Escherichia coli strains were aerobically cultivated in Lysogeny Broth (LB) medium (Sambrook et al., 1989), supplemented with required antibiotics at 37°C with shaking at 180 rpm. E. coli strain DH5α (Promega, Madison, WI, USA) was used for plasmid construction and replication, and Rosetta (DE3) (Novagen, Madison, WI, USA) was used for protein expression. E. coli strain AW3110 (DE3) [ΔarsRBC; ArsR-repressor, ArsB-As(III) efflux pump, ArsC-As(V) reductase] was used for arsenic resistance and biotransformation assays (Carlin et al., 1995). Nostoc (also known as Anabaena), kindly provided by Professor Wen-Li Chen, Huazhong Agricultural University, was grown in BG11 medium without nitrate and cultured as previously described (Yan et al., 2015). MAs(III) was produced by reduction of MAs(V) using Na₂S₂O₃, Na₂S₂O₅, and H₂SO₄ (Yoshinaga and Rosen, 2014). All other used reagents were purchased from commercial sources, and were of analytical grade or better.

NsarsM (alr3095, accession number HQ891148) and NsarsI (alr1104, accession number BAB73061) have been identified in previous studies (Yin et al., 2011; Yan et al., 2015). The NsarsM or NsarsI was cloned into the expression vector pET28a (Novagen, Madison, WI, USA) or pET22b (Novagen, Madison, WI, USA) to generate the plasmid pET28a-NsarsM, pET28a-NsarsI, or pET22b-NsarsI; and the primers used for amplification were listed in Table 1. The two pET28a plasmids were transformed independently into E. coli AW3110 to construct strains expressing NsarsM or NsarsI. The plasmids pET28a-NsarsM and pET22b-NsarsI were co-transformed into E. coli AW3110. Although the two-plasmid system using the same origin of replication are commonly believed not to exist in one E. coli cell, several similar approaches with two incompatible plasmids were successfully used if under the selection pressure of two different antibiotics (Yang et al., 2001; Wang et al., 2007; Zhang et al., 2011). Thus, the strain co-expressing NsarsM and NsarsI was selected by growth on LB agar plate containing 100 μg mL^-1 ampicillin, 50 μg mL^-1 kanamycin, and 30 μg mL^-1 chloramphenicol (pET28a-NsarsM and pET22b-NsarsI plasmids confer kanamycin and ampicillin resistances, respectively).

Western blots were used to detect expression of NsarsM or/and NsarsI in the E. coli AW3110 strains. E. coli AW3110 cells in exponential phase were induced by 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG). The proteins from cell lysate were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE; gradient 15%), and transferred to polyvinylidene fluoride (PVDF) membranes (Pall Corporation, East Hills, NY, USA) using the Trans-Blot Turbo Transfer System (Bio-Rad, Hercules, CA, USA). Immunoblot analyses were carried out with anti-His tag (D3I1O, Cell Signaling, Beverly, MA, USA) by incubating at 4°C overnight. Membranes were washed, and incubated with appropriate peroxidase-conjugated secondary antibody. Specific bands were visualized by WesternBright ECL HRP substrate (Advansta, Inc., Menlo Park, CA, USA), and finally scanned on a Kodak image station 4000 mm Pro (Carestream Health, New Haven, CT, USA).

As(III) and MAs(III) resistance assays of E. coli AW3110 strains bearing pET28a vector, pET28a-NsarsM, pET28a-NsarsI, or pET28a-NsarsM+pET22b-NsarsI plasmids were performed as described previously (Ye et al., 2014; Yan et al., 2015). The cells were cultured in LB medium containing As(III) (0, 10, 30, 50, 70, 90, and 110 μM) at 37°C or in ST medium (10-fold concentrated ST 10^-1 medium) (Maki et al., 2006) containing MAs(III) (0, 1, 2, 3, 4, 6, and 8 μM) at 30°C. After incubating for 24 h, optical density at 600 nm (OD_{600 nm}) was measured using an ultraviolet-visible spectrophotometer (UV-6300 double beam spectrophotometer, Mapada, Shanghai, China). In addition, the four E. coli AW3110 strains were treated with 35 μM As(III) and 2 μM MAs(III) at the same time in ST medium, and OD_{600 nm} was monitored at 0, 3, 6, 12, 18, 24, 30 h.

The IPTG-induced E. coli AW3110 cells were cultured in LB medium containing 25 μM As(III) at 37°C for 24 h or ST 10^-1 medium containing 1 μM MAs(III) or 1 μM As(III) and 0.5 μM MAs(III) at 25°C for 1 h. Same amounts of arsenicals were also added to LB medium or ST 10^-1 medium without cells as non-inoculated controls. All samples were centrifuged at 13400 g for 2 min, and the supernatants were collected. Arsenic speciation was determined by high-performance liquid chromatography (HPLC, 1200, Agilent Technologies, Palo Alto, CA, USA)-inductively coupled plasma-mass spectrometry (ICP-MS, 7500a, Agilent Technologies, Palo Alto, CA,USA) to analyze arsenic biotransformation.

