Introduction

Perchlorate (ClO4 −) is widely used in rocket fuel, fireworks, and explosives due to its high energetic value as a strong oxidant and its chemical stability (USEPA 2005). Its release into water resources causes contamination. Perchlorate can interfere with the synthesis and secretion of thyroid hormone, which affects human metabolism and hinders human growth and development (USEPA 2008). Though the US Environmental Protection Agency (USEPA) has not yet established a nationwide maximum contaminant level (MCL) for ClO4 −, some states have set up cleanup target levels from 2 to 18 μg/L in drinking water (Gu and Coates 2006). Nitrate (NO3 −) is a co-contaminant with ClO4 − in groundwater (USEPA 2001). Nitrate has been regulated by the USEPA at 10 mg N/L in drinking water (USEPA 2009).

Commonly used technologies for removing ClO4 − and NO3 − from water include membrane filtration, ion exchange, chemical reduction, and biological reduction (Coates and Achenbach 2004; Matos et al. 2008; Zhao et al. 2014). Membrane filtration and resin regeneration in ion exchange produce highly saline ClO4 − and NO3 − brines that need to be further treated or disposed of. Biological reduction can be used to directly treat ClO4 −- and NO3 −-contaminated water or brines produced during membrane filtration or ion exchange. Compared to chemical reduction, biological reduction consumes less energy and requires less chemical addition because enzymes in microbes are used as the catalysts (Ye et al. 2012). During biological reduction, the perchlorate reductase (pcrABCD) reduces ClO4 − to chlorite (ClO2 −) (Kengen et al. 1999), and chlorite dismutase (cld) catalyzes the disproportionation of chlorite (ClO2 −) to chloride (Cl−) and oxygen (O2) (Van Ginkel et al. 1996), which is subsequently reduced to H2O. During the NO3 − reduction, reductions of nitrate to nitrite (narGHJI), nitrite to nitric oxide (nirSJFD/GH/L), NO to N2O (norZ), and N2O to N2 (nosZDFY) are performed by various reductases (Ettwig et al. 2010).

Perchlorate-reducing bacteria (PRB) and denitrifiers can use a variety of electron donors to gain energy (Miller and Logan 2000; Sahu et al. 2009; Son et al. 2006; Yu et al. 2006). When the electron donor is a gas, such as hydrogen and methane, membrane biofilm reactors (MBfRs) are one of the most widely used reactors because they have close to 100 % gas utilization efficiency and high NO3 − and ClO4 − removal rates (Zhao et al. 2013; Tang et al. 2012a, b; Luo et al. 2015; Chen et al. 2016).

Compared with CH4-based MBfRs, H2-based MBfRs have been extensively studied and used at the field scale for NO3 − and ClO4 − removal. The effects of salinity on nitrate and perchlorate reduction in H2-based MBfRs are well understood. H2-based MBfRs reach simultaneous reduction of nitrate and perchlorate, but the reduction rate decreases as the salt concentration increases. Chung et al. (2007) reported that nitrate and perchlorate reduction was not affected by 2 % salinity (20 g/L NaCl) in a H2-based MBfR, while 4 % salinity (40 g/L NaCl) decreased reduction by 40 %. Van Ginkel et al. (2008) investigated the kinetics of ClO4 − and NO3 − reduction in a H2-based MBfR fed with brines. They found that the NO3 − removal fluxes and H2 pressure were roughly first order but roughly zero order with NO3 − concentration in H2-based MBfR. NO3 − and ClO4 − were reduced simultaneously in a CH4-based MBfR (Luo et al. 2015), but the effects of salinity on NO3 − and ClO4 − reduction in CH4-based MBfRs are unknown.

CH4-based MBfRs have been much less studied compared to their H2-based counterparts. Previous research has demonstrated the potential of using CH4-based MBfRs to simultaneously remove NO3 − and ClO4 − from water (Luo et al. 2015; Chen et al. 2016); however, these studies were all conducted using synthetic fresh waters. Therefore, it is of interest to investigate how salinity affects biological perchlorate and nitrate reduction in a CH4-based MBfR. The overall objective of this work is to investigate the effects of salinity on the simultaneous reduction of NO3 − and ClO4 − in a CH4-based MBfR.

