Introduction

Transport of mercury (Hg) in groundwater is gaining increasing attention as an important source of contamination to surface and drinking waters. Submarine groundwater discharge, the subsurface mixing of fresh groundwater with seawater, has been identified as a substantial source of Hg in the Waquoit Bay in Massachusetts, USA (Bone et al. 2007), the Seine Bay on the south coast of the English Channel, France (Laurier et al. 2007), and the waters of the central California Coast, USA (Black et al. 2009). Hg contamination of private wells has been reported in Southern New Jersey (Barringer et al. 1997b; Barringer and Szabo 2006) and in Tuscany (Grassi and Netti 2000; Protano et al. 2000) where the proposed cause has been mobilization of Hg from geological deposits due to the penetration of seawater (Protano et al. 2000).

Little is presently known about the role of microorganisms in the biogeochemical cycle of Hg in subsurface environments. While microbial interactions with Hg are well established in lacustrine ecosystems (Barkay et al. 2003; Barkay and Wagner-Dobler 2005), microbe-Hg interactions in groundwater and subsurface sediments may be very different due to the effects of limited oxygen and a scarcity of electron donors on microbial physiology and microbial communities. In addition, sorption of Hg to the sediment matrix, in particular to clays and iron oxides, may limit the bioavailability of Hg in the subsurface (Grigal and Nord 1983).

Nitrate is another prevalent contaminant in groundwater, causing drinking water contamination globally (Rivett et al. 2008). Nitrate contamination in groundwater in urban areas includes leaking sewage pipelines and septic systems (McMahon and Böhlke 2006), and in agricultural areas, nitrate contamination can be caused by fertilizer runoff and leaching from manure lagoons (Inwood et al. 2005). Denitrifying bacteria are ubiquitous in subsurface environments (Rivett et al. 2008), and microbial denitrification in groundwater is an important process for the protection of drinking water (McMahon and Böhlke 2006) and the prevention of eutrophication of surface waters such as lakes and estuaries (Beman et al. 2005; Bowen et al. 2007).

Understanding the interactions between denitrifying microorganisms and Hg in the subsurface is particularly pressing with regard to the remediation of nuclear waste in contaminated subsurface environments. For example, approximately 250,000 kg of Hg was disposed of at East Fork Poplar Creek (Brooks and Southworth 2011) on the Oak Ridge Reservation, TN, resulting in contamination of sediment and surface waters and accumulation of methylmercury in fish (Southworth et al. 1995, 2000). A portion of this Hg is presumed to be transported via groundwater (Southworth et al. 2000). Indeed, spikes of Hg concentrations have been recorded in groundwater at the nearby Bear Creek (NABIR FRC Public Database), where a mixed waste plume consisting of high concentrations of uranium and nitrate penetrated the aquifer from waste collection ponds.

The USDOE has established an integrated field-research challenge (IFRC) site at the Bear Creek site to test bioremediation strategies for the treatment of the mixed waste plume (Lloyd and Renshaw 2005). The residuum at this site is contaminated with radionuclides such as Uranium, Technietium-99, Thorium-230, as well as non-radioactive metals such as barium, copper, lead, mercury, nickel, vanadium and zinc, and organics such as acetone, methylene chloride, tetrachloroethylene, and toluene (USDOE 1997). Nitric acid was used for the disposal of nuclear wastes at this site, thus resulting in low pH (<4.0), and high concentrations of nitrate that reach as high as 25,000 mg/L in groundwater plumes in some areas (Brooks 2001). Concentrations of nitrate in the groundwater at a background area were 0.0683 mg/L and uranium concentrations were beneath the limit of detection (Pena et al.). This high concentration of nitrate in the contaminated area of the IFRC leads to the dominance of denitrification as the terminal electron accepting process when oxygen concentrations in the aquifer decline. In many studies, Uranium(VI) reduction does not occur until nitrate is depleted, since nitrate is an energetically more favorable terminal electron acceptor in microbial respiration (Edwards et al. 2007; Finneran et al. 2002; Wu et al. 2006a, 2006b). However, in one study, Uranium(VI) and nitrate were reduced concurrently (Istok et al. 2004). In addition, U(IV) can be re-oxidized by intermediates of the denitrification process (Senko et al. 2002). Thus, remedial strategies in the extreme environment of the subsurface at the IFRC depend to a great extent on the activities of nitrate reducing microorganisms and their response to the presence of toxicants and inhibitors in their environment. Although the IFRC sediments contain many other contaminants, none is as toxic to microorganisms and their activities as Hg (Nies 1999) as indicated by its impact on the diversity of microbial communities (Gans et al. 2005). Understanding how toxic metals such as Hg affect denitrifying microbial communities is therefore a key issue in assessing the potential for bioremediation. This may also lead to the identification of microbes that can overcome low pH and metal toxicity to sustain remedial action at the site.

