Introduction

As a worldwide environmental issue, exposure to high arsenic (As) groundwater has affected more than 140 million people from countries such as Bangladesh, China, Vietnam and the USA (Barringer et al. 2011; Harvey et al. 2002; Li et al. 2014; Winkel et al. 2011). Groundwater arsenic can originate from geogenic As-laden iron (Fe) oxide minerals (Fendorf et al. 2010) or anthropogenic utilization of As-containing fertilizers as well as mining of As-bearing minerals (Neidhardt et al. 2012). Consumption of As containing groundwater as drinking water is harmful to human health and can cause many serious diseases including vascular and respiratory disease, melanosis, and even skin and heart cancer (Ferreccio and Sancha 2006; Maleki et al. 2015).

Numerous studies have been devoted to investigate the release and transformation mechanisms of As in groundwater aquifers (Fan et al. 2008; Jiang et al. 2014; Winkel et al. 2011) and results have shown that several processes may lead to the mobilization of As into groundwater, including reductive dissolution of iron (Fe) oxide minerals, oxidation of sulfide minerals, anion competition for desorption sites and microbial reduction of arsenate (As (V)). Among these processes, the microbially mediated As (V) reduction is mainly caused by two kinds of microorganisms including dissimilatory As (V)-respiring prokaryotes (DARPs) and As (V)-resistant microbes (ARMs) (Tsai et al. 2009; Liao et al. 2011). The DARPs can use arsenate (As (V)) as terminal electron acceptor and gain energy for metabolic growth under anaerobic conditions. The arrA gene, encoding the respiratory arsenate reductase subunit ArrA, is a credible biomarker for detection of respiratory As-reducing bacteria. The resistance of ARMs is encoded by the ars operon which usually contains the As (V) reductase ArsC and the regulatory protein ArsR (Branco et al. 2008). Different from the DARPs, the ARMs can survive in both aerobic and anaerobic conditions without energy conservation (Macy et al. 2000; Saltikov et al. 2005). To date, numerous phylogenetically diverse prokaryotes capable of As (V) reduction have been isolated from terrestrial and aquatic habitats (Chang et al. 2012; Handley et al. 2009; Niggemyer et al. 2001). ARMs are widespread almost in all bacterial phyla, while DARPs are affiliated mainly in Proteobacteria, Firmicutes, Aquificae, Deferribacteres, Chryosiogenetes and Archaea (Cavalca et al. 2013).

Hetao plain of Inner Mongolia is one of China’s representative high As affected area with the highest As concentration in groundwater up to 1.74 mg/L and more than 250,000 residents under As threat (Deng et al. 2009; Deng et al. 2011). Previous studies have showed that As release in Hetao Plain is the consequence of several biogeochemical processes such as reductive dissolution of As-rich Fe(III) oxyhydroxides and dissimilatory reduction of As (Cavalca et al. 2013; Deng 2008; Deng et al. 2009; Guo et al. 2011). Results of our previous studies have demonstrated that bacterial communities (16S rRNA genes based) in high As groundwater of Hetao Plain are dominated by Acinetobacter, Pseudomonas, Brevundimonas, Geobacter, Thiobacillus and large quantity of unidentified bacterial groups (Li et al. 2013; Wang et al. 2016). However, until now, only several As (V)-resistant strains have been isolated from the Hetao Plain, including two aerobic bacteria affiliated with Pseudomonas sp. and Bacillus sp. and three anaerobic bacteria belonged to Bacillus sp. Desulfitobacterium sp. and Exiguobacterium (Cai et al. 2016; Guo et al. 2015). Besides, studies on the community composition and diversity of indigenous As-reducing bacteria are still limited in the high As aquifers.

Therefore, the main objectives of the present study were to (1) isolate functional bacteria capable of As reduction and detect their arsenic-metabolizing gene clusters and (2) investigate the diversity and composition of indigenous As(V)-reducing bacterial communities via arrA and arsC genes from high As groundwater of the Hetao Plain, Inner Mongolia, China.

