A molecular switch in sulfur metabolism to reduce arsenic and enrich selenium in rice grain AbstractRice grains typically contain high levels of toxic arsenic but low levels of the essential micronutrient selenium. Anthropogenic arsenic contamination of paddy soils exacerbates arsenic toxicity in rice crops resulting in substantial yield losses. Here, we report the identification of the gain-of-function arsenite tolerant 1 (astol1) mutant of rice that benefits from enhanced sulfur and selenium assimilation, arsenic tolerance, and decreased arsenic accumulation in grains. The astol1 mutation promotes the physical interaction of the chloroplast-localized O-acetylserine (thiol) lyase protein with its interaction partner serine-acetyltransferase in the cysteine synthase complex. Activation of the serine-acetyltransferase in this complex promotes the uptake of sulfate and selenium and enhances the production of cysteine, glutathione, and phytochelatins, resulting in increased tolerance and decreased translocation of arsenic to grains. Our findings uncover the pivotal sensing-function of the cysteine synthase complex in plastids for optimizing stress resilience and grain quality by regulating a fundamental macronutrient assimilation pathway. IntroductionRice is the dominant source of the carcinogenic arsenic (As), accounting for around 60% of the dietary intake of inorganic As for some Asian populations1,2,3,4. Widespread contamination of paddy soils with As due to mining activities and irrigation of As-laden groundwater not only increases As accumulation in rice grain5,6, but also results in As toxicity in rice plants and substantial yield losses7,8. Worryingly, future climate changes are likely to exacerbate both the problems of As accumulation in rice grain and yield loss induced by As toxicity9. It is therefore important to develop rice crops that are more resistant to As and accumulate less As in the grain.A key mechanism of As detoxification in plants is via the complexation of arsenite [As(III)] by thiol compounds, such as glutathione (GSH) and phytochelatins (PCs), and subsequent sequestration into the vacuoles10,11,12,13,14. This mechanism also reduces As translocation to rice grain12,15,16,17. The metal(loid) stress-induced production of GSH and PCs is limited by the availability of its precursor cysteine (Cys) in plants18,19,20. Cys is synthesized in all subcellular compartments with own protein biogenesis by a two-step pathway: the formation of O-acetylserine (OAS) from serine and acetyl CoA catalyzed by serine acetyltransferase (SAT) and the condensation of OAS and sulfide to Cys catalyzed by O-acetylserine(thiol)lyase (OAS-TL)21,22. This two-step pathway is conserved in bacteria, alga, and higher plants, and serves as a checkpoint for the integration of sulfur, carbon, and nitrogen. The enzymatic functions of three isoforms of OAS-TL (OAS-TL A, B, and C in the cytosol, plastids, and mitochondria, respectively) in Arabidopsis thaliana and a cytosolic OAS-TL A in rice have been characterized23,24,25,26,27. Knockout of each of the three isoforms of OAS-TL in Arabidopsis did not affect growth, because the substrates for Cys biosynthesis can be exchanged efficiently between the cytosol and the organelles25. SAT and OAS-TL form the hetero-oligomeric cysteine synthase complex (CSC), and SAT is active only within this complex21,22. The product of SAT, OAS, has a dissociation effect on the CSC22,28,29, thus providing a feedback inhibition on SAT activity. Synthesis of OAS by SAT is also the rate-limiting step of Cys biosynthesis, whereas the total activity of OAS-TL is in excess of that required for Cys biosynthesis22. The association of the CSC stimulates the SAT activity and thus provides a mechanism for metabolic regulation of Cys biosynthesis by sensing the supply of