The bacterial strains and plasmids used in this study are presented in Table 1. The strains of A. caldus were cultured at 40°C shaken at 150 rpm in liquid Starkey-S⁰ medium (pH 2.5) or on solid Starkey-Na₂S₂O₃ medium (pH 4.8) (Jin et al., 1992). The liquid Starkey medium contained S⁰ (8 g/L, boiling sterilized) as the energy source. Kanamycin (200 μg/ml), streptomycin (200 μg/ml), or chloromycetin (68 μg/ml) was used in liquid Starkey-S⁰ when required, and kanamycin (80 μg/ml), streptomycin (80 μg/ml), or chloromycetin (27.2 μg/ml) was used in the solid Starkey-Na₂S₂O₃ medium for selection. Escherichia coli strains were grown at 37°C shaken at 170 rpm in liquid Luria–Bertani (LB) broth or on solid LB medium, and ampicillin (100 μg/ml), kanamycin (100 μg/ml), streptomycin (100 μg/ml), or chloromycetin (34 μg/ml) was added when required.

NCBI BLASTP (http://blast.ncbi.nlm.nih.gov/Blast.cgi) was used to search for GR homolog in the sequenced genome of A. caldus MTH-04 (CGMCC 1.15711). The protein molecular masses of the homologous proteins and the isoelectric points (pI) were predicted by using an ExPASy Compute pI/Mw tool (http://web.expasy.org/compute_pi/). ClustalX version 1.81 was used to perform the multiple sequence alignment. The phylogenetic tree was constructed by ClustalX version 1.81 and MEGA version 5, with a p-distance distribution, pairwise deletion, and bootstrap analysis of 10,000 repeats. Subcellular localization of protein was predicted using SignalP 4.1 (http://www.cbs.dtu.dk/services/SignalP/), Softberry ProtCompB tool (http://linux1.softberry.com/berry.phtml?topic=pcompb&group=programs&subgroup=proloc), and PSORTB v3.0 (http://www.psort.org/psortb/).

General molecular biological techniques, including restriction enzyme digestion, ligation, gel electrophoresis, and transformation of the plasmids, were conducted according to the standard protocols (Sambrook and Russell, 2001). The genomic DNA of A. caldus was isolated using the TIANamp Bacteria DNA Kit of TIANGEN. Plasmids were isolated using the TIANprep Mini Plasmid Kit of TIANGEN. DNA fragments were recovered from agarose gels using the OMEGA E.Z.N.A.^® Gel Extraction Kit of Omega Bio-Tek. DNA polymerase, restriction enzymes, and T4 DNA ligase were purchased from TaKaRa, and primers were generated by Invitrogen.

The strains and plasmids used to clone the potential GR gene are presented in Table 1, and the primers are presented in Table 2. To express and purify the recombinant protein of F0726_RS04210, the coding sequence was amplified using primers F-gr and R-gr. The fragment was digested by NdeI and XhoI and inserted into NdeI-XhoI-treated pET28a, to generate recombinant plasmid pET28a-gr. Successful insertion of the coding sequences of target genes was confirmed by sequencing; then, the recombinant plasmids were transformed into E. coli BL21(DE3) cells. Isopropyl-β-D-thiogalactopyranoside (IPTG) was added to the final concentration of 0.4 mM to induce the expression of recombinant protein at 25°C for at least 5 h. Recombinant proteins were analyzed using 10% SDS-PAGE and purified using HisTrap™ HP Crude Columns (GE Health) and AmiconUltra-15 Centrifugal Filter Units with Ultracel-3 membranes (Merck Millipore). Finally, the protein concentrations were determined using a Pierce^®BCA Protein Assay Kit.

