Isolation and identification of arsenic-resistant strain M20
Strain M20 was isolated from aquaculture sewage under the pressure of 30 µM roxarsone, an organic compound of arsenic. MIC analysis revealed that strain M20 was also resistant to inorganic arsenic compounds, such as NaH2AsO3 and NaH2AsO4 (Table S3). On the basis of phylogenetic affiliation analysis, strain M20 was identified as Brevundimonas sp. (Fig. 1A). Brevundimonas spp. are opportunistic pathogens [29]. Brevundimonas sp. M20 harbors a 3.315-Mb chromosome (67.56% G + C mol%) (Fig. 1B). The genome annotation included 1,208 hypothetical proteins and 2,052 proteins with functional assignments. The proteins with functional assignments included 758 proteins with Enzyme Commission (EC) numbers, 648 with Gene Ontology (GO) assignments, and 567 proteins that were mapped to Kyoto Encyclopedia of Genes and Genomes pathways (Table S4). Subsystem analysis of PATRIC annotation indicted that strain M20 contains 97 stress response, defense, and virulence genes, and 53 membrane transport-related genes (Fig. 1C).
Identification of an ars cluster and a met operon
Genome annotation of Brevundimonas sp. M20 revealed a novel ars cluster, which contained putative arsenic resistance genes arsH, arsR, arsN, arsB, and arsC (Fig. 2). In this 3,011-bp cluster, arsH is an organoarsenical oxidase-encoding gene, which was considered to confer resistance to roxarsone and methylarsenite [19]. arsR, encoding a regulatory protein, is responsible for regulating ars cluster gene expression. arsB and arsC are arsenic detoxification genes; arsenate is reduced to arsenite by arsenate reductase (ArsC), followed by efflux of arsenite by the arsenite transporter ArsB [7]. ArsN detoxifies arsenate by acetylation of the α-amino group of arsinothricin [9]. BLAST analysis of the ars cluster revealed high similarity with the ars cluster in Caulobacterales spp. (JAFLCT010000083), Brevundimonas sp. Bin7 (JACVCC01000001), and B. nasdae strain Au-Bre29 (CP080034), although the ars cluster in the latter does not contain arsN (Fig. 2A).
No arsenic methylation gene was found in the ars cluster of Brevundimonas sp. M20. Downstream of the ars cluster, a 5649-bp met operon related to methionine biosynthesis was identified. The met operon was located 21.5 kb downstream of the ars cluster and contained four genes, arsRM (encoding a transcriptional regulator fused with a methyltransferase), metF (methylenetetrahydrofolate reductase), HMT-1 (5-methyltetrahydrofolate-homocysteine methyltransferase), and metH (methionine synthase) (Fig. 2B). The DNA sequence of the arsR fragment in arsRM showed very low similarity with the arsR gene in the ars cluster, although the similarity of the amino acid sequence was 21.78%. The met operon showed high similarity with that in B. nasdae Au-Bre29, which was isolated in Fujian Province, China, in 2022 [12]. However, the gap between the ars cluster and the met operon in strains M20 and Au-Bre29 was 21.5 kb and 2.1 Mb, respectively. The function of ArsRM has not yet been explained.
ArsR M increased the arsenite resistance of recombinant E. coli
No arsenic methylation gene was found in the ars cluster of Brevundimonas sp. M20, while ArsRM, encoded by arsRM of the met operon, was identified as a transcriptional regulator fused with a methyltransferase. To analyze the functions of ArsRM, the complete arsRM gene (960 bp) was cloned and inserted into pET-15b (Fig. 3A). The resulting plasmid was transferred into E. coli BL21 (DE3), and the resulting strain E. coli ARM3 was verified by PCR (Fig. 3B).
After induction by IPTG, the MIC of NaH2AsO3 toward E. coli ARM3 was increased to 1.5 mM compared with that in the uninduced control (Table S3). This enhancement indicated that ArsRM likely contributes to arsenite resistance in strain M20. On the basis of similarities with various arsenite methyltransferases in different species (Figure S1), the arsenite resistance mediated by ArsRM may be due to arsenite methylation.
ArsRM is a bifunctional fusion protein
To further investigate its function, the protein structure of ArsRM was homology modeled. ArsRM from Brevundimonas sp. M20 formed a dimer (Fig. 4A). The two monomers adopt different orientations (Fig. 4B). Each monomer contains a transcriptional regulator domain (TRD; α1–α5 and β1–β2), a methyltransferase domain (MTD; α6–α9 and β3–β7), and a 11-amino acid-residue loop linker (residues E111–A121) (Fig. 4C).
Unlike the typical ArsR [Protein Data Bank Code (PDB): 1R1T], which contains five α-helixes and two antiparallel β-sheets [30, 31], the TRD of ArsRM is composed of six short α-helixes and two antiparallel β-sheets. The DNA-binding domain (DBD) in the TRD (TRD has a diameter of approximately 36.6 Å) is composed of three helices (α1–α3) that form the core of the TRD and a C-terminal β-hairpin (β1–β2) (Fig. 4D). Compared with the tight domain in ArsR, these shortened α-helices in the DBD of ArsRM produce a flexible structure that may lead to a different regulatory mechanism.
The MTD of ArsRM has a mixed structure consisting of α-helices (α6–α9) and β-strands (β3–β7). This domain (residues 122–320) was compared with arsenite methyltransferases in bacteria (including Thermosediminibacter oceani, Streptomyces barringtoniae, Chloroflexi spp., Alteripontixanthobacter maritimus, and Hymenobacter roseosalivarius); Homo sapiens (residues 1–279); and the unicellular red alga Cyanidioschyzon sp. (complete protein). A glycine-rich sequence, “DLGTGSG,” is conserved as the hallmark of the SAM-binding motif [30] (Figure S1). This glycine-rich motif presented a circular, open shape and was located among α1, α2, β3, β4, and β5 (Fig. 4E). The As(III)-binding site (ABS) has three modular components in ArsR (PDB: 1R1T) [30], while arsenite methyltransferases in different species are variable (Figure S1). The ABS of ArsRM, adjacent to the SAM-binding motif, was composed of residue C134 (equivalent to C72 in PDB: 1R1T) and residues in the region Q224-L229 (Fig. 4F).
ArsRM binds its own promoter region
Molecular docking analysis showed that the TRD of the fusion protein ArsRM from Brevundimonas sp. M20 could bind to its own gene promoter region (ParsRM). Similar to the DNA binding in ArsR [31], the dimer formed wing regions and helix α5 (equivalent to α4 in ArsR, PDB: 1GXP) interact with DNA in Helix-turn-helix (HTH)–DNA complex structures (Fig. 5A). The DNA-binding sequence of ArsRM in the ParsRM region included a region that contained a palindromic sequence “CTTTATATAAAG” located upstream of the initiation codon of arsRM (Fig. 5A). This region is similar to the A/T-rich binding sites of ArsR. The interaction residues include R20 and E42, and, in helix α5, N96, A97, A98, D99, D100, L103, and E104 (Fig. 5B).
We tested biotin-labeled probes of the arsRM promoter region (215-bp long) in EMSAs: ParsRM-1 (containing the predicted target sequence), ParsRM-2, and ParsRM-3 (Fig. 5C). Sufficient quantities of recombinant full-length ArsRM–His6 protein were produced in E. coli BL21 (DE3). The soluble protein was eluted from a Ni-NTA column in buffer containing 150 mM imidazole (Fig. 5D). Probe ParsRM-1, containing the predicted binding sequence of ArsRM to ParsRM, was shifted in EMSAs by the addition of purified ArsRM (Fig. 5E). This result confirmed the binding of ArsRM to its own promoter region, which might regulate the transcription of the methionine biosynthesis gene cluster in Brevundimonas sp. M20.
ArsRM contributes to the methylation of As(III)
The process of arsenic methylation was originally proposed to be a detoxification mechanism, by which trivalent inorganic arsenic is biotransformed to a less toxic pentavalent methylated form. The MTD of ArsRM was predicted to interact with SAM and trivalent arsenic (Fig. 6A). Amino acid residues C134, H223, Q224, H227, F254, R263, H268, and P296 were identified to be the SAM-binding residues (Fig. 6A). Unlike in the arsenite methyltransferases shown in Figure S1 where Cys interacts with As(III), in ArsRM, Q224 and Y228 were the As(III) interacting residues, at a distance of 2.3–2.4 Å (Fig. 6B). These distances are similar to those for residues (C174 and C224) in C. merolae ArsM (2.2–2.3 Å) [30].
The methyltransferase activity of ArsRM was measured using a Methyltransferase Activity Assay Kit. Activity was detected in the presence of a mixture of SAM, arsenious acid, and ArsRM (Fig. 6C). This confirmed that arsenious acid can be methylated to monomethylarsonic acid by ArsRM with 0.87U/mg, although the transformation rate was low.