Article

Nature Biotechnology 20, 1140 - 1145 (2002)
Published online: 7 October 2002; | doi:10.1038/nbt747

Engineering tolerance and hyperaccumulation of arsenic in plants by combining arsenate reductase and -glutamylcysteine synthetase expression

Om Parkash Dhankher¹, Yujing Li¹, Barry P. Rosen², Jin Shi², David Salt³, Julie F. Senecoff¹, Nupur A. Sashti¹ & Richard B. Meagher¹

¹ Department of Genetics, University of Georgia, Athens, GA 30602.

² Department of Biochemistry and Molecular Biology, Wayne State University, Detroit, MI 48201.

³ Center for Plant Environmental Stress Physiology, 1165 Horticulture Building, Purdue University, West Lafayette, IN 47907.

Correspondence should be addressed to Richard B. Meagher meagher@arches.uga.edu

We have developed a genetics-based phytoremediation strategy for arsenic in which the oxyanion arsenate is transported aboveground, reduced to arsenite, and sequestered in thiol−peptide complexes. The Escherichia coli arsC gene encodes arsenate reductase (ArsC), which catalyzes the glutathione (GSH)-coupled electrochemical reduction of arsenate to the more toxic arsenite. Arabidopsis thaliana plants transformed with the arsC gene expressed from a light-induced soybean rubisco promoter (SRS1p) strongly express ArsC protein in leaves, but not roots, and were consequently hypersensitive to arsenate. Arabidopsis plants expressing the E. coli gene encoding -glutamylcysteine synthetase (-ECS) from a strong constitutive actin promoter (ACT2p) were moderately tolerant to arsenic compared with wild type. However, plants expressing SRS1p/ArsC and ACT2p/-ECS together showed substantially greater arsenic tolerance than -ECS or wild-type plants. When grown on arsenic, these plants accumulated 4- to 17-fold greater fresh shoot weight and accumulated 2- to 3-fold more arsenic per gram of tissue than wild type or plants expressing -ECS or ArsC alone. This arsenic remediation strategy should be applicable to a wide variety of plant species.

Arsenic is an extremely toxic metalloid pollutant that adversely affects the health of millions of people worldwide¹. Inorganic arsenic species, which are considered very hazardous for human health, are classified as group A human carcinogens and cause skin lesions, lung, kidney, and liver cancers, and also damage to the nervous system^{2, 3} (US Environmental Protection Agency (EPA), 1996: http://www.epa.gov/ogwdw/ars/arsenic.html). Hundreds of Superfund sites in the United States are listed on the National Priority List (NPL) (http://www.epa.gov/superfund/sites/nl/info.html) as having unacceptably high levels of arsenic and recommended for cleanup. In the majority of cases, arsenic-contaminated sites are not cleaned up because the cost in both dollars and environmental damage is too high. Physical remediation methods involving soil removal and burial are expensive, impractical on the scale that is needed, and environmentally destructive.

Higher plants can extract pollutants from the soil or water through their normal root uptake of nutrients⁴. They can store and concentrate pollutant in their cells and/or convert toxic pollutants to less toxic forms⁵. Plant-based phytoremediation strategies for heavy metals rely on plant roots to extract, plant vascular systems to transport, and leaves as a sink to concentrate arsenic aboveground for harvest and processing. A native fern indigenous to the southern United States has recently been characterized that hyperaccumulates arsenic to very high levels⁶. However, the genetic basis for its activity is unknown, and hence the enzymes responsible for hyperaccumulation are not yet available for manipulation into other species with wider geographic and ecological distribution and greater biomass.

Most arsenic in surface soil and water exists primarily in its oxidized form, the oxyanion arsenate (AsO₄^3-), which is an analog of phosphate. Arsenate can potentially be taken up from soil and translocated up the plant vascular system along with phosphate^{7, 8}. In contrast, the reduced form of arsenic, arsenite (AsO₃^3-), has a strong affinity toward thiol groups and, once formed aboveground in leaf and stem tissues, should be trapped as peptide−thiol complexes such as those formed by -glutamylcysteine (-EC; refs 9,10) as shown in Figure 1. Our working hypothesis was, therefore, that controlling the electrochemical state of arsenic in aboveground tissues and increasing thiol sinks throughout the plant would result in both resistance and hyperaccumulation of arsenic. As a first step toward testing this hypothesis, we examined the effects of coexpressing two bacterial genes, arsenate reductase (arsC) and -glutamylcysteine synthetase (-ECS), in Arabidopsis plants.