GR can catalyze the reaction as follows: NADPH + H ⁺ + GSSG → NADP ⁺ + 2GSH, and the oxidation of NADPH can be measured by the decrease in absorbance at 340 nm (ε = 6.22 × 10³M⁻¹ · cm⁻¹). Purified GR was used in GR activity measurement using the method reported by Greer with minor modifications (Greer and Perham, 1986). In brief, the assay mixture (200 μl) contained 0.1 mM NADPH, 1.2 mM GSSG, and an appropriate amount of purified GR in 1 mM phosphate buffer (pH 7.0) at room temperature. Values of absorbance at 340 nm were recorded every 1 min and continued for at least 5 min. The reaction was started by the addition of NADPH. All reactions were performed with three independent biological replicates. Heat inactivated purified GR was used as the control for each reaction. Specific GR activity was measured as units of GR activities per mg of purified GR protein. Moreover, 1 unit of GR activity is defined as that can reduce 1 μmol of GSSG in 1 min at room temperature.

The effects of pH and temperature on GR activity were determined using the method described above. The optimum pH of GR was determined using purified GR in 1 mM citrate buffer solution (pH 3–5), 1 mM phosphate buffer solution (pH 6–8), or 1 mM Tris–HCl buffer solution (pH 9–10), respectively. The purified GR was incubated in the above buffers for 60 min at room temperature before measuring the activity at various pH conditions. The optimum temperature of GR was measured using purified GR within the range of 20–50°C at pH 7.0. The purified GR was incubated at the same temperature for 60 min before the activity measurement at each temperature.

The effect of metal ions (Cu²⁺, Zn²⁺, Cd²⁺, Ag⁺, Fe²⁺, Co²⁺, Mg²⁺, Sn²⁺, and Mn²⁺) on GR activity was determined by measuring its specific activity in the presence of 0.5 mM, 1 mM, or 1.5 mM of each reagent. The purified GR was incubated in the above reagent for 60 min before the GR activity measurement. The p-value was calculated using a t-test.

Markerless gr knockout mutant was generated as described previously with minor modifications (as shown in Figure 1) (Wang et al., 2016). The plasmids used in generating markerless gr knockout mutants are presented in Table 1, and the primers are presented in Table 2.

The upstream and downstream homologous arms of gr were amplified by PCR using primer pairs of grUF/grUR [819 bp upstream homologous arm (UHA)] and grDF/grDR [933 bp downstream homologous arm (DHA)]. The amplified sequences were inserted into the suicide vector pSDUDI after digestion with appropriate restriction enzymes. The sequences of the resulting plasmids pSDUDI-gr were confirmed by sequencing.

The suicide vector pSDUDI-gr was transformed into E. coli SM10, and the generated transformant was used as the donor. Acidithiobacillus caldus MTH-04 was used as the recipient to construct Δgr. Plasmid pSDUDI-gr was transferred from E. coli SM10 to A. caldus by conjugation, as described earlier (Wu et al., 2017). Colonies on selective Starkey-Na₂S₂O₃ plates containing kanamycin were selected and analyzed using colony PCR. Primer pairs oriTF/oriTR were used to detect the single crossover of Δgr, where a fragment of 465 bp was expected in the case of a single crossover, and no fragment was expected in wild type or Δgr. Colonies with the correct PCR fragments were inoculated into a liquid Starkey-S⁰ medium, and the genomic DNA of each selected colony was isolated for PCR analysis to confirm the single-recombination event. Primer pairs grIF/grIR were used to confirm the single crossover of Δgr, where both fragments of 3,088 bp and 1,054 bp were expected in the case of a single crossover.

Plasmid pSDU1-I-Sce I was then transferred to the single crossover cells of A. caldus to induce a second homologous recombination, thereby generating the knockout mutants or wild type individuals. The Δgr strain was identified using colony PCR based on screening colonies grown on selective Starkey-Na₂S₂O₃ plate containing chloromycetin using primer pairs grIF/grIR, as described above. A 1,054 bp fragment was expected in the Δgr strain, and a 3,088 bp fragment was expected for wild type cells. Primer pairs oriTF/oriTR, grIF/grIR, and grOF/grOR were used to confirm the deletion of gr. For the primer pairs grOF/grOR, a 2,963 bp fragment was expected in gr knockout mutant, and a 4,325 bp fragment was expected in the wild type cells. The amplified fragment from gr knockout mutant using primer pairs grOF/grOR was sequenced to confirm the mutants.