Material and methods

Groundwater sampling and geochemical analysis

Four representative shallow groundwater samples IMCUGGW05 (N40.908°, E106.829°), IMCUGGW07 (N40.865°, E106.884°), IMCUGGW13 (N41.202°, E107.362°) and IMCUGGW17 (N41.070°, E107.159°) with differentiated geochemical characteristics were collected from Hangjinhouqi County, an As threatened area located in the Hetao Plain of western Inner Mongolia. Groundwater samples were pumped out from local domestic wells (depth: 20-25 m) for about 10 min to get a stabilized Eh values. Then, about 10 L fresh groundwater samples were filtered through 0.2 μm filters (Millipore, Germany). The filter membranes were collected into sterilized tubes and frozen on dry ice for further analysis. For strain isolation, the prepared serum bottles with sterile and deoxygenized chemically defined medium (CDM) were injected with approximately 1 mL groundwater samples and kept at room temperature during transportation.

Field and laboratory geochemical analyses including temperature, pH and oxidation-reduction potential (ORP) of groundwater samples as well as As and Fe speciation were performed with same procedures as in our previous studies (Wang et al. 2015; Li et al. 2017). Details were shown in Supporting Information (SI).

Strain isolation and identification

For enrichment and isolation, CDM was used (Weeger et al. 1999). It contained the following (g/L): NaCl 0.46, MgSO4•7H2O 0.117, KH2PO4 0.225, K2HPO4 0.225 and (NH4)2SO4 0.225. Additional 20 mM lactate, 0.1% yeast extract and 0.1% resazurin solution (1%) were added into the medium as carbon source, nutrient supplement and dissolved oxygen indicator, respectively. The medium was divided into 100 mL serum bottles, deoxygenated with pure N2, and then autoclaved. Arsenate (1 mM), L-cysteine (1 mM) and NaHCO3 (5 mM) were filtered through 0.22 μm filters and then added into the medium. After three times of transfer, diluted suspension was inoculated into anaerobic tubes containing CDM-As and purified agar (1.8%, w/v) following the Hungate technique (Hungate 1969). Single clones were picked out and inoculated into new medium. After several times of isolation, three pure cultures were obtained (IMARCUG-11, IMARCUG-C1 and IMARCUG-12).

Growth experiments of the isolated strains were conducted with 40 mL CDM in 100 mL serum bottles in triplicate. Gradient temperature and pH were set to detect the most suitable growth conditions. Different carbon sources with the same carbon concentrations of 20 mM were added into the medium to test the potential utilization of carbon sources. 1 mM As (V) was added into the pre-cultured stain suspensions to confirm the ability of As (V) reduction. Mediums without strain suspensions were cultured as controls. All experimental sets were performed in triplicate at 30 °C under anaerobic conditions. At different time intervals, 1 mL of the mixture were taken out from serum bottles and equally divided into two aliquots. One aliquot was filtered (0.22 μm) for the detection of As speciation, while the other aliquot was analyzed for protein concentration to monitor the bacterial growth (Lv and Yu 2013). All the experiments taken above were accomplished in the anaerobic glove box.

Genomic DNA of the isolates was extracted using the FastDNA spin kit for soil (MP Bio Laboratories, USA) following the manufacturer’s procedures. The 16 S rRNA gene fragment of the isolates were amplified using universal primer pair 27 F and 1492 R (Table 1). The PCR programs were the same as in our previous studies (Dai et al. 2016). Successful PCR products were purified and then commercially sequenced. The obtained sequences were blasted in NCBI to search sequences with high similarities and construct neighbor-joining tree with MEGA 7.0 software. The 16 S rRNA gene sequences of strain G-11, G-C1 and G-12 were submitted to the GenBank under accession numbers MK027325, MG198846 and MK027326, respectively.