The plasmid pET28a-NsarsM was transformed into E. coli strain Rosetta (DE3) for purification of NsArsM. His-tagged NsArsM expression was induced by addition of 1 mM IPTG when OD_{600 nm} of E. coli Rosetta (DE3) cells reached 0.5–0.8. The induced cells were ruptured in a French-press at 10 MPa. NsArsM was purified by Ni-NTA agarose column (Qiagen, Hilden, Germany) according to the manufacturer’s instruction. The purified NsArsM was concentrated by 10-kDa cutoff Amicon Ultrafilter (Millipore, Bedford, MA, USA), and identified using SDS-PAGE.

In order to determine kinetic parameters, in vitro As(III) and MAs(III) reaction systems of NsArsM were established according to previous study with a few modifications (Qin et al., 2006). For the methylation reaction, 8 μM purified NsArsM was added in a buffer consisting of 50 mM MOPS buffer (pH 7.5) with 125 mM NaCl, 1 mM S-adenosylmethionine chloride (SAM), 8 mM reduced glutathione (GSH), and the indicated concentrations of As(III) or MAs(III). After the reaction was performed at 37°C for 1 h, the assay was immediately terminated by boiling, and added 3% H₂O₂ to oxidize arsenic. Enzyme activity was measured by determining arsenic species. The mole equivalents of methyl groups (-CH₃) transferred from SAM to arsenic were used to approximate the apparent rates of methylation. Since ArsM catalyzes three separate methylation reactions, the overall rates were lumped together as one mole of SAM methyl groups to methylate one mole of As(III) to MAs, two moles of SAM to methylate As(III) to DMAs, and three moles of SAM to methylate to TMAs (Walton et al., 2003; Wang et al., 2012). Non-linear regression analysis was performed with OriginPro 8.5.

Samples were filtered through 0.22 μm filters (Millipore, Bedford, MA, USA), and analyzed by HPLC-ICP-MS using previously established instrument parameters (Zhu et al., 2008). Arsenic species were determined with either a 10-μm PRP-X100 anion exchange column (250 mm × 4.1 mm ID, Hamilton, Reno, NV, USA) eluted isocratically with a mobile phase (pH 6.2) consisting of 10 mM ammonium di-hydrogen phosphate and 10 mM ammonium nitrate (Ye et al., 2014) or a Jupiter 5 μ C18 300A reverse-phase column (250 mm × 4.6 mm, Phenomenex, Torrance, CA, USA) using the mobile phase (pH 5.95) with 3 mM malonic acid, 5 mM tetrabutylammonium hydroxide, and 5% methanol (Yoshinaga and Rosen, 2014). The flow rate for HPLC was 1.0 mL min^-1, and ICP-MS was tuned for monitoring of m/z 75 (arsenic). Arsenic species in samples were identified by retention times which were compared with those of the standards. The arsenic was quantified by external calibration curves with peak areas integrated by using WinFASS.

Nostoc at the mid-exponential growth phase was cultured with or without arsenic for 6 h. Total RNA was extracted from Nostoc cells treated with As(III) (0, 1, 5, 10, 40, and 100 μM) or MAs(III) (0, 0.2, 1, 3, 6, and 12 μM) by using TRIzol reagent (Invitrogen Life Technologies, Gaithersburg, MD, USA) following the manufacturer’s recommendations. Contaminating genomic DNA was removed from total RNA using DNase I (Promega, Madison, WI, USA) at 37°C for 45 min. The total RNA was further purified with RNA clean kit (Tiangen, Beijing, China). About 40 ng purified RNA was used for RT-qPCR with GoTaq^® 1-Step RT-qPCR System (Promega, Madison, WI, USA) in a 20-μL volume. The primers for NsarsM and NsarsI were listed in Table 1. qPCR was performed on a LightCycler 480 (Roche Applied Science, Indianapolis, IN, USA). Each reaction was carried out in triplicate with the housekeeping gene rnpB as the internal standard (Vioque, 1992; Latifi et al., 2005). The PCR efficiencies of the targets (NsarsM and NsarsI) and reference (rnpB) were calculated from the slope of their standard curves (E = 10^[-1/slope]), respectively, and the relative transcript levels of NsarsM and NsarsI were calculated using the formula (E_target)^ΔCptarget^{(control-sample)} × (E_ref)^ΔCpref^{(sample-control)} (Pfaffl, 2001).

