Arsenic (As) contamination of groundwater is a serious public health problem in Bangladesh and eastern part of India exposing over 100 million people to unsafe levels of As in drinking water (Raessler 2018). The region geologically contains high levels of As sorbed onto soil particles. Dissimilatory microbial reduction of soil-bound arsenate (AsV) (Malasarn et al. 2004; van Geen et al. 2004; Farooq et al. 2012) and/or reductive dissolution of arsenic-bearing minerals such as Fe, Mn and Al (Islam et al. 2004; Hery et al. 2010; Paul et al. 2015) followed by As reduction in deep soils majorly govern the As release in alluvial aquifers. The reduced form of As (arsenite, AsIII), having less affinity for soil minerals and clay, is released and transported to the groundwater through subsurface hydrologic gradient (van Geen et al. 2004; Harvey et al. 2006).

Besides biogeochemical cycling of As in deep aquifer soils, microbial As cycling also occurs in surface aquatic environments (Oremland and Stolz 2003; Bandopadhyay et al. 2017; Sultana et al. 2011; Goswami et al. 2015). In Mekong River floodplains of Cambodia, there is >60–90-fold increase in dissolved As concentration within a few metres depth of water-sediment interface of wetlands; this is coupled with organic matter oxidation suggesting role of the wetland sediment microbiota in the mobilization process (Polizzotto et al. 2008). Bengal Delta has thousands of such ponds and wetlands that significantly contribute to the groundwater arsenic contamination in the area by providing labile organic matter for microbial arsenic reduction in subsurface soil (Fendorf et al. 2010; Mladenov et al. 2010; Lawson et al. 2013), as well as hydrologic gradient for transport of the solutes to the groundwater (Harvey et al. 2006). Besides, ponds in the area also contain a significant amount of As (Chowdhury et al. 2015) and might act as active sites of As mobilization similar to that in the Mekong River wetlands. Farooq et al. (2012) observed significant As mobilization from the upper 2.6 m of pond sediment, under influence of organic matter loading from jute retting. We had earlier observed As reduction by microbiota present in top layers of pond sediment in partial to true anoxic conditions (Bandopadhyay et al. 2017). Interestingly, the As-reducing bacteria in top layers of pond sediments were distinct from those observed in deep aquifers (Sultana et al. 2011; Paul et al. 2015; Bandopadhyay et al. 2017) suggesting possible role of diverse groups of microbiota or even different biogeochemical processes in As cycling. This paper has isolated AsV-reducing anaerobic bacteria by enrichment method and examined their As-reduction ability in deep layers of pond sediments. The observation emphasizes the role of aquatic sediments in As reduction and mobilization and fills a void in information on bacteria involved in As transformation between top layers of aquatic sediments and deep aquifer soils.

Sediment samples were collected, with help of a sediment sampler (Eijkelkamp), from the middle and two sides of four numbers of ponds used for aquaculture and household purposes (Bandopadhyay et al. 2017) in arsenic-affected villages of Chakdaha and Haringhata blocks of West Bengal, India. Samples from 150–170 cm sediment depth were incised out with help of a scalpel-blade, quickly filled up to the brim in sterile bottles avoiding air pockets, tightly capped and brought to laboratory under ice cover. For estimation of total As contents, sediment samples were air dried, grounded and sieved through 100 mesh sieve. These were digested with tri-acid mixture (HNO3:HClO4:H2SO4::10:4:1), filtered and As content measured by atomic absorption spectrometer (GBC Avanta Sigma) coupled with a hydride generator following method described in detail by Chowdhury et al. (2015). Total As level ranged between 13.58 and 34.72 mg kg−1 (mean 21.73 ± 2.48 mg kg−1) and was significantly higher than that at top layers of the sediment (10.34 ± 0.4 mg kg−1) (Bandopadhyay et al. 2017) indicating higher As levels in deep sediment layer either as geological deposition, or use of contaminated groundwater for re-filling of ponds during dry season (Chowdhury et al. 2015) and downward mobilization of the metalloid. This was similar to the increasingly higher As level with sediment depth in wetlands of the Mekong River (Polizzotto et al. 2008).