The pSDU1-I-Sce I plasmid in mutant cells was eliminated by spontaneous loss, as described earlier (Wu et al., 2017).

A Southern blot analysis was performed, as described earlier (Wen et al., 2014), using Sac I digests of the genomic DNA of the Δgr mutant and wild type of A. caldus. The downstream homologous arm of gr gene was labeled with digoxigenin and used as the probe.

The strains and plasmids used to construct the gr gene overexpression strain of A. caldus are presented in Table 1, and the primers are presented in Table 2. The coding sequence of gr along with its SD sequence was amplified using primer pairs O-F/O-R. The amplified fragment was first cloned into plasmid pUC19 for sequencing, after digestion with BamH I and EcoR I. The fragment pJRD215-tac was obtained from plasmid pJRD215-tac-cup after digestion with BamH I and EcoR I. Then, the confirmed gr-coding sequence was obtained from its sequenced recombinant plasmid by digestion with BamH I and EcoR I and ligated with pJRD215-tac to produce pJRD215-tac-gr. Plasmid pJRD215-tac-gr and the control plasmid pJRD215 were transformed into E. coli SM10. The plasmids were then transformed into A. caldus MTH-04 through conjugation as described earlier (Wu et al., 2017).

The A. caldus MTH-04 wild type, Δgr, control strain (wild type carrying plasmid pJRD215), and gr overexpression strain (wild type carrying plasmid pJRD215-tac-gr) were grown in the Starkey-S⁰ medium. All the strains were grown in the Starkey-S⁰ medium for 7 days, and then the cells were collected by centrifugation and adjusted to the same cell concentration (OD₆₀₀ = 20.0). An aliquot (150 μl) of the treated cells was inoculated into 150 ml of fresh Starkey-S⁰ medium, added with different amounts of CuSO₄ with the final concentrations of 0 mM, 5 mM, 10 mM, and 20 mM or different amounts of ZnSO₄ with the final concentrations of 0 mM, 40 mM, 80 mM, and 160 mM, respectively, cultivated at 40°C, and shaken at 150 rpm. After low-speed centrifugation at 400 × g for 5 min to remove the solid sulfur (Wang et al., 2016), the growth of A. caldus cultured in different concentrations of CuSO₄ or ZnSO₄ was monitored by measuring the optical density at 460 or 600 nm, respectively. All measurements were performed in triplicate, and error bars correspond to the standard deviations.

Wild type, Δgr, the control strain (wild type carrying plasmid pJRD215) and the gr overexpression strain (wild type carrying plasmid pJRD215-tac-gr) of A. caldus were grown for 7 days in Starkey-S⁰ medium without heavy metals and used to measure the transcriptional levels of the glutathione-related genes. The cells were first collected by centrifugation at 12,000 × g for 5 min, resuspended in RNAlater^® Solution (Ambion), and harvested from the RNAlater^® suspension by centrifugation after overnight storage at 4°C.

To study the responses of the glutathione-related genes to the heavy metal stress at different time points, above A. caldus strains grown for 7 days without heavy metals were supplemented with CuSO₄ (final concentration: 5 mM) or ZnSO₄ (final concentration: 40 mM), respectively, incubated at 40°C and shaken at 150 rpm. Cells were collected at two time intervals (1 h and 2 h) and treated as described above.

Overall, 100 μl of lysis buffer (1 mg/ml lysozyme, 10 mM Tris, and 1 mM EDTA, pH 8.0) was used to resuspend the cells and then incubated at 26°C for 6 min. TRIzol^® Reagent (Ambion) was used to extract total RNA following the manufacturer's instruction. Denaturing formaldehyde agarose gel electrophoresis was used to examine RNA quality, and a NanoDrop-1000 spectrophotometer (NanoDrop Technologies) was used to determine the concentration of RNA. A PrimeScript™ RT Reagent Kit with gDNA Eraser (Perfect Real Time; TAKARA) was used to remove the genomic DNA and synthesize the cDNA.

