The bacterial strains and plasmids used in this study are listed in Supplementary Table S1, and the primers are listed in Supplementary Table S2. Alishewanella sp. WH16-1, E. coli, and their derivative strains were cultured at 37°C in Luria-Bertani (LB) medium unless otherwise noted. Stock solutions of rifampin (Rif, 50 mg mL⁻¹), kanamycin (Km, 50 mg mL⁻¹), chloramphenicol (Cm, 25 mg mL⁻¹), tetracycline (Tet, 5 mg mL⁻¹), K₂CrO₄ (1 M), Na₂SO₄ (1 M), and Na₂S (0.1 M) were added when required.

To identify the molecular mechanism of Cr(VI) detoxification of strain WH16-1, Tn5 transposon mutagenesis was used for screening Cr(VI) resistance genes. Transposon insertion mutants were generated with a suicide plasmid pRL27 (Larsen et al., 2002) transferred from the donor strain E. coli S17-1 to the recipient strain WH16-1 using the filter mating method (Smith and Guild, 1980). After conjugation, the Tn5 (Km^r) transposon was randomly inserted into the chromosome DNA of strain WH16-1, generating a library of insertion mutants. Selection was done on LB plates with Rif (50 μg/mL) and Km (50 μg/mL) to obtain strains in which transposition had occurred. The transconjugants were then plated on two LB plates with or without 20 mM K₂CrO₄, and the colonies that were unable to grow in the presence of K₂CrO₄ were reserved and subjected to further analyses. Cloning of genes neighboring the Tn5 transposon was performed according to the plasmid rescue method described before (Chen F. et al., 2015). The resulting neighboring sequences were searched against the whole genome of strain WH16-1 (Xia et al., 2016) using the NCBI BLAST server.

To identify the function of Cytbd, a complemented strain was constructed. The whole cytbd operon was cloned into the pCT-Zori plasmid using SacI and HindIII restriction enzyme sites. The generated plasmid was transferred into the mutant strain Δcytbd by conjugation from E. coli S17-1 to obtain a complemented strain, Δcytbd-C.

For analysis of cytbd conservation, homologous operon sequences from members of the Alteromonadaceae were selected from their genomes. They were Alteromonas macleodii HOT1A3^T (NZ_CP012202), Alteromonas marina AD001^T (NZ_JWLW01000010), Alteromonas sp. Mex14 (CP018023), Alteromonas sp. Nap 26 (LSMP01000036), Alteromonas australica H17^T (NZ_CP008849), Salinimonas chungwhensis DSM 16280^T (NZ_KB899391), Glaciecola pallidula DSM 14239^T (NZ_AUAV01000023), Alishewanella agri BL06^T (AKKU01000001), Alishewanella jeotgali KCTC 22429^T (AHTH01000001), Paraglaciecola arctica BSs20135^T (NZ_BAEO01000055), and Lacimicrobium alkaliphilum YelD216^T (NZ_CP013650). Phylogenetic analysis was carried out based on the cytbd operon (GbsR family regulator CydE, CydA, and CydB) amino acid sequences. The analysis was performed by MEGA 6.0 (Tamura et al., 2013) with a neighbor joining algorithm, and 1,000 bootstrap repetitions were computed to estimate the reliability of the tree. In addition, the operon arrangement in these strains was also analyzed.

For co-transcription analysis, strain WH16-1 was incubated to an OD₆₀₀ of approximately 0.3 in 100 mL LB broth, followed by incubation with 1 mM K₂CrO₄ for 3 h. Total RNA was extracted by Trizol reagent (Invitrogen), and DNA was removed by digestion with DNase I (Takara). Reverse transcription was conducted with a RevertAid First Strand cDNA Synthesis Kit (Thermo) with 300 ng total RNA for each sample. The resulting cDNA was used as a template to amplify the fragments between genes in the cytbd operon. Genomic DNA was used as a positive control. The total RNAs of strain WH16-1 and ddH₂O were used as negative controls. Primers are shown in Supplementary Table S2.