Anaerobic As-reducing bacteria in sediment enrichments

Bacterial isolation and As-reduction studies by the isolates were conducted in gas mixture (80% N2:10% CO2:10% H2) environment in an anaerobic workstation (Ruskinn Bugbox). Ten percent (w/v) of suspensions of freshly collected sediment samples were made in 0.85% saline solution, serially diluted (10−1 to 10−3) and 500 μL volume from different dilutions were enriched in 20 mL culture medium [K2HPO4 0.225g, KH2PO4 0.225g, NaCl 0.46g, (NH4)2SO4 0.225g, MgSO4 0.117g, yeast extract 0.5g, distilled water to 1000 mL, pH 7.5] supplemented with 27.3 mM sodium acetate and 10 mM sodium arsenate (Na2HAsO4·7H2O) as electron donor and acceptor, respectively. As-transforming prokaryotes use a wide range of carbon substrates such as glucose, lactate, acetate, butyrate and citrate as electron donor (Oremland and Stolz 2003; Paul et al. 2015). Among these, acetate has been widely used to stimulate growth of As-transforming microbes and As reduction/release from deep soil or pond sediment (van Geen et al. 2004; Lear et al. 2007; Hery et al. 2010; Bandopadhyay et al. 2017) and was used as electron donor in present experiment. Unlike As-resistant bacteria which can tolerate very high concentrations of the metalloid, As-respiring prokaryotes have low level of resistance, generally up to 10 mM As: as such, 10 mM AsV concentration was used to favour growth of As-respiring microbes. The enrichment broths were incubated at 30 °C for 15 days when cultures were transferred to fresh medium containing 10 mM AsV. After 15 days of further incubation, the cultures were streaked several times on to the AsV-containing medium solidified with 1.5% agar for pure culture isolation.

A total of 17 presumptive arsenic-reducing bacteria were isolated from sediment enrichments. Identification of the bacterial isolates was done through PCR amplification and sequencing of 16S rRNA gene using 8F and 1492R universal bacterial primers. Bacteria were tentatively identified by finding similarity of the sequences with SSU sequences in the NCBI GenBank, Greengenes (http://greengenes.lbl.gov/cgi-bin/nph-index.cgi) and RPD (http://rdp.cme.msu.edu) databases. The evolutionary phylogenetic tree was prepared by minimum evolution method (Rzhetsky and Nei 1992) with 1000 bootstrap using MEGA7 (Kumar et al. 2016) bioinformatics software. Shigella strain was further identified through a battery of biochemical tests following Bergey’s manual (Strockbine and Maurelli 2005). About 41% of the strains were identified as Paraclostridium benzoelyticum, 23.6% strains as P. bifermentans, 17.6% strains as Clostridium and Paraclostridium sp.; only one strain was identified as Shigella boydii and 2 strains belonged to the genus Exiguobacterium (Fig. 1). Clostridium has earlier been detected in anaerobic enrichments of polluted river sediment in a gold-mine area (Suhadolnik et al. 2017) as well as As-affected Bengal Basin soil (Islam et al. 2004) signifying role of Clostridium in As transformation in diverse environments. Paraclostridium is a newly coined genus with reclassification of Clostridium bifermentans (Sasi Jyothsna et al. 2016). Paraclostridium benzoelyticum gen. nov was reported from marine sediments (Sasi Jyothsna et al. 2016); we detected the species in freshwater environments suggesting its wide habitat distribution. Members of Clostridium avidly ferment organic matter and are known to reduce a number of metals and metalloids including arsenic (Oremland and Stolz 2003; Suhadolnik et al. 2017). Total organic carbon of dried sediment samples, measured in TOC analyser (OI Analytical), was 1.65 ± 0.12% and might have favoured growth of Clostridium and Paraclostridium in pond environments. Exiguobacterium is a well-known arsenic-resistant bacterium and has been reported from As-rich soil of Chhattisgarh, India (Pandey and Bhatt 2016).