Some evidence to suggest that Hg has unique effects on subsurface microorganisms comes from studies on the structure and function of the mercuric reductase, encoded by merA, which allows bacteria to reduce Hg(II) to Hg(0). Novel merA phylotypes were found in anoxic environments (Ni Chadhain et al. 2006). In the facultative, denitrifying bacterium Pseudomonas stutzeri OX, Schaefer et al. (2002) found that the mer operon was induced at higher concentration of inorganic Hg (Hg[II]) under anaerobic conditions relative to aerobic conditions, presumably because anaerobiosis limited Hg(II) transport into the cells.

The goals of our study were to quantify the effect of Hg on denitrification, ascertain Hg-dependent changes to the structure of denitrifying microbial communities, and identify the key microorganism(s) that are selected during the denitrification process in presence of Hg. We used microcosm incubations of subsurface sediments from the IFRC in Bear Creek, TN, to show sensitivity of denitrification to Hg toxicity and that microorganisms that may alleviate this toxicity. The results are relevant to environmental remediation efforts that are based on the activities of denitrifying microbes in mixed waste contaminated subsurface environment.

Materials and methods

Sampling and the primary and secondary enrichment

Two sediment cores were obtained with sterilized drilling equipment from borehole FB627, near field well FW301, at the background area of the USDOE Oak Ridge IFRC on Nov. 16, 2005. Cores from the background area consist of saprolized shale (Driese 2002). The cores were put in a sterile container flushed with argon and then stored in a refrigerator. A section representing a depth of 5.8–8.2 m below surface was shipped on blue ice to Rutgers University on March 6, 2006 (Core I). Material from a second core sampled at the same depth was shipped to Rutgers University on May 9, 2007 (Core II), and were stored at 4 °C until processing. Subsamples were removed from the cores in a polymer glove box (Coy Laboratories, Grass Lake, MI) with an atmosphere of approximately 95 % nitrogen and 5 % hydrogen. Core sediments were prepared by breaking up the material in the glove box using a sterile mortar and pestle. Ground sediments were placed in a beaker and homogenized by stirring with a sterile spatula. 5 g (wet weight) of homogenized sediments were placed in a serum bottle with 65 mL of pipes buffered artificial ground water medium (PiAGW), designed to mimic the groundwater chemistry at a monitoring well proximal to the borehole (NH4Cl—250 mg/L, MgSO4—24 mg/L, NaHCO3—360 mg/L, NaH2PO4—4.2 g/L, Na2HPO4—0.18 g/L, Wolfe’s Minerals Solution) (Atlas 2004), pH 6.7. Nitrate (5 mM as sodium nitrate) was provided as a terminal electron acceptor (TEA) for sediment enrichments of both cores, and ethanol (10 mM) was provided as an electron donor for Core I. Enrichments of Core II were established with sodium acetate (10 mM). Ethanol and acetate were selected as electron donors as they had been used as donors in previous bioremediation experiments at the IFRC (Istok et al. 2004). After 30 days of incubation at 28 °C, two active primary enrichment cultures from the same core were pooled to increase the amount of starting material, then diluted 1:40 to construct a series of secondary enrichments containing PiAGW with nitrate (5 mM), acetate (10 mM), 10 % sterilized homogenized sediment, and increasing concentrations of Hg (53–24 μM), provided as HgCl2. Secondary enrichments were constructed in triplicate for each concentration of Hg.

Pure culture isolation and identification

The primary nitrate sediment enrichment from Core I was serially diluted and plated on PiAGW medium amended with 1.5 % washed agar, 5 mM nitrate, 10 mM ethanol, and 10 or 20 μM Hg as HgCl2. Washed Agar consisted of 10 g of Bacto Agar that was stirred with 3 liters of MilliQ water overnight, filtered through a Whatman Grade 595 filter paper 150 mm (4–7 μm particle retention), then spread on aluminum foil to dry. Plates were incubated in the anaerobic chamber at room temperature for one month until colonies appeared. For each unique colony morphology, eight representative colonies were streaked three consecutive times before inoculating into liquid media for DNA extraction. The 16S rRNA genes were amplified by PCR with primer set 27F and 519R (Lane 1991) from purified DNA from at least three cultures representing each morphological type. PCR products were purified with the QIAquick PCR purification kit (Qiagen Sciences, MD) and sent to Genewiz (South Plainfield, NJ) for sequencing.