RT-qPCR was performed with the LightCycler^®480 system (Roche), following the manufacturer's instruction with SYBR^®Premix Ex Taq (TaKaRa). In this study, all RT-qPCR reactions were performed in triplicate with at least three independent biological replicates. All primers used for RT-qPCR are presented in Table 3. Gene alaS was used for normalization, and transcriptional results were calculated and shown in 2^−ΔΔCT (Livaka and Schmittgenb, 2001; Wu et al., 2017). The p-value was calculated using a t-test. The relative mRNA levels of these genes were measured using the gr knockout strain with the wild type as the control or using the gr overexpression strain with the wild type carrying vacant pJRD215 plasmid as the control. Fold change ≥ 2 and p-value ≤ 0.05 were considered to be upregulated, while fold change ≤ 0.5 and p-value ≤ 0.05 were considered to be downregulated.

We found a putative glutathione reductase encoding gene F0726_RS04210 in the genome of A. caldus MTH-04 by conducting an online BLASTN search. The ORF of F0726_RS04210 encoades a 453 amino acid glutathione disulfide reductase with a theoretical isoelectric point of 6.58 and molecular mass of 49.2 kDa. F0726_RS04210 shareas sequence identity with known GRs. These include Dickeya dadantii (53% identity), E. coli K12 (54% identity), E. coli VRa50 (53% identity), Francisella noatunensis subsp. orientalis str. Toba 04 (45% identity), Nostoc sp. PCC 7120 (40% idaentity), Phaeospirillum molischianum DSM 120 (41% identity), Caulobacter sp. AP07 (38% identity), and Sinorhizobaium meliloti (38% identity), respectively. An unrooted phylogenetic tree was constructed with F0726_RS04210 and the above mentioned GRs to better uanderstand the relationship between the reported GRs and the putative GR detected in A. caldus MTH-04. The evolutionary relationship of F0726_RS04210 is close to the glutathione reductase from E. coli K12, one of the best-understood GR, with a high identitay (54%; Figure 2). By alignment with known GRs, F0726_RS04210 has the highly conserved functional motifs characterized for GR, indicating the close relationship among GRs (Figure 3).

F0726_RS04210 was then expressed and purified from E. coli BL21(DE3) (Figure 4). We observed the GR activity of recombinant protein F0726_RS04210 in vitro. The GR activity of F0726_RS04210 was 415.8 U/mg, which was similar to that from E. coli (Nigel et al., 1987). Based on the activity we detected in the in vitro assays, we designated F0726_RS04210 as glutathione reductase. The optimum pH for the GR was 7.0 (Figure 5A), and the optimum temperature was 30°C (Figure 5B). The effect of metal ions on GR activity was also studied. Cu²⁺, Zn²⁺, Cd²⁺, Ag⁺, Fe²⁺, and Co²⁺ were found to strongly inhibit the activity of GR (P < 0.05), while GR activity was not obviously affected by Mg²⁺, Mn²⁺, and Sn²⁺ (Figure 5C). Similar phenomena were also reported in rainbow trout liver, E. coli, Phaeodactylum tricornutum, and Spinacia oleracea L. Leaves (Asnis, 1955; Michail and James, 1977; Diego et al., 2010; Ekinci and Sentürk, 2013).

In addition, the predicted subcellular localization of GR was analyzed using ProtCompB, SignalP 4.1, and PSORTB v3.0, and no signal peptide or transmembrane region was found, suggesting that it may be cytoplasmic protein, which is consistent with the neutral optimum pH of GR.

To better understand the function of GR in A. caldus, we used a markerless gene knockout system to generate gr knockout mutant Δgr (as shown in Figure 1). We identified the candidate Δgr mutant firstly by observing fragment size in a PCR analysis, and all observed PCR fragments were in accordance with the predicted sizes as described in the Materials and Methods section (Figure 6A). Then, Southern blot hybridization was performed using the downstream homologous arm as the probe. After digestion by using Sac I, the expected bands of 1,683 and 3,045 bp were obtained for Δgr mutant and the wild type, respectively, which indicates the successful knockout of the gr gene from A. caldus MTH-04 (Figure 6B). Finally, the obtained Δgr mutant was confirmed by sequencing the mutated region. The gr overexpression strain and a control strain (wild type carrying plasmid pJRD215) were also successfully constructed using the method described in the previous section.

