Aquaglyceroporin AqpS from Sinorhizobium meliloti conducts both trivalent and pentavalent methylarsenicals

Abstract

Arsenic is a toxic metalloid that enters cells adventitiously via uptake systems for phosphate transporters, aquaglyceroporins (AQPs) or sugar permeases. However, transport of highly toxic methylarsenite (MAs(III)) and relatively nontoxic methylarsenate (MAs(V)) by bacterial AQPs has not been characterized. MAs(V) has a history of use as an herbicide. Here we used whole genome sequence analysis of AQPs in arsenic resistance (ars) operons. The aqp genes are frequently located next to MAs(III) resistance genes such as arsH, which suggests that they could be involved in MAs(III) uptake. Bacterial AQPs encoded by ars operons can be classified into two subgroups. One subgroup includes AqpS from the plant symbiont Sinorhizobium meliloti 1021. Our data suggests that AqpS has a substrate selectivity filter different from that of other bacterial AQPs Both Escherichia coli GlpF and AqpS conduct MAs(III) efficiently, but GlpF conducts the MAs(V) anion poorly, so E. coli takes up MAs(V) inefficiently. In contrast, AqpS conducts MAs(V) under physiological conditions. A homology model of AqpS indicates that it has a substrate channel with a selectivity filter containing the nonpolar residue Val177 instead of the charged arginine residue found in other AQPs. While the selectivity filter in most AQPs prevents movement of anions, Val177 is predicted to allow movement of the MAs(V) anion through the channel. We propose that AqpS is a component of an MAs(III) resistance pathway in which MAs(III) enters cells of S. meliloti via AqpS, is oxidized by ArsH to MAs(V), which exits the cells via AqpS.

Graphical Abstract

1. Introduction

Arsenic exists in either of two oxidation states, pentavalent arsenate (As(V)) and trivalent arsenite (As(III)). In neutral solution arsenite is the uncharged molecule As(OH)_3, which is sufficiently chemically similar to glycerol to be taken into cells through aquaglyceroporin (AQP) channels (Mukhopadhyay et al., 2010). For example, As(III) is taken into cells by the E. coli glycerol channel GlpF (Meng et al., 2004) by the Saccharomyces cerevisiae aquaglyceroporin Fps1p (Wysocki et al., 2001) and the mammalian aquaglyceroporins AQP9 and AQP7 (Liu et al., 2004) In humans, 13 mammalian AQP isoforms have been identified (Aikman et al., 2018). Based on structural and substrate selectivity, these bidirectional channel proteins are divided into two main groups: orthodox aquaporins involved only in water permeability and aquaglyceroporins, facilitating transport of small uncharged solutes including water, glycerol, urea and As(OH)₃. In plants a number of genes encoding nodulin 26-like intrinsic proteins (NIPs) (Bienert et al., 2008; Wallace et al., 2006) are classified as a subfamily of the plant aquaporin family (Danielson and Johanson, 2010). The Oryza sativa (rice) OsNIP2;1 gene, also named OsLsi1 because of its primary function as a silicon transporter, is a pathway for entry of As(III) into roots (Ma et al., 2008). Other NIP channel proteins such as rice OsNIP1;1, OsNIP2;2 (also named OsLsi6) and OsNIP3;1, as well as Arabidopsis thaliana AtNIP5;1 and AtNIP6;1, also mediate As(III) uptake (Zhao et al., 2009).

As(III) is methylated to methylarsenite (MAs(III)) and dimethylarsenite (DMAs(III)) and, to a lesser extent, to trimethylarsine (TMAs(III)) by the enzyme As(III) S-adenosylmethionine methyltransferase (AS3MT in mammals and ArsM in microbes) (Zhu et al., 2014). MAs(III) and DMAs(III) are more toxic than As(III), so methylation increases arsenic toxicity (Styblo et al., 2000). In mammalian cells trivalent methylarsenicals are taken up more readily than the pentavalent species and inorganic arsenicals (Dopp et al., 2005). The order of uptake in mammalian cells for trivalent methylarsenicals is DMAs(III)>MAs(III)>As(III), and for pentavalent arsenicals As(V)>MAs(V)>DMAs(V)>TMAsO (Dopp et al., 2004). The higher uptake of the trivalent methylated arsenicals may contribute to its greater cyto/genotoxicity in humans. Those studies did not define the pathways of uptake, but a few aquaglyceroporins can transport MAs(III) (McDermott et al., 2010). Rat AQP9 conducts MAs(III) at a higher rate than inorganic As(III) (Liu et al., 2006). Human AQP9 and rat AQP7 also transport MAs(III) at a rate 5-fold higher than As(III) (McDermott et al., 2010). Surprisingly, human AQP7 conducts As(III) but does not appear to transport MAs(III) (Liu et a;/ 2006). The much larger aquaglyceroporin Fps1 from S. cerevisiae, which functions physiologically as a glycerol transporter, also mediates uptake of arsenite and MAs(III) (Wysocki et al., 2001). These results illustrate the diversity of substrate selectivity in AQPs.

Aquaglyceroporins exclude most charged compounds. AQP6 appears to be an exception as a novel chloride channel, but the permeability pathway for the anion appears to be in a different location than other AQPs (Rambow et al., 2014). Pentavalent MAs(V), with pKa values of 4.1 and 9.02 (Ahrens and Edwards, 1994), is predominantly an anion at physiological pH and would be expected to be excluded by most AQPs. Yet both hAQP9 and rice OsLsi1 have been reported to transport MAs(V) (Li et al., 2009; McDermott et al., 2010). The rate of transport was higher at nonphysiological acidic pH than at physiological pH, where MAs(V) is taken up less efficiently than either inorganic As(V) or As(III) (Zhao et al., 2009). At lower pH, the molecule is predominately in the protonated neutral form, which could explain its uptake by OsLis1 or hAQP9. Some AQPs also conduct carboxylic acids, but this property also seems to be related to the protonation of those acids (Rambow et al., 2014).

Aquaporins contain six transmembrane α-helices, one of which is composed of two half-helices, each with a highly conserved Asn-Pro-Ala (NPA) motif that form a central pore (Agre et al., 2002). The asparagine side chains of the two NPA sequences point into the pore and form a constriction that generates an electrostatic barrier for proton exclusion (Fu and Lu, 2007; Savage et al., 2003; Savage et al., 2010). A second constriction site formed by four residues is located close to the extracellular end of the pore. This region is termed the aromatic/arginine (ar/R) constriction based on the presence of a conserved arginine residue in loop E (LE) and the prevalence of aromatic residues in helix 2 (H2) (de Groot and Grubmuller, 2001). The ar/R constriction forms the narrowest part of the pore and is generally assumed to serve as a substrate selectivity filter. In orthodox aquaporins the pore aperture at the ar/R selectivity filter is 3 Å, preventing movement of molecules larger than water (2.8 Å). In aquaglyceroporins the pore aperture can be as large as 3.4 Å, which permits movement of larger molecules such as glycerol.(Aikman et al., 2018) The positive charge of the conserved arginine residue in the ar/R selectivity filter prevents entry of charged molecules such as MAs(V) into the channel (de Groot and Grubmuller, 2001).

In this study we focused on transport of MAs(V) by bacterial AQPs. We observe that genes for aquaglyceroporins are frequently found in ars operons (Fig. 1). These aqp genes are usually associated with MAs(III) resistance genes such as arsH and arsI, which leads us to predict that their gene products facilitate uptake of MAs(III) into cells, followed by detoxification by the ArsH MAs(IIII) oxidase or the ArsI C-As lyase. The AQP channels encoded by genes in ars operons fall into two groups. One group include those that have an arginine residue in the NPA motif, as is found in most AQP channels. We designate those ars operon-encoded aquaglyceroporins with an NPAR motif as Aqp_NPAR to differentiate them from members of the second group, including the S. meliloti 1021 AqpS (accession number: WP_010969025), which have an NPAV (Asp174-Pro175-Ala176-Val177) motif in which Val177 replaces the arginine residue in the ar/R selectivity filter. In addition, the sequence of AqpS shows low protein sequence similarly to other AQPs such as hAQP9 or OsLsi1 that places it in a phylogenetically divergent group from other aquaglyceroporins (Fig. 2). In this study, both GlpF and AqpS are shown to facilitate uptake of MAs(III). In contrast, AqpS efficiently conducts MAs(V) at physiological pH, while E. coli GlpF does not. Thus MAs(V) is taken up at a higher rate in S. meliloti compared with E. coli. Reciprocally, AqpS exhibits lower uptake of As(III) or MAs(III) compared with GlpF. AqpS residue Val177 is required for MAs(V) uptake, indicating that this divergent aromatic/arginine (ar/R) selectivity filter contributes to the unusual substrate selectivity of AqpS. We propose that AqpS plays two roles in detoxification of MAs(III), first by facilitating uptake of MAs(III), which is then oxidized to MAs(V) by ArsH, and second by conducting the MAs(V) product out of the cells.

Fig. 1. The aquaporin genes disturbed in ars operons.

Shown are representative ars operons (accession numbers in parentheses) containing either aqpS or aqp_NPAR genes (white fill). Sinorhizobium meliloti 1021 (NC_003047.1), Rhizobium giardinii (NZ_ARBG00000000.1), Bradyrhizobium sp. WSM 1791 (NZ_JAAVLX000000000.1), Xanthobacteraceae bacterium (WOUY00000000.1), Ensifer adhaerens (NZ_CP015880.1), Scytonema hofmannii (NZ_ANNX00000000.2), Fischerella sp. PCC 9605 (NZ_ALVT00000000.1), Cylindrospermum stagnale (NC_019757.1), Oscillatoria nigro-viridis (NC_019729.1),Tychonema bourrellyi (NZ_NXIB00000000.2), Fischerella muscicola (NZ_AJLK00000000.1), Nostoc calcicole (NZ_MRBZ00000000.1), Chamaesiphon polymorphus (NZ_PVWO00000000.1), Calothrix rhizosoleniae (NZ_FYBH00000000.1), Scytonema tolypothrichoides (NZ_JXCA00000000.2), Pararhizobium polonicum (NZ_LGLV00000000.1).

Fig. 2. AqpS exhibits a divergent evolution.

A neighbor-joining phylogenetic tree showing the evolutionary relationships of bacteria AqpS protein with aquaporin proteins from members of other kingdoms. AqpS from S. meliloti in this study, OsLsi1 and hAQP9 which showed MAs(III) transport activity are indicated by black triangles.

2. Materials and Methods

2.1. Chemicals

Unless otherwise indicated, chemicals were purchased from Sigma-Aldrich. MAs(V) was chemically reduced to MAs(III) as described (Reay and Asher, 1977). The reduced products were not thiolated, as determined by simultaneous As and S analysis by high pressure liquid chromatography (HPLC) coupled with inductively coupled mass spectroscopy (ICP-MS) (ELAN DRC-e; Perkin-Elmer, Waltham, MA) (Qin et al., 2006).

2.2. Strains, medium and growth conditions

E. coli Stellar™ (Clontech Laboratories, Mountain View, CA) (F⁻, endA1, supE44, thi-1, recA1, relA1, gyrA96 phoA, Φ80d lacZΔ M15, Δ(lacZYA-argF)U169, Δ(mrrhsdRMS-mcrBC), ΔmcrA, λ–) was used for plasmid DNA construction and replication. E. coli AW3110 (Δars::cam F−IN(rrn-rrnE) (Qin et al., 2006), which is hypersensitive to As(III), was used for complementation studies. E. coli strain OSBR1 (Δars::cam F−IN(rrn-rrnE, ΔGlpF), which was derived from AW3110, has a TnphoA insertion in glpF and exhibits low levels of As(III) uptake and resistance to As(III) (Yang et al., 2005a) was used for used for complementation studies. For most experiments, cultures of E. coli were grown aerobically at either 30 °C or 37 °C in either lysogeny broth (LB) medium (Sambrook et al., 1989) or low phosphate medium(Oden et al., 1994) (with 0.2% glucose as a carbon source, as noted), supplemented with 125 μg mL⁻¹ ampicillin, 50 μg mL⁻¹ kanamycin or 34 μg mL⁻¹ chloramphenicol, as required (Chong, 2001). S. meliloti strain Rm1021 (SU47; str-21) and S. meliloti strain SmK956 (in-frame SMc02648 (aqpS) deletion, Sm^r)(Yang et al., 2005a), which shows increased resistance to As(III), were used for MAs(III) resistance and uptake assays. S. meliloti strains were grown aerobically at 30 °C in either LB medium or low phosphate medium.

2.3. Phylogenetic analysis

Multiple alignment of aquaglyceroporin homolog sequences was performed using Clustal Omega.(Sievers and Higgins, 2018) Aquaglyceroporins in ars operons were selected for phylogenetic analysis. Acquisition of sequences was performed by searching a list of reference organisms or from the National Center for Biotechnology (NCBI) protein database by BLASTP search (Johnson et al., 2008). Phylogenetic analysis was performed to show the evolutionary relationship among the aquaglyceroporins of various organisms. The phylogenetic tree was constructed using the Neighbor-Joining method using MEGA 6.0.1 (Tamura et al., 2013). The statistical significance of the branch pattern was estimated by conducting a 1000 bootstrap (Saitou and Nei, 1987).

2.4. Plasmid construction

An E. coli glpF gene corresponding to the sequence of the cDNA clone was chemically synthesized with 5’-NcoI and 3’-SalI sites at each end and cloned into vector plasmid pBAD/myc-HisA (Life Technologies, Grand Island, NY), creating plasmid pBAD-GlpF. The expression of GlpF is controlled by the arabinose promoter (GenScript, NJ, USA). The S. meliloti aqpS gene was expressed in plasmid pBAD-AqpS, which we constructed previously (Yang et al., 2005a). The fidelity of the two synthetic aqp genes in the recombinant plasmids was confirmed by DNA sequencing using Sequetech DNA Sequencing Services (Mountain View, CA).

2.5. Metalloid resistance assays

For metalloid resistance assays in liquid medium, E. coli or S. meliloti cells were grown in LB medium at 30 °C to an A_600nm of 2.0. Overnight cultures were diluted 100-fold in low phosphate medium (pH 7.0). Cells were cultured for 12 h with shaking at 30 °C with arsenicals at the indicated concentrations. To analyze GlpF and AqpS activities in E. coli OSBR1, 0.2% arabinose was added as an inducer, and 0.5% glycerol (v/v) replaced glucose as carbon source.

2.6. In vivo assays of arsenic uptake

For in vivo uptake assays, E. coli cells or S. meliloti were grown to A_600nm = 2 at 37 °C with shaking in LB medium. The cells were harvested and suspended in buffer A (75 mM HEPES-KOH, pH 7.5, 0.15 M KCl and 1 mM MgSO₄) at A_600nm = 10. To initiate the transport reaction, 20 μM of As(III), MAs(III), MAs(V) or DMAs(V), as indicated, was added to 1 mL of cell suspension. Portions (0.2 mL) from the cell suspension were withdrawn at the indicated times, filtered through nitrocellulose filters (0.2 μm pore diameter; EMD Millipore, Billerica, MA) and washed twice at room temperature with 5 mL of buffer A. The filters were digested with 0.3 mL of concentrated HNO₃ (68–70%) overnight at room temperature. The dissolved filters were incubated for 10 min at 70 °C, allowed to cool to room temperature and diluted with HPLC-grade water (Sigma-Aldrich) to produce a final HNO₃ concentration of 2%. Arsenic was quantified by ICP-MS. Standard solutions were made in the range of 0.5–50 ppb in 2% nitric acid using arsenic standard (Ultra Scientific, N. Kingstown, RI). Data are the mean ± SE (n=3). The statistical significance between AqpS wide type and mutants for uptake of the two substrates are represented by asterisk, where *, p < 0.05 and **, p < 0.01.

2.7. Mutagenesis of the aqpS gene

Mutations in S. meliloti aqpS were introduced by site-directed mutagenesis using QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA). The mutagenic oligonucleotides used for both strands and the respective changes introduced (underlined) are as follows: V177RF, 5’-CGCCAATCCCGCCAGGGCACTGGCGCGG-3’ and V177RR, 5’-CCGCGCCAGTGCCCTGGCGGGATTGGCG-3’; V177IF, 5’-CGCCAATCCCGCCATAGCACTGGCGCGGT-3’ and V177IR, 5’-ACCGCGCCAGTGCTATGGCGGGATTGGCG-3’; T49WF, 5’-GCCAACACGATCGCCTGGGGCGCAATCCTCGTC-3’ and T49WR, 5’-GACGAGGATTGCGCCCCAGGCGATCGTGTTGGC-3’; T49FF, 5’-CCAACACGATCGCCTTCGGCGCAATCCTCG-3’ and T49FR, 5’-CGAGGATTGCGCCGAAGGCGATCGTGTTGG-3’. Each mutation was confirmed by commercial DNA sequencing (Sequetech, Mountain View, CA).

2.8. Homology modeling of the AqpS structure

The AqpS homology model was built using SWISS-MODEL online server (http://swissmodel.expasy.org/) on the structure of aquaporin AqpM (PDB ID: 2F2B) as template (Lee et al., 2005). The sequence identity and similarity between the model and the template are 26.5 and 32%, respectively. The QMEAN and GMQE scores of the model are −3.65 and 0.66, respectively, indicating the quality of the model.(Kiefer et al., 2009) The solvent channel of the AqpS model was calculated using MOLEonline server (https://mole.upol.cz/online/) (Berka et al., 2012). Introduction of the mutations in the ar/R filter residues, including Thr49 and Val177, was done using COOT software (Emsley and Cowtan, 2004). Structural models were generated using PYMOL, Molecular Graphics System, Version 1.3, Schrodinger LLC (http://www.pymol.org/).

3. Results

3.1. Distribution of aquaglyceroporin in ars operon

Comparative genetic analysis of chromosomal ars operons from different bacterial reveals widely distributed aquaglyceroporin genes encoding either aqpS or aqp_NPAR genes that encode aquaglyceroporins different from AqpS in sequence and function (Fig. 1). AqpS, a GlpF homolog, was first reported in the legume symbiont S. meliloti and is the only known aquaglyceroporin with a physiological function in arsenate resistance. When S. meliloti is exposed to As(V), As(V) enters into the cells via phosphate transporters and is reduced by ArsC to As(III), which subsequently flows out of the cells down its concentration gradient via AqpS. The X-ray structures of aquaporins indicates that four amino acids constitute an aromatic/arginine (ar/R) pore constriction known as the selectivity filter. Its size and charge govern its ability to transport multiple substrates. In ars operons, some aqp_NPAR gene products have the conserved arginine residue in the LE2 position that allows for As(III) conductivity (Fig. S2). hAQP9 and OsLsi1, both of which have that arginine residue (Liu et al., 2004; Zhao et al., 2010), facilitate As(III) movement. In contrast, AqpS has the non-polar hydrophobic residue Val177 in the position where other AqpS have the conserved arginine residue (Fig. S2). We predict that this valine residue at position LE2 both widens the pore aperture in the ar/R region and eliminates the positive charge that restricts movement of charged substrates. This is similar to plant TIP1 homologs that have a valine residue in the LE2 position (Azad et al., 2012). These have a pore aperture of 6.26 Å in the ar/R constriction, while aquaglyceroporins that have an arginine residue in that position have a much smaller aperture. However, it is not known whether TIP1 homologs conduct MAs(V) or other anions.

The substrate selectivity of bacterial aquaglyceroporins encoded by ars operons has not been investigated. The different characteristics in the ar/R region suggests that the function and substrates of AqpS may differ significantly from other Aqps. Many of these genes are potentially co-transcribed with either an arsI or arsH MAs(III) resistance determinant. Although MAs(III) has been reported to be transported by mammalian aquaglyceroporins such as AQP9.(Liu et al., 2006) and the silicon transporter, OsLsi1 (Li et al., 2009), to our knowledge, there are no reports of uptake of methylated arsenicals by aquaglyceroporins in the bacterial kingdom, the largest group in nature. A phylogenetic analysis conducted with aquaporin sequences from both inside and outside of ars operons as well as representative eukaryotic AQPs shows that there are a number of diverse groups of aquaglyceroporins (Fig. 2). On the whole, Aqp_NPAR and AqpS_NPAV aquaglyceroporins in ars operons from bacteria are different and form separate groups from bacterial aquaglyceroporins outside of ars operons or eukaryotic aquaglyceroporins. Aqp_NPAR channels encoded in bacterial ars operons are most closely related to aquaglyceroporins from plants, many of which facilitate As(III) uptake (Zhao et al., 2009). The AqpS group is also distinct from Aqp_NPAR proteins encoded by aqp genes within other ars operons. This suggests that AqpS may have different arsenic transport properties compared with aquaglyceroporins such as GlpF and hAQP9.

3.2. Resistance to and uptake of trivalent and pentavalent arsenicals in S. meliloti and E. coli

Cells of E. coli AW3110(Δars) are sensitive to As(III), while cells of E. coli OSBR1(ΔarsΔglpF) are resistant to trivalent metalloids (Fig. 3A) (Meng et al., 2004; Sanders et al., 1997). Cells of E. coli OSBR1 are also resistant to MAs(III) compared to cells of strain AW3110, which suggests that GlpF also facilitates MAs(III) uptake. Wild type S. meliloti wild type is tolerant to As(V), but S. meliloti ΔaqpS has been shown to be hypersensitive to As(V) due to reduction to As(III) by the ArsC reductase (Yang et al., 2005b). Without AqpS, the resulting As(III) is accumulated and kills the cells. Cells of S. meliloti ΔaqpS are also resistant to As(III) and MAs(III) compared with the parental strain, indicating that AqpS also facilitates uptake of both As(III) and MAs(III). In agreement with the resistance assays, cells of E. coli AW3110 accumulated higher amounts of MAs(III) than cells of E. coli OSBR1 cells (Fig. 3B), indicating that GlpF facilitates uptake of MAs(III). Similarly, deletion of aqpS in S. meliloti resulted in increased resistance to both As(III) and MAs(III), and low As(III) and MAs(III) uptake in cells, showing that AqpS, like GlpF, facilitates uptake of both trivalent inorganic and methylarsenicals.

Fig. 3. Uptake of arsenicals in cells of S. meliloti and E. coli.

(A): Cells with deletion of glpF (E. coli OSBR1(ΔarsRBCΔglpF)) or aqpS in S. meliloti are more resistance to As(III) and MAs(III). E. coli and S. meliloti strains were grown in low phosphate medium containing arsenicals, as indicated, at 30 °C with shaking, as described in Material and Methods. (B): Cells of E. coli exhibit low uptake of pentavalent methylated arsenicals. Arsenicals accumulation in cells of E. coli or S. meliloti was assayed with 20 μM, final concentration, of the indicated arsenicals, as described in Material and Methods. Data are the mean ± SE (n=3).

In contrast to the results with the trivalent species, cells of all the E. coli and S. meliloti strains exhibited resistance to either 100 μM MAs(V) or DMAs(V). This could due to low toxicity of the pentavalent methylarsenicals or from lack of uptake. To differentiate between these possibilities, uptake of trivalent and pentavalent species was compared (Fig. 3B). Cells of E. coli OSBR1 accumulated little As(III) but high levels of As(V). Cells of either E. coli AW3110 or OSBR1 accumulated low amounts of MAs(V) or DMAs(V), indicating that E. coli is relatively impermeable to the pentavalent methylated arsenicals. In contrast, wild type S. meliloti accumulated both MAs(V) and DMAs(V), while S. meliloti ΔaqpS accumulated much less, suggesting that AqpS facilitates uptake of the pentavalent anions.

3.3. AqpS but not GlpF facilitates uptake of pentavalent methylated arsenicals

Aquaglyceroporins such as hAQP9 and OsLsi1 transport MAs(V) at low pH, where the charge on the pentavalent species is mostly neutralized. To examine whether bacterial aquaglyceroporins can transport the pentavalent species at physiological pH, the glpF and aqpS genes were cloned into vector plasmid pBAD, creating plasmids pBAD-GlpF and pBAD-AqpS, which were expressed in E. coli strain OSBR1. Cells expressing either aqpS or glpF genes exhibited sensitivity to both As(III) (Fig. S1A) and MAs(III) (Fig. S1B). Since neither MAs(V) nor DMAs(V) are particularly toxic, complementation studies do not answer whether the arsenicals are taken up, so accumulation was assayed in the complemented strains (Fig. 4). Cell of E. coli OSBR1 took up little arsenic no matter the species. Cells expressing glpF accumulated more As(III) (Fig. 4A) and MAs(III) (Fig. 4B) than cells expressing aqpS. However, cells expressing glpF accumulated low amounts of either MAs(V) (Fig. 4C) or DMAs(V) (Fig. 4D) compared with cells expressing aqpS. These results demonstrate that GlpF transports MAs(III) and As(III) but not MAs(V) or DMAs(V). In contrast, AqpS transports both the trivalent and pentavalent species at physiological pH.

Fig. 4. AqpS and GlpF facilitate arsenical uptake.

Uptake of arsenicals by E. coli OSBR1 carrying either vector plasmid pBAD (□), pBAD-AqpS (○) or pBAD-GlpF(▽), was assayed with 20 μM, final concentration, of (A) As(III), (B) MAs(III), (C) MAs(V) or (D) DMAs(V), as described in Material and Methods. Data are the mean ±SE (n=3).

AqpS channels have the sequence for six transmembrane helices and two conserved NPA motifs that define AQPs. The ar/R selectivity filter of AqpS consist of four residues (Thr49, Tyr164, Ser171 and Val177), in the 2nd and 5th helices and loop E that place it in a distinctly different group of AQPs (Fig. S2). The key residues in the ar/R filter are variable in aquaglyceroporins. For example, they are Gly, Ser, Gly and Arg in OsLsi1, Trp, Gly, Phe and Arg in GlpF and Phe, Ala, Cys and Arg in hAQP9 (Fig. 5). In most other aquaglyceroporins the fourth key residue in Loop E is a positively charged arginine, while in AqpS it is a neutral valine residue. In some plant tonoplast aquaporins (TIPs) such as TIP1 from A. thaliana, rice and maize a neutral isoleucine residue replaces the arginine residue (Mitani-Ueno et al., 2011). From the crystal structures of GlpF and AQP1 (de Groot and Grubmuller, 2001), the positive charge on the arginine residue serves as a filter that prevents passage of charged substrates. Substitution of the conserved arginine residue in rat AQP9 eliminates MAs(III) and As(III) uptake, consistent with the concept that the arginine residue contributes to the selectivity filter for entry to the channel (Mitani-Ueno et al., 2011). Substitution of the ar/R phenylalanine residue in rAQP9 did not affect transport of MAs(III) or As(III), suggesting that a hydrophobic residue is not required for selectivity in that position. From the crystal structures of AQP1 and GlpF, the ar/R Phe residue in helix 2 helps to orient the water molecules for hydrogen bonding to the arginine residue. Although Phe64 is not crucial in rAQP9, mutation to Ala or Thr lowered uptake of As(III), while conservative replacement to a tryptophan residue had much less effect (Liu et al., 2004). In a similar analysis of OsLsi1, in which smaller residues (Gly, Ser, Gly and Arg) compose the ar/R region, As(III) uptake was unaffected by a Gly to Ala substitution at H2 of OsLsi1 (Mitani-Ueno et al., 2011). Many plant NIPs possess small residues in the ar/R filter, resulting in a larger pore diameter (Wallace and Roberts, 2005). For example, a Ser to Ile mutation in helix 5 of OsLsi1 strongly reduced transport of both As(III) and silicon (Mitani-Ueno et al., 2011). Isoleucine is larger than serine, so a Ser to Ile substitution in helix 5 reduces the size of the pore aperture at the ar/R region. Substitution of polar serine with nonpolar isoleucine could also affect pore selectivity. Thus, the size, hydrophobicity and charge of the four residues in the ar/R filter alter the properties of the selectivity filter.

Fig. 5. A top view into the aquaglyceroporin constriction.

Four key residues are seen from the extracellular aspect along the axis of the ar/R selectivity pores of (A) AqpS, (B) OsLsi1, (C) GlpF (PDB ID: 1FX8) or (D) hAQP9. The AqpS, OsLsi1 and hAQP9 homology models were constructed based on the crystal structures PDB IDs 2F2B, 2ZZ9 and 6F7H respectively, as described in Material and Methods.

We hypothesize that non-aqpS genes in ars operons that encode AQPs with an NPAR motif in the ar/R filter are ancestral to aqpS. A single substitution in the fourth codon would replace the arginine residue with a valine to produce the NPAV sequence in the AqpS ar/R filter. In ars operons the genes for Aqp_NPAR channels are frequently adjacent to an arsH gene (Fig. 1). However, it is reasonable to assume that an Aqp_NPAR channel cannot efficiently transport MAs(V), the oxidation product of ArsH, out of cells at physiological pH and would accumulate MAs(V) to high levels which might be toxic or produce osmotic difficulties. We propose that such challenges provided selective pressure for mutation of the NPAR motif to the NPAV of AqpS, which allows MAs(V) to flow out of the cells at physiological pH. In addition, decreased intracellular MAs(V) could reduce product inhibition of ArsH, accelerating oxidation of MAs(III). Thus, coupling AqpS with ArsH may synergistically confer high-level in MAs(III) tolerance.

3.4. Val177 in the ar/R filter of AqpS permits MAs(V) transport

In AQPs with a canonical arginine residues in the ar/R filter, the positive charge prevents entry of As(V), MAs(V) and DMAs(V). Since AqpS has a hydrophobic valine residue in place of the arginine residue in the ar/R selectivity filter, we examined the possibility that AqpS has a different substrate specificity than other aquaglyceroporins. Val177 was changed to either Ile or Arg (Fig. S3A), and the transport activities for MAs(III) and MAs(V) were assayed (Fig. 6). The V177R substitution increased MAs(III) transport activity, leading to lower resistance to MAs(III) (Fig. 6A and Fig. S4A). A V177I substitution decreased MAs(III) transport activity and resulted in higher resistance to MAs(III). In contrast, a V177R substitution resulted in nearly complete loss of MAs(V) transport, while conservative replacement with Ile decreased MAs(V) transport activity only slightly (Fig. 6B). These results support our hypothesis that a hydrophobic residue in the position of the conserved arginine of the ar/R sequence permits transport of the MAs(V) anion and restricts movement of neutral MAs(III), which makes AqpS distinctly different from other AQPs.

Fig. 6. Val177 in the ar/R selectivity of AqpS is involved in MAs(V) uptake.

Uptake of 20 μM (A) MAs(III) or (B) MAs(V) by cells of E. coli OSBR1 expressing wild-type or mutant AqpS genes was performed as described in Material and Methods. Data are the mean ± SE (n=3). Differences in uptake are represented as *, p < 0.05 and **, p < 0.01.

In addition, Thr49 in helix 2 was changed to either Phe or Trp (Fig. S3B). Both mutants exhibited increased uptake of both MAs(III) and MAs(V), and higher sensitivity to MAs(III) (Fig. 6 and Fig. S4B). Thus, changing this polar residue to larger aromatic residues appear to facilitate entry of substrates into the channel. Consistent with this result, when Phe64 in rAQP9 was changed to either an alanine or threonine residue, channel activity decreased, suggesting that a bulky hydrophobic residue is preferred at this position in helix 2 over a smaller or polar residue (Liu et al., 2002).

4. Discussion

Arsenic contamination is a global problem. Millions of people worldwide are at danger of arsenic poisoning through consumption of arsenic-contaminated food and water. Methylarsenicals are introduced into environment both biogenically and anthropogenically. Inorganic arsenic is methylated to trivalent methyl-, dimethyl- and trimethylarsenicals by soil and marine microbes. In air those are oxidized to the corresponding pentavalent species. In addition, pentavalent monosodium methylarsenate (MSMA) has a long history of use as a post-emergent herbicide to control a range of annual grass-type weeds. Its use in the United States is now limited to cotton fields, new golf courses and highway medians, and it is still applied world-wide as an herbicide on rice, cotton, fruit trees and coffee in a number of Asian countries (Burlo et al., 1999). Even though the levels of MAs(V) are low in soil, their historical use as pesticides, as well as their presence in poultry litter that is used as fertilizer may elevate its levels in specific soils (Zhao et al., 2013). MAs(V) is found in paddy soil derived in part from a combination of microbial and algal biomethylation and from past uses of the MSMA herbicide. MAs(V) is taken up through the roots and translocated to shoots and grain in the rice plant, a staple food for much of the population of the world. Yet how bacteria, animals and plants transport MAs(V) is an unanswered question. Our results show that both GlpF from E. coli and AqpS from S. meliloti conduct MAs(III) better than As(III). Aquaglyceroporins encoded by ars operon, including AqpS, can catalyze MAs(IIII) uptake and confer resistance by coupling to detoxifying enzymes such as ArsH or ArsI. We demonstrate here that AqpS is unusual in its ability to conduct MAs(V) at physiological pH as a consequence of having a non-polar valine residue in the ar/R filter. This may explain how MAs(V) moves through the bacterial membrane, but how does it get into plant tissues? At physiological or alkaline pH, OsLsi1 showed little ability to conduct either MAs(V) or DMAs(V) (Li et al. 2009). On the other hand, tonoplast intrinsic proteins (TIPs), which are abundant in plants such as Arabidopsis and rice, may have the ability to conduct MAs(V) and DMAs(V) (Azad et al, 2012). Like AqpS, both OsTIP1;1 and OsTIP1;2 have an NPAV motif. It is reasonable to speculate that they could be responsible for accumulation of pentavalent methylarsenicals in rice. MAs(V) transport by TIPs has not been measured, and the effect of substitution of the arginine residue to valine in the aromatic/arginine selectivity filter on MAs(V) transport profiles of TIPs needs to be determined. However, if OsTIP1;1 and OsTIP1;2 do indeed conduct MAs(V), then changing tonoplast AQPs to an NPAR motif might be predicted to reduce accumulation of MAs(V) and DMAs(V) in the rice plant. In rice grain, methylated arsenic, especially DMA(V) can account up to 80% of total arsenic. Therefore, reducing methylated arsenic in plant is an alternative and promising strategy to reduce arsenic content and increase food safety. Genetic engineering rice with TIPs in which the NAPV motif is changed to NPAR might be predicted to decrease MAs(V) and DMAs(V) uptake, but not As(III), B or Si and reduce total arsenic accumulation without affecting crop yield. Further studies are needed to explore plant TIPs for its uptake of methylated arsenic, especially DMAs(V), to achieve this goal.

Finally, what are the environmental implications of the presence of the aqpS gene in S. meliloti? We propose that the ability to conduct the MAs(V) anion confers a selective advantage for S. melliloti to grow in the presence of environmental methylarsenicals (Fig. 7). In microbial communities some bacteria methylate As(III) to MAs(III), and other bacteria reduce MAs(V) to MAs(III), using this highly toxic species as an antibiotic (Chen et al., 2019). S. meliloti 1021 has a four-gene ars operon including arsR encoding an As(III)-responsive transcriptional repressor, arsC encoding an As(V) reductase, arsH encoding a MAs(III) oxidase and aqpS, encoding the AqpS aquaglyceroporin. When S. meliloti is exposed to As(V), As(V) is reduced by As(V) reductase (ArsC) in cells to As(III), which flows out of the cells down its concentration gradient by AqpS, conferring resistance to As(V) (Yang et al., 2005a). However, S. meliloti can also be exposed to MAs(III) produced by other members of microbial communities, which would enter the cells adventitiously via AqpS. S. melliloti could detoxify it via ArsH oxidation to MAs(V), but the pentavalent product would accumulate to high levels, which could be deleterious. AqpS provides a release mechanism for MAs(V), preventing its buildup inside the cells. We propose that a need for MAs(III) detoxification may have been the driving force for acquisition of the anion-conduction properties of this novel aquaglyceroporin. Of what usefulness is AqpS to crop plants and the world food supply? Since Rhizobia are critical nitrogen-fixing symbionts for plants, including rice (Chi et al., 2005), the presence of the aqpS gene in these bacteria can protect them from biogenic and anthropogenic MAs(V), enhancing nitrogen fixation and plant growth.

Fig. 7. Pathway of MAs(III) resistance in S. meliloti.

In microbial communities bacteria with an arsM gene methylate As(III) to MAs(III), which is extruded by ArsP, to kills MAs(III) sensitive bacteria. In air MAs(III) is oxidized to MAs(V), and MAs(V) is also introduced into the environment as the MSMA herbicide. Other members of microbial communities reduce both biogenic and anthropogenic MAs(V) to MAs(III) to give them a competitive advantage over MAs(III)-sensitive community members. MAs(III) flows into cells of S. meliloti down its concentration gradient via AqpS, where it is oxidized to MAs(V) by ArsH, which then flows out of the cells down its concentration gradient also by AqpS. Thus the ability of AqpS to conduct bidirectional movement of MAs(III) and MAs(V) coupled to ArsH oxidation provides a novel pathway for resistance to toxic MAs(III).

Highlights:

• Our objective was to identify how bacteria transport methylarsenicals

• We analyzed genes for aquaglyceroporins (AQPs) in arsenic resistance (ars) operons

• The products of aqp genes located next to MAs(III) resistance genes transport methylarsenite

• AqpS from Sinorhizobium meliloti 1021 transports methylarsenate

• Val177 in the AqpS selectivity filter allows movement of methylarsenate

Abbreviations:

MAs(IIII)

methylarsenite

MAs(V)

methylarsenite

DMAs(III)

dimethylarsenite

DMAs(V)

dimethylarsenate

MSMA

monosodium methylarsenate