gene (arsB) was detected in 24 strains affiliated mostly to the genera Pseudomonas, Brevundimonas, Rhodococcus, Staphylococcus, etc. Sequences of arsB gene obtained from Brevundimonas CAS4005i, Acinetobacter BAS108i, Rhodococcus CAS922i and Arthrobacter CAS411i showed relatedness with putative As³⁺ efflux pump from Serratia, Staphylococcus, etc. (Fig. 8). Arsenite transporter gene acr3(2) was detected in 14 strains belonging to genera Pseudomonas, Staphylococcus, Arthrobacter, Rhizobium, etc. Sequences of acr3(2) gene detected in these strains showed lineages with As³⁺ transporters/efflux pumps reported from genera Ochrobactrum, Rhizobium, etc. (Fig. 8). The aioB gene encoding As³⁺ oxidase was present in five bacterial strains affiliated to genera Rhodococcus, Acinetobacter, Staphylococcus and Microbacterium. Nucleotide sequences of this gene showed their closeness with As³⁺ oxidase of Bacillus, Achromobacter and with several uncultured β- and α-proteobacterial members (Fig. 8). Dissimilatory As⁵⁺ reductase gene arrA was relatively less abundant and could be amplified successfully from only seven strains affiliated to genera Rhodococcus, Pseudomonas and Brevundimonas. The sequence obtained from Rhodococcus CAS922i revealed its close lineage with arrA of Shewanella sp. Noticeably, a number of strains showed simultaneous presence of arsC, arsB and acr3(2) or arsC and arrA genes. Possible relation among the presence of all these genes detected within the bacterial strains and their As-resistance and -transformation phenotypes was studied by UPGMA (Fig. 9A). Presence of arsC gene was found to be strongly related with As⁵⁺ reductase activity and together with acr3(2) gene, a close agreement between As resistance, transformation and presence of both these As related genes was evident.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Phylogenetic tree of deduced amino acid sequences of genes encoding As⁵⁺ reductase (arsC and arrA), As³⁺ oxidase (aioB), and As³⁺ efflux [arsB, and acr3(2)].

Genes detected in this study are shown in bold font. Percentage values on each branch represent the corresponding bootstrap probability values obtained in 500 replications. Significant bootstrap values (>50%) of major branch points are shown in the tree.

https://doi.org/10.1371/journal.pone.0118735.g008

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Correlation among the various metabolic properties and As related genes detected within selected 44 isolates.

(a) UPGMA cluster analysis based on selected genetic and metabolic characters (viz., As related genes, As-resistance and -transformation abilities) and (b) plot of PCA scores on utilization of different C sources and alternate terminal electron acceptors, As³⁺ oxidase and As⁵⁺ reductase activities, As resistance genes, motility and siderophore production of the isolates. In 9(b), each bacterial strain is represented by numbers referring to their names (S5 Table); arrows indicate corresponding parameters for which the isolates with higher eigen values have clustered.

https://doi.org/10.1371/journal.pone.0118735.g009

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Details of the results obtained for As³⁺ and As⁵⁺ resistance, utilization of different hydrocarbon compounds and HCO₃ as carbon source, As³⁺ as electron donor, siderophore production, motility and presences of As related genes within selected strains isolated from six As contaminated groundwater samples.

https://doi.org/10.1371/journal.pone.0118735.t004

Ability to grow autotrophically using HCO₃⁻ or heterotrophically utilizing a range of long chain aliphatic and polyaromatic hydrocarbons was found to be ubiquitous within the bacterial strains. Comparatively, a fewer strains showed their ability to use As³⁺ as electron donor. In particular, Rhodococcus strains from AS9, Arthrobacter strains from AS41 and a few Pseudomonas, Rhizobium strains were capable of chemolithotrophic metabolism using As³⁺. Among the test hydrocarbons, except docosane rest all were metabolized by most (>86%) of the strains. Motility was detected within 63% of the bacterial isolates. Ability to produce siderophore was scarce and three isolates namely, Arthrobacter CAS4101i and CAS4117i and Rhodococcus CAS4021i were found positive.

Interrelationship among the various metabolic properties tested within the bacterial strains was further studied by euclidean PCA (Fig. 9B, S5 Table). Based on their metabolic profiles, bacterial strains found to be grouped into distinct clusters and distributed within four quadrants (Fig. 9B). Quadrant 1 and 4 accommodated maximum strains and properties. Within quadrant 1 bacterial strains affiliated to genera Pseudomonas, Rhizobium, Rhodococcus, Brevundimonas and Staphylococcus showed higher level of agreement with respect to presence of arsC and acr3(2) genes, utilization of As⁵⁺, Fe³⁺, NO₃⁻ and SO₄²⁻ as TEA and use of acetate, lactate, starch, anthracene, naphthalene cyclohexane and bicarbonate, as carbon source (Fig. 9B). On the other hand, As³⁺ oxidase activity, aioB gene, utilization of As³⁺ as electron donor and docosane as carbon source by a few Pseudomonas, Rhodococcus, Arthrobacter and Brevundimonas strains represented the fourth quadrant (Q4).