the carbon/nitrogen-containing precursor OAS and the sulfur precursor sulfide.In this study, we isolated a gain-of-function arsenite tolerant 1 (astol1) rice mutant. We show that ASTOL1 encodes the chloroplast-localized OAS-TL. Surprisingly, the astol1 gain-of-function mutation (S189N) inactivates the catalytic activity of OAS-TL by inhibiting OAS binding. We provide direct evidence that the inactive ASTOL1 protein forms a non-dissociable CSC, which permanently stimulates endogenous SAT activity explaining the dominant inheritance of the astol1 mutation. Uncoupling of the OAS-sensing function of the plastidic CSC from its regulatory impact on SAT resulted in (1) enhanced S uptake and assimilation, (2) increased tolerance to As and (3) decreased As accumulation in rice grain. Moreover, the OsASTOL1S189N聽mutation significantly enhanced the rice grain content of selenium (Se), an essential micronutrient for which up to a聽billion people worldwide have insufficient dietary intake30,31. Besides the potential biotechnological applications, our findings uncover the pivotal sensing-function of the plastid CSC in the regulation of S and Se metabolism and tolerance to metalloid toxicity.ResultsScreening and characterization of the astol1 mutantWe screened more than 4000 individuals derived from an EMS-mutagenized library of the rice cv. Kasalath for arsenite [As(III)] tolerance using a root elongation assay at 20鈥壜礛 As(III) (approximately the half-inhibitory concentration, Supplementary Fig.聽1a). This screen revealed only one As(III) tolerant mutant (Supplementary Fig.聽1b) that we named arsenite tolerant 1 (astol1).The root growth of astol1 was indistinguishable from the wild type (WT) in the absence of As(III). However, upon As(III) treatment astol1 grew 2-times longer roots than WT (Fig.聽1a, b). When grown hydroponically in a growth chamber (light intensity ~300鈥壩糾ol m鈭? s鈭?), astol1 segregated into two phenotypes, one with decreased growth compared with WT and the other dying of leaves from tips inwards leading to eventual plant death (Fig.聽1c, d, and Supplementary Fig.聽1c鈥揻). The two phenotypes were later confirmed to represent the heterozygous (astol1(+/鈭?) and homozygous (astol1(+/+)) mutant, respectively, based on genotyping of the causal gene. Thus, we used astol1(+/鈭? in most of the experiments. The root elongation assay also revealed that astol1(+/+) was slightly more tolerant to As(III) than astol1(+/鈭? (Fig.聽1h). Moreover, the roots of astol1(+/鈭? and astol1(+/+) were more tolerant to arsenate (As(V)), compared to WT (Supplementary Fig.聽2a, b). Upon As(III) or As(V) treatment, the shoot As concentration of astol1(+/鈭? was not significantly different from WT (Supplementary Fig.聽2c, d). However, astol1(+/鈭? accumulated significantly more As in the roots than WT (Supplementary Fig.聽2c, d), implying that the increased As tolerance in astol1 is not caused by a decreased As uptake.Fig. 1: Phenotypes of astol1 mutant and MutMap-based cloning of OsASTOL1.a The phenotype of WT and astol1 mutant treated with 0 or 20鈥壩糓 As(III) for 2 days. Three plants for each genotype are shown. Scale bars, 20鈥塵m. b Relative root elongation of WT and astol1 mutant treated with 0 or 20鈥壩糓 As(III) for 2 days. Asterisks indicate significant difference by two-sided Student鈥檚 t test: ***P鈥?lt;鈥?.001. c Growth phenotype of 5-week-old WT, astol1(+/鈭? and astol1(+/+) mutant grown hydroponically. Two plants for each genotype are shown. Scale bars, 10鈥塩m. d Biomass of 5-week-old WT, astol1(+/鈭? and astol1(+/+) mutant grown hydroponically. e Wild-type rice and astol1 mutants were grown in a paddy field until maturity. f Chromosomal location and gene structure of OsASTOL1. White and black boxes represent untranslated regions and exons, respectively, and the lines between them represent introns. Red line