To study the effect of gr gene on heavy metal tolerance of A. caldus, Δgr, the gr overexpression strain and the control strains of wild type and the wild type carrying plasmid pJRD215 of A. caldus were grown in the Starkey-S⁰ medium with different concentrations of CuSO₄ or ZnSO₄. Without heavy metals, knockout or overexpression of gr did not affect the growth curves on S⁰ compared with the control strains (Figure 7A), suggesting that GR did not play a key role in the growth on S⁰.

All strains showed obvious growth lags under low concentrations of copper ions, and the inhibition was increased with the increase in copper ion concentrations. The growth of Δgr mutant was completely inhibited under 20 mM copper ions, while the gr overexpression strain showed a growth advantage over the wild type (Figure 7). The high sensitivity to copper ions of the Δgr mutant and the enhanced tolerance to copper ions of the gr overexpression strain indicated the involvement of gr gene in copper tolerance in A. caldus MTH-04.

The zinc tolerance of A. caldus strains was investigated as well. The increased growth inhibitions with the concentration of zinc ions were also observed in all A. caldus strains. Moreover, the Δgr mutant grew lowest with a longer growth delay, while the gr overexpression strain grew highest with a shorter growth delay under the same concentration of zinc ions (Figure 8). The above results indicated the important role of the gr gene in zinc tolerance in A. caldus MTH-04.

To further understand how gr may influence the expression patterns of glutathione-related genes in A. caldus, the relative mRNA levels of these genes were measured using the gr knockout strain with the wild type as the control or using the gr overexpression strain with the wild type carrying vacant pJRD215 plasmid as the control using the RT-qPCR measurement method. Without heavy metal stress, gr knockout resulted in increased expression levels of glutathione synthetase and glutathione peroxidase, while gr overexpression resulted in increased expression levels of glutathione reductase, thioredoxin reductase, and heterodisulfide reductase subunit C and decreased expression levels of peroxiredoxin, glutathione synthetase, and glutathione S-transferase (Table 4). These results suggest that the GR activity was involved in the glutathione system in A. caldus during elemental sulfur oxidization.

When the strains were incubated with CuSO₄, as shown in Table 5, most of the investigated genes were apparently upregulated in Δgr compared with the wild type. On the other hand, the increased expression levels of the genes were somewhat different from others. For instance, the expression levels of peroxiredoxin, thioredoxin-disulfide reductase, and heterodisulfide reductase subunit C were increased after incubation for either 1 or 2 h, while the expression levels of thiol peroxidase and sulfur dioxygenase only increased after incubation for 1 h. When the gr overexpression strain was compared with the control strain (wild type carrying plasmid pJRD215), the expression levels of glutathione reductase and heterodisulfide reductase subunit C were increased after incubation for 1 and 2 h, respectively, while the expression levels of thioredoxin-disulfide reductase and sulfur dioxygenase were decreased after incubation for 1 h, and the expression levels of peroxiredoxin were decreased after incubation for 2 h, respectively.

Table 6 presents the responses of the selected genes to ZnSO₄. It shows that when compared with the wild type, the deletion of gr resulted in the upregulation of peroxiredoxin and heterodisulfide reductase subunit C and the downregulation of thioredoxin reductase, glutathione S-transferase, and sulfur dioxygenase after incubation for 1 h. The absence of gr also resulted in the upregulation of thiol peroxidase, peroxiredoxin, and thioredoxin reductase after incubation for 2 h. When compared with the control strain (wild type carrying plasmid pJRD215), the overexpression of gr resulted in the upregulation of glutathione reductase and heterodisulfide reductase subunit C after incubation for either 1 or 2 h and downregulation of peroxiredoxin after incubation for 2 h. The above results suggest that the GR activity was involved in heavy metal tolerance in A. caldus.