Clone library construction and restriction fragment length polymorphism (RFLP) analysis

Total DNA was extracted from 10 g IFRC sediment with methods described by Hurt et al. (2001), or from 15 mL of enrichment cultures by MoBio PowerSoil DNA isolation kit (Mo Bio Laboratories Inc., Carlsbad, CA). The 16S rRNA genes were amplified as follows: in a final volume of 25 μL, reactions contained 1 × buffer, 1.5 mM MgCl2, 0.2 mM of each dNTP, 0.4 μM of forward primer 27F and reverse primer 519R, and 1 U Taq DNA polymerase (Denville, NJ). The reactions were carried out under the following amplification conditions: 95 °C, 5 min; 28 cycles of 95 °C, 1 min; 53 °C, 20 s; 72 °C, 30 s and were terminated by 10 min at 72 °C.

PCR products (493 bp) were purified from agarose gel slices (QIAquick Gel Extraction Kit, Qiagen Sciences, MD) and cloned into pGEMT-easy vector (pGEMT-easy Vector System, Promega, Madison, WI) according to the manufacturer’s instructions. The ligation mixture was transformed into Escherichia coli DH10B competent cells (Invitrogen, USA), and transformants were selected on LB agar plates supplemented with ampicillin (100 μg/mL) and X-gal (40 μg/mL). White colonies were picked and inoculated in LB broth with ampicillin (100 μg/mL). Plasmids were extracted from these cultures to screen for insert-containing clones. For RFLP analysis, the clones were digested with EcoRI and HaeIII and run on 1 % agarose gel. DNA from at least 3 clones of each RFLP pattern was sent to Genewiz for sequencing. NCBI accession numbers: HM193541 to HM193543, HM123764.

PCR amplification of merA genes

Two degenerate primer sets known to amplify a broad diversity of merA genes (Wang et al. 2011) were used to amplify the merA gene from the nitrate enrichments and pure cultures. The first primer set, forward primer (Nsf): ATC CGC AAG TNG CVA CBG TNG G and reverse primer (rev): CGC YGC RAG CTT YAA YCY YTC RRC CAT YGT, amplified a 308 bp PCR product with the transposon Tn501 DNA as a template and was used to amplify merA from the HgCl2 treated enrichments or pure cultures. The second primer set targeted high GC mol% merA, forward primer (4 high GC-for): CGT SAA CGT SGG STG CGT GCC STC CAA G and reverse primer (4 high GC-rev): CGA GCY TKA RSS CYT CGG MCA KSG TCA GGT AGG, amplified a 1,205 bp merA product with Thermus thermophilus HB27 DNA as template and was used to amplify a long merA product. PCR programs for both primer sets included 95 °C, 5 min, followed by 30 cycles of 95 °C, 1 min, different annealing and extension temperatures and times for each primer set (59 °C, 30 s; 72 °C, 30 s for primer set 1 and 64 °C, 60 s; 72 °C, 90 s for primer set 2). At the end of the cycling a final extension of 10 min at 72 °C was followed by storage at −20 °C until analysis.

Nitrate reduction assays

The occurrence of nitrate reduction was monitored by the disappearance of nitrate and/or appearance of nitrite. Nitrate and nitrite were measured using a DX-100 ion chromatograph (Dionex, Sunnyville, CA) with sodium bicarbonate (0.954 g/L) as an eluent. Standard curves consisting of known concentrations of sodium nitrate and sodium nitrite were established for quantitation.

Terminal restriction fragment length polymorphism (tRFLP) analysis

tRFLP was performed as described previously (McGuinness et al. 2005). Briefly, fluorescently labeled 16S rRNA gene PCR products were obtained using primers 27F* (labeled with 6-carboxyfluorescien, Applied Biosystems, Foster City, CA) and 519R, and 5 μL of each PCR product was separated on a 1 % agarose gel to check DNA concentration. The concentration of DNA was determined by comparison to the intensity of bands in a λ HindIII digest by image analysis. Each PCR product (15 ng) was digested with MnlI (New England Biolab, Beverly, MA) for 6 h at 37 °C. Digests were precipitated and fingerprinting was carried out on an ABI 310 genetic analyzer (Applied Biosystems, Foster City, CA) using Genescan software.

Mercury analysis

Mercury was extracted from sediment slurries using aqua regia digestion by the following protocol (Wogelius 2005). Sediment slurries (1 mL) were mixed with 5 ml of freshly prepared Aqua Regia (3:1 12 M hydrochloric acid: 16 M nitric acid) and allowed to stand at room temperature overnight. Subsequently, they were heated to 90 °C for 3 h, with vortexing every 20 min. Samples were diluted 1:10 in 2 % hydrochloric acid and analyzed via Cold Vapor Atomic Absorbance Spectroscopy (CVAAS) using a Leeman Labs Hydra AA.