To investigate arsenic resistance and biotransformation when NsArsM and NsArsI co-existed in the same cell, E. coli AW3110 strain co-expressing NsarsM and NsarsI genes was constructed. The E. coli AW3110 strain bearing plasmid pET28a served as negative control. The expression levels of NsarsM and NsarsI in the single or co-expressed E. coli AW3110 cells were estimated by western blot using anti-His antibodies that recognize NsArsM or NsArsI (Figure 1). The results show that both NsarsM and NsarsI were expressed, separately and together. When co-expressed, more NsArsI was produced than NsArsM.

Resistance assays were conducted using E. coli AW3110 expressing NsarsM and/or NsarsI. When E. coli AW3110 cells were cultured in arsenic-free medium, the growth rates of all four E. coli AW3110 strains were nearly the same (Figure 2). In As(III) resistance assays, the growth of all the E. coli AW3110 cells was stimulated by 10 μM As(III), but inhibited when the As(III) concentration was over 30 μM (Figure 2A). The reason for this apparent growth stimulation by As(III) is not known, but it has been observed in another alga (Zhang et al., 2013). Furthermore, both the E. coli AW3110 cells expressing NsarsM and cells co-expressing NsarsM and NsarsI grew much better than those bearing pET28a or pET28a-NsarsI at 50–110 μM of As(III) (t-test, P < 0.05). For MAs(III) resistance assays, 1 μM MAs(III) inhibited growth of all of E. coli AW3110 cells (Figure 2B). E. coli AW3110 expressing NsarsI grew best at concentrations of MAs(III) between 2 and 8 μM, while E. coli AW3110 expressing NsarsM only grew better than the vector alone at 2 μM MAs(III). E. coli AW3110 co-expressing NsarsM and NsarsI exhibited less resistance to MAs(III) than that expressing NsarsI, and significantly higher resistance than that expressing NsarsM. In addition, the growth rates of E. coli AW3110 exposed to mixed 35 μM As(III) and 2 μM MAs(III) monitored over 30 h (Figure 2C), showed that co-expression of NsarsM and NsarsI conferred higher arsenic resistance than expression of NsarsM, but lower resistance than that of NsarsI.

To elucidate the biotransformation pathways when NsarsM or/and NsarsI were expressed in E. coli AW3110, arsenic species in As(III) and/or MAs(III)-containing media with or without cells were determined (Figure 3). When treated with 25 μM As(III), E. coli AW3110 bearing pET28a or pET28a-NsarsI did not change arsenic species in the media compared to the non-inoculated control. The detection of As(V) in the control may come from the trace contaminant of reagent and/or oxidation by air. While the cells expressing NsarsM or co-expressing NsarsM and NsarsI transformed As(III) into methylated arsenic (Figure 3A). It is worthwhile to note that the methylated species produced were different for the two strains. DMAs(V) and TMAsO were the predominant arsenic species, and MAs(V) was undetectable in the medium of cells expressing NsarsM. However, MAs(V) was the main arsenic species in the medium culturing E. coli AW3110 co-expressing NsarsM and NsarsI. When treated with 1 μM MAs(III) (Figure 3B), MAs(V) was detected in the controls, and it was probably coming from the spontaneous chemical oxidation of MAs(III) under aerobic conditions. DMAs(V) and As(III) were the primary arsenic species in the culture media of E. coli AW3110 cells expressing NsarsI and NsarsM, respectively. Both As(III) and DMAs(V) were detected in the medium with co-expression AW3110 strain. Similarly, under the exposure to a mixture of As(III) and MAs(III) (Figure 3C), DMAs(V) was only detected in the medium when NsarsM was expressed alone or co-expressed with NsarsI, and the percentage of As(III) increased in the latter.

Apparent kinetic constants were determined with purified NsArsM from the rate of methyl transfer from SAM. The relationship between the substrate As(III) or MAs(III) and the enzyme NsArsM fit conventional Michaelis–Menten kinetics (Figure 4). The affinity for As(III) was sevenfold greater than MAs(III), with a K_m of 5 ± 1 μM for As(III) and 37 ± 4 μM for MAs(III). The V_max for As(III) and MAs(III) were 60 ± 5 pmol CH₃ h^-1 mg^-1 and 167 ± 6 pmol CH₃ h^-1 mg^-1, respectively.

The transcript levels of NsarsM and NsarsI were analyzed by RT-qPCR after Nostoc exposed to MAs(III) or As(III) at indicated concentrations for 6 h. Nostoc under identical cultivation conditions without arsenic was used as a control. As shown in Figure 5, the transcript levels of NsarsM had no significant difference (P > 0.05) between the Nostoc cultures with and without arsenic. The transcript levels of NsarsI were not significantly increased (P > 0.05) by As(III) less than 40 μM or MAs(III) less than 6 μM, while they were significantly enhanced (P < 0.05) when 40 or 100 μM As(III), or 6 or 12 μM MAs(III) was added in the cultures.