Figure 1. ArsC- and -ECS-catalyzed reactions.
	The bacterial arsenate reductase (ArsC) catalyzes the electrochemical reduction of arsenate to arsenite. The bacterial -glutamylcysteine synthetase (-ECS) catalyzes the formation of -glutamylcysteine (-EC) from the amino acids glutamate and cysteine and is the committed step in the synthesis of glutathione (GSH) and phytochelatins (PCs; indicated by three arrows). Reduced arsenite can bind organic thiols (RS) such as those in -EC, GSH, and PCs. Full Figure and legend (6K)

Results
Transgenic plants expressing ArsC aboveground.
Bacterial resistance to arsenic is acquired by first reducing arsenate to arsenite using the arsenate reductase (ArsC) enzyme in a glutathione-dependent reduction¹¹, as shown in Figure 1. Because the prokaryotic arsC gene has been previously shown to confer resistance in the eukaryote Saccharomyces cerevisiae¹², it was reasonable to consider that it would functionally express in plants. To test the first part of our working hypothesis that deals with controlling the electrochemical state of arsenic aboveground, the gene encoding this same bacterial ArsC enzyme was expressed under control of the regulatory sequences from the well-characterized soybean ribulose bisphosphate carboxylase small-subunit (rubisco) SRS1 gene¹³, which shows strong light-induced expression in leaves and stems. The arsC gene was cloned as a translational fusion to the ATG initiation codon of SRS1p and nos terminator to make SRS1p/ArsC. For comparison, arsC was also cloned into a well-characterized constitutive plant viral cauliflower mosaic virus (CaMV) 35S promoter and nos terminator (35Sp/ArsC). Arabidopsis thaliana was transformed with both constructs using vacuum infiltration, and the T₁ generation seeds were screened for a linked kanamycin resistance marker. Five kanamycin-resistant Arabidopsis plants containing each arsC construct were randomly selected after transformation. The lines containing SRS1p-driven arsC were designated SRS1p/ArsC2, SRS1p/ArsC7, SRS1p/ArsC8, SRS1p/ArsC9, and SRS1p/ArsC10, and the lines containing 35Sp-driven arsC were designated 35Sp/ArsC3, 35Sp/ArsC4, 35Sp/ArsC7, 35Sp/ArsC8, and 35Sp/ArsC13. These transgenic plants did not show any phenotypic differences from the untransformed Arabidopsis plants when grown without arsenic.

The organ-specific expression of ArsC protein was examined on western blots for both SRS1p/ArsC and 35Sp/ArsC constructs. Three of the five lines for both SRS1p and 35Sp promoter constructs are shown in Figure 2A and B, respectively. ArsC protein from the SRS1p/ArsC transgene was expressed only in leaf tissues and not in roots (Fig. 2A), whereas protein from the 35Sp/ArsC construct was expressed equivalently in leaves and roots (Fig. 2B). Prolonged exposure of these and additional blots analyzing SRS1p/ArsC expression still did not detect ArsC in the roots.

Figure 2. Immunodetection of ArsC and -ECS proteins in transgenic plants.
	(A) Strong ArsC (16 kDa) protein expression was observed on western blots from leaves, but not roots of three independent lines transformed with the SRS1p/ArsC construct. Protein extracts from E. coli expressing ArsC from pNA1 plasmid and wild-type (WT) plant extracts serve as positive and negative controls, respectively. (B) Strong expression of ArsC protein was observed in both leaves and roots of three independent transgenic lines transformed with the 35Sp/ArsC construct. (C) Both -ECS (top panel) and ArsC (bottom panel) are assayed on western blots of protein extracts from transgenic ArsC9 parental plant line expressing ArsC alone, a ACT2p/ECS1 line expressing g-ECS (57 kDa) alone, and ten lines generated by transforming the ArsC9 parental line with ACT2p/-ECS (lines ArsC9 + ECS1−10). The western membrane was cut into two strips and reacted separately with ArsC- and -ECS-specific antisera. For (A−C), equal amounts (10 g) of total protein were resolved on a 12% (wt/vol) polyacrylamide gel by SDS−PAGE and blotted to membrane. Western blots of plant extracts were developed as described in Bizily et al.^34 after reacting with polyclonal antisera to ArsC^35 and monoclonal antibodies to -ECS^31. Equal loading of samples was confirmed by Coomassie staining of parallel samples on a separate gel. Full Figure and legend (61K)

The strong aboveground specificity of ArsC expression and the lack of belowground expression were important in the proposed arsenic remediation strategy to allow arsenate to be transported aboveground. A transcriptional fusion between SRS1p and the -glucuronidase (GUS) reporter gene also supported the aboveground leaf-specific expression of the SRS1p (Fig. 3A). Very strong SRS1p/GUS expression was seen in all leaf and most stem tissues, but no expression was detected in any hypocotyl or root tissues even after very close inspection of several independent transgenic lines stained for GUS activity for prolonged periods.

