To describe the diversity and metabolic potential of microbial communities in uranium mine tailings characterized by high pH, high metal concentration and low permeability.
To describe the diversity and metabolic potential of microbial communities in uranium mine tailings characterized by high pH, high metal concentration and low permeability.
To assess microbial diversity and their potential to influence the geochemistry of uranium mine tailings using aerobic and anaerobic culture-based methods, in conjunction with next generation sequencing and clone library sequencing targeting two universal bacterial markers (the 16S rRNA and cpn60 genes). Growth assays revealed that 69% of the 59 distinct culturable isolates evaluated were multiple-metal resistant, with 15% exhibiting dual-metal hypertolerance. There was a moderately positive correlation coefficient (R = 0·43, P < 0·05) between multiple-metal resistance of the isolates and their enzyme expression profile. Of the isolates tested, 17 reduced amorphous iron, 22 reduced molybdate and seven oxidized arsenite. Based on next generation sequencing, tailings depth was shown to influence bacterial community composition, with the difference in the microbial diversity of the upper (0–20 m) and middle (20–40 m) tailings zones being highly significant (P < 0·01) from the lower zone (40–60 m) and the difference in diversity of the upper and middle tailings zone being significant (P < 0·05). Phylotypes closely related to well-known sulfate-reducing and iron-reducing bacteria were identified with low abundance, yet relatively high diversity.
The presence of a population of metabolically-diverse, metal-resistant micro-organisms within the tailings environment, along with their demonstrated capacity for transforming metal elements, suggests that these organisms have the potential to influence the long-term geochemistry of the tailings.
This study is the first investigation of the diversity and functional potential of micro-organisms present in low permeability, high pH uranium mine tailings.
Mining and processing of uranium are associated with the generation of large quantities of residual materials containing heavy metals (Palmisano and Hazen 2003). Micro-organisms have evolved mechanisms that enable them to expand their habitat range so as to grow in a variety of apparently inhospitable environments. An example of such an environment is the uranium tailings management facility (the Deilmann Tailings Management Facility; DTMF) at Key Lake, northern Saskatchewan, Canada, which is characterized by high pH, high metal concentrations and low permeability. Tailings management facilities have been engineered to stabilize elements of concern (e.g. As, Se, Mo, U, Ra-226), but have only been in operation for several decades. In contrast, the processes that may affect their geochemical evolution and stability will occur over significantly longer periods of time, and thus, contaminant migration into the natural environment cannot be ruled out (Camus et al. 1999). As off-site migration of metal contaminants from uranium mine tailings is well documented and a subject of much environmental concern, it is critical to identify factors that may initiate or accelerate these processes (Pyle et al. 2002; Thompson et al. 2005; Abdelouas 2006; Muscatello et al. 2008). Off-site migration may involve both geochemical and biogeochemical factors; thus, modelling of the stability of tailings waste facilities should consider the possible contribution of both eventualities (Wilkin 2008).
As indicated by Rastogi et al. (2010), various studies have examined microbial community composition in uranium-contaminated environments (Fields et al. 2005; Akob et al. 2006). However, relatively few of these studies have involved actual uranium mining or tailings deposition sites (Wolfaardt et al. 2008; Rastogi et al. 2010; Islam et al. 2011). Furthermore, most uranium tailings are acidic and unsaturated; little microbiological research has been conducted on high pH, saturated, uranium tailings deposition sites (Wolfaardt et al. 2008) such as the DTMF located at Key Lake, Saskatchewan. Geochemical studies have been conducted to characterize heavy metal behaviour in the DTMF uranium tailings, and the results indicate that their long-term stability, in terms of immobilizing metals and oxyanions such as arsenic, selenium and nickel, is dependent upon maintenance of oxic and alkaline conditions within the deposited tailings mass (Dughan-Essilfie et al. 2011; Shaw et al. 2011).
Microbial processes are important and sometimes even dominant factors in determining the fate and transport of contaminants in various subsurface environments. In general, microbial metabolism or resistance mechanisms result in redox transformation reactions leading to immobilization or increased mobility of the metal/metalloid element (Gadd 2010). Dissimilatory Fe (III) reduction is an important biogeochemical process, as it may determine both iron distribution and the fate of iron-associated elements in subsurface systems (Lovley 2006). Evidence suggests that reductive dissolution of ferrihydrite-bound arsenic leads to an elevated concentration of soluble arsenic (Islam et al. 2011). Similarly, arsenic and selenium speciation and mobility can readily be changed through microbial dissimilatory reduction or resistance mechanisms (Stolz et al. 2006). Microbial metabolism may also result in the production of acidic micro-environments, causing a local decrease in pH that may promote metal mobilization if the buffering capacity of the immediate environment is exceeded (Al-Hashimi et al. 1996).
