The diversity of membrane transporters encoded in bacterial arsenic-resistance operons

ars operon data collection and verification

The sequence of Bacillus sp. CDB3 ars cluster 1 (Bhat et al., 2011) was used for mining putative ars operon data using the tblastn program and searching the NCBI genome database (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi). Putative ars operons (updated to October 2013) were then verified by cross-checking in the ProOpDB (predicted) operon database (http://operons.ibt.unam.mx/OperonPredictor/) (Taboada et al., 2012). Data from some literature-reported functional ars operons, whose host full-genome sequences are not available, were also used in this study.

Classification of transporter proteins

All membrane transporter-like sequences were extracted from the operon database based on their homologies to other known transporters. The PSIPRED Protein Sequence Analysis Workbench (http://bioinf.cs.ucl.ac.uk/psipred/) and TMpred (Hofmann & Stoffel, 1993) were used to verify the transmembrane span of extracted sequences.

These were then classified into putative family groups based on BLAST results in the Transporter Classification (TC) system (http://www.tcdb.org/) and the NCBI protein database. The HHpred protein prediction tool (http://toolkit.tuebingen.mpg.de/hhpred) was applied when BLAST could not provide sufficient information for such family classification. WebLogo version 2.8.2 (Crooks et al., 2004) was used to further identify conserved regions within different protein families.

Phylogeny analysis

The MAFFT multiple sequence alignment program was used for sequence alignment (http://mafft.cbrc.jp/alignment/software/) and a rough phylogeny tree was generated using UPGMA (Unweighted Pair Group Method with Arithmetic Mean) algorithms (Sneath & Sokal, 1973). Neighbor-joining trees were applied in further analysis of separated transporter families.

Classification of membrane transporters encoded in ars operons

Of about 2500 bacterial genomes surveyed (full-genome data published before October 2013), 685 ars operons were identified and 717 putative membrane transporters were extracted. In addition, a number of experimentally characterized arsenic transporters were included in our study, including ArsB (Chen et al., 1986; Bröer et al., 1993; Bruhn et al., 1996; Ryan & Colleran, 2002), ArsY (ACR3) (Hu et al., 2011; Bhat et al., 2011), Aqps (Yang et al., 2005) and GlpF (Meng, Liu & Rosen, 2004). The large majority of these membrane proteins are significantly homologous to known arsenic transporters (Table 1) with the most abundant being ArsY (484; 66.5%) followed by ArsB (111; 15.2%), confirming previous findings. There were a large number of putative ars proteins annotated as AR (arsenic resistance) in the genome database due to inadequate identification. These were found to be either ArsY (mostly) or ArsB proteins and are reclassified in this study. Three other types of transporters were also observed to occur quite frequently, being the 49 homologues of the novel permease ArsP (Wang et al., 2009; Shen et al., 2014), 35 candidates of Major Facilitator protein Superfamily (MFS) and 18 putative Major Intrinsic Proteins (MIPs). In addition, 20 other putative membrane transporter proteins were also found, including 5 ATP-Binding Cassette transporters (Table 1).

An UPGMA tree was constructed using data from the five major protein families (Table 1). Distinct clusters of ArsBs, ArsYs and MIPs were found whilst ArsP and MFS proteins were divided into two sub-clusters (Fig. 1). Two putative ArsY and three MFS homologues were not clustered closely with their respective families and thus were not included in this tree.

Major facilitator protein superfamily

The MFS proteins represent a large family of secondary transporters carrying small solutes (Pao et al., 1998). To date at least 74 sub-families have been classified, each of which transports a specific substrate (Reddy et al., 2012), although no arsenic-specific sub-family has been nominated. A human MFS member, the liver sugar porter Glut2, was previously proposed as a bi-directional arsenic transporter (Drobná et al., 2010). Only two MFS proteins have been reported encoded in bacterial ars clusters (Chauhan et al., 2009; Drewniak et al., 2013); however, no experimental evidence yet demonstrates their involvement in arsenic resistance. A total of 35 putative MFS proteins were identified in this study, accounting for nearly 5% of the membrane transporters encoded by the surveyed ars operons. In the phylogeny tree they were divided into at least two sub-clusters (Fig. 1), and proteins from MFS sub-cluster 2 showed greater conservation compared with MFS sub-cluster 1 (data not shown). Certain homologies were also found between MFS sub-cluster 2 proteins and the N-terminal half (residues 50–250) of human liver protein, Glut2 (Fig. 2). Multiple conserved glycines detected within transmembrane helix regions were suggested as key structures of these proteins, for their motif homologies with previously reported GXXXGXXXG glycine zippers (Kim et al., 2005). Moreover, following the proposed glycine zipper structure, an additional highly conserved glycine was located in the coil region between two TMDs, suggesting a possible substrate-binding site for these MFS proteins. Assumingly, these ars operon-encoded MFS proteins function in their host bacteria against arsenic stress, but experimental proof is required. Whether the two groups of MFS transporters have different substrate specificities also warrants investigation.

Major intrinsic proteins

Composed of aquaporins (AQPs) and glycerol-permeable aquaglyceroporins (GLPs), MIPs are water channels, some of which can facilitate the efflux and intake of arsenite (Bienert, Schüssler & Jahn, 2008). Arsenite-transporting AQPs have been identified in numerous eukaryotes including plants, yeasts, fish, mice and humans (Bienert et al., 2008; Maciaszczyk-Dziubinska, Wawrzycka & Wysocki, 2012; Yang et al., 2012). Apart from arsenite, some of the AQP transporters in plants and animals are also permeable to methylated arsenicals (Maciaszczyk-Dziubinska, Wawrzycka & Wysocki, 2012). In bacteria, two MIPs have been proved capable of arsenite transport: Aqps in Sinorhizobium meliloti (Yang et al., 2005) and GlpF in E. coli (Meng, Liu & Rosen, 2004). Sequence alignments of the 18 putative MIPs recognized in this study revealed two highly conserved NPA motifs confirming the transportation role of these MIPs (Fig. 3). Previous studies attributed their arsenic transportation capability to molecular similarity between As(OH)₃ and glycerol (Yang et al., 2005; Mukhopadhyay, Bhattacharjee & Rosen, in press). It is assumed that all of the 18 ars operon-encoded MIPs serve the same arsenic transportation role as exemplified by the Sinorhizobium Aqps (Yang et al., 2005).

ArsP permeases

Forty-nine of the putative membrane proteins showed good homology to the ArsP protein, a novel permease encoded by an ars operon in Campylobacter jejuni, which has been recently proved as an organic arsenic (roxarsone and nitarsone) transporter (Shen et al., 2014). These ArsP homologues were found amongst 11 bacterial phyla, indicating a wide occurrence of this permease in bacteria that tolerate arsenic toxins, particularly the organic form. Furthermore, homologous proteins were also found widely in archaea with a few in eukaryotes (Yang et al., 2013, unpublished data) suggesting an ancient origin.

All of the surveyed ArsP homologues have a conserved C(S/T)C motif located on the transmembrane domain (Fig. 4), The well-conserved C(S/T)C motif in these ArsP proteins may assume an important role in the transporter function as suggested by Shen et al. (2014). Many studies have indicated the significant role of conserved cysteine residues in arsenic resistance-associated proteins including ArsA, ArsR, ArsC and ArsD (Rosen, 1999; Bhat et al., 2011). A highly conserved cysteine has also been identified in the pore helix of ArsY, although its biological role remains unknown (Fu et al., 2009).

Unsurprisingly, given the divergence within ArsPs noted in Fig. 1, at least two ArsP clusters were distinguished. Based on the ClustlW (Fig. 4) and WebLogo (Fig. S1) analyses, three distinctive sequence differences between the two clusters were observed. The cluster 1 proteins contain a conserved motif of Hv (L/I)XClv PAf FIAG towards their N-termini that is absent from cluster 2. In addition, directly upstream of the universal C(S/T)C motif, LAV is conserved in cluster 1 compared with t PF in cluster 2. It is also interesting to note the conservation of another cysteine residue in the N-terminal region among cluster 1 ArsPs. Whether this additional conserved cysteine and the associated motifs relate to different substrate specificity or affinity has also yet to be investigated.

The phylogeny tree of ArsP sequences does not match that of 16S rDNA sequences (Fig. 5), indicating that the divergence between two ArsP clusters was not due to division of bacterial species and thus may be function-related.

Minor groups of transporters

For the other 20 putative transmembrane proteins recognized (Table 1), only the five ABC transporters appear likely to have arsenic-transporter functions. ABC transporters facilitating arsenic extrusion have been found in several eukaryotes (Cole et al., 1994; Kala et al., 2000; Xie et al., 2004; Song et al., 2010; Wawrzycka et al., 2010; Wysocki et al., 2001; Mukhopadhyay et al., 2002; Tan et al., 2014). ABC transporters normally contain 12 transmembrane regions and are one of the biggest membrane superfamilies across all domains of life (Johnson, Lewinson & Rees, 2009). The fact that less than 1% of the membrane transporter proteins surveyed in this study that were ABC transporters suggests a less important role played by this type of transporter in bacterial arsenic resistance. These bacterial transporters may also carry arsenic non-specifically as do their counterparts in eukaryotes (Song et al., 2010; Xie et al., 2004; Kala et al., 2000; Wysocki et al., 2001; Wawrzycka et al., 2010).

The other membrane proteins in Table 1 may or may not relate directly to arsenic transportation. CHR family proteins are known to confer chromate resistance (Díaz-Magaña et al., 2009); CDF family proteins are known to transport a range of heavy metals (Co, Zn and Cd) but not arsenic (Paulsen & Saier, 1997) and the DMT family is involved in export of a wide range of drugs and metabolites (Jack, Yang & Saier, 2001). The ars operon-encoded protein homologues may have retained the same functions or diverged with arsenic-carrying ability. The presence of these transporter genes within ars operons might have resulted from survival pressures or random genomic translocation events.

Co-existence of the membrane transporters

While ArsY and ArsB were found to be dominant in the ars operon-encoded membrane transporters (Table 1), it was also observed that there is no case of co-existence of ArsB and ArsY in a single operon. There have been a few reported cases of their co-existence with other transporters (Chauhan et al., 2009; Wang et al., 2009). This study revealed that out of the 122 other transporters identified, 84 (68.9%) are located in operons which encode either ArsY (89.3%) or ArsB (10.7%) (Table 2). All of the 18 MIPs were found standalone; that is, none of the ars operons bearing MIPs encode another membrane transporter, which further supports the proposition that these MIPs play an arsenic transportation role. It was interesting to note that around 40% of surveyed ArsP (including members of both sub-clusters) are also encoded alone. All the other types of membrane transporters identified in this study were encoded together with either ArsB or ArsY in a particular ars operon.

The reason for the co-existence of two membrane transporters may be explained by their specificities to carry different arsenicals. It has been reported that two such transporters, ArsY and ArsP encoded by the four-gene C. jejuni ars operon (Wang et al., 2009; Shen et al., 2014), coordinate to transport inorganic and organic arsenic toxins, respectively. Thus, the co-existence of two or more membrane transporter genes in a single ars operon may have evolved as a defense mechanism due to increased levels and chemical complexity of arsenicals in a certain environment.