indicates the mutation site. g Validation of knock-out OsASTOL1/ OsASTOL1S189N protein in four independent CRISPR lines in the astol1(+/鈭? background using immunoblot analysis. OsASTOL1 protein (lower band) is detected by AtOAS-TL A1 antibody (anti-OAS-TLA) and the Rubisco large subunit is used as a loading control. The experiment was repeated three times and similar results were obtained. h Relative root elongation of WT, astol1(+/鈭?, astol1(+/+), and four independent CRISPR lines in the astol1(+/+) background. Plants were treated with 20鈥壩糓 As(III) for 2 days. i Biomass of 5-week-old WT, astol1(+/鈭?, astol1(+/+), and CRISPR lines in astol1(+/鈭? or astol1(+/+) background. DW, dry weight. WT, wild type. astol1(+/鈭?, astol1 heterozygote. astol1(+/+), astol1 homozygote. Data in b, d, h, and i are shown as means鈥壜扁€塻.d., n鈥?鈥?0 for WT (control) and astol1 (As(III) treatment), 13 for astol1 (control) and WT (As(III) treatment) (b), 5 (d), 6 for astol1(+/+), 10 for WT and all CRISPR lines/astol1(+/+), 19 for astol1(+/鈭? (h) or 3 (i) biological replicates; each biological replicate represents an individual plant. Different letters in d, h, and i indicate significant differences (P鈥?lt;鈥?.05) using one-way ANOVA followed by Tukey鈥檚 test.Full size imageWe then investigated the relationship between the phenotypes of growth inhibition observed in astol1 and the light intensity in a growth chamber. Increasing light intensity from ~300 to ~800鈥壩糾ol m鈭? s鈭? increased the growth of both WT and astol1(+/鈭?, and decreased the difference in plant biomass between the two genotypes from 63% to 31%, but did not improve the growth of astol1(+/+) (Supplementary Fig.聽3a, b). When grown in an open paddy field, where light intensity often exceeds 1000鈥壩糾ol m鈭? s鈭?, astol1(+/鈭? was indistinguishable from WT in all agronomic traits including grain yield (Fig.聽1e, Supplementary Fig.聽4a鈥揻). Remarkably, in open paddy fields astol1(+/+) also grew to maturity but produced smaller grain yield due to a low percentage of filled spikelets. These results indicate that the slow growth phenotype of astol1 can be reversed under normal light conditions in paddy fields. OsASTOL1 encodes a chloroplast-localized OAS-TL proteinTo clone the causal gene for the astol1 phenotype, we backcrossed astol1(+/鈭? (paternal) with WT (maternal). Among the 30 F1 plants, 40% showed the As(III) tolerance phenotype (not significantly different from the expected 50%), indicating that the mutation in OsASTOL1 has a dominant inheritance for As(III) tolerance. In the F2 progenies a segregation pattern with a ratio of 58 WT: 166 astol1(+/鈭? :15 astol1(+/+) was obtained, suggesting a semi-dominant effect of the mutation. However, the determined ratio was inconsistent with the expected segregation ratio of 1: 2: 1 (蠂2鈥?鈥?1.66鈥?gt;鈥壪?sup>20.05,2), likely due to partial sterility and poor seed germination of the homozygous mutant.We conducted genomic resequencing mapping and MutMap analysis32 on the backcrossed F2 progeny with astol1(+/+) phenotype. One SNP (SNP-26700810) was identified with a SNP index of 1 (Supplementary Fig.聽5). This SNP was located in the fifth exon of the gene (Os12g0625000) encoding a putative OAS-TL that leads to a Ser189Asn mutation (codon AGC鈥夆啋鈥堿AC, Fig.聽1f). The point mutation in this OsASTOL1 gene was verified by DNA sequencing and a derived cleaved amplified polymorphic sequence (dCAPS) marker (Supplementary Fig.聽6a, c, d).Since astol1 is a semi-dominant mutant, we generated transgenic plants expressing the mutant Osastol1 gene under the control of the maize Ubiquitin promoter in WT. When grown in an open paddy field, all transformed T0 individual lines (n鈥?鈥?) that over-expressed the mutant聽OsASTOL1S189N protein produced sterile seeds (Supplementary Fig.