Mercury reduction assay

Prior to conducting Hg reduction experiments, the anaerobic pure culture of Bradyrhizobium sp. FRC01 was grown at 28 °C in PiAGW amended with sodium acetate (10 mM) as an electron donor and potassium nitrate (5 mM) as an electron acceptor. Cell growth was monitored by measuring optical density (OD) at 660 nm, which usually ranged from 0.01 initially to a final reading of about 0.20 within 3 days. 20 mL of PiAGW in foil-wrapped 65 mL serum bottles was spiked anaerobically with Hg as HgCl2 to a final concentration of 0.6 mM, followed by inoculation (1:10 dilution) with log-phase anaerobic cultures. Reaction vessels were incubated statically at 28 °C for 2 days in the dark. Heat-killed controls were obtained by heating an aliquot of the culture for 30 min at 80 °C. The efficiency of heat pasteurization was verified by a constant OD value observed throughout the course of the experiment, as well as by conventional viable count (i.e., spreading the cells on plates containing solid medium). On day 2, the cultures were purged with Hg free nitrogen for 45 min and elemental Hg (Hg[0]) was collected in a trapping solution consisting of (V/V) 5 % of concentrated sulfuric acid, 5 % concentrated nitric acid, 0.75 % potassium permanganate, and 0.4 % potassium persulfate. The setup of the purge-and-trap system was similar to those described previously (Allard and Arsenie 1991; Wiatrowski et al. 2006). Preliminary results showed 96–103% recovery of Hg(0) using this protocol. Mercury was analyzed by CVAAS as described above.

Statistical analysis

Values represent means, and errors are standard deviations of the means. Comparison of nitrate reduction rates were made by one-way analysis of variance (ANOVA) test followed by post hoc testing by the method of Tukey–Kramer HSD, using the program JMP version 6.0 (SAS Institute Inc.)

Results

Diverse communities in IFRC subsurface sediment cores and in the primary enrichments

Microbial community analyses of sediment material sampled from two cores obtained from the IFRC background site showed different bacterial communities (Table 1). For Core I, a clone library consisting of 45 clones of 16S rRNA genes was constructed and all clones were analyzed by digesting with the restriction enzymes EcoRI and HaeIII. Three different digestion patterns, designated as pattern I, II and III, emerged from this RFLP analysis. At least 3 clones from each pattern were then sequenced. The sequence of pattern I, representing 30 clones (66.7 % of all clones in the library, NCBI accession number HM193541), was most similar (97 % identity) to Duganella zoogloeoides strain R2-14 (AB495152.1). 11 clones (24.4 % of the library) belonged to RFLP pattern II (HM193542), which was most similar (96 % identity) to Herbaspirillum seropedicae SmR1 (CP002039.1), and 4 clones (8.9 % of the clones) belonged to RFLP pattern III and were most closely related (99 % identity) to an uncultured Enterobacteriales bacterium (FN298001.1).

Table 1 Community analysis of the IFRC background sediments derived from Core I and Core II and of the primary enrichment (5 mM nitrate and 10 mM ethanol) from Core I sediments
Full size table

To examine how nitrate enrichment affected community structure, a clone library of 16S rRNA genes was constructed to represent the community in the primary nitrate-enriched microcosm from Core I (Table 1). RFLP analysis of 148 clones were divided to three RFLP digestion patterns, 84.5 % belonged to pattern I (125 of all clones), 8.1 % to pattern II (12 clones), and 7.4 % to pattern III (11 clones). The DNA sequences of pattern I colonies were most similar to D. zoogloeoides (97 % identity), of pattern II to H. seropedicae (95 % identity) and of pattern III (HM193543) to Variovorax sp. S24561 (D84645.2) (98 % identity). This result shows that reduction of 5 M nitrate caused enrichment of D. zoogloeoides-like bacteria from 66.7 to 84.5% of the community, and Variovorax sp. S24561-like organisms, a known denitrifier (Kim et al. 2006), from non-detectable to 7.4 % of the community.

The community in sediment Core II was different from that of Core I (Table 1). Thirty-six clones were screened and sequenced. The results indicated that colonies with pattern I were most similar to three different groups of bacteria: 5.6 % (2 clones) of the library was related to Bradyrhizobium elkanii, 19.4 % (7 clones) to uncultured Proteobacteria, and 5.6 % (2 clones) to the Deltaproteobacteria. Sequences belonging to pattern II, represented by 11.1 % of the community (4 clones), were most similar to that of Acinetobacter sp. Pattern III dominated the clone library, representing 50 % of the community (18 clones) and sequences were most similar to uncultured Actinobacterium. Pattern IV represented 8.3 % of the community (3 clones) and was most closely related to Firmicutes (Table 1).