Materials and methods

MBfR setup

A two-column MBfR system similar to Zhao et al. (2011) was used in this study. The MBfR had composite hollow fibers (hydrophobic microporous polyethylene fiber, 280-μm o.d., 180-μm i.d., and 0.1–0.15-μm pore size) manufactured by Mitsubishi Rayon (Model MHF-200TL, Mitsubishi, Ltd., Japan). One column contained 32 fibers, and the other contained 20 fibers as “coupon fibers” for sampling purposes. Each fiber ran all the way through the column, with one end glued into a CH4 supply manifold and the other end sealed to avoid gas leakage. The total volume of the MBfR was 65 mL, and the total membrane surface area was 90.24 cm2. The MBfR was completely mixed via recirculation with a peristaltic pump (Longer Pump, Model 1515X, Longer Precision Pump Co, Ltd., China) at 100 mL/min.

Inoculum source and culture medium

We inoculated the MBfR with a 10-mL culture from a CH4-based MBfR (Luo et al. 2015). The medium contained the following mineral salts (analytical grade or purer) per liter of deionized water: 1 mg CaCl2, 0.3 g NaHCO3, 5 mg MgSO4∙7H2O, 0.2 g KH2PO4, 0.4 g NaHPO4∙12H2O, 1-mL acid trace element solution (100 mM HCL, 2.085 g FeSO4·7H2O, 68 mg ZnSO4·7H2O, 14 mg H3BO3, 120 mg CoCl2·6H2O, 500 mg MnCl2·4H2O, 320 mg CuSO4, 95 mg NiCl2·6H2O per liter), and 1-mL alkaline trace element solution (10 mM NaOH, 67 mg SeO2, 50 mg Na2WO4·2H2O, 242 mg Na2MoO4·2H2O per liter). The medium was degassed with N2 for 15 min to maintain anaerobic conditions. The medium pH was adjusted to 7.5 ± 0.2 with hydrochloric acid. The influent feeding rate was 0.5 mL/min (hydraulic retention time (HRT) of 130 min), the CH4 pressure was 10 psig (0.69 bar), and the temperature was 29 ± 1 °C for all experiments.

Start up and continuous operation of the MBfR

After inoculation, the MBfR was fed with NO3 − and ClO4 − at ∼15 and ∼ 10 mg/L, respectively, and maintained for more than 20 days to allow sufficient biomass to accumulate. After the startup stage, we fed the reactor with 2 mg/L of perchlorate from stage 1 to stage 5. To investigate the effects of salinity on ClO4 − and/or NO3 − reduction, we set up salinity at 1 % (10 g/L of NaCl) in stage 2 and 0.2 % (2 g/L of NaCl) in stages 5 and 6. To further investigate if nitrate inhibition plays a role in the inhibition mechanisms, we varied the nitrate concentrations in different stages: 10 mg NO3 −/L in stages 1, 2, and 3 and 5 mg NO3 −/L in stage 6. We allowed each stage to reach steady state, which was defined as when effluent concentrations were stable (<10 % variation) at least three times the HRT.

Analyses

Liquid samples were taken from the MBfR with 5-mL syringes and filtered immediately through 0.2-μm membrane filters (LC + PVDF membrane, Shanghai Xinya, China). NO3 − and NO2 − were assayed using ion chromatography (Metrohm 833 Basic IC plus, Switzerland) with an A-Supply-5 column and an eluent containing 3.2 mM NaHCO3, 1.0 mM Na2CO3, and 5 % acetone at a flow rate of 1 mL/min. ClO4 − was measured using ion chromatography (DIONEX ICS-1000, USA) with an AS 16 column and AG 16 precolumn, an eluent concentration of 35 mM KOH, and a 1.0 mL/min flow rate. The pH values of the influent and effluent were measured using a pH meter (Seven Easy, Mettler Toledo, Switzerland) and were between 7.5 and 7.7 for all stages. The bacterial morphology on the surface of membrane at the end of each stage was analyzed using scanning electron microscopy (SEM).