The putative promoter and promoter-cydE regions (Supplementary Figure S1A) were each PCR amplified from genomic DNA of strain WH16-1. Each DNA fragment was then cloned into plasmid pLSP-kt2lacZ using EcoRI–BamHI restriction enzyme sites. The resulting constructs were designated as pLSP-promoter-lacZ (Supplementary Figure S1B) and pLSP-promoter-cydE-lacZ (Supplementary Figure S1C). E. coli DH5α containing pLSP-promoter-lacZ or pLSP-promoter-cydE-lacZ was incubated in LB medium. Overnight cultures were diluted 100 times with fresh medium and incubated for approximately 4 h (OD₆₀₀ approximately 0.3). Next, Na₂SO₄ (0, 5, 25, and 50 mM) and K₂CrO₄ (0, 1, and 5 mM) were added to the cultures. The cultures were then distributed into tubes after a 6-h incubation. β-Galactosidase enzymatic assays were performed using the method described by Li et al. (2015).

Strains WH16-1, Δcytbd, and Δcytbd-C were each inoculated into 5 mL LB and incubated at 37°C with shaking at 150 rpm. When the OD₆₀₀ reached approximately 0.8–1.0, the strains were each inoculated into 100 mL LB with the presence of 500 mM Na₂SO₄, 200 μM Na₂S, 3 mM K₂CrO₄, or no addition. Na₂SO₄ powder was added to LB medium before sterilization, while K₂CrO₄ and Na₂S were added from stock solutions. In addition, LB plates with 0 or 3 mM K₂CrO₄ were used for Cr(VI) sensitivity analysis. For observing Cr(VI) reduction, strains were incubated in LB broth with 1 mM K₂CrO₄. To maintain consistent growth conditions, K₂CrO₄ was added when OD₆₀₀ reached 0.6. At designated times, culture samples were taken for measuring OD₆₀₀ and chromate amounts by spectrophotometry (DU800, Beckman) and atomic absorption spectrometry (AAS; 986A, Beijing Puxi General Instrument Co., Beijing, China), respectively.

To analyze the effects of Cytbd on oxidative stress, membrane proteins of strains WH16-1, Δcytbd, and Δcytbd-C were extracted to react with H₂O₂ and chromate. The membrane protein extraction was performed as described previously by Das et al. (2005), and the protein concentration was determined by the Lowry method (Lowry et al., 1951). Then, 10 mg/L of membrane proteins was reacted with 10 mM hydroquinone and 10 mM H₂O₂ or K₂CrO₄ in Tris–HCl (pH 8.5) buffer for 30 min under anoxic conditions in a N₂ chamber. The residual H₂O₂ was measured by the Amplex red/horseradish peroxidase assay (Mishin et al., 2010), and chromate concentrations were determined as mentioned above.

To gain more insight into the effects of Cytbd on cellular oxidative stress, cellular H₂O₂ contents of the WH16-1, Δcytbd, and Δcytbd-C strains were determined. The strains were incubated to an OD₆₀₀ of approximately 0.3 in 100 mL LB. K₂CrO₄ was then added to the cultures until the final concentrations reached 1 mM. Cells were centrifuged and washed twice with potassium phosphate buffer (50 mM, pH 7.7). The pellets were lysed via sonication on ice for 3 min and centrifuged for 5 min at 12,000 rpm to remove particulate materials. H₂O₂ amounts were measured as mentioned above.

Moreover, inhibition zone and H₂O₂ sensitivity tests were performed. For the inhibition zone test, cultures of each strain (OD₆₀₀ approximately 0.8–1.0) were added to LB agar medium, and 200 μL of 3% H₂O₂ was added to the Oxford cup (Wang et al., 2009). For the H₂O₂ sensitivity assay, 5 μL overnight cultures of WH16-1, Δcytbd, and Δcytbd-C (OD₆₀₀ 0.8–1.0) were dropwise added onto LB agar media containing various amounts of H₂O₂ (0, 0.05, 0.1, 0.5, and 1 mM).

Chemically defined medium (CDM) was selected to test the effects of sulfate on Cr(VI) reduction and resistance in strain WH16-1. The components of the CDM medium were the same as previously described (Weeger et al., 1999), except for replacing sodium lactate, magnesium sulfate and sodium sulfate with maltose, magnesium, and sodium chloride, respectively. This medium contained 0.12 mM SO₄⁻² and no other forms of sulfur. Strain WH16-1 was incubated with or without 100 μM Cr(VI) and additional 0, 5, or 10 mM sulfate in CDM medium. The remaining Cr(VI) in the medium was determined as mentioned above.