Table 1 Primers used in this study for amplification of 16S rRNA, arsB, arrA and arsC genes
Full size table

Amplification of functional genes of isolates

Based on conservative sites from several arsC gene-related sequences in the genus Paraclostridium, a primer pair PB2-F and PB2-R (Table 1) was designed to amplify the arsC gene of strain G-11 with the target length of 202 bp. The arsB gene of strains G-C1 and G-12 were amplified with primer pair darsB1F and darsB1R (Achour et al. 2007) (Table 1). Amplification details of arsC and arsB gene were showed in the SI. Based on acquired arsB gene sequences, three primer pairs (Z540F/Z155R, K551F/K38R, 53BF/53BR, Table 1) were designed and two kinds of restriction endonucleases (EcoR I and Hind III, TaKaRa) were selected according to the inverse-PCR method (Ochman et al. 1988). Different from the common way, the forward primers here were designed at the end of the obtained arsB gene sequences while the reverse primers were designed at the head. Besides, it must be sure that the arsB gene could not be cut off by selected endonucleases. Restriction digests were carried out using 2 μg of extracted DNAs of strain G-C1 and G-12, which were treated with 15 U EcoR I and 15 U Hind III in a 50 μL mixture at 37 °C for 5 h followed by an inactivation of endonucleases at 65 °C for 20 min. For circularization, digested DNAs were purified using method mentioned above and the ligation reaction was performed by the addition of T4 DNA ligase (TaKaRa) to a concentration of 1 U/μL and the reaction was proceeded at 25 °C for 5 h and the circularized DNAs were purified again for PCR amplification. The PCR was performed manually in reactions containing 5 x PrimeSTAR GXL Buffer, 200 μM dNTP mixture, 0.2 μM designed primers, 50 ng circularized DNA and 1 U of PrimeSTAR GXL DNA Polymerase (TaKaRa, Japan). We used 30 cycles of 98 °C for 10 s, 55 °C for 15 s, and 68 °C for 1 min according to manufacturer’s guide. The PCR products were fractionated at 1% agarose gel, purified and then formed sticky ends with a 20 μL mixture containing 0.5 U of Taq DNA polymerase and 200 μM dNTP mixture at 72 °C for 20 min. Finally, the extended DNA products were purified again for ligating, transforming as described above and several single clones were picked out and sequenced.

PCR amplification and clone library analyses of arsenate reductase genes in groundwater samples

Total genomic DNA was extracted and purified from about one-quarter of each filter membrane of water samples using FastDNA spin kit for soil (MP Bio Laboratories, USA). The concentration and quality of extracted DNA were checked using Nanodrop 2000 (Thermo Fisher Scientific, USA). The arrA gene and arsC gene of groundwater samples were amplified with primer sets arrA-CVF1/arrA-CVR1(Mirza et al. 2016) and P52F/P323R (Bachate et al. 2009) (Table 1), respectively. Detailed information of PCR mixture and protocol was shown in SI. After clone screening procedure, several clones were picked out and sequenced. Data from the obtained raw sequences were processed following our previous methods (Li et al. 2014; Wang et al. 2018). The GenBank accession number of the arrA and arsC gene sequences were MK038511-MK038604.

Results and discussion

Isolation, identification and As (V) reduction

Three As (V)-reducing strains were obtained from this study. Phenotypic analysis found strain G-11 to be a gram-positive, strictly anaerobic bacterium, while strain G-C1 and G-12 were gram-negative, facultatively anaerobic bacteria. Strain G-11 and G-C1 had a pH range of 4.0–9.0 with the optimum pH of 8.0 and the optimum growth temperature of 37 °C, while strain G-12 could survive at the wide pH range of 3.0-10.0 (optimum at 8.0) and at the optimal growth temperature of 35 °C. All these strains could utilize different carbon sources including maltose, fructose, glucose, sucrose, galactose, sodium citrate, sodium acetate and sodium lactate.