Microcosms study

Further to investigate the role of inhabitant bacteria in As mobilization, eight strains were used in sediment microcosm study (Table 4). Considering the metabolic properties and taxonomic affiliation of predominant populations representative strains were selected. All these strains were endowed with As reductase activity, arsC, arsB and/or acr3(2) genes. Rhodococcus CAS922i, Brevundimonas CAS4005i and Pseudomonas CAS907i contain arrA gene as well. All the strains except Pseudomonas CAS907i and Acinetobacter BAS123i could use As⁵⁺ as TEA while Staphylococcus BAS108i and Arthrobacter CAS4101i have additional ability to utilize Fe³⁺ as TEA. Biotic microcosms showed that all the bacterial strains could maintain their viability during incubation (up to 300 days) with orange sand and cause As release (S6 Table). Concentration of released As varied with respect to the bacterial strain used; slightly higher As mobilization (>14 μg/l) was noted for Rhodococcus CAS922i, Brevundimonas CAS4005i and Staphylococcus BAS108i strains (Fig. 10). Noticeably, in each biotic microcosm, aqueous As concentration reached to its maximum within the first 7 days and 80^_90% of released As was found to be in the form of As³⁺. It was noted that strains with arrA gene and ability for anaerobic growth utilizing As⁵⁺ as TEA (Rhodococcus CAS922i and Brevundimonas CAS4005i) could mobilize relatively higher concentrations of As. In contrast, Pseudomonas CAS907i, that showed presence of arrA gene but unable to use As⁵⁺ during anaerobic growth showed release of lesser concentration of As.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Concentrations of As (a and b), Fe (c and d) and other metals (Mn, K and Al) (e) detected within the aqueous phase of microcosms (after 300 days).

Error bars indicate standard deviations (n = 3). Bacterial strains represented in X axis are as follows: CAS922i (Rhodococcus sp.), CAS4005i (Brevundimonas sp.), BAS108i (Staphylococcus sp.), BAS224i (Phyllobacterium sp.), CAS4101i (Arthrobacter sp.), BAS323i (Pseudomonas sp.), BAS123i (Acinetobacter sp.) and CAS907i (Pseudomonas sp.).

https://doi.org/10.1371/journal.pone.0118735.g010

Coupled with As, release of Fe [predominantly (90^_95%) in the form of Fe²⁺] was observed in all the biotic microcosms, however, pattern of Fe and As release did not correlate well among the isolates. During the initial days (1^_7 days) of incubation, it was noted that Fe reducing strains Arthrobacter CAS4101i and Staphylococcus BAS108i could lead to release of significantly higher amount of Fe over the other sets and this occurred before the appearance of significant amount of As³⁺ in the aqueous phase. In contrast to this, microcosms with other As⁵⁺ reducing strains (not capable of Fe³⁺ reduction) Rhodococcus CAS922i and Brevundimonas CAS4005i showed appearance of As³⁺ followed by Fe³⁺. Compared to all the biotic microcosms, abiotic control sets showed negligible release of As (1.9 μg/l) and Fe (0.08 mg/l) even upto the end of incubation period (300 days) (Fig. 10).

Together with As and Fe, release of other trace elements (Al, Mn, K, etc.) was also observed in case of the biotic microcosms. Analyses of both solid and aqueous phases indicated mobilization of several elements through possible dissolution of host minerals of orange sand (Fig. 10E, S7 Table). Release of Al was observed in microcosms with strain Acinetobacter BAS123i, Rhodococcus CAS922i and Pseudomonas BAS323i. Release of Mn (1.4^_5.2 mg/l) and K (2.7^_3.8 mg/l) was found in microcosms with Brevundimonas CAS4005i, Staphylococcus BAS108i, Pseudomonas CAS907i and Pseudomonas BAS323i. X-ray fluorescence (XRF) analysis of sediment before and after incubation with bacteria corroborated well with the elemental analysis data of aqueous samples indicating reduced abundance of Fe, Al, K, Mn, etc. following bacterial incubation (S7 Table). Although release of As was not strongly (S8 Table) correlated (p≤0.05) with that of other trace elements tested, nevertheless, in all higher As releasing microcosms mobilization of one or more constituting elements of the sand was obviously noted. Mineralogical analysis confirmed that the orange sand was composed of several As bearing minerals like Fe oxyhydroxide based goethite [FeO(OH)], albite (NaAlSi₃O₈) and glauconite (K[(Fe,Al)₂(Si,Al)₄O₁₀(OH)₂]). XRD data indicated that following incubation with bacteria there was distinct mineralogical changes (Table 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Mineralogical composition (as determined by XRD analysis) of orange sand before and after 300 days incubation with selected bacterial strains under anaerobic condition.

https://doi.org/10.1371/journal.pone.0118735.t005

Since mineral dissolution is often caused by microbially produced organic acids, concentrations of such acids were further estimated within the aqueous phase of microcosms. It was found that oxalic acid is the only organic acid that could be detected at significantly higher concentration (p≤0.05) from the aqueous part of microcosms by IC analysis. Although several strains could produce this acid (115^_562 mg/l), Brevundimonas BAS4005i yielded maximum concentration (S3 Fig.). Low concentration of oxalic acid was also detected in abiotic control which could probably be resulted from abiotic transformation of glucose in presence of minerals as also reported earlier by Frey et al. [50].