Biofilm sampling and DNA extraction

When the reactor reached a steady state, biofilm was collected. N2 gas was sparged at the sampling point to preclude O2 exposure. We cut off one 5-cm-long section from a coupon fiber and then sealed the remaining fiber by tying the end into a knot. The DNA was extracted using a DNeasy Blood & Tissue Kit (Qiagen, USA) as previously described by Zhao et al. (2011).

Quantification of bacterial 16S rRNA and functional genes

Plasmids containing target fragments were used as positive controls and to produce calibration curves (Zhao et al. 2011). The primers and qPCR conditions were the same as previously described: pcrA for ClO4 − reductase (Nozawa-Inoue et al. 2008), nirS for NO2 − reductase (Throbaeck et al. 2004), mcrA for the formation of methane by most of methanogens and reverse methanogenesis (Steinberg and Regan 2008), pMMO for CH4 monooxygenase (Paszczynski et al. 2011), and 16S ribosomal RNA (rRNA) gene for bacteria (Maeda et al. 2003). We used the SYBR Premix Ex Taq Kit (Takara Bio Inc., Japan) and performed qPCR as previously described by Zhao et al. (2011). The copy numbers of each gene were calculated by comparison to standard curves. Negative controls included water instead of template DNA in the PCR reaction mix. We performed triplicate PCR reactions for all samples and negative controls. The functional genes (pcrA, nirS, pMMO, mcrA) and 16S rRNA gene abundances in different stages are shown in Fig. 3.

Results and discussion

Perchlorate and nitrate reduction

CH4 pressure was set at 10 psig throughout the experiment, which made CH4 available in all stages (Table S1). The influent and effluent concentrations of ClO4 − and NO3 − for the MBfR are plotted in Fig. 1. The reactor was operated for 23 days in batch mode before stage 1 to accumulate sufficient biomass to reduce up to 5 mg/L of perchlorate and 15 mg/L of nitrate.

Fig. 1
figure 1

The NO3 − and ClO4 − concentrations in the influent and effluent of the MBfR. Stage 2 and stage 5 were operated with 1 and 0.2 % salinity, respectively. Salinity was defined as the quantity of dissolved salt content of the water

Full size image

The NO3 − and ClO4 − surface loadings in stage 1 were 0.171 g N/m2-day and 0.091 g/m2-day, respectively. Without salinity, the nitrate was fully removed and 57.2 % of the ClO4 − was reduced after the reactor reached steady state. When 1 % salinity was applied in stage 2, the NO3 − removal percentage dropped to ∼50 %, and ClO4 − reduction was completely inhibited. Obviously, the perchlorate and nitrate reductions were both severely affected by the salinity in the CH4-based MBfR. Van Ginkel et al. (2008) reported that in a H2-based MBfR, NO3 − removal was not affected at NaCl concentrations less than 20 g/L (i.e., 2 %). Chung et al. (2007) demonstrated that both nitrate and perchlorate were reduced in the presence of NaCl ranging from 10 to 40 g/L (i.e., 1–4 %) in a H2-based MBfR. By comparison, microorganisms using methane as the electron donor are thus more sensitive to salinity than the microorganisms using H2.

To investigate if the biofilm could be recovered after salinity shock, we removed salinity in stage 3 so that the operating conditions in stage 3 were the same as in stage 1. To our surprise, the perchlorate and nitrate removal reached only 28.3 and 64.4 %, respectively. This removal is slightly higher than that observed in stage 2 but much lower than that in stage 1. The biofilm was possibly irreversibly damaged by the high salinity. Park et al. (2001) reported that the activity of denitrifying bacteria decreased after adding salt to the system. Gingras and Batista (2002) reported that the ClO4 − reduction rate declined to 10 % when 1–1.5 % salt concentrations were present in culture.

To investigate if NO3 − contributed to the inhibition of ClO4 − reduction, we removed NO3 − from the influent in stage 4. Without feeding NO3 −, the ClO4 − bioreduction improved to around 70 %. Therefore, it is possible that, in stage 2, the dual inhibitions of nitrate and salinity caused the absence of ClO4 − reduction. In a model, Tang et al. (2012a, b) explained that in a H2-MBfR simultaneously reducing NO3 − and ClO4 −, a high load of NO3 − (>0.6 g N/m2-day) strongly inhibits ClO4 − reduction.