聽7). This finding supports our previous observation that astol1(+/+) showed a low percentage of filled spikelets in the field and induces plant death when grown in growth chambers (Fig.聽1d, Supplementary Fig.聽4d). To provide direct evidence that mutated OsASTOL1 is responsible for the astol1 phenotypes, we knocked out the mutant Osastol1 gene in astol1 using the CRISPR-Cas9 gene-editing method (Supplementary Fig.聽6b). Four independent T0 CRISPR-induced knockout lines in the astol1(+/鈭? background with both OsASTOL1 and Osastol1 being homozygously knocked out were obtained and confirmed by immunoblot using an antibody against the full-length Arabidopsis OAS-TL A (Lower band in Fig.聽1g, with the upper band likely to be another isoform of OAS-TL). In the T1 generation of these lines, three types of plants including homozygous knockout of OsASTOL1 and/or Osastol1 in the backgrounds of WT, astol1(+/鈭? or astol1(+/+) were isolated by using the dCAPS marker (Supplementary Fig.聽6d). The relative root elongation of all CRISPR lines in the astol1(+/+) background showed the same As(III) sensitive phenotype as WT (Fig.聽1h). Furthermore, all CRISPR lines in both the astol1(+/鈭? and astol1(+/+) backgrounds showed a complete rescue of the astol1 mutation-induced phenotypes in growth and root As concentration (Fig.聽1i and Supplementary Fig.聽6c鈥揺). These results demonstrate that the S189N point mutation in the OsASTOL1 protein is the cause for the astol1 phenotypes.OsASTOL1 shares high homology with Arabidopsis cytosolic OAS-TL A (Supplementary Fig.聽8). However, in contrast to AtOAS-TL A, OsASTOL1 contains 87 extra amino acid residues at the N-terminus (Supplementary Fig.聽9), which are predicted to generate a chloroplastidic transitpeptide33 (Supplementary Table聽2). OsASTOL1 is predominately expressed in photosynthetic tissues (Supplementary Fig.聽10a鈥揹). When transiently expressed in tobacco leaves and rice protoplasts, both wild-type OsASTOL1-YFP and mutant OsASTOL1S189N-YFP fusion proteins localize in the chloroplasts (Fig.聽2a and Supplementary Fig.聽10e). These findings demonstrate that the astol1 point mutation does not affect the chloroplastic localization of OsASTOL1 in rice.Fig. 2: OsASTOL1 encodes a chloroplast-localized O-acetylserine(thiol)lyase (OAS-TL).a Subcellular localization of OsASTOL1. Rice protoplasts were expressed with the eYFP (top), OsASTOL1-eYFP (middle), or Osastol1-eYFP (bottom) driven by the cauliflower mosaic virus 35S promoter. Left to right: YFP fluorescence, chlorophyll autofluorescence, bright-field images, and merged images. Scale bars, 10鈥壜祄. At least five protoplasts were investigated with similar results to the images shown in a. b Validation of OsASTOL1/ OsASTOL1S189N protein (lower band) levels in 3-week-old WT, astol1(+/鈭?, and astol1(+/+) using immunoblot analysis. OsASTOL1/ OsASTOL1S189N protein levels are detected by AtOAS-TL A1 antibody (anti-OAS-TL聽A) and Rubisco large subunit is used as a loading control. The experiment was repeated two times and similar results were obtained. c Total OAS-TL enzyme activity in whole soluble protein extracts of the shoots of 3-week-old WT, astol1(+/鈭?, and astol1(+/+). Data are shown as means鈥壜扁€塻.d., n鈥?鈥? biological replicates; each biological replicate represents an individual plant. d Expression of recombinant mature OAS-TL proteins in E. coli. Proteins were separated on SDS-PAGE and visualized by Coomassie Brilliant Blue (CBB) staining (upper) and by immunoblot with anti-His antibody (bottom). M, marker. The experiment was repeated two times and similar results were obtained. e In vitro OAS-TL enzyme activity of purified mature OsASTOL1 and AtOAS-TL A and their corresponding mutant proteins. Data are shown as means鈥壜扁€塻.d., n鈥?