In the DTMF, the ability of ferrihydrite to complex with, and thus control the solubility and mobility of, various heavy metals and metalloids is strongly dependent upon the oxic/alkaline condition of the tailings, which may be influenced by microbial activity. Initial studies have shown that bacteria were present in all core samples (obtained at 1 m intervals over a 1·0–60 m depth profile) throughout the tailings body, with numbers ranging from 2·1 × 102 to 2·2 × 108 cfu g−1 wet tailings material (V.F Bondici, N.H. Khan, J.R. Lawrence, D.R. Korber, unpublished data). The goal of this study was to assess the functional and biological diversity of these bacteria to develop a better understanding of potential geomicrobiological processes that might occur within the tailings over extended periods of time (on the order of 1000's of years). Culture-dependent and culture-independent methods were therefore used to characterize the microbial diversity in the tailings in conjunction with a panel of assays addressing metabolic transformation and resistance to metals to characterize potential functional roles of culturable organisms. This knowledge base will improve our understanding of the microbial ecology of these environments and provides an improved point of reference for microbial effects research on tailings geochemistry, and may also benefit efforts to mitigate the environmental impact of tailings management facilities.
The Key Lake operation is located in the southern region of the Athabasca Basin, Canada, and is an important supplier of uranium ore for the global nuclear industry (Natural Resources Canada 2004). Mining commenced in 1983 and the mined-out Deilmann pit became the mill tailings deposition site (DTMF) in 1996. Currently, the Key Lake mill processes uranium ore extracted from the McArthur River mine. As this ore is of a very high purity (c. 34% UO2), it is mixed with lower-grade ore and waste rock during milling. This process generates millions of tons of uranium tailings that are deposited on top of the original Deilmann tailings. For example, in 2004, more than 30 million tons were produced (Natural Resources Canada 2004). The DTMF is 1000 m by 600 m by 60 m deep and is currently covered with a c. 45 m layer of water. The average concentration of various metals along the depth profile of the tailings sediment has recently been reported (Shaw et al. 2011). In general, the concentration of the various elements in the two tailings zones differ; the lower Deilmann tailings have a higher elemental concentration than the overlying McArthur River tailings (Table 1). These differences are attributed to the fact that the Deilmann uranium ore body contained higher concentrations of Ni- and As-bearing sulfides; however, the pH (mean c. 10) of the tailings does not vary with depth between the two zones (Shaw et al. 2011).
|Parameter (mean)||Deilmann tailings (<410 masl)||McArthur tailings (>410 masl)|
|pH||c. 10||c. 10|
|T (oC)||c. 10||c. 10|
|Na (μg g−1)||175||35|
|As||5·9 × 103||440|
|Ni||6·1 × 103||551|
A vertical borehole was cored in the DTMF east cell to a depth of 122·6 m from water surface using a sonic track-mounted drill rig placed on a floating barge. Core sampling over the entire tailings depth (c. 60 m) was conducted using a 75 mm (inner diameter) by 3·1-m-long core barrel. Each tailings core brought to the surface was immediately aseptically subsampled (using presterilized 10-ml syringes as coring devices) at 1·0-m intervals, with duplicate samples collected at each depth for subsequent aerobic and anaerobic storage (in separate, sterile Whirlpak™bags; Fisher Scientific Ltd., Toronto, ON, Canada) and handling. Samples for anaerobic storage and processing were purged with an excess of O2-free nitrogen gas before sealing, on site. The 120 samples (60 each for aerobic and anaerobic conditions) were then stored at 4°C until processing, which was initiated immediately after receiving the shipment. The tailings core material, which was slightly radioactive, was packed in a leak-proof primary container and shipped by truck to the laboratory and analysed within 10 days from the time of sampling.
Approximately 0·2 g (+/−0·01 g) of each tailings sample was resuspended in 1·0 ml of sterile Tris–EDTA (TE) buffer at pH 8. Four types of media were used for primary isolation of aerobic, facultative anaerobic and potential strictly anaerobic bacteria: 5% tryptic soy agar (TSA), Reasoner's 2A media (R2A), R2A10 (pH 10) containing 18·2 g R2A agar and Na2CO3 with a final concentration of 0·1 mol l−1 (Schmidt et al. 2006), and a modified 0·2% TSA media that contained the major chemical constituents in the tailings pore water, as determined in 2008. This chemically-modified TSA medium was designed to improve the isolation of strains that were highly adapted to in situ DTMF conditions and hence difficult to cultivate on standard agar media. To eliminate fungal growth on bacterial plates, cycloheximide (10 mg l−1) was added to the media used for initial plating of the tailings samples (Irisawa and Okada 2009).
To enhance the isolation of potential psychrophiles, samples collected for aerobic bacteria isolation were incubated at 5°C for 3 weeks following plating (Gow and Mills 1984). An anaerobic glovebox with an O2-free atmosphere consisting of 10% CO2, 80% N2 and 10% H2 was utilized for the isolation of anaerobic bacteria at room temperature (21 ± 2°C); anaerobic media (the same four media formulations as used for aerobic cultivation) was supplemented with cysteine as a reducing agent (Meng et al. 2001). Plates were incubated under anaerobic conditions for 3 weeks. The streak plate method was employed to isolate bacterial colonies from the mixed bacterial populations growing on each media type. Distinct colonies, based on colour, size and morphological type, were selected for isolation and subsequent molecular identification. To ensure that pure cultures were obtained, the bacterial isolates were subcultured at least three times.
Isolated bacterial colonies were subjected to DNA extraction and partial 16S rRNA gene (the first three variable regions) polymerase chain reaction (PCR) amplification was performed using the 8F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 531R (5′-ACGCTTGCACCCTCCGTATT-3′) primer set (Hirkala and Germida 2004), yielding an c. 500-bp PCR product. The final concentrations of the reagents used for a 50-μl PCR reaction mixture were as follows: 1·0 μmol l−1 each of the forward and reverse primers, 1·0 U Taq polymerase (Invitrogen, Carlsbad, CA, USA), 1·0 μmol l−1 MgCl2, 1·0 X PCR buffer (both provided with the Taq polymerase) and 4·0 μmol l−1 of deoxyribonucleotide triphosphates (dNTPs). The thermal cycling profile was set according to Hirkala and Germida (2004). For sequencing, the PCR amplicons were purified using Qiagen's purification kit (Qiagen Sciences, Germantown, MD, USA) according to the manufacturer's instructions. For identification and phylogenetic analysis, 16S rRNA gene sequences were compared to all type strains using seqmatch within the Ribosomal Database Project (RDP) database (Cole et al. 2009). Sequence alignments were edited manually using the GeneDoc program v 2.7 (Nicholas and Nicholas 1997). Representative isolates from groups of different isolates having identical (100%) gene sequences were chosen (there were 59 genetically-distinct isolates in total) for further phenotypic characterization. Phylogenetic trees of the aligned unique sequences and the closest type strain sequences from the RDP database were constructed by the neighbour joining method (Saitou and Nei 1987) using MEGA5 with 1000 bootstrap replicates (Tamura et al. 2011).
An agar plate dilution method was employed to test each of the 59 different bacterial isolates for salt and metal/metalloid tolerance (Lim and Cooksey 1993). A 10 μl aliquot of a 1·0 × 108 cells ml−1 cell suspension was spotted, in duplicate, onto a series of 5% TSA plates amended with increasing concentrations of analytical grade metal salts: NaCl, NiCl2 • 6H2O (Analar, BDH, Ltd., Poole, UK), Na2SeO3 (Alfa Aesar, Ward Hill, MA, USA), Na2HAsO4•7H2O (Sigma-Aldrich, St Louis, MO, USA) and NaAsO2 (J.T. Baker, Phillipsburg, NJ, USA). An appropriate incubation period necessary for conclusive determination of growth success and phenotype of each isolate was empirically determined by observation at 2, 5, 10 and 14 days. The plates were incubated at room temperature, and growth inhibition was determined by comparison with the 5% TSA control plates. To test whether a correlation between resistance to metals and enzyme expression (De Souza et al. 2007) existed, enzymatic activity assays for oxidase, catalase, amylase, nitrate reductase, gelatinase, urease and proteinase were conducted for each distinct bacterial isolate. Standard media for enzymatic testing were prepared according to the Difco BD (Sparks, MD, USA) instructions. The biochemical tests were carried out at room temperature, and the incubation period was 1 week. The correlation analyses were conducted in R (version 2.8.1) (R Development Core Team, 2008).
The ability of bacterial isolates to mediate redox transformations of metal elements was determined by culture using the following liquid media in 24-well microtiter plates: amorphous iron (Lovley 2006), sulfate (Postgate 1951), molybdenum (Shukor et al. 2008) and 5% TSA amended media with 2·5 mmol l−1 As(III) and 5 mmol l−1 As(V). The biogeochemical potential of the bacterial isolates was determined visually, based on the appearance of a specific colour corresponding to the reduction of arsenate to arsenite (yellow), arsenite to arsenate (brown), ferric to ferrous iron (black), sulfate to sulphide (black) or molybdate to molybdenum (blue). The redox state of arsenic was determined by flooding the media with 0·1 mol l−1 of AgNO3 (Simeonova et al. 2004). Iron and sulfate enrichment broth were aseptically transferred into 24-well microtiter plates and inoculated with the bacterial isolates. The degree of oxygen requirement of each isolate was determined using thioglycollate broth, and those isolates that did not grow in this type of medium were streaked onto 5% TSA media and incubated in an anaerobic glovebox with an atmosphere of 10% CO2, 80% O2-free N2 and 10% H2. All isolates were tested for arsenic and molybdenum redox transformation potential under aerobic conditions, and only facultative isolates were subjected to anaerobic growth on arsenic, iron and sulfate enrichment media. Incubation periods for colour development were determined empirically; 4 weeks for Fe (III) and SO4 enrichment media, and 2 weeks for As(V), As(III) and Mo(VI) transformations. All tests were carried out in duplicate, and the positive control organisms for the iron and sulfate reduction tests were Geobacter metallireducens (ATCC 53774) and Desulfovibrio desulfuricans (ATCC 29577), respectively.
The FastDNA SPIN kit (MP Biomedicals, Solon, OH, USA) for soil was used to extract DNA from the tailings core samples by following the supplier's recommended protocol with minor modification. Aliquots of 0·5 g tailings sediment from each of the 60 core samples were subjected to DNA extraction. Equivalent volumes of extracted DNA from individual tailings samples were then combined, resulting in three DNA pools representing composited material from the upper, middle or lower layers of the DTMF, hereafter referred to as the 0–20, 20–40 and 40–60 m pools.
Ion Torrent sequencing and data processing were carried out using Torrent adapter A (forward) and adapter P1 (reverse), which were annealed to sequencing primers specific for the V5 region of the 16S rRNA gene. The forward primer also included a key tag and a multiplex identifier (MID). The sequence of these primers were F5′-CCATCTCATCCCTGCGTG TCTCCGACTCAGMIDGATTAGATACCCTGGTAG and R5′-CCTCTCTATGGGCAGTCGGT-GATCCGTCAAT-TCCTTTRAGTTT, respectively, and the single underlined sequence represents the target region-specific primer, the double underlined sequence is the key tag, and the sequences in bold are the adapters. The MIDs used for each pooled DNA sample were as follows: ACGCT (0–20 m), AGACG (20–40 m), and AGCAC (40–60 m). PCR reactions were carried out in 50 μl volumes containing 2 μl of template DNA, 1·0 μmol l−1 each primer, 1·0 U Taq polymerase (Invitrogen), 1·0 μmol l−1 MgCl2, 1·0 X PCR buffer (both provided with the Taq polymerase) and 4·0 μmol l−1 of deoxyribonucleotide triphosphates (dNTPs). Cycling conditions consisted of a 5-min denaturing step at 95°C followed by 40 cycles of 30 s at 95°C, 30 s at 57°C and 45 s at 72°C, and a final elongation step of 10 min at 72°C. The PCR amplicons were purified using a Qiagen purification kit (Qiagen Sciences) according to the manufacturer's instructions and quantified using a NanoDrop spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Pooled, MID-tagged amplicon libraries were sequenced using an Ion Torrent Personal Genome Machine (Life Technologies, Carlsbad, CA, USA) with a 314 chip at the Biotechnology Research Institute (Montreal, Quebec, Canada). Sequence data were processed using RDP pyrosequencing pipeline tools. The pipeline Initial Process Tool was used to trim and remove low-quality sequences, and sequences shorter than 100 bp. Quality sequences were submitted to the RDP classifier with a bootstrap cut-off value of 50%. Unclassified DNA sequences were further examined using Basic Local Alignment Search Tool (BLAST) to confirm that they were the 16S rRNA-encoding gene. Once non-16S rRNA sequences were removed, Unifrac analysis was used to compare the three DNA libraries based on phylogenetic information obtained from the GreenGene core data set (Hamady et al. 2010).
The cpn60 clone libraries were assembled according to Schellenberg et al. (2009) based on pooled DNA from the 0–20, 20–40 and 40–60 m regions of the DTMF. A total of 2800 white colonies were randomly picked, and sequences were processed according to Schellenberg et al. (2009). Trimmed cpn60 sequences were submitted to an RDP classifier previously trained on a set of cpn60 UT sequences, which consisted of all type I chaperonin sequences.
DNA sequences obtained in this study were submitted to the EMBL database. The accession numbers for the partial 16S rRNA gene sequences are HE650716 to HE650774. The accession numbers for the cpn60 sequences are KC552251 to KC553978.
A total of 199 bacterial isolates were obtained, of which 51 and 148 were recovered under anaerobic and aerobic conditions, respectively. Analysis of partial 16S rRNA sequences revealed the presence of 59 unique bacterial isolates, of which 25 were affiliated with the phylum Proteobacteria, with eight, nine and eight isolates belonging to the α-, β- and γ-Proteobacteria, respectively. Twenty-three isolates belonged to the Actinobacteria, nine to the Firmicutes, and two to the Bacteriodetes (Fig. 1). The number of representative genera was found to be the highest for the Proteobacteria phylum (25 isolates) and lowest for the Bacteriodetes (two isolates). Proteobacteria was the proportionately-dominant phyla in the upper and middle zones of the DTMF, and Actinobacteria was dominant in the lowest zone, respectively. Despite the fact that the majority of the unique isolates showed >97% sequence similarity to previously cultured type strain bacteria reported in the 16S rRNA database, 25% exhibited divergence of greater than 3%, indicating that they represented putatively novel species (Table 2) (Keswani and Whitman 2001).