BLAST analysis of the 16 s rRNA gene sequences indicated that these three strains belonged to genera Paraclostridium, Citrobacter and Klebsiella, respectively. Strain G-11 presented highest similarities with two microorganisms referred to as P. benzoelyticum strain JC272 (100%) isolated from marine sediment (Jyothsna et al. 2016) and P. bifermentans strain HYN0063 (99.93%) isolated in South Korea (NCBI database, accession number MF988694). Strain G-C1 showed high similarity (99.93%) with C. freundii strain S1 which was isolated from soil (Kary and Alizadeh 2016), while G-12 presented high similarity (99.57%) with K. variicola strain XF4 (Zhang and Kong 2014) (Fig. 1). Several previous studies reported that Klebsiella and Citrobacter were capable of As (V) reduction (Maeda et al. 1992; Shakoori et al. 2010; Wang et al. 2017), with the maximum tolerated concentration (MTC) of As for Klebsiella oxytoca was 3.2 mM (Shakoori et al. 2010). The three strains isolated in the present study could tolerate As (V) up to 30 mM (Fig. S1), indicating their strong As resistance. To date, there have been rare reports on reduction of As (V) by Paraclostridium while it was reported to be involved mainly in hydrogen or alcohol production under strictly anaerobic conditions (Liu et al. 2014; Yang et al. 2019).

Fig. 1
figure 1

Neighbor-joining tree of 16 S rRNA gene sequences of strains G-11, G-C1 and G-12. The sequence of Bacteroides fragilis strain ATCC 25285 was used as the outgroup. Only bootstrap values above 50% (for 1000 iterations) are shown. The GenBank accession number for each reference strain is shown before the strain name

Full size image

In CDM containing 1 mM As (V), strain G-11, G-C1 and G-12 could reduce > 99% of As (V) to As (III) in 60 h, 72 h and 84 h, respectively, suggesting strain G-11 possessed the strongest ability with least amount of biomass (Fig. 2). The As (V) reduction abilities of the strains in the present study were comparable with the results of a previous study (Saleh and Ekram 2013) where Klebsiella oxytoca and Rahnella aquatilis could reduce 1.33 mM As (V) by percentages of 75 and 69%, respectively.

Fig. 2
figure 2

As(V) reduction and microbial growth of strains G-11, G-C1 and G-12 in CDM. The initial As(V) concentrations were 1 mM. Error bars are the SD of duplicate values

Full size image

Arsenic-reducing clusters of As(V) reduction

In order to investigate the arsenic-metabolizing gene clusters of the isolated strains, putative arsC gene sequence (Accession No. MK038613) was amplified from strain G-11 using the primer set PB2-F/PB2-R. However, no arsC gene sequence could be amplified from strain G-C1 and G-12 with any reported primer sets (Table S1) (Bachate et al. 2009; Macur et al. 2004; Sun et al. 2004). Fortunately, the arsB gene sequences (750 bp) were obtained from genomic DNA of strain G-C1 and G-12. Using the reverse PCR technique with designed reverse PCR primers (Table 1, Z540/Z155R for strain G-C1, K551F/K38R and 53BF/53BR for strain G-12), the flanked sequences by the arsB gene from strain G-C1 and G-12 were successfully amplified. Finally, the arsC family genes including arsR gene (188 bp, MK038605), arsB gene (1290 bp, MK038606), arsC gene (426 bp, MK038607) were obtained from genomic DNA of strain G-C1.

For strain G-12, two incomplete ars operons consisting of arsA gene (888 bp, MK038608), arsB gene (1290 bp, MK038609) and arsC gene (426 bp, MK038610) for operon 1, and arsB gene (1119 bp, MK038611) and arsC gene (426 bp, MK038612) for operon 2 were successfully amplified (Fig. 3). It was worth mentioning that the genome of G-12 presented two arsB and arsC genes. Previous studies showed that more than one ars operon might be involved in the microbe genome such as strain Citrobacter freundii UMH16 (Anderson et al. 2018) and Pseudomonas putida BIRD-1 (Matilla et al. 2011). The transcription of ars operon might be regulated by environmental factors such as temperature (Páez‐Espino et al. 2015; Wang et al. 2016). Further investigations are needed to find the possible key environmental factors regulating ars systems in strain G-12.