Fig. 1
figure 1

Phylogenetic relationships of arsenic-reducing bacteria from deep pond sediment, inferred from 16S rDNA sequences. Strains in the present study are indicated in bold

Full size image

No true chemolithotroph was isolated in present study probably due to limitation of the culture protocol followed. However, some other studies have also observed absence of chemolithotrophs and abundant presence of aerobic, facultative anaerobic or denitrifying populations of Proteobacteria, Bacteroidetes, Firmicutes and Actinobacteria suggesting involvement of diverse groups of heterotrophic bacteria in As reduction in shallow and deep aquifers of Bengal Delta (Islam et al. 2004; Sutton et al. 2009; Sultana et al. 2011; Paul et al. 2015). The bacteria identified from deep pond sediment enrichments were different from those isolated from pond water and superficial sediment (Chryseobacterium sp., Pseudomonas spp., Acinetobacter spp., Microbacterium oleivorans and Comamonas aquatica) (Goswami et al. 2015; Bandopadhyay et al. 2017) suggesting a likely shift in As-reducing bacterial community with sediment depth. However, culture-dependent study can identify only a small fraction of the total microbiota and metagenomic approach might reveal the wider spectrum of microbial assemblages along the sediment depths.

Arsenic reduction by bacterial strains

Arsenic reduction by different bacterial strains was estimated in culture medium supplemented with 10 mM arsenate and 10 mM sodium acetate following method described before (Bandopadhyay et al. 2017). Briefly, 50 μL of individual bacterial culture (8.83 × 109–1.12 × 1010 CFU) was added to 20 mL culture medium and incubated at 30 °C. Thirty microlitre volume of culture was taken in duplicate tubes at different time intervals (0, 1, 2, 7, 10, 15 and 20 days of incubation) for As estimation: one set of tubes was treated with reducing agent (1.4% Na2S2O3:14% Na2S2O5:3.5 N H2SO4::2:2:1) for 12 min for complete reduction of AsV (“reduced”) and the other set was acidified with 25 mM HCl (“unreduced”). Both the “reduced” and “unreduced” samples were treated with mixed reagent (containing ammonium molybdate, ascorbic acid and antimony potassium tartrate) for 10 min at 78 °C, centrifuged at 10,000 × g for 3 min and supernatant read at 865 nm. Both AsV and phosphate form coloured complexes with molybdenum, while AsIII does not: so only phosphate is detected in “reduced” samples while AsV and phosphate are detected in “unreduced” samples. AsV remaining in the sample was calculated by subtracting the “reduced” sample value from that of the “unreduced”. The amount of AsV reduction was obtained by subtracting the amount of AsV remaining in culture medium from 10 mM AsV added to the culture medium. Sterile culture medium amended with AsV and acetate was kept as negative control to assess chemical As reduction in anaerobic culture environment. Arsenic reduction by each bacterial strain was examined in two occasions keeping 5 replicate tubes every time. Bacterial growth was measured photometrically at 600 nm and correlated with As reduction. Data are presented as mean ± S.E. Pearson’s correlation test was applied to examine relationships between bacterial growth and As reduction. One-way ANOVA was applied to find significant difference in growth or As reduction among strains.

All the bacterial isolates caused ≥ 5 mM AsV reduction (Fig. 2). Among these, AnRBC1a and AnRBC3 strains caused 8.05–9.22-mM reduction in 7–10-day period; other strains took 15–20 days to reduce 5.14–8.19 mM As. Species-wise, majority of the Paraclostridium benzoelyticum strains were more efficient (p < 0.05), reducing 5.95–9.22 mM AsV in 7–10 days. Strains belonging to Paraclostridium bifermentans, Paraclostridium sp. and Clostridium sp. transformed 5.2–7.75 mM As in 15–20 days. Exiguobacterium and Shigella boydii caused about 6-mM reduction in 10 and 20 days, respectively. All the strains took 10–15 days to reach highest density of growth, with comparatively higher growth achieved by P. benzoelyticum strains. Bacterial growth was significantly correlated with AsV reduction (p < 0.05) with minor strain to strain variations: less number of cells of P. benzoelyticum strain AnRBC1a but more number of cells of Shigella boydii strain AnRBC15 were required for sufficient reduction. The rate of reduction was comparable to that observed in Cambodian sediment (Lear et al. 2007) but slower as compared that observed by Paul et al. (2015) in Bengal Delta sediment or by As-respiring or resistant strains from other geographical region (Li and Krumholz 2007), probably because our medium had limited energy supply. However, AsV reduction rates of the present bacteria were higher than that caused by isolates from superficial sediments and water (Bandopadhyay et al. 2017) showing higher As transformation ability of bacteria at sediment depths.

Fig. 2
figure 2figure 2

Arsenate reduction and growth curve of bacterial strains from deep pond sediments. Figure title indicates control or strain ID. Horizontal axis represents incubation period in days. Extent of arsenate reduction (out of initial 10 mM arsenate) is in primary Y-axis, and bacterial growth (OD) in secondary Y-axis. Solid line indicates AsV reduction; broken line indicates bacterial growth (OD)

Full size image

Identification of the arsenic reduction gene(s)

Microbes reduce AsV by two distinct mechanisms: arsenic respiration facilitating growth in anoxic environments with help of periplasmic/membrane associated respiratory arsenate reductase (Arr) or cytoplasmic As-reduction and expulsion mechanism (Ars) (Silver and Phung 2005). While the former is considered to be the principle mechanism of As release in aquifer soil, the latter is a part of As resistance strategy found in a wide diversity of microbes from contaminated environments (Oremland and Stolz 2003). Presence of As-reducing genes in the genome and plasmid was examined through PCR amplification of the dissimilatory AsV-respiring genes (ArrA and ArrB) and cytoplasmic AsV-reducing gene (ArsC) of Ars operon using primers described in Table 1. The genomic DNA and plasmid DNA of individual bacteria were extracted separately using commercially available kits (Sigma-Aldrich and Invitrogen) and taken as template in PCR amplification following the reaction conditions given in references in the Table 1. Bacteria, viz., Pseudomonas sp., Comamonas aquatica, Acinetobacter junii and Chryseobacterium sp., isolated earlier from superficial sediments (Bandopadhyay et al. 2017) were also included in the study. The PCR amplicons were gel-purified and sequenced. Some of the PCR products were cloned in Escherichia coli DH5α using pGEM®-T vector system (Promega) and sequenced following standard protocol (Sambrook and Russell 2001). The gene sequences were edited by Sequence Scanner v1.0 software (Applied Biosystems, Inc.) and BLAST searched in NCBI database: Blastn programme was used for finding the closest nucleotide matches, blastp for protein blast and blastx for protein databases using a translated nucleotide query. Aligned nucleotide sequences were also transformed in to protein sequences by ExPASy (https://web.expasy.org/translate) translation tool to obtain the converted protein sequences and similarity of the gene. The phylogenetic relationships of the As reduction genes with those from other bacteria was derived using MEGA7 software by maximum likelihood method based on the JTT matrix-based model and 1000 bootstrap replications (Jones et al. 1992).