|Distinct isolates||Accession number||Best 16S rRNA gene match||Family||% identity|
|AET17H||HE650749||Polaromonas vacuolata 34-P||Comamonadaceae||95|
|AER18D||HE650759||Polaromonas vacuolata 34-P||Comamonadaceae||95|
|AER14F||HE650747||Massilia niastensis 5516S-1||Sphingomonadaceae||96|
|AER11A||HE650757||Paenibacillus chondroitinus DSMZ5051T||Paenibacillaceae||90|
|ANr04A||HE650717||Alkalibacterium iburiense M3||Staphylococcaceae||94|
|ANr55A||HE650734||Bacillus akibai 1139||Bacillaceae||94|
|ANr41/42||HE650730||Bacillus circulans T||Bacillaceae||96|
|AER05A||HE650729||Hymenobacter norwichensis NS/50||Bacteroidetes||94|
|AER37C||HE650737||Modestobacter versicolor CP153-2||Geodermatophilaceae||96|
|AET51J||HE650755||Georgenia muralis 1A-C||Bogoriellaceae||96|
|AET35A||HE650763||Arthrobacter aurescens DSM 20124||Micrococcaceae||96|
|AET51I||HE650743||Leifsonia antarctica SPC20||Microbacteriaceae||93|
|AET52A||HE650745||Plantibacter flavus P 297/02||Staphylococcaceae||96|
|AER22D||HE650752||Yonghaparkia alkaliphila KSL-113||Microbacteriaceae||96|
|AER22/23A||HE650754||Pedobacter cryoconitis A37||Sphingobacteriaceae||92|
The majority of the 59 representative isolates showed tolerance to elevated NaCl and metal/metalloid salts concentrations (Fig. 2). The resistance profiles varied significantly; nickel and arsenite were inhibitory at relatively lower concentrations in comparison with arsenate and selenite. The overall toxicity of the metal ions was as follows: Ni(II) > As(III) > Se(IV) > As(V). Of the 59 cultures tested, 48% exhibited tolerance to 7% NaCl (w/v), 70% to 10 mmol l−1 Se(IV), 73% to 100 mmol l−1 As(V), 32% to 1·0 mmol l−1 Ni(II) and 35% to 5·0 mmol l−1 As(III). These concentrations corresponded to, or encompassed the range of, concentrations previously employed to discriminate metal/metalloid-resistant from metal/metalloid-sensitive bacteria (Burton et al. 1987; Sabry et al. 1997; Drewniak et al. 2008). A high proportion of the tested strains were resistant to multiple-metal ions: 14% of the tested isolates were tetra-, 31% were tri-, and 27% were dual-metal resistant. The multiple-metal-resistant isolates predominantly belonged to the genera Pseudomonas, Arthrobacter, Massilia, Micrococcus, Staphylococcus, Rhodococcus and Bacillus. As the concentration of metal salts increased, the percentage of isolates capable of growth decreased; however, 15% of all bacterial isolates exhibited dual-metal hypertolerance (Table 3). These isolates not only showed tolerance to 300 mmol l−1 Se(IV) and As(V) metalloids, but the majority could also grow at a relatively high NaCl concentration (15%). All dual-metal hypertolerant isolates demonstrated extreme tolerance to As(V) (400 mmol l−1), and three isolates exhibited tolerance to Se(IV) (400 mmol l−1). Furthermore, two isolates, which were closely related to Rhodococcus fascians (AET15H) and Arthrobacter aurescens (AET35A), showed tolerance to 20 mmol l−1 Ni(II) and 500 mmol l−1 Se(IV), respectively.
|Name of the isolates||Closest match in RDP||% NaCl (w/v)|
Enzymatic assay (i.e. catalase, oxidase, amylase, protease, gelatinase, urease and nitrate reductase) results were correlated with the metal ion resistance profiles of the distinct bacterial isolates, as conducted previously by De Souza et al. (2007). In general, bacterial isolates exhibiting multiple-metal resistance had a considerably wider range of enzymatic activities (Fig. 3). A correlation coefficient (R = 0·43, P < 0·05) indicated that there was a moderate positive correlation between the two variables. Thus, multiple-metal-resistant isolates tended to express a broader range of enzymatic potential than organisms that were resistant to fewer metals.
Seventeen isolates reduced amorphous iron, and none of the isolates reduced sulfate, arsenate or oxidized arsenite under anaerobic conditions. The majority of the iron-reducing micro-organisms were closely related to the genus Pseudomonas and the rest belonged to the genera Pectobacterium, Microbacterium, Sphingomonas, Methylobacterium, Paenibacillus, Pedobacter and Janthinobacterium. Furthermore, metal redox change was detected only for molybdate and arsenite under aerobic conditions, with 22 isolates reducing molybdate and seven isolates oxidizing arsenite to arsenate. Isolates that were able to oxidize arsenite were closely related to Herminiimonas arsenicoxydans, along with the genera Arthrobacter, Micrococcus, Polaromonas, Massillia and Sphingomonas.
To complement the culture-based methods, and to increase the sensitivity of detection of microbial diversity existing within the DTMF, two culture-independent techniques targeting two universal bacterial targets were employed. The same DNA template was used for both 16S rRNA amplicon Ion Torrent sequencing and cpn60 clone library Sanger sequencing. A total of 21 640 high-quality reads with an average length of 173 bp were retrieved following Ion Torrent sequencing, including 10 169, 4152 and 7517 sequences corresponding to the 0–20, 20–40 and 40–60 m DNA libraries, respectively. From the clone library data sets, a total of 920, 952, 693 full-length cpn60 universal target reads were generated for the DNA extracted from the 0–20, 20–40 and 40–60 m zones, respectively. The combined data were classified into 415 genera representing 21 phyla. Furthermore, the two methods revealed slightly different taxonomic profiles, although the most abundant taxa in each data set were from the same eight phyla (Fig. 4). Seven phyla (Gemmatimonadetes, Chloroflexi, Deferribacteres, Deinococcus-Thermus, Verrucomicrobia, Aquificae, Thermotogae) were detected only in the cpn60 libraries; whereas, five classified phyla (Cyanobacteria/Chloroplast, Planctomycetes, Chlamydia, Fusobacteria, Tenericutes) and one unclassified group were detected only by 16S rRNA sequencing. The most prevalent phylum detected by both approaches was Proteobacteria. The relative abundance of this phylum varied between the two sequencing techniques, but exceeded 50% relative abundance in either case. The other major constituents of the sequence data sets included Firmicutes, Actinobacteria and Bacteroidetes (Fig. 4).
As the shared phyla of the two data sets included the most abundant taxa, the results were compared at the higher resolution of family and genus levels. A total of 151 families were detected of which 47% was shared, 21% unique to the cpn60 clone libraries, and 32% unique to the 16S rRNA libraries (Fig. 5). It was remarkable to observe that only a relatively low percentage (13%) of the phylotypes were shared at the genus level, with 27% unique to the cpn60 clone library and 60% to the 16S rRNA method, respectively. Furthermore, the shared phylotypes were minor members of the bacterial community with relatively low frequency, except for Rhodoccocus, Acidovorax, Oxalicitibacter, Polaromonas, Propionibacter, Pseudomonas, Variovorax. At the fine-scale analyses of the overlapped reads, Ion Torrent sequencing generally not only detected the common phylotypes at higher frequency, but was also able to identify them in more samples in comparison with the cpn60 clone library.
Consistent with observations of culturable bacteria, sequence libraries from all three depths were largely dominated by Proteobacteria, Firmicutes, Actinobacteria and Bacteroidetes. Comparison of the phylogenetic composition of the 16S rRNA sequence libraries with Unifrac distance statistical analyses showed highly significant differences (0·001 < P < 0·01) between the upper zones (0–20 m and 20–40 m) and lower zone (40–60 m) DNA libraries; whereas, the upper zone was significantly different (0·01 < P < 0·05) from the middle zone (Table 4).
|Ion Torrent libraries of the tailings zone||0–20 m||20–40 m||40–60 m|
Taxonomic profiling by both sequencing methods revealed the presence of bacterial groups that are well documented in terms of biogeochemical potential for transforming elements that are abundant in the DTMF tailings and whose redox transformation may directly or indirectly influence mobilization of elements of concern (i.e. As, Se, Mo). Fifteen and nine genera were captured that belong to recognized sulfate-reducing and iron-reducing bacterial groups, respectively. Four phylotypes, including Ferruginibacter, Georgfuschia, Dethiobacter and Albidiferax, were detected consistently in all three pooled libraries with a maximum frequency reaching 0·9% in the lower zone (40–60 m).
Elevated concentration of various heavy metals and oxyanions (Ni, Se, As, Mo) and high pH (10) dominate the DTMF environment. In general, the presence of heavy metals tends to inhibit bacterial activity and growth; therefore, micro-organisms persisting in metal-contaminated subsurface systems are likely to possess a high degree of genetic flexibility (Romero and Palacios 1997) aiding in their adaptation to DTMF conditions. The positive correlation between bacterial potential activity and cultivability may imply that the specialized, readily cultivable microbiota play an important role in this unique ecosystem in terms of biogeochemical function (Ellis et al. 2003). In the DTMF tailings, representatives from the Proteobacteria, Actinobacteria and Firmicutes phyla were most frequently isolated by culture-based techniques. However, earlier studies have demonstrated that these phyla are, in general, the predominant culturable bacterial groups (Fredrickson et al. 2004; Zhang et al. 2007; Jose et al. 2011), and thus, they tend to be among the numerically most abundant taxa recovered at various contaminated sites (Akob et al. 2006; Rastogi et al. 2011).
It was somewhat remarkable that 69% of the DTMF isolates tested exhibited multiple-metal resistance. Other studies of multiple-metal resistance genes demonstrated the adaptive ability of a microbial community indigenous to mixed-contaminant waste sites (Sobecky and Coombs 2009). There are three main microbial mechanisms of metal ion resistance: efflux, sequestration and redox modification (Ordonez et al. 2005). Multiple-metal resistance may require a combination of these mechanisms. In the present study, we also found that nine dual-hypertolerant bacterial isolates could grow in the presence of at least 300 mmol l−1 of Se(IV) and As(V). The majority of these isolates were retrieved from the middle zone (20–40 m) of the tailings, which includes the interface between the McArthur River and Deilmann mine tailings. Selenium (IV) and As(V) are widespread in nature, and therefore, bacterial resistance would not be confined to contaminated areas. However, such a high level of tolerance to these elements is, based on the literature, rare (Drewniak et al. 2008). In our study, isolates closely related to Rhodococcus fascians and Arthrobacter aurescens exhibited hypertolerance to arsenic, selenium and nickel (Table 3).