The DNA binding activity of CydE was tested in vivo with a bacterial one-hybrid system (Guo et al., 2009). The cydE coding sequence was amplified and cloned into the pTRG vector using BamHI–EcoRI restriction enzyme sites to obtain a plasmid pTRG-cydE. The promoter sequence of the cydE (Supplementary Figure S1A) was amplified and inserted directly into XcmI site of pBXcmT, yielding the pBX-promoter plasmid. The next steps were followed as previously described (Guo et al., 2009; Shi et al., 2017). The pTRG-cydE and pBX-promoter plasmids were co-transformed into E. coli XL1-Blue and grew on selective screening medium plates (Guo et al., 2009). In addition, E. coli XL1-Blue containing the pBX-MthspXp and pTRG-Rv3133c plasmids served as the positive controls, while E. coli XL1-Blue containing the empty vectors pBXcmT or pTRG was used as negative controls (Guo et al., 2009; Shi et al., 2017).

The CydE coding sequence was also amplified from DNA of strain WH16-1 using specific primers (Supplementary Table S2) that were designed to contain the restriction sites for BamHI and HindIII. The PCR product was digested with these enzymes and cloned into pET28a generating plasmid pET28a-cydE. After DNA sequencing confirmation, the plasmid was introduced into E. coli BL21 (DE3) cells. CydE was overexpressed by adding 0.1 mM IPTG to cells at an OD₆₀₀ of 0.3–0.4 that were further cultured for 4 h at 28°C. The cells were then harvested by centrifugation (8,000 rpm for 10 min at 4°C). After washing twice with 50 mM Tris–HCl (pH 8.0), the pellets were lysed via French Press at 120 MPa. Next, the soluble supernatant was mixed with 1 mL pre-equilibrated Ni-NTA His Bind Resin (7sea Biotech) and gently agitated at 4°C for 1 h. The resin was transferred into a 10-mL gravity-flow column and washed with 4 mL Tris–HCl with 200 mM imidazole to elute the miscellaneous proteins. The His-tagged CydE protein was eluted in 1 mL Tris–HCl with 500 mM imidazole, and the eluted fractions were analyzed with sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). The quality and quantity of the proteins were assessed with spectrophotometry (NanoDrop 2000, Thermo) and SDS–PAGE.

The DNA probe of cytbd promoter sequence (Supplementary Figure S1) was generated using the primer pair PromoterF/PrompterR (Supplementary Table S2). The PrompterR primer was labeled by fluorophore FAM when needed. In general, DNA binding assay was performed in a 20 μL reaction volume containing 2 μL 10× binding buffer (10 mM Tris, pH 7.5, 10 mM EDTA, 1 mM KCl, 1 mM DTT, 50% glycerol, and 0.1 mg/mL BSA), 100 ng FAM-labeled probe, and different concentrations (0, 0.1, 0.2, and 0.4 μg) of the purified CydE. For competition assay, 0.2, 1, and 2 μg unlabeled probes were added to reaction mixtures containing 0.4 μg CydE and the 100-ng labeled probe. All reaction mixtures were incubated at 37°C for 30 min before being loaded onto an 8% native polyacrylamide gel (Shi et al., 2017). After 1 h of electrophoresis at 120 V in 0.5× TGE buffer (6 mM Tris, 47.5 mM glycine, 0.25 mM EDTA, pH 8.0), gels were exposed to a phosphor imaging system (Fujifilm FLA-5100). For derepression analysis, 0.4 μg CydE was incubated with different concentrations of Na₂SO₄ (0, 1, 10, and 100 mM) and K₂CrO₄ (1, 10, and 100 mM) for 15 min, and then, 100 ng FAM-labeled probe and other components were added. Gel analysis was carried out as described above.

One hundred nanograms of FAM-labeled DNA probe was incubated with 0, 0.2, and 0.4 μg CydE, respectively (the reaction system was the same as in EMSA), then digested by 6 × 10⁻⁴ U/μL DNase I (New England Biolabs) for 10 min at room temperature. Next, the reaction was stopped by addition of 50 mM EDTA and incubation in a water bath at 65°C for 10 min. The digested DNA fragments were purified with a PCR clean-up Gel extraction kit (Macherey-Nagel). Samples were analyzed in a 3730 DNA Analyzer (Applied Biosystems, Foster City, CA, United States), and the electropherograms were aligned with GeneMapper v3.5 (Applied Biosystems, Foster City, CA, United States). For verification of the binding sites, a short-fragment EMSA test was used. The DNA sequences of the two binding sites identified by DNA footprinting were synthesized by Tsingke (Biological Technology Company, Beijing, China). The process of short-fragment EMSA was performed as described above. The final gel was stained by ethidium bromide.