Fig. 3
figure 3

Map position of putative ars gene clusters of strains G-11, G-C1 and G-12

Full size image

Several universal primer sets from previous studies (Fisher et al. 2008; Kulp et al. 2006; Malasarn et al. 2004; Mirza et al. 2016; Song et al. 2009) and specific primer sets (designed in the present study from arrA genes of the same genera as the isolated strains) (Table S1) were chosen but failed to amplify any arrA related genes, implying that even the same genus could present different genomic compositions (Escudero et al. 2013).

Biodiversity of in situ As(V)-reducing bacterial communities

A total of 160 arrA and 302 arsC gene sequences were obtained to construct phylogenetic trees. The coverage values of arrA and arsC gene clone libraries were 0.80-0.89 and 0.90-0.91, respectively. With a 97% sequence similarity, the arrA gene sequences were classified into 8, 14, 12 and 10 OTUs, while the arsC gene sequences were classified into 11, 13, 13 and 11 OTUs, respectively (Table S2). Shannon and Chao1 indices of arrA and arsC gene clone libraries ranged from 1.00 to 2.35, 1.45 to 1.84 and from 13.00 to 20.00, 16.75 to 41.00, respectively (Table S2). The diversity indices of the functional genes and geochemical characteristics (Table S3) did not present significant correlations according to statistical correlations analysis (Table S4), indicating that diversity of As-reducing bacteria in groundwater were affected by various geochemical variables.

The amino acid sequences derived from the arrA gene sequences were phylogenetically clustered into Betaproteobacteria, Deltaproteobacteria and an unidentified group, with Deltaproteobacteria being dominant by the percentage of 53.13% (Fig. 4a). In the class Deltaproteobacteria, 48.75% and 4.37% of sequences were related to arrA amino acid sequences of Desulfuromonadales and Desulfobacterales, respectively. The cluster associated with Desulfuromonadales were mainly composed of Geobacter and showed high similarities to the arsenate respiratory reductase of Geobacter sp. OR-1, Geobacter thiogenes and Geobacter uraniireducens, which were previously reported isolated from sediments or paddy soils (Ohtsuka et al. 2013; Qiao et al. 2017; Upadhyaya et al. 2012). Previous studies showed that Geobacter sp. was also the dominant Fe(III)-reducing bacteria in many anaerobic environments and relatively high percentages of Geobacter sp. in microbial communities might be correlated with Fe(III) reduction and As(III) release in aquifers (Li et al. 2013; Rowland et al. 2007). Furthermore, G. uraniireducens was capable of reducing sorbed As(V) on Fe(III) oxyhydroxides to more mobile As (III) (Héry et al. 2010). These results suggested that Geobacter sp. could play an important role in As mobilization in groundwater via either direct or indirect mechanisms. Seven clones belonged to Desulfobacterales were similar with sequences of Desulfopila aestuarii DSM 18488, Desulfofustis glycolicus DSM 9705 and Desulfobulbaceae sp. A2 in sulfidic groundwater-fed fountain (Sharrar et al. 2017). The second cluster associated with Betaproteobacteria accounted for 11.88% of clones and was highly similar to ArrA like protein of Betaproteobacteria and Sulfuritalea hydrogenivorans sk43H from Rifle groundwater or freshwater (Anantharaman et al. 2016; Watanabe et al. 2014). Besides, 35% of sequenced clones were affiliated with putative ArrA proteins from New Jersey groundwater (Mumford et al. 2012), paddy soil (Zhang et al. 2015) or Cache Valley Basin (Mirza et al. 2014) and showed extremely low similarity with previously identified arrA gene sequences from cultured isolates, suggesting the possible existence of unique dissimilatory As(V)-reducing bacteria in groundwater of the Hetao Plain.