Figure 3. Arsenic-sensitive phenotype of ArsC-expressing Arabidopsis.
	(A) Leaf-specific expression of SRS1p/GUS reporter fusion (A3 and A4) is compared to GUS-stained wild type (WT: A1 and A2), confirming the strong aboveground light-induced expression from the rubisco SRS1 promoter in two-week-old plants. Stained enlarged roots (inset box) from WT and SRS1p/GUS plants are shown in panels A2 and A4, respectively. (B, C) ArsC-expressing Arabidopsis are hypersensitive to arsenate in comparison with wild type. (B) Arsenate sensitivity of the transgenic line SRS1p/ArsC9 expressing ArsC from the SRS1p promoter compared with wild type (WT) on increasing concentrations of sodium arsenate (0, 75, and 150 M) in 0.5 MS medium. (C) Enlargement of wild-type and transgenic line SRS1p/ArsC9 grown vertically on 150 M arsenate. Plants in (B) and (C) were grown for three weeks on half-strength MS medium with the arsenic concentrations indicated. Full Figure and legend (94K)

ArsC plants show enhanced arsenate sensitivity.
Arabidopsis wild-type and three independent SRS1p/ArsC transgenic lines with low (SRS1p/ArsC2), medium (SRS1p/ArsC7), and high (SRS1p/ArsC9) levels of ArsC protein expression were tested for arsenic resistance or sensitivity by germination on medium containing 0, 75, and 150 M arsenate. All three ArsC transgenic lines were hypersensitive to arsenic, whereas wild-type plants remained relatively healthy at these concentrations, as shown for SRS1p/ArsC9 in Figure 3B and C. Leaves of transgenic plants grown on 75 M arsenate developed more slowly than wild-type controls and turned yellow. When challenged with 150 M arsenate, the leaves expressing the SRS1p/ArsC transgene were extremely stunted, whereas root growth was not significantly inhibited, as shown in Figure 3C (right panel). All of the ArsC plants were extremely chlorotic after three weeks on 150 M arsenate. Although wild-type plants grew at reduced rates, they all survived at these two concentrations of arsenate and had two- to threefold higher fresh weights than the transgenic ArsC plants after three weeks of growth (Fig. 4A). There was no significant difference in fresh weight between transgenic lines and wild type when grown on medium not supplemented with arsenate. The sensitivity of the SRS1p/ArsC transgenic leaves to arsenate suggests that the bacterial arsenate reductase arsC gene is functional and, considering its activity in bacteria, it makes leaves more sensitive to arsenate (AsO₄^3-) by electrochemically reducing it to the more toxic thiol-reactive form arsenite (AsO₃^3-)¹⁴. The control 35Sp/ArsC lines expressing ArsC constitutively were also more sensitive to arsenate than wild type, but generally less sensitive than the SRS1p/ArsC lines (not shown).

Figure 4. Relative growth inhibition and arsenic speciation of ArsC-overexpressing plants.
	(A) Comparative growth inhibition of three SRS1p/ArsC lines (SRS1p/ArsC2, SRS1p/ArsC7, and SRS1p/ArsC9) and the wild-type plants grown on the indicated concentrations (the average and s.e. values of three replicates of 30 seedlings for each line). (B) The percentage relative concentration of free AsO4^3-, AsO3^3-, and As(III) tris-glutathione in leaf samples as determined by XANES at the Stanford Synchrotron Radiation Laboratory (SSRL) following the methods described by Pickering et al^9. Leaf tissues of wild-type plants and the four SRS1p/ArsC Arabidopsis lines 2, 7, 8, and 9 grown for three weeks on 75 M arsenate were examined. Full Figure and legend (35K)

Plants coexpressing ArsC and -ECS.
Another system used by bacteria and other organisms to resist thiol-reactive ions like arsenite involves increasing the thiol-rich peptide content of cells and tissues. In particular, the enzymes in the pathway leading to the biosynthesis of glutathione (GSH), such as -glutamylcysteine synthetase (-ECS), shown in Figure 1, play a role in metal ions resistance. For example, arsenite-resistant Leishmania often show amplification of the -ECS locus along with other arsenic resistance genes¹⁵. Related work has shown that transgenic plants expressing bacterial -ECS contained higher levels of both peptides -EC (-glutamylcysteine) and GSH (glutathione) and accumulate more of the thiol-reactive metal ion Cd(ii) than wild-type plants^{16, 17}. Building on this work for cadmium, we set out to test the second part of our working hypothesis, enhancing the arsenite-binding thiol−peptide sink in plants by expressing -ECS.

It was recently shown that plants expressing the bacterial -ECS gene under the control of a strong constitutive actin (ACT2) gene promoter (ACT2p/-ECS) are moderately resistant to mercury and arsenic (Y. Li, O.P. Dhankher, and R.B. Meagher, manuscript in preparation). We have repeated these results for arsenate, selecting for the kanamycin resistance marker on the ACT2p/-ECS construct. One of the strong arsenate-resistant lines, ACT2p/ECS1, was selected as a control for further experiments. When the ACT2p/ECS1 line was grown on medium containing 200 M arsenate, these plants grew substantially better than wild type, as shown in Figure 5B (left lower quadrant on each plate).

Figure 5. Arsenic resistance of plants expressing ArsC9 and -ECS.
	(A) Selection of SRS1p/ArsC9 (ArsC9) transgenic plants retransformed with ACT2p/