Previous investigations have also reported exceptionally high levels of metal tolerance (hypertolerance) or multiple-metal resistance for bacterial isolates retrieved from contaminated environments; for example, Arthrobacter spp. from the Savannah River Site, South Carolina, and the Hanford Site, Washington, with resistance to Pb(II), Hg(II) and Cr(VI) (Benyehuda et al. 2003), and a Rhodococcus sp. from a gold mine with tolerance to As(V) (Drewniak et al. 2008). Members of the coryneform group are recognized for their high tolerance to elevated concentrations of As(V) (Ordonez et al. 2005; Drewniak et al. 2008); in addition, the arsenic resistance determinants of a Corynebacterium glutamicum strain have been fully-elucidated (Ordonez et al. 2005). However, little research has been conducted on the metal tolerance of Arthrobacter spp. (Benyehuda et al. 2003). The Arthrobacter isolates recovered in the present study with demonstrated involvement in redox reactions could have biotechnological relevance because metal tolerance is also considered to be a prerequisite characteristic for bioremediation strategies. Micro-organisms with potential to successfully bioremediate a pollutant must first survive in the presence of both target and nontarget toxic metals (Haferburg et al. 2007).
Previous studies on Antarctic bacterial isolates showed an inverse correlation between multiple-metal-resistant bacteria and enzymatic potential (De Souza et al. 2007). Overall, their results demonstrated that the metal-resistant strains tended to have lower enzymatic potentials. In contrast, the DTMF isolates displayed a moderately positive correlation between multiple-metal resistance and enzymatic activity, suggesting that multiple-metal resistance did not compromise general, or overall, metabolic activity.
It was previously recognized that organic carbon tends to make heavy metals less bio-available (Kim et al. 1999); therefore, bacterial isolates retrieved from environments with relatively high organic carbon content would tend to be metal sensitive and possess a wider range of enzyme profiles. Furthermore, it was previously observed that isolates from environments characterized by lower organic carbon concentrations exhibited multiple-metal resistance with lower enzymatic potential (Ordonez et al. 2005). The DTMF tailings contain relatively high concentrations of numerous heavy metals and the solid-phase organic carbon concentrations may reach 0·9%, a relatively high carbon content. It is thus speculated that the microbiota in the DTMF environment may not be required to sacrifice enzymatic functionality due to the abundance of both factors.
Microbial iron reduction is a ubiquitous and well-documented process that may influence the geochemistry of subsurface settings. Phylogenetically diverse Fe(III)-reducing organisms have been isolated and described from various environments spanning a wide range of chemical and physical conditions (Weber et al. 2006). The most common and comprehensively studied micro-organisms that conserve energy to support growth from Fe(III) reduction belong to the Geobacteriaceae (genera: Geobacter, Desulfuromonas, Desulfuromusa and Pelobacter, Acidobacteria and Shewanella spp.) (Lovley 2006; Weber et al. 2006). Other iron-reducing bacterial groups have been previously reported; for example, Pseudomonas spp. (Nevin et al. 2003). In a microcosm study conducted on high salinity, uranium-contaminated sediment samples, phylogenetic profiling demonstrated that micro-organisms closely related to Pseudomonas species were involved in nitrate, followed by Fe(III) and U(VI), reduction (Nevin et al. 2003).
In our study, the well-known iron-reducing bacterial groups were not recovered because we did not use specific liquid media for them during initial culturing and isolation. However, our observation that the readily culturable bacterial members besides the well-known iron reducers exhibited reductive dissolution extends our understanding of the indigenous microbial community's biogeochemical potential and their potential effect on system stability. Furthermore, characterization of these organisms may allow the identification of physiological factors that may control the rate of iron reduction in this type of environment. Reductive dissolution of Fe on a large scale within the DTMF tailings would be a concern, as this process could result in the release of adsorbed and co-precipitated As into the aqueous phase, with subsequent migration of this element into the environment.
The fact that none of the tailings bacteria isolated during this study were able to reduce arsenate under anaerobic or aerobic conditions suggests that arsenate did not serve as an electron acceptor during respiration and that whatever detoxification process is utilized by these organisms, it does not involve redox transformation to arsenite. However, in the case of the seven isolates that were able to oxidize arsenite under aerobic conditions, oxidation was possibly due to a detoxification process rather than use of arsenite as an energy source (Oremland et al. 2004). Several isolates have been reported to oxidize arsenite; for example, Thiomonas (Casiot et al. 2003), Pseudomonas (Chang et al. 2010) and Polaromonas (Osborne et al. 2010). Bacterial conversion of As(III) to As(V) is also of considerable interest for bioremediation applications, as this process generates a less-toxic and less-soluble arsenic species (Oremland et al. 2004). Micro-organisms isolated from the DTMF with the ability to convert specific redox-sensitive elements to an insoluble state would thus have potential application in bioremediation. This is particularly true for bacterial isolates that are well adapted to contaminated systems, such as Arthrobacter aurescens, which exhibited extreme tolerance to As(V) (400 mmol l−1) and Se(IV) (500 mmol l−1), and may act to stabilize the DTMF tailings through conversion of As(III) to As(V).
To minimize any potential biases of the 16S rRNA-based approach, techniques based on protein-encoding universal targets used in parallel with culture-based methods have been suggested (Donachie et al. 2007; Na et al. 2011; Roux et al. 2011). Comparative studies have demonstrated that cpn60 fulfils the criteria of being an excellent molecular target and provides greater discriminatory power when closely related species are compared (Goh et al. 1996; Hill et al. 2005). In our study, the integration of these two molecular approaches led to the capture of greater bacterial diversity (21 phyla) relative to the use of either molecular technique employed alone. The complementary nature of these two sequencing methods targeting two different universal targets was further evidenced by the fact that at the family-level, there was a relatively high fraction (47%) of taxa overlap; whereas, at higher taxonomic resolution only 13% overlap existed between the two data sets. A similar pattern has been observed in other comparative studies where 16S rRNA Sanger and pyrosequencing data sets were compared (Na et al. 2011). The low number of bacterial taxa shared at the genus level can most likely be attributed to data set size, amplification bias of ‘universal’ PCR primers, as well as library construction method (clones vs. amplicon sequencing) (Schellenberg et al. 2009; Roux et al. 2011). Ion Torrent sequencing is a relatively new platform; however, reliability has recently been validated in a study that provided near-identical output as the 454 sequencing approach (Yergeau et al. 2012). It is noteworthy that this study represents the first time that the RDP classifier has been used for cpn60 UT sequences. With an 80% cut-off value, the taxonomic identification of the cpn60 sequences obtained with the trained classifier was in good agreement with the output of the wateredBlast.
It has previously been demonstrated that concentrations of metals/metalloids could be considered as a strong determining factor of bacterial diversity at uranium-contaminated habitats (Fields et al. 2005; Akob et al. 2006). The significant difference in microbial diversity across the tailings depths can be primarily attributed to the difference in geochemical composition of the two ore bodies. In general, the concentration of the various elements (e.g. As, Ni) in the upper zone of the Deilmann solids is c. 100X lower than in the deeper zones (Shaw et al. 2011).
The major lineages (Proteobacteria, Firmicutes, Actinobacteria and Bacteroidetes) captured by both molecular techniques were also detected by culturing. Only Pectobacterium, Modestobacter, Georgenia, Planococcus, Paenisporosarcina, Hymenobacter, Kinococcus were isolated, but not detected, by the culture-independent methods. The fact that the microbial diversity spectrum captured by the molecular techniques can be extended by the parallel application of culture-based methods is well documented (Donachie et al. 2007). It has previously been suggested that one of the potential biases relates to the difficulty in obtaining DNA template from bacterial groups with spore-forming ability (Donachie et al. 2004).
Of particular importance is that the readily-culturable bacterial isolates closely related to, for example, Rhodococcus, Pseudomonas, Arthrobacter, Massillia, Micrococcus, Polaromonas, Hydrogenophaga and Bacillus, were also consistently observed at high frequency in the metagenomic data sets. In addition to relative high abundance and culturability, these micro-organisms exhibited a high degree of tolerance to metals/metalloids or reducing/oxidizing capability. This further suggests an important role of the abundant and readily culturable micro-organisms in terms of their biogeochemical function in the DTMF system.
The most abundant lineages detected solely by sequence-based methods were also affiliated to Proteobacteria and Firmicutes that are known to survive harsh environments, such as extreme pH and heavily contaminated areas (Akob et al. 2006). Less-abundant, but important, phylotypes that are closely related to well-known sulfate-reducing and iron-reducing bacteria were also identified. Representatives of sulfate-reducing and iron-reducing lineages included Dethiobacter, Desulfovibrio, Desulfosporosinus, Desulfomicrobium and Geobacter, Geoalkalibacter, Georgfuchsia, Ferruginibacter, Ralstonia, respectively. These genera have been reported previously from uranium-, heavy metal- and hydrocarbon-contaminated sites (Sitte et al. 2010; Zavarzina et al. 2006; Rastogi et al. 2011). Bioremediation studies have demonstrated that members of these genera may become numerically- and functionally-important when ferric- and sulfate-rich sediments are amended with various carbon sources, resulting in reductive dissolution (Vrionis et al. 2005; Na et al. 2011).
This is the first examination of microbial diversity and functional potential of culturable bacterial populations within high pH, metal-rich, low permeability mining-impacted environments such as the DTMF. High (and multi-) metal tolerance, as well as broad enzymatic potential, indicate that a significant proportion of the resident micro-organisms are well adapted to this unique environment. The fact that the uranium tailings support the growth of well-known iron-reducing and sulfate-reducing bacteria, along with detection of culturable bacterial populations that exhibit metal-reducing capability, indicates the potential for Fe and S biogeochemical cycling. Overall, the results of this study also emphasize the value of using multiple approaches for the characterization of microbial functional and taxonomic diversity in complex ecological systems like the DTMF.
Cameco Corporation, Natural Sciences and Engineering Research Council of Canada-Collaborative Research and Development (NSERC-CRD) and NSERC are acknowledged for their financial support. Matthew Links is acknowledged for his assistance with the training and application of the RDP Classifier for use with cpn60 sequences. Monique Haakensen is acknowledged for isolating DNA from tailings samples. Tom Kotzer, Harm Matthius and Virgil Guran are thanked for their technical assistance.
The authors declare that no conflict of interest exists relating to this work.