Fig. 4
figure 4

The phylogenetic tree of deduced amino acid sequences based on the arrA and arsC gene sequences of four samples from the in situ high As groundwater. One representative type within each OTU was shown. The GenBank accession number for each reference strain is shown before the strain name. Numbers in parentheses represent sequences numbers obtained for each OTU. Only bootstrap values above 50% (for 1000 iterations) are shown. □: IMCUGGW05; ■: IMCUGGW07; ☆: IMCUGGW13; ★: IMCUGGW17. (a) arrA gene based; (b) arsC gene based

Full size image

Unlike the results of arrA genes, sequences derived from arsC genes were widely distributed among the microbial phyla, indicating that the detoxification was the most common mechanism of As reduction, which was consistent with the result of our previous study (Li et al. 2017). The translated arsC amino acid sequences were divided into three clusters: Omnitrophica, Deltaproteobacteria and Firmicutes (Fig. 4b). The largest cluster which belonged to Omnitrophica (88.08%) was similar to those bacteria of Candidatus Omnitrophica with arsenate reductase ArsC in groundwater (Anantharaman et al. 2016). Candidatus Omnitrophica which was previously known as OP3 belonged to the PVC superphylum and had not get any pure cultures yet (Devos and Ward 2014). What differed from the result of a previous study which indicated that OP3 was more abundant in low As groundwater samples (Wang et al. 2016), results of the present study showed that OP3 endowed with cytosolic arsenate reductase gene might be widely distributed in groundwater samples with different As concentrations. Twenty-one sequenced clones belonged to Syntrophobacterales which showed similarity with ArsC protein of Syntrophus aciditrophicus and Syntrophobacter sp. SbD1 (NCBI database, accession numbers WP011417387 and WP106821703). Two clones belonged to Bacillaceae showed similarities with Oceanobacillus chungangensis which was isolated from a sand dune (Lee et al. 2013). Besides, several sequences which belonged to Proteobacteria and Firmicutes have been proved to be involved in the As reduction processes (Cavalca et al. 2013). The cluster associated with Geobacter accounted for 4.3% of arsC amino acid sequences and was related to arsenate reductase ArsC of Geobacter such as Geobacter. sp. metabat 2.715 which was isolated from wetland sediments (Martins et al. 2018). These results once again proved the important role of Geobacter sp. in As cycling of the high As groundwater.

Collectively, results of the indigenous As (V) reducing communities showed that Geobacter and Omnitrophica were the major populations of DARPs and ARMs, respectively. Meanwhile, Geobacter was ever reported to be the dominant genus of microbial communities in high As groundwater of the Hetao Plain (Li et al. 2013). Previous studies found that Geobacter species were one kind of typical dissimilatory iron-reducing bacteria with the ability of extracellular electron transfer (Liu et al. 2018; Shi et al. 2009). In high arsenic groundwater, arrA-harboring Geobacter species could transfer electrons through cytomembrane using their specific extracellular electron transport system. The electrons might be sequestrated by ArrA reductase during the electrons being transferred through periplasmic space, leading to arsenate reduction. Besides, consistent with the failed amplification of arrA genes from isolated strains of the present study, no Gammaproteobacteria and Firmicutes were detected in the analysis of arrA based As (V) reducing bacterial communities, suggesting that the isolates in this study might be ARMs.

Conclusion

Three As (V)-reducing bacterial strains with strong As (V)-reducing abilities were isolated from high As groundwater of the Hetao Plain and identified as Paraclostridium, Citrobacter and Klebsiella, respectively. This was the first report of As (V) reduction by genus Paraclostridium. These strains might present different detoxifying strategies of As. The predominant As (V)-reducing populations in indigenous high As groundwater were diverse, which were dominated with Geobacter by targeting the functional arrA genes and dominated with Omnitrophica as well as Geobacter by targeting the arsC genes, respectively. The study might further expand our understanding of microbially-mediated As reduction and release in high As groundwater aquifers.