Random mutants were generated by mobilization of the suicide plasmid pRL27 from the donor strain E. coli S17-1 into the recipient strain WH16-1. Approximately 8,000 Km- and Rif-resistant clones were randomly chosen and initially tested for their ability to grow on LB plates containing 20 mM K₂CrO₄. After 48 h of incubation, 40 Cr(VI)-sensitive mutants were obtained. Mutation sites were identified in 13 mutants with decreased Cr(VI) resistance. These mutant genes encoded CydB, ChrB, ferredoxin, iron transporter, DNA repair proteins (UvrC, UvrD, RecA, RecB, RecC, and YebC), and three other proteins (ComEC, ScpA, and a hypothetical protein). The cydB mutant (Δcytbd) was selected for this study since it may reveal potentially novel Cr(VI) detoxification mechanisms. The BLAST results showed that the Tn5 was inserted in the middle of the cydB (AAY72_09260) gene.

The cytbd operon sequence is conserved in the Alteromonadaceae strains based on phylogenetic analysis (Figure 1A). Moreover, the gene arrangement is also similar in these strains (Figure 1A). The genes coding for CydE (AAY72_09270) and CydA (AAY72_09265) were identified adjacent to cydB. The operon was located in contig 1 of the genome sequence. To gain more insight, RT-PCR was carried out. The forward and inverse primers used for RT-PCR were designed to overlap each two adjacent genes. The results of RT-PCR showed that DNA fragments between the three genes (cydE/cydA and cydA/cydB) were amplified with DNA and cDNA templates. It implied that cydE, cydA, and cydB were co-transcribed in an operon (Figure 1B). To verify the function of Cytbd, a complementation experiment was carried out. The complete cytbd operon including cydE, cydA, and cydB was introduced into the mutant strain Δcytbd and confirmed by PCR using primers cydBF/cydBR (Figure 1C) and DNA sequencing. This generated the complemented strain Δcytbd-C.

Escherichia coli DH5α-pLSP-promoter-lacZ and E. coli DH5α- pLSP-promoter-cydE-lacZ were constructed (Supplementary Figure S1) to test the expression of Cytbd protein. When cells were incubated in LB medium for 4 h without sulfate, the β-galactosidase activity was higher without CydE, indicating that CydE repressed the activity of the cydE promoter (Figure 2A). Furthermore, the β-galactosidase activity of E. coli DH5α-pLSP-promoter-cydE-lacZ was upregulated when sulfate was added (Figure 2A). However, Cytbd was constitutively expressed when chromate or sulfide were added (data not shown).

The wild-type, mutant, and complemented strains were used in sulfate- and sulfide-sensitivity tests. Cultures containing corresponding strains without sulfate and sulfide were used as controls (Figure 2B). The results showed that with the addition of 500 mM Na₂SO₄ or 200 μM Na₂S, the wild-type strain grew almost as well as the ones without the addition of Na₂SO₄ or Na₂S (Figures 2C vs. B). However, the growth of strain Δcytbd was partially inhibited with the addition of sulfate (Figure 2C) and completely inhibited with the addition of sulfide (Figure 2D), and the complemented strains were partially recovered to the wild-type levels. These results indicated that Cytbd was weakly associated with sulfate resistance, but it was very essential for sulfide resistance in strain WH16-1. Sulfate appeared not to be very toxic to strain WH16-1 since the addition of 500 mM sulfate had almost no effect on its growth.

A chromate sensitivity test was also performed. The chromate sensitivity test was performed on LB plates and in LB medium. The results showed that the chromate resistance of the mutant strain was noticeably weaker than in the wild-type and the complemented strain (Figures 3A,B). The chromate minimal inhibition concentration (MIC) of the mutant strain was 3 mM, while for the wild type, it was 45 mM (Xia et al., 2016). In addition, the Cr(VI) reduction ability of the mutant strain was also somewhat weaker than the wild-type and complemented strain (Figure 3D) under similar growth conditions (Figure 3C).