鈥? technical replicates. WT, wild type. astol1(+/鈭?, astol1 heterozygote. astol1(+/+), astol1 homozygote. Asterisks in c, e indicate significant difference by two-sided Student鈥檚 t test: **P鈥?lt;鈥?.01, ***P鈥?lt;鈥?.001.Full size imageSer189 of OsASTOL1 is crucial for OAS-TL activityTo test whether the OsASTOL1S189N mutation affects OAS-TL activity in planta, we determined the total OAS-TL activity and OAS-TL protein abundance in the WT and the astol1 mutant. Remarkably, the abundance of the mutated OsASTOL1S189N was unaffected in astol1 (+/+), but the total extractable OAS-TL activity was decreased by 32% in astol1(+/+) when compared to WT (Fig.聽2b, c). These results suggest that Ser189Asn mutation does not destabilize the OsASTOL1S189N聽protein but impairs its enzymatic activity.OAS-TL proteins are highly conserved in prokaryotes and eukaryotes (Supplementary Figs.聽9 and 11). In particular, the Ser189 of OsASTOL1 is conserved in all analyzed OAS-TLs (Supplementary Fig.聽9). In order to compare the OAS-TL activity of the mature OsASTOL1 and the mutated OsASTOL1S189N, we deleted the 87 amino acid residues encoding the transitpeptide and expressed them in E. coli. Because AtOAS-TL A has been well characterized in enzymatic analysis23,26 and is highly homologous to OsASTOL1 protein (Supplementary Fig.聽8), it was applied as a positive control for a canonical OAS-TL enzyme (Fig.聽2d). Expression of AtOAS-TL A in E. coli Cys- auxotrophic strain NK334, which lacks the cysteine synthase gene and consequently is unable to grow on medium without cysteine supplement, restored its growth in Cys-free medium within 2 days of incubation (Supplementary Fig.聽12a). In comparison, expression of OsASTOL1 restored the growth of NK3 strain in Cys-free medium after 7 days of incubation, whereas OsASTOL1S189N聽failed to complement NK3 strain (Supplementary Fig.聽12a). Moreover, the purified mature OsASTOL1 protein clearly displayed OAS-TL activity, but its specific activity was only about 1/4 of AtOAS-TL A (Fig.聽2e). In contrast, the mutated OsASTOL1S189N displayed no OAS-TL activity (Fig.聽2e). This is consistent with the complementation assay in NK3 (Supplementary Fig.聽12a). Moreover, size exclusion chromatography (SEC) analysis shows that both the OsASTOL1 and OsASTOL1S189N were present as dimers as other OAS-TLs (Supplementary Fig.聽12b). Furthermore, the replacement of the corresponding Ser102 in AtOAS-TL A with Asn fully inactivated the OAS-TL AS102N mutant protein (Fig.聽2e). The protein crystal structure35 of AtOAS-TL A indicates that Ser102 is located at the opening of the binding site for the substrate OAS (Supplementary Fig.聽13a, b).To investigate the role of this conserved Ser, we changed the Ser189 of OsASTOL1 to different amino acid residues by site-directed mutagenesis and determined the in vitro OAS-TL activities of the respective protein mutants (Supplementary Fig.聽14a, b). All mutation alleles lost their enzymatic activity, except OsASTOL1S189T that displayed 41% of the ASTOL1 activity (Supplementary Fig.聽14b). Because both the mutation alleles of S189A (a constitutively nonphosphorylable form) and S189D (a mimic of the constitutively phosphorylated form) lost the OAS-TL activity completely, it is unlikely that phosphorylation of S189 is required for the enzyme activity. Thus, we conclude that the conserved Ser189 in OsASTOL1 is crucial for OAS-TL activity, and mutation of this residue in OsASTOL1S189N leads to a total loss of OAS-TL activity.Loss of OAS-TL activity is irrelevant to astol1 phenotypeTo investigate the causal relationship between the phenotypes of the astol1 mutation and the OAS-TL activity in the mutant, we generated CRISPR lines and overexpression lines of OsASTOL1 in the WT background. The CRISPR-induced loss-of-ASTOL1 function resulted in three independent lines in a decrease to 50鈥?3% of the total in vivo OAS-TL activity when compared to WT (Supplementary Fig.聽15a, b). In contrast, overexpression of wