1. Arsenate uptake
Arsenate (As(V)) is the main As species in aerobic soils. It has a strong affinity for iron oxides/hydroxides in soil; thus the concentrations of arsenate in soil solutions are usually low. Wenzel et al. (2002) reported ≤ 53 nM arsenate in the soil solutions from a range of uncontaminated and moderately contaminated soils and up to 2.3 µM in a highly contaminated soil. Many hydroponic studies have used much higher concentrations of arsenate than those found in soil solution, and their environmental relevance has been questioned (Fitz & Wenzel, 2002).
Physiological and electrophysiological studies have shown that arsenate and phosphate share the same transport pathway in higher plants, with the transporters having a higher affinity for phosphate than for arsenate (e.g. Asher & Reay, 1979; Ullrich-Eberius et al., 1989; Meharg et al., 1994). The uptake mechanism involves cotransport of phosphate or arsenate and protons, with stoichiometry of at least 2H+ for each or (Ullrich-Eberius et al., 1989). A number of phosphate transporters have been characterized in plants (Rausch & Bucher, 2002; Bucher, 2007). There are over 100 phosphate transporters in the Phosphate transporter 1 (Pht1) family, most of which are strongly expressed in roots and are likely to be involved in phosphate uptake from the external medium (Bucher, 2007). In Arabidopsis thaliana, two phosphate transporters, Pht1;1 and Pht1;4, play a significant role in phosphate acquisition from both low- and high-phosphorus (P) environments (Shin et al., 2004). The A. thaliana double mutant pht1;1Δ4Δ was much more resistant to arsenate than the wild type, indicating that Pht1;1 and Pht1;4 mediate arsenate uptake (Shin et al., 2004). In the A. thaliana mutant defective in phosphate transporter traffic facilitator 1 (PHF1), the trafficking of the Pht1;1 protein from the endoplasmic reticulum to the plasma membrane is impaired (González et al., 2005). This mutant was much more resistant to arsenate than the wild type, further supporting a role of Pht1;1 in arsenate uptake. Recently, Catarecha et al. (2007) identified an arsenate-tolerant mutant of A. thaliana, pht1;1-3, which harbours a semidominant allele coding for the high-affinity phosphate transporter PHT1;1. Rather intriguingly, the pht1;1-3 mutant displays the dual phenotypes of decreased arsenate uptake in the short-term and increased As accumulation over a longer period of growth. As the wild-type plants suffered from severe As toxicity, it is perhaps not surprising that their As accumulation capacity was curtailed compared with the mutant. Acquisition of knowledge about phosphate transporters and their regulation in plants will undoubtedly lead to a better understanding of the arsenate uptake mechanisms in plants. Specifically, it would be interesting to determine the relative selectivity of different transporters for phosphate and arsenate, and to examine allelic variation in this selectivity.
Reduced uptake of arsenate is a well-known mechanism of arsenate resistance employed by many plant species, which is achieved through a suppression of the high-affinity phosphate/arsenate uptake system in the resistant plants (reviewed by Meharg & Hartley-Whitaker, 2002).
2. Arsenite uptake
Arsenite (As(III)) is the dominant As species in reducing environments such as flooded paddy soils (Marin et al., 1993; Takahashi et al., 2004; Xu et al., 2008). Thermodynamically, reduction of arsenate to arsenite can occur quite readily at intermediate redox potentials (Inskeep et al., 2002). Flooding of paddy soils leads to mobilization of arsenite into the soil solution and enhanced As bioavailability to rice plants (Xu et al., 2008). The arsenite concentration in soil solutions from flooded paddy soils typically varies from 0.01 to 3 µM; these concentrations are generally higher than those of arsenate found in uncontaminated aerobic soils. Arsenous acid (As(OH)3) has a pKa of 9.22. Therefore, arsenite, despite its name referring to an oxyanion, is actually present in solution predominantly as an undissociated neutral molecule at pH < 8.
Little was known about the mechanisms of arsenite uptake in plants until recently. Research on arsenite uptake mechanisms in plants has benefited greatly from the knowledge gained from microbial studies. In Escherichia coli, yeast and humans, some aquaglyceroporins, a subfamily of the aquaporin superfamily with larger pores to allow passage of neutral molecules such as glycerol, can transport arsenite (reviewed by Bhattacharjee & Rosen, 2007). These include the E. coli glycerol facilitator GlpF, the yeast glycerol channel protein Fps1p, and the mammalian aquaglyceroporins AQP7 and AQP9. An alternative mechanism, other than that involving aquaglycerinporins, has also been identified in yeast. The addition of glucose inhibited arsenite uptake by 80%, and the deletion of hexose permease genes led to a much reduced uptake of arsenite in yeast, suggesting that hexose permeases catalyse the majority of arsenite uptake (Liu et al., 2004b).
Plant roots are capable of rapidly taking up arsenite from the external medium. Short-term (20 min) uptake experiments with excised rice roots showed that the maximum (Vmax) of arsenite influx was comparable to that of arsenate in the absence of phosphate, but the concentration at which the influx is half Vmax (Km) was higher (Abedin et al., 2002b). At higher concentrations (> 100 µM), that is, in the low-affinity range, arsenite influx was substantially faster than arsenate (Abedin et al., 2002b; Meharg & Jardine, 2003). Furthermore, unlike arsenate uptake, arsenite uptake was inhibited by glycerol and antimonite, but not by phosphate. Based on competition experiments, Meharg & Jardine (2003) suggested that arsenite may be taken up by aquaporin channels in plant roots.
Recently, evidence that some plant aquaporin channels can mediate arsenite influx has been obtained from three independent studies (Bienert et al., 2008b; Isayenkov & Maathuis, 2008; Ma et al., 2008). Bienert et al. (2008b) expressed a number of plant genes encoding the nodulin26-like intrinsic proteins (NIPs), a subfamily of the plant aquaporin family, in yeast. They found that the expression of AtNIP5;1 and AtNIP6;1 from A. thaliana, OsNIP2;1 and OsNIP3;2 from rice, and LjNIP5;1 and LjNIP6;1 from Lotus japonicus increased the sensitivity of yeast to arsenite and antimonite, as well as As accumulation in the yeast cells. Interestingly, AtNIP5;1 has been identified as a boric acid transporter essential for boron (B) uptake by A. thaliana roots (Takano et al., 2006). Despite the arsenite transport activity of AtNIP5;1 and AtNIP6;1 when expressed heterologously in yeast (Bienert et al., 2008b), the A. thaliana T-DNA insertion lines of these two genes showed no significant difference from the wild type in growth in the presence of elevated concentrations of either arsenite or arsenate, suggesting that these two proteins do not contribute significantly to arsenite transport in planta (Isayenkov & Maathuis, 2008). These authors identified AtNIP7;1 as a possible candidate for arsenite transport in A. thaliana. The T-DNA insertion lines of AtNIP7;1 were more resistant to arsenite than the wild type, and accumulated approx. 25% less As when grown in agar plates containing 7 µM arsenite. Expression of AtNIP7;1 in yeast increased arsenite sensitivity.
Arsenite uptake is of particular importance for rice and other aquatic plants with their roots growing in anaerobic or semi-anaerobic environments. Recently, Ma et al. (2008) have identified OsNIP2;1, also named Lsi1 because of its primary function as a silicon (Si) transporter (Ma et al., 2006), as a major pathway for the entry of arsenite into rice roots. Expression of Lsi1 in Xenopus laevis oocytes and in yeast markedly increased the uptake of arsenite, but not of arsenate. Mutation of Lsi1 in rice (lsi1 mutant) resulted in a c. 60% loss in the short-term (30-min) arsenite influx to roots compared with wild-type rice. These data indicate that arsenite shares the Si transport pathway for entry into rice root cells. This is not surprising because arsenite and silicic acid have two important similarities: both have a high pKa (9.2 and 9.3 for arsenous acid and silicic acid, respectively); and both molecules are tetrahedral with similar sizes. Lsi1 is strongly expressed in rice roots and its expression is further enhanced in plants not supplied with Si (Ma et al., 2006). The Lsi1 protein is localized on the plasma membrane of the distal side of both exodermis and endodermis cells, where Casparian strips occur.
Ma et al. (2008) showed that, in addition to Lsi1, three other NIP channel proteins in rice, OsNIP1;1, OsNIP2;2 (also named Lsi6) and OsNIP3;1, are also able to mediate arsenite influx into X. laevis oocytes expressing these genes. While OsNIP2;2 is permeable to silicic acid, OsNIP1;1 and OsNIP3;1 are not (Mitani et al., 2008); the latter (OsNIP3;1) has been shown to mediate B uptake in rice roots (Takano et al., 2008). However, unlike Lsi1, OsNIP1;1, OsNIP2;2 and OsNIP3;1 are expressed at very low levels in rice roots, and thus are unlikely to play a significant role in arsenite influx. Indeed, the loss of function lines of OsNIP2;2 did not show a significant decrease in arsenite uptake (Ma et al., 2008).
NIPs represent one of the four subfamilies of the plant major intrinsic proteins (MIPs), commonly called aquaporins, the other three being plasma membrane intrinsic proteins (PIPs), tonoplast intrinsic proteins (TIPs), and small basic intrinsic proteins (SIPs) (Chaumont et al., 2005; Maurel et al., 2008). NIPs have low to no water permeability and the ability to transport multiple uncharged solutes of varying sizes including glycerol, urea, ammonia, boric acid and silicic acid (Wallace et al., 2006), as well as arsenite (Bienert et al., 2008b; Isayenkov & Maathuis, 2008; Ma et al., 2008). They are sometimes called aquaglyceroporins (Wallace et al., 2006), although some of the NIPs (e.g. OsNIP2;1) have little permeability to glycerol (Ma et al., 2006; Mitani et al., 2008), and there is no direct evidence for a physiological role in plants of glycerol transport through NIPs (Bienert et al., 2008b). Phylogenetic studies suggest that NIPs were acquired early at the beginning of plant evolution by horizontal gene transfer of a bacterial homologue of aquaporins, whose founding member is the bacterial GlpF that can also transport arsenite (Zardoya et al., 2002; Wallace et al., 2006). There are nine and 10–13 members of the NIP subfamily in the A. thaliana and rice genomes, respectively (Forrest & Bhave, 2007; Maurel et al., 2008). The substrate selectivity of aquaporins is mainly controlled by two pore constrictions, one formed by the highly conserved asparagine–proline–alanine (NPA) boxes and the other the aromatic/arginine (ar/R) selectivity filter (Wallace et al., 2006; Maurel et al., 2008). Based on homology modelling of pore structures at the ar/R selectivity filter, NIPs have been subdivided into two (Wallace et al., 2006) or three subgroups (Mitani et al., 2008) (Fig. 1). The NIP I subgroup includes the archetype nodulin 26 and is permeable to water, glycerol and lactic acid. The members of NIP II subgroup have a predicted larger pore size than those of the NIP I subgroup, and are permeable to larger solutes such as urea, formamide and boric acid, but with a much reduced water permeability (Wallace et al., 2006). The NIP III proteins transport silicic acid; the ar/R region of these proteins contains residues of smaller size, thus forming a larger constriction site compared with other NIP subgroups. However, it appears that arsenite permeability is a property widespread in all NIP subgroups (Fig. 1), suggesting that transport of arsenite is not controlled by the ar/R selectivity filter. It is likely that more members of the NIP protein family will be found to be permeable to arsenite.
Figure 1. Phylogenetic tree of plant nodulin26-like intrinsic protein (NIP) channel proteins. The NIPs that have been shown to be permeable to arsenite are shown in bold. At, Arabidopsis thaliana; Os, Oryza sativa; Zm, Zea mays; Lj, Lotus japonica; Gm, Glycine max; Cp, Cucurbita pepo; rAQP9, mammalian aquaglyceroporin 9. I, II and III represent three subgroups of the NIP proteins. The amino acid sequences of NIPs were aligned by Clustal W (http://www.ebi.ac.uk/Tools/clustalw2).
Download figure to PowerPoint
To date, there is no report of arsenite permeability in the PIP, TIP and SIP channel proteins in plants. Whether these proteins can transport arsenite remains to be investigated. PIPs have a narrow pore structure typical of orthodox, water-selective aquaporins (Maurel et al., 2008) and are, therefore, not likely to be permeable to arsenite. By contrast, it is possible that some TIP channels may be permeable to arsenite and contribute to arsenite transport into the vacuoles.
While Lsi1 transports arsenite into rice root cells, a different transporter, Lsi2, has been found to mediate arsenite efflux in the direction of xylem (Ma et al., 2008) (Fig. 2). Lsi2 was initially identified as an Si efflux transporter (Ma et al., 2007). Lsi2 is also localized at the exodermis and endodermis of rice roots, but, in contrast to Lsi1, at the proximal side. Therefore, the pathway of Si transport from the external medium to the stele involves the influx of silicic acid mediated by the aquaporin channel Lsi1 (Ma et al., 2006), and the efflux of Si towards the stele mediated by Lsi2 (Ma et al., 2007). Mutation in Lsi2 in two independent rice mutants led to a marked decrease (66–75%) in As accumulation in shoots compared with wild types (Ma et al., 2008). Arsenite concentrations in xylem sap from the mutants were much lower than those in xylem sap from wild-types. Moreover, addition of Si to the nutrient solution inhibited arsenite transport to the xylem and accumulation in the shoots in the wild-type rice, but not in the two lsi2 mutants. Rice is a strong accumulator of Si, with Si concentration in the shoots typically varying from 5 to 10%. The efficient Si uptake pathway in rice also allows inadvertent passage of arsenite, thus explaining why rice is efficient in accumulation of As.
When both the lsi1 and lsi2 mutants and their wild-types were grown to maturity in a field experiment, the lsi2 mutants were found to contain significantly lower concentrations of As in straw and grain than the wild-type rice, whereas the differences between the lsi1 mutant and its wild-type were not statistically significant (Ma et al., 2008). Therefore, although loss of function of Lsi1 affects short-term arsenite influx and As accumulation in rice, the efflux of arsenite toward xylem mediated by Lsi2 is the crucial step in controlling As accumulation in rice shoots and grain over a longer growth period. It has been reported that additions of silicate inhibited As accumulation by rice when arsenate was the form of As added to the nutrient solution; yet this effect was not attributable to a direct competition between Si and arsenate because they do not share the same transporters (Guo et al., 2005, 2007). These observations can now be explained by the involvement of Lsi2. Arsenate taken up by rice roots is reduced in the root cells to arsenite, which is transported towards the xylem via the Si/arsenite effluxer Lsi2 and is subject to competitive inhibition from Si. Applying Si fertilizers to rice crops may prove to be an effective way of mitigating the problem of excessive transfer of As from paddy soil to rice grain. In addition, Si application can increase rice yield by alleviating biotic and abiotic stresses (Ma & Yamaji, 2006).
4. Efflux of As species
Following uptake of arsenate by roots, some of the arsenate is lost from the cells via efflux to the external medium (Xu et al., 2007); this is similar to the situation for phosphate, which can also be lost via efflux especially under high-P conditions (Mimura, 1999). The mechanism of arsenate efflux is not known, but may be similar to that of phosphate efflux which is thought to be via anion channels (Mimura, 1999).
Xu et al. (2007) showed that arsenate added to the aerated nutrient solution was rapidly converted to arsenite by the roots of tomato (Lycopersicon esculentum) and rice. Microbes living in the nutrient solution or root exudates contributed little to arsenate reduction to arsenite. This is surprising because arsenate is expected to be stable in the aerobic environment. Phosphate inhibits arsenate uptake and the subsequent production of arsenite in the external medium, suggesting that the arsenite is extruded by root cells following arsenate reduction inside the cells. Indeed, efflux of both arsenate and arsenite was observed when tomato roots preloaded with arsenate were transferred to an As-free medium. Furthermore, the protonophore carbonylcyanide m-chlorophenylhydrazone (CCCP) inhibited the efflux of arsenite, suggesting that the efflux is linked to the proton gradient across the plasma membranes, or is metabolically dependent. Arsenite efflux by roots has been observed in other plant species, including A. thaliana, Holcus lanatus, wheat (Triticum aestivum), barley (Hordeum vulgare) and maize (Zea mays), but not in the As hyperaccumulator Pteris vittata (discussed in more detail Section VII). Within 24 h of exposure to arsenate, arsenite efflux was approximately 3 times the amount of As accumulated in the plants, suggesting rapid cycling of As between plant roots and the medium (F. J. Zhao et al., unpublished). It appears that arsenite efflux by roots is very rapid immediately following arsenate uptake, and diminishes once the arsenate supply is withheld, possibly because cellular arsenite is complexed with thiols and sequestered in the vacuoles (see Section IV). The rapid conversion of arsenate to arsenite in the external medium raises a question regarding previous hydroponic studies where As speciation was not monitored.
In aerobic soils, arsenite is oxidized rapidly to arsenate either chemically by reactions with manganese oxide (e.g. Oscarson et al., 1981) or by arsenite-oxidizing microbes (Macur et al., 2004). Thus, soil, plant roots and microbes are likely to be engaged constantly in the reduction–oxidation cycle of arsenate–arsenite. Studies using rhizoboxes to enable measurement of As speciation in the rhizosphere showed accumulation of arsenite close to the vicinity of sunflower (Helianthus annuus) and maize roots (Ultra et al., 2007a,b; Vetterlein et al., 2007), suggesting that efflux of arsenite occurs in soil-grown plants.
The mechanisms of arsenite efflux from plant roots remain to be elucidated. In microbes, arsenate reduction followed by arsenite efflux is a common and important mechanism of As detoxification (Bhattacharjee & Rosen, 2007). In E. coli, arsenite efflux is mediated by either ArsB or the ArsAB complexes. ArsB is a secondary efflux protein coupled to the proton-motive force. ArsB can also associate with ArsA, an ATPase, to form a pump that is much more efficient than ArsB alone at extruding arsenite from the cells (Dey et al., 1994). In yeast and fungi, a different type of efflux carrier protein, Acr3p, is responsible for arsenite efflux, although the mechanism of Acr3p is possibly similar to that of ArsB, relying on the proton-motive force for energy (Wysocki et al., 1997). The CCCP sensitivity of arsenite efflux by tomato roots (Xu et al., 2007) suggests a possibility of ArsB- or Acr3p-like carriers for the efflux. However, direct evidence has yet to be obtained.
Another possible mechanism responsible for arsenite efflux from plant roots involves aquaporin channels, some of which allow bidirectional passage of solutes (Mitani et al., 2008). Unlike the efflux carriers or efflux pumps described above, aquaporin-mediated arsenite efflux occurs through diffusion when the internal arsenite concentration exceeds that in the external medium. In the legume symbiont Sinorhizobium meliloti, the arsenic resistance (ars) operon includes an aquaglyceroporin (aqpS) in place of arsB, which confers arsenate resistance possibly through arsenite efflux (Yang et al., 2005). Bienert et al. (2008b) showed that expression of AtNIP5;1, AtNIP6;1, OsNIP2;1, OsNIP3;1 and LjNIP6;1 in yeast significantly enhanced its tolerance to arsenate. This is interpreted as the NIP channels mediating efflux of arsenite, which was produced by the reduction of arsenate inside the yeast cells. Similarly, Isayenkov & Maathuis (2008) reported that expression of AtNIP7;1 in the yeast acr3Δ mutant resulted in a small but consistent increase in arsenate tolerance, suggesting that AtNIP7;1 may mediate arsenite efflux in the absence of ACR3. Bienert et al. (2008a) argued that aquaporins/aquaglyceroporin channel-mediated extrusion of arsenite is an ancient mechanism for As detoxification. However, it is not clear whether a similar mechanism is involved in plants.
Is arsenite efflux by plant roots a detoxification mechanism, as has been shown for microbes? This question can only be answered unequivocally when the transporters responsible for the efflux are identified. Knockout or knockdown lines of these transporters would then allow a detailed examination of their roles in As accumulation, efflux and tolerance. In a study comparing arsenite efflux in arsenate-resistant and nonresistant ecotypes of H. lanatus, Logoteta et al. (2008) found that arsenite efflux was proportional to arsenate uptake in both ecotypes, with the resistant ecotype having a much lower arsenate uptake, as has been demonstrated before (Meharg & Macnair, 1992). This finding suggests that arsenite efflux is not enhanced in the resistant ecotype, which has evolved as a result of the selection pressure of high As availability in soil. However, this observation does not rule out the possibility that arsenite efflux is a constitutive, rather than adaptive, mechanism of As detoxification. Although arsenite can be re-absorbed by roots, the presence of rapid efflux machinery would logically lead to a decreased As burden in the root cells.