To achieve a better understanding of how Cytbd contributes to chromate resistance, a series of experiments were carried out. First, the membrane protein of the wild-type, mutant, and complemented strain was extracted and reacted with H₂O₂ and chromate. The H₂O₂ decomposition activity of the Δcytbd membrane protein was noticeably lower than the wild type and Δcytbd-C (Supplementary Figure S2A), while chromate reduction showed no significant difference (data not shown). Furthermore, the cytoplasmic H₂O₂ contents were measured to reflect the cellular oxidative stress. Without K₂CrO₄, there were no significant differences in H₂O₂ content among strains WH16-1, Δcytbd, and Δcytbd-C (Supplementary Figure S2B). When exposed to 1 mM K₂CrO₄, the H₂O₂ contents of all three strains were increased (Supplementary Figure S2B). However, the H₂O₂ content in the mutant strain was higher than that in the wild-type and the complemented strain with K₂CrO₄ (Supplementary Figure S2B). A similar result was obtained based on the inhibition zone test for H₂O₂ sensitivity. The diameter of the inhibition zone of the mutant strain was visibly larger than for the other two strains (Supplementary Figure S2C). This finding means that the mutant strain was more sensitive to H₂O₂. In addition, the MIC to H₂O₂ of the mutant strain was 0.5 mM, which was lower than that of the wild-type and the complemented strain (Supplementary Figure S2D).

Interestingly, the mutant strain lost partial capability for H₂O₂ decomposition. Hence, it was more sensitive to H₂O₂ compared to the wild-type and complemented strain. As a result, we inferred that the Cytbd catalyzes the reduction of cytoplasmic oxidative stress to enhance Cr(VI) resistance in strain WH16-1, but it does not directly catalyze chromate reduction.

The effects of sulfate on chromate resistance and reduction were examined since sulfur metabolism is relevant to chromate metabolism in many bacteria (Viti et al., 2014), and the above results showed that sulfate and chromate resistance are both associated with Cytbd. The growth of strain WH16-1 showed no significant difference with or without additional Na₂SO₄ in the absence of added K₂CrO₄ (data not shown) but was affected after adding K₂CrO₄ (Figure 4A). However, growth was much better with the addition of Na₂SO₄ (Figure 4A). Additionally, the Cr(VI) reduction ability of strain WH16-1 increased with increasing concentrations of Na₂SO₄ (Figure 4B).

CydE is homologous to the GbsR-type regulator based on the results of BLASTP in NCBI. It shares 18.8% and 19.4% similarities with GbsR and OpcR, respectively. Next, we aligned CydE with the two reported GbsR regulators (Figure 5A). The results showed that CydE harbored the same conserved amino acids as the GbsR-type regulators. These conserved amino acid residues may be involved in DNA binding. GbsR-type regulators usually act as repressors of gene expression (Nau-Wagner et al., 2012; Lee et al., 2013).

To examine the regulation function of CydE, we first used a bacterial one-hybrid system to test the protein–DNA interaction based on the transcriptional activation of HIS3 (imidazoleglycerol-phosphate dehydratase gene involved in histidine biosynthesis) and aadA (streptomycin resistance gene) (Guo et al., 2009). The promoter of cydE (Supplementary Figure S1A) was cloned into upstream of HIS3–aadA in the reporter vector pBXcmT, while the CydE coding region (Supplementary Figure S1A) was introduced into the pTRG vector. Both constructed vectors were then transferred into a histidine synthesis defective and streptomycin (Str) sensitive strain. The generated strain and positive control strain grew well on the screening plate (Guo et al., 2009) containing 3-amino-1,2,4-triazole (3-AT) and Str, while the negative control strain did not grow. The results demonstrated that CydE could interact with the promoter of the cytbd operon in vivo (Figure 5B).

Next, the purified His-tag CydE (Supplementary Figure S3) and cydE promoter DNA (Supplementary Figure S1A) were used to test the interaction in vitro using EMSA. With increasing amounts of CydE, the free DNA substrates gradually disappeared, while the intensity of the shifted DNA band increased (Figure 5C). Moreover, the unlabeled DNA substrate could competitively inhibit the binding of CydE to the labeled DNA substrate (Figure 5C).

To identify the binding site of CydE, DNA footprinting was carried out. With increasing amounts of CydE, decreases in two sites of the peaks were observed (Figure 6A), indicating that there are two binding sites. This is consistent with the EMSA results. The sequences of the binding sites were TATTTCAGAAATTTCTGAAAGTTCA and GGGATGCGCATATGCAAAT (Figure 6B). The two binding site sequences were synthesized and then incubated with CydE. The EMSA result showed that both binding sites could interact with CydE (Figure 6C).

To investigate the repression ability of CydE, an EMSA derepression experiment was performed. The results showed that the free DNA increased when more sulfate was added (Figure 6D). However, the phenomenon was not observed when chromate was added (Figure 6D). These results are coincided with those of the lacZ reporter assay (Figure 2A). All these results demonstrated that CydE can repress the expression of the cytbd operon and sulfate addition results in derepression.