In order to get a better understanding on bacterial role in As mobilization, particularly the type of transformation taking place within the microcosms a UPGMA based analysis was performed, considering a set of metabolic and genetic characters and concentrations of As and Fe released (Fig. 11). A good agreement between As mobilization and bacterial ability to use As⁵⁺ as terminal electron acceptor, As⁵⁺ reductase activity, presence of arsC gene and production of oxalic acid was evident. Release of Fe, was distantly related to As release, but remained strongly correlated with bacterial ability to use Fe³⁺ as TEA and presence of arrA gene.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 11. Correlation between the metabolic-genetic characteristics of bacterial strains used for microcosm study and release of As and Fe within different microcosms.

https://doi.org/10.1371/journal.pone.0118735.g011

Statistical analysis

Interrelation among the samples with respect to their geochemical properties, community composition and metabolic characteristics of inhabitant bacteria was studied using a number of statistical analyses. PCA performed on selected geochemical parameters indicated a resemblance between AS40, AS41, AS2 and AS1, while AS9 was distantly related to these four samples (Fig. 12A). Based on the geochemical parameters used for this analysis, AS3 remained distinctly separated from rest of the samples. In order to understand the relationship between geochemical properties and bacterial community composition (as detected by both culture-dependent and -independent approaches) Canonical Correspondence Analysis (CCA) was performed (Fig. 12B). Axes 1 and 2 of the resulting biplot accounted for 35.07% and 32.39%, respectively of total variability. The CCA data showed that a few specific groups like Methyloversatilis, Methylobacterium, Rhizobium and Acidovorax correlated to the As content. Fig. 12C represented cumulative metabolic characteristics of the strains isolated obtained from each of the samples and their interrelations. The heat map developed based on relative abundance of individual properties tested within the bacterial strains obtained from each of the samples was presented in Fig. 12C(i). The data indicated that a number of characteristics e.g., As⁵⁺ reductase activity, utilization of diverse carbohydrates and use of As⁵⁺ as TEA were more or less consistently present across the bacteria from different samples. In contrast, As³⁺ oxidase activity, ability to use Fe³⁺, NO₃⁻ and SO₄²⁻ as TEAs remained more sample specific. UPGMA dendogram represented that based on overall metabolic properties of indigenous bacteria the sampling sites can be divided into a few distinct clades (Fig. 12C(ii)). It could be noted that AS40 and AS41 as well as AS2 and AS1 were strongly related to each other forming two distinct clades. As also observed during geochemical parameters based analysis (Fig. 12A), samples AS9 and AS3 were distantly related to rest of the samples even with respect to metabolic properties of inhabitant bacteria. A second UPGMA was performed to analyse the interrelation between different metabolic characteristics tested within the bacterial isolates. It was evident that utilization of Fe³⁺, S₂O₃²⁻and SO₃²⁻ as terminal electron acceptor was closely related, whereas ability to use NO₃⁻ and SO₄²⁻ was coupled together. Use of different sugars as carbon source showed higher level of similarity, and ability to use of As⁵⁺ or Se⁶⁺ as TEA was related to As⁵⁺ reductase activity [Fig. 12C(iii)]. Overall analyses suggest that there was a good agreement among the local geochemical parameters and composition vis a vis metabolic activities of indigenous bacterial groups, thereby highlighting the possible role of local geochemical conditions in shaping the bacterial communities and their metabolic characteristics.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 12. PCA, CCA and UPGMA based ordination plots on samples’ microbiological and geochemical properties.

(a) Principal Component Analysis (PCA) of the samples based on geochemical parameters; (b) Canonical correspondence analysis (CCA) of samples geochemical properties and bacterial community compositions (c) ordination plots representing (i) comparative analysis of metabolic properties of the isolates obtained from six As contaminated samples, (ii) cumulative metabolic profiles represented by UPGMA for the isolates obtained from individual samples and (iii) UPGMA cluster analysis among the test metabolic properties.

https://doi.org/10.1371/journal.pone.0118735.g012

Conclusions

Overall study indicates that bacterial communities in As rich groundwater of West Bengal are constituted by diverse taxonomic groups, metabolically as well as genetically well equipped to survive and function within the aquifer environment. Arsenic mobilization in the aquifer may be an outcome of collective metabolic reactions of these indigenous populations. Unlike preponderance of any particular event, release of sediment bound As could possibly occurred due to a cascade of catabolic processes namely dissolution of host minerals, followed by bacterial As⁵⁺ reducing activities or direct action of bacterial reductive machinery on host minerals resulting elevated As³⁺ in aqueous phase. These processes are not mutually exclusive but may incorporate each other. Furthermore, our orange sand based microcosm study indicates that significant mobilization of As can occur, provided suitable bacterial strains interact with them. So conclusively, though the present perception reveals relatively safe nature of orange sand (Pleistocene aquifer) as a source of As free drinking water our study indicates its potential vulnerability.