To test the perchlorate or nitrate reduction under lower salinity, we reduced the salinity to 0.2 % NaCl in stages 5 and 6. ClO4 − was the only electron acceptor in stage 5, and NO3 − was the only electron acceptor in stage 6. The ClO4 − and NO3 − removals were 53.3 and 59.2 %, respectively. This performance is much worse than that in stage 1, which can be explained by the 0.2 % salinity inhibition and the irreversible inhibition in stage 2. The inhibition mechanisms will be further explored in “Microbial morphology” section.

Microbial morphology

To investigate the bacterial morphological change, we analyzed biomass samples from each stage at steady state using SEM. Figure 2 compares the microbial morphologies at stages 1 (Fig. 2a), 2 (Fig. 2b), 3 (Fig. 2c), and 6 (Fig. 2d). Stage 2 had the highest salinity (1 %). Correspondingly, it has the most wire-like structures and the cell surfaces had the most wrinkles (Fig. 2b). These bacterial filaments may be a way for the microorganisms to accommodate the change in osmotic pressure and protect the cells from salt shock. The damaging effect of salinity on bacteria leads to cell content emission and the formation of extracellular polymeric substances (EPS)/soluble microbial products (SMP) (Di Bella et al. 2013; Jang et al. 2013; Reid et al. 2006). Some studies argue that the accumulation of cell contents and intermediate products and bacterial autolysis are the responses to salt toxicity (Sharghi et al. 2014; Pendashteh et al. 2011; Sun et al. 2010). However, stage 6 had a medium salinity (0.2 %) and the biofilm also had some wire-like structures. Similarly, Jin et al. (2007) observed that under higher osmotic stress, the structure of a nitrifying bacterial community was disrupted and microbial diversity decreased. Therefore, salinity inhibition likely played a role in the decrease of nitrate removal in stage 6.

Fig. 2
figure 2

The SEM images of biomass samples from different stages. a Stage 1 at 10,000 magnification (10 mg/L NO3 −, 2 mg/L ClO4 −). b Stage 2 at 10,000 magnification (10 mg/L NO3 −, 2 mg/L ClO4 −, and 10 g/L NaCl). c Stage 3 at 10,000 magnification (10 mg/L NO3 −, 2 mg/L ClO4 −). d Stage 6 at 10,000 magnification (5 mg/L NO3 −, 2 g/L NaCl)

Full size image

Functional community structure

Figure 3 shows the functional and 16S rRNA gene copy numbers of the microbial community at all stages. The bacterial community was very dense in stage 1, illustrated by the highest copy numbers of the 16S rRNA gene (2.84 × 1012 copies/m2 fiber). The functional respiration genes such as pcrA and nirS and methane oxidation gene pMMO reached the highest in stage 1, indicating a very good correlation between methane oxidation and nitrate/perchlorate reduction. When 1 % salinity was added in stage 2, the abundances of all tested genes dropped by three orders of magnitude and reached their lowest points, which was consistent with the drop in nitrate and perchlorate bioreduction. All the functional genes increased in stages 3 and 4, concurrent with the removal of NaCl, and decreased in stages 5 and 6 when 0.2 % salinity was added to the influent. Throughout the whole experiment, the functional gene abundance correlated very well with the salinity.

Fig. 3
figure 3

The functional genes (pcrA, nirS, pMMO, mcrA) and 16S rRNA gene abundances in different stages. Abundance refers to gene copy numbers per square centimeter fiber

Full size image

Microbial community structure

We analyzed the diversity and structure of the MBfR microbial communities using pyrosequencing of the 16S rRNA gene from the MBfR biofilm samples. Figure 4 shows the relative abundance of dominant microbial phylotypes at the genus level, while Fig. S1 shows the class level. Methylomonas was the dominant (31.3 %) bacterium in stage 1. Its abundance sharply decreased to 5.9 % in stage 2 when 1 % salinity was introduced and remained at low levels in the latter stages. Methylomonas are known methanotrophs and sensitive to salinity. For example, Sherry et al. (2016) reported the disappearance of Methylomonas when NaCl was higher than 1 g/L. Cunliffel et al. (2008) analyzed samples from surface (lower salinity) and subsurface (higher salinity) water of the Blyth estuary and found that Methylomonas was present in the surface samples but not in subsurface samples. In our MBfR, Methylococcus emerged in stage 2 (1.1 %), and its abundance continued to increase in the rest of stages. Methylococcus grows on CH4 under low O2 concentrations (Ward et al. 2004). Some studies show that Methylococcus is highly abundant in coastal saline soil (Mishra et al. 2012; Yousuf et al. 2012). Another methanotroph, Methylocystis, was present at a low abundance (2.1 %) in stage 1 but increased to ∼10 % in stage 2 at 1 % salinity and further increased to 21.3 % in stage 3 when the salinity was removed. Its abundance decreased to ∼10 % in stage 4 when nitrate was absent and perchlorate was the only electron acceptor. Methylomonas and Methylocystis reduce NO3 − using O2 to activate CH4 (Dam et al. 2013; Kits et al. 2015).

Fig. 4
figure 4

Clustering based on relative abundance analysis of the microbial community structure at the genus level

Full size image

Methylophilus (Betaproteobacteria) abundance reached ∼9 % in stage 2 but remained at a low abundance in other stages. Giri et al. (2013) reported that Methylophilus sp. grow well at a NaCl concentration ≤200 mM. Methylophilus is phylogenetically close to known methanol assimilation denitrifiers. Figure 5 shows that Methylophilus is phylogenetically close to Dechloromonas agitata, Dechloromonas aromatica, and Azospira restricta, which are all known PRB. Therefore, Methylophilus probably played an important role in the reduction of NO3 − and ClO4 − under salinity stress. Denitratisoma abundance remained stable (∼10 %) through all stages. It is phylogenetically close to the known denitrifier Denitratisoma oestradiolicum, which reduces NO3 − to a mixture of N2O and N2 (Fahrbach et al. 2006). Unclassified Xanthomonadaceae, phylogenetically close to Thermomonas haemolytica, existed in all the stages and was significantly enriched in stages 2 and 3. It is possibly a denitrifier that can live under saline conditions. Mergaert et al. (2003) isolated two species of Thermomonas from a denitrifying reactor and reported that they could reduce NO3 − and NO2 −. Schmalenberger et al. (2013) reported the existence of Xanthomonadaceae in high salinity.

Fig. 5
figure 5

Phylogenetic analysis of the dominant bacterial genera in the MBfR biofilm and two from outgroups

Full size image

Figure S2 shows the PICRUSt analysis results of the functional genes. Genes encoding transporters (particularly ATP-binding cassette (ABC) transporters) and bacterial motility proteins were the most highly abundant. The abundance of these dominant genes decreased when the salinity increased in stage 2. Glycine betaine maintains osmotic pressure balance and bacterial morphology in salt environments, and ABC transporters are a member of the major betaine transporter family (Youssef et al. 2014; Silke et al. 2007). When NaCl appeared in stage 2, genes predicted to encode ABC transporters dropped to the lowest point. It is possible that the high salinity inhibited betaine absorption and caused changes in the microbial morphology and death, which explained the decreased NO3 − and ClO4 − reduction in stage 2.

Conclusions

In summary, salinity significantly affected the simultaneous ClO4 − and NO3 − reduction in a methane-based MBfR. Salinity at 1 % completely inhibited ClO4 − reduction and significantly reduced NO3 − reduction. Under saline conditions, the number of functional bacteria and total microorganisms decreased, wire-like cell structures became prevalent, and the cell surfaces developed wrinkles. In addition, the diversity of methanotrophs and denitrifying bacteria decreased significantly, and unclassified Xanthomonadaceae became more dominant when the salinity increased to 1 %. The saline environment changed cell osmotic pressure, which influenced the cell membranes and enzymes.