Table 1 Primers used for detection of ArrA, ArrB and ArsC genes
Full size table

Earlier studies have identified respiratory arsenate reductase, coded by the arr operon, as a principal mechanism As reduction in deep soils of South and SE Asia (Krafft and Macy 1998; Afkar et al. 2003). However, despite use of 6 sets of commonly used primers, we did not detect presence of Arr genes in any of the sediment isolates. Although AsV reduction was positively correlated with bacterial growth, absence of Arr genes suggests that these bacteria were not using AsV for respiration. Degenerate primers of Arr genes yielded non-specific amplifications but none of those was identified as ArrA or ArrB gene following cloning and sequencing (raw data not given). The study, however, found presence of the ArsC gene. Among 3 sets of ArsC primers used, only amlt successfully amplified the ArsC gene with amplicon length of 390 bp, which is partial sequence of ArsC gene. Among all the strains, only 3, viz., P. benzoelyticum (strain AnRBC1a, AnRBC5b) and Shigella boydii strain AnRBC15, carried the ArsC gene in their chromosome and plasmid respectively (Table 2, Fig. 3) suggesting that neither all the As-reducing bacteria nor all the members of a species carried the gene. None of the other strains belonging to Paraclostridium, Clostridium sp. and Exiguobacterium sp. carried the gene and their As-reduction mechanism is presently undefined. However, the amlt primers were originally designed from Enterobacteriaceae (Sun et al. 2004) which might be a reason why only a few strains of Firmicutes harboured the gene. This is also a likely reason of phylogenetic closeness of the identified ArsC genes with those from the family Enterobacteriaceae, but distant from Firmicutes (Fig. 3). ArsC is one of the most common detoxification mechanisms among diverse groups of prokaryotes with early origin and rapid evolution (Fekih et al. 2018) that demands use of a wide range of primers for their detection. While glutaredoxin reductase-dependent ArsC is common among Enterobacteriaceae, thioredoxin-coupled ArsC is found in Gram-positive bacteria and in a wide range of soil arsenic concentrations (Escudero et al. 2013). As such, our study based on enterobacterial related primers does not rule out a possibility of presence of thioredoxin-ArsC among the members of Firmicutes.

Table 2 Arsenic reduction by the bacterial isolates from deep pond sediments and comparison with bacteria from pond water and superficial layers of the sediment
Full size table
Fig. 3
figure 3

Phylogenetic relationships of ArsC in this study with that of other bacteria. Strains in present study are indicated in bold

Full size image

An earlier study of contaminated groundwater has also detected dominance of arsenic reductase (ArsC) over dissimilatory arsenic reductase (Arr) gene (Paul et al. 2015). They detected the Arr gene in few strains only which also harboured Ars genes, and the extent of As release in sediment microcosm was independent of Arr gene; ArsC, rather than the Arr, was more closely correlated with As release. Goswami et al. (2015) also observed absence of Arr gene in arsenic-reducing Acinetobacter and Microbacterium from surface waters in Bengal Delta. Our study, in association of others (Paul et al. 2015; Goswami et al. 2015), indicates cytosolic arsenate reductase as a possible As-reduction mechanism in the Bengal Basin.

As summarized in Table 2, the level of AsV reduction by bacteria having ArsC gene was not significantly higher than by those lacking the gene. Thus, neither the presence of the identified ArsC was stringently associated with As reduction, nor an absence of Arr and ArsC genes disables an organism from As reduction suggesting non-exclusive role of these reductases in As transformations. Interestingly, Clostridium, the major group of anaerobes found here, has been shown to methylate As (Wang et al. 2015: Zeng et al. 2018) using thiols such as GSH as a reductant (Bentley and Chasteen 2002). Microbial methylation in anaerobic sediments is an important As detoxification mechanism with significance in the global biogeochemical cycles of the metalloid (Zeng et al. 2018). Whether these sediment microbiota cause significant As methylation as a part of their resistance strategies using thiols and other reducing agents in labile organic matter-rich aquatic environments needs further study.

Conclusion

Paraclostridium and Clostridium isolated from the anaerobic pond sediment enrichments showed strong AsV reduction activity. Such As reduction by aquatic bacteria might contribute significantly to the groundwater arsenic contamination in Bengal Basin inundated with thousands of ponds and wetlands.