Effect of Hg(II) on nitrate removal in cultures enriched with nitrate and acetate

To determine a minimum inhibitory concentration of Hg(II) for denitrification, we split a primary denitrifying enrichment culture (5 mM of sodium nitrate and 10 mM of sodium acetate) with sediment from Core II into secondary enrichment cultures containing 10% autoclaved sediment and Hg concentrations ranging from 176 ± 30 nM to 24.5 ± 0.3 μM. Concentrations were selected based on the observation that pure cultures of anaerobic microorganisms without a mer operon can grow without inhibition in the presence of 200 nM Hg, and organisms with a mer operon can often grow in concentrations in excess of 10 μM Hg(II) (Schaefer et al. 2002; Wiatrowski et al. 2006). Nitrate removal was complete on the 13th day of incubation in the unexposed enrichments (Table 2). However, nitrate removal was not complete until the 28th day in the enrichments containing 176 ± 30 nM and the 55th day in the enrichments containing 312 ± 17 nM Hg(II). At 593 ± 15 nM Hg, one enrichment culture removed nitrate in 77 days, one in 83 days, and, in one enrichment, no nitrate-was removed after a period of 220 days. Enrichment cultures with concentrations greater than 1.1 ± 0.02 μM Hg did not remove nitrate even on the 220th day. These results suggested that Hg inhibited denitrification in a direct relationship to its concentration, although at concentrations lower than anticipated.

Table 2 Effect of Hg(II) on time of complete nitrate removal in secondary enrichments containing increasing concentrations of HgCl2
Full size table

To determine the rate of nitrate removal, we then constructed a second series of secondary enrichment cultures with sediments from Core II at lower Hg concentrations. In unexposed controls, nitrate removal commenced at a rate of 0.70 ± 0.05 mmol nitrate per day, and was completed in 5 of 6 enrichments by day 14 (Table 3, Fig. 1a). In enrichments containing 52.7 nM of Hg, nitrate removal also commenced before the sixth day of incubation, all nitrate was consumed by the 20th day of incubation, and the rate of removal, 0.62 ± 0.06 mmol/day, was not significantly different than in the unexposed enrichments. In the secondary enrichments containing 62.4 nM Hg, nitrate removal was significantly slower than in unexposed enrichments, with a rate of 0.44 ± 0.04 mmol/day, and reached completion in 2 of 3 bottles by day 20. At 97.0 nM Hg, nitrate removal was not completed until day 39, with a rate of 0.30 ± 0.01 mmol/day, significantly lower than rates at lower Hg concentrations. At 189.0 nM Hg, two cultures removed nitrate in 64 days, and one in 107 days. The enrichments containing 220 and 293 nM Hg did not remove nitrate in 114 days (data not shown). These results confirmed that, as observed at higher concentrations (Table 2), Hg inhibited the onset of nitrate removal and, furthermore, nitrate was removed at lower rates relative to rates in unexposed controls in a direct relationship to the concentration of the Hg spike (Fig. 1a; Table 3).

Table 3 Effect of Hg(II) on the nitrate removal rate
Full size table
Fig. 1
figure 1

Effect of Hg on nitrate removal in denitrifying enrichments derived from IFRC Core II. Changes in nitrate concentrations during incubations is presented in panel a and changes in nitrite concentrations in panel b. Squares unexposed (no Hg added); circles 52.7 ± 2.5 nM Hg; triangles 62.4 ± 3.4 nM Hg; and diamonds 97.0 ± 10.6 nM Hg

Full size image

Production of nitrite was also monitored during the incubations. For all the enrichments without or with different levels of Hg, maximal levels of nitrite were detected slightly after nitrate removal was completed (Fig. 1b). For example, nitrate removal in unexposed enrichments was completed by day 14 of incubation, while maximum level of nitrite was detected (2.15 mM) on day 16. Nitrate removal in 2 of 3 bottles of enrichments containing 62.4 nM Hg was completed by the 20th day of incubation, and the maximum level of nitrite (2.18 mM) was detected on day 22. Nitrate removal in enrichment containing 97 nM Hg was completed by day 39, when maximum level of nitrite (about 2.20 mM) detected on the same day. No nitrite data was collected after day 39 as these enrichments were sacrificed for tRFLP analysis. Of note is that we observed less than stoichiometric conversion of nitrate to nitrite. This is most plausibly explained if we presume that the nitrite was further reduced to either ammonia or nitrogen gas.

Effect of Hg(II) on the structure of denitrifying microbial communities

Selected denitrifying enrichments were subjected to tRFLP analysis to examine the effect of Hg on the composition of the denitrifying subsurface microbial communities. tRFLP analysis could not be performed on DNA extracted from the background sediments prior to the primary enrichment because of low DNA yields. The communities of the primary enrichment, untreated secondary enrichments, and of those treated with Hg at 53, 176 and 312 nM were analyzed. Increasing concentration of Hg decreased the number of species in the enrichments as indicated by the number of major peaks in the tRFLP chromatogram (Fig. 2). More than 14 such peaks were present in the primary enrichment, eight peaks in the untreated secondary enrichments, seven peaks in the enrichment with 53 nM Hg, six in the 176 nM Hg treatment, and only one single dominant peak remained in the enrichment with 312 nM Hg. The sequenced PCR product of the16S rRNA gene (27F*-519R) from the 312 nM Hg enrichment had a 99 % identity to that of an uncultured Bradyrhizobium sp. (EF658679) from sawfly larvae (Graham et al. 2008).

Fig. 2
figure 2

tRFLP patterns of subsurface sediment nitrate enrichments derived from IFRC Core II and exposed to increasing concentrations of Hg. Peaks representing individual species in the enrichments are labeled by numbers. Primary enrichment—the primary nitrate enrichment culture of sediment to which Hg was not added; secondary enrichment—the subculture of the primary enrichment, to which Hg was not added; 53, 176, and 312 nM—subcultures of the primary nitrate enrichment treated with 53, 176 and 312 nM of Hg

Full size image

A 308 bp merA gene fragment was amplified with degenerate primer set 1 from the DNA extract of the 312 nM Hg treated enrichment. The amino acid sequence deduced from the sequence of this PCR product was 92 % identical to merA of Pseudomonas sp. KHg3 (ACN60171.1). Together, these results suggest that in enrichments that were exposed to 312 nM HgCl2, a single merA containing denitrifier, most closely related to Bradyrhizobium spp., was selected in the secondary enrichments of IFRC subsurface sediments.

Pure culture isolation and merA gene detection

An organism closely related to the one present in the 312 nM enrichment was isolated by plating diluted primary nitrate enrichments of Core I on medium supplemented with 10 or 20 μM of HgCl2. The 16S rRNA gene sequence showed similarity to Bradyrhizobium sp. GSM-467 (FN600560.2, 99 % identity). Another group of morphologically distinct colonies were obtained, and one representative isolate was purified and its 16S rRNA gene was found most similar to that of Paenibacillus sp. (HM563047, 99 % identity). Bacterial strains belonging to these two genera have been previously documented as denitrifiers (Behrendt et al. 2010; Fernandez et al. 2008). The two new isolates were designated strains Bradyrhizobium sp. FRC01 and Paenibacillus sp. FRC02.

Since the isolates were obtained in presence of relatively high concentrations of HgCl2 (μM levels), we tested them for the presence of the merA gene. PCR reactions with Bradyrhizobium sp. FRC01 DNA produced a 1,226 bp amplicon (HM123764) when amplified with degenerate primer set 4high GC-for and 4high GC-rev (Fig. 3) and a 308 bp amplicon with degenerate primer set Nsf and rev (data not shown). Top BLASTX hits, performed using the 1,226 bp amplicon (HM123764) as a query, were all annotated as merA. Several sequences recovered in this search showed 97 % identity to our query, and represented by merAs obtained from a large group of Bacteria, including Betaproteobacteria (Cupriavidus metallidurans CH34 [YP_581983.1]), Gammaproteobacteria (P. aeruginosa [YP_789374.1], Acinetobacter sp. 6013113 [ZP_06784112.1], Salmonella enteric subsp. [ABQ57371.1], Shigella flexneri 5a Tn501 [NP_085424.1], Proteus mirabilis [ACA35072.1]) and Bacillus sp. KHg2 (ACN60170.1) of the Firmicutes phylum.

Fig. 3
figure 3

Agarose gel of the merA gene amplified from Bradyrhizobium sp. FRC01. Lane 1: 1 kb ladder; lane 2: positive control: PCR product of merA amplified with chromosomal DNA from Thermus thermophilus HB27; lane 3: PCR product of merA amplified with ten-folds dilution of Bradyrhizobium sp. FRC01 DNA; lane 4: PCR product of merA amplified with 100-folds dilution of Bradyrhizobium sp. FRC01 DNA; lane 5: negative control (no template DNA)

Full size image

The sequence of the 308 bp amplicon from Bradyrhizobium sp. FRC01 was identical to that of the 308 bp amplicon obtained from the 312 nM HgCl2 treated nitrate enrichment. Likewise, alignment of the enrichment 308 bp merA amplicon with that of the 1 226 bp merA sequence obtained from strain Bradyrhizobium sp. FRC01 was identical at the overlapping region 3′ end of merA. This result suggests that the isolated pure culture of Bradyrhizobium sp. FRC01 represented the population that was enriched to dominance in the 312 nM HgCl2 treated subsurface sediment nitrate enrichment (Fig. 2). merA genes were not detected by PCR in Paenibacillus sp. FRC02.

Hg(II) reduction activity of Bradyrhizobium sp. FRC01

To determine whether the newly isolated merA-carrying Bradyrhizobium sp. FRC01 could reduce Hg, we performed a Hg reduction assay with live and heat killed cells. A culture of live Bradyrhizobium sp. FRC01 was grown in the presence of 237 ± 35 nM Hg (total mass, 6.15 ± 0.20 μg). After 24 h, the Hg(0) that had accumulated in the headspace was flushed into acidified potassium permanganate. After flushing, 1.32 ± 0.20 μg, equal to 23.5 % of the initial Hg (Fig. 4a), was found in the potassium permanganate trap and 4.29 ± 0.20 μg of Hg remained in the culture, together accounting for a mass balance of 91.22 ± 9.81 % (Fig. 4b). In comparison, the heat-killed cells produced only 0.47 ± 0.25 μg of Hg(0), 7.64 % of the initial Hg that was added to the killed cell suspension, after 24 h of incubation (Fig. 4a). These results suggest that Bradyrhizobium sp. FRC01 reduces Hg(II) to Hg(0).

Fig. 4
figure 4

Mercury reduction by Bradyrhizobium sp. FRC01. Growing cultures were exposed to 0.6 μM HgCl2 for 2 days following which Hg(0) was purged from the incubations and collected in a trapping solution. The error bars indicate standard deviation of the mean of triplicate cultures. a Mass of Hg(0) in the trapping bottle that was reduced by the live cells (white bar) and heat killed cells (dark bar) of FRC01. b Mass balance of Hg obtained after 1 day of incubation of live cells of FRC01. Day 1 results: mass of Hg(II) remaining in reaction bottles (white bar) and Hg(0) collected in traps (dark bar). Mass balance shows a 91.2 ± 9.8 % recovery

Full size image

Discussion

Our study demonstrates that the onset and rate of nitrate removal in enrichments of the subsurface sediment from the OR-IFRC are highly susceptible to Hg toxicity. Higher Hg concentration is directly associated with delayed onset and inhibited rates of nitrate loss (Tables 2, 3; Fig. 1a). Due to the less than stoichiometric conversion of nitrate to nitrite, and subsequent loss of nitrite, the most likely process to have taken place in these enrichment cultures would be denitrification. The other two possible processes could be anaerobic ammonium oxidation (anammox) and reduction of nitrate to ammonium, both of which would be favored in incubations with high organic matter concentrations (Maier et al. 2009). In addition, anammox is not a likely pathway for the loss of nitrate in our enrichments, as this activity has only been observed in members of the phylum Planctomyces, and 16S rRNA genes homologous to those members of this phylum were not detected in this study (Table 1) and rarely occur in subsurface communities.

Denitrifiers in the subsurface of Oak Ridge are sensitive to Hg at levels that often occur in Hg contaminated groundwater at other sites. For example, mercury at 62.4 nM (about 12 μg/L), a concentration only sixfold above the US Environmental Protection Agency’s MCL of 2 μg/L for drinking water, was found to inhibit both the rate and onset of nitrate loss in enrichment cultures (Fig. 1a; Table 3). Comparable levels of Hg have been found in groundwater in New Jersey (Barringer et al. 1997a, 2005; Barringer and Szabo 2006) and in Tuscany (Grassi and Netti 2000; Protano et al. 2000) and in mine tailings in New Brunswick, Canada (Shaw et al. 2006). Additionally, Hg concentrations of 1.1 μM, comparable to what has been observed in a groundwater Hg plume at a wood treatment plant in southern Germany (Bollen et al. 2008), were found to completely inhibit nitrate removal for a period greater than 220 days (Table 2). This level of Hg sensitivity is much lower than anticipated, since anaerobically growing P. stutzeri OX can tolerate 200 nM of Hg (Schaefer et al. 2002). In retrospect, this high level of Hg resistance in P. stutzeri may be the consequence of growth in a rich culture medium, which might contain components that decrease the bioavailability of Hg (Farrel et al. 1993). Given the prevalence and importance of denitrification in the subsurface, it is notable to observe such a profound effect of low concentrations of Hg on the onset and rate of denitrification.

Although a clear relationship was demonstrated between Hg concentration and the onset and rates of denitrification (Table 2, 3; Fig. 1), slight difference existed between different batches of experiments. For example, nitrate removal was complete within 28 days in the enrichment with 176 nM of Hg (Table 2), while it took 39 days to completion in enrichment with 97 nM of Hg (Table 3). This contradiction is possibly caused by slight differences in the primary enrichment. While the same primary enrichment was used to construct the secondary enrichments, and nitrate and acetate were continually added to the primary enrichment, several months passed between constructions of the two sets of secondary enrichments, thus the biomass in the primary enrichment might have changed with time.

Our results showed that with additions of up to approximately 593 nM Hg, onset of nitrate removal was delayed rather than permanently inhibited (Table 2), most likely due to the enrichment of Hg resistant denitrifying bacteria such as Bradyrhizobium sp. FRC01 (an Alphaproteobacterium) (Fig. 2) that contain mer systems (Figs. 3, 4a, b). Barkay et al. (2010) have recently hypothesized that denitrification is the only anaerobic respiratory process that occurs at redox conditions high enough for inorganic Hg to exist as Hg(II). The selection of strains that are resistant to Hg via the mer system under denitrifying conditions supports this hypothesis. Acclimation to Hg, the enrichment of Hg resistant bacteria upon exposure to Hg and their roles in restoring activities in Hg impacted communities, was previously documented in natural waters (Barkay 1987) and soils (Rasmussen et al. 2000). This study is the first report on acclimation to Hg in subsurface communities and could impact future management of contaminated aquifers and subsurface environments.

The deduced amino acid sequence of the merA gene amplified from Bradyrhizobium sp. FRC01 has a high degree of sequence identity (97 %) to merA in a large group of bacteria including Gammaproteobacteria, Betaproteobacteria and Firmicutes, which suggests a cross phylum horizontal gene transfer (HGT) event. In many organisms, the mer operon is located on plasmids and transposons (Barkay et al. 2003, 2010), indicating the potential for spread by HGT. Moreover, HGT has been previously implicated in the prevalence of mer (De Lipthay et al. 2008; Oregaard and Sørensen 2007), and other metal-resistance genes (Coombs and Barkay 2004; Martinez et al. 2006; Sobecky and Coombs 2009) in soils and sediments from the Oak Ridge IFRC.

The role of merA in the Hg biogeochemical cycle has been documented in environments contaminated by atmospheric deposition of Hg, including lakes and polar regions (Poulain et al. 2007; Schaefer et al. 2004) and in industrial bioreactors (Felske et al. 2003). More recently, the presence of the mer system has been reported in geothermal environments where high Hg concentrations are the result of geothermal activities, such as hot springs (Chatziefthimiou et al. 2007; Simbahan et al. 2005; Wang et al. 2009) and deep sea hydrothermal vents (Vetriani et al. 2005). This is the first report for the occurrence of merA in subsurface communities and cultured isolates suggesting that adaptation to Hg in this environment may be due to the reduction of Hg(II) to Hg(0).

As of June 2009 there were 20 merA homologs in complete alphaproteobacterial genome sequences available in databases (Barkay et al. 2010) and few environmental Hg resistant Alphaproteobacteria isolates were reported. Oregaard and Sørensen (2007) reported the isolation of four alphaproteobacterial Hg resistant strains from the flood plain soils of the Hg contaminated Lower East Fork Poplar Creek Floodplain (LEFPCF) in Oak Ridge (TN, USA). All these strains contained homologs to the same merA gene (EF455065) which had 62–66 % identity to the merA of the soil bacterium Xanthobacter (ZP_01199704). Rasmussen et al. (2008) reported a merA with most similarity to a locus in Sphingomonas spp. However, to date no mer system from any representative of the Alphaproteobacteria has been genetically or biochemically characterized. The amino acid sequence of merA from Bradyrhizobium sp. FRC01 has only 36 % identity to merA from the four alphaproteobacterial isolates from LEFPCF based on a ClustalW2.0 alignment. In this study, we provided direct evidence to show that the pure culture of Bradyrhizobium sp. FRC01, which contains a merA gene, reduced Hg(II) to volatile Hg(0). This is the first time a pure culture of Alphaproteobacteria was shown to reduce Hg in vivo.

Alphaproteobacteria, including members of the genus Bradyrhizobium, were consistently detected during uranium and nitrate remediation field tests in a radionuclide contaminated area in the IFRC (Hwang et al. 2008; Vishnivetskaya et al. 2011). Thus, this specific bacterial taxon may be a key to the environmental management of contaminated subsurface sediments.

In summary, our work shows that Hg affects both onset and rate of denitrification as well as microbial community structure in subsurface sediments. The Hg resistant Bradyrhizobium spp., though initially found in subsurface sediments at low abundance, were enriched in the presence of Hg and nitrate. We hypothesize that these bacteria reduced Hg toxicity by reducing Hg(II) to Hg(0), sustaining denitrification in presence of Hg. The presence of nitrate and Hg in enrichments of subsurface sediments selected for Hg resistant Bradyrhizobium spp. with horizontally transferred merA genes. We propose that Hg resistance and detoxification by bradyrhizobia may be central to bioremediation in nitrate and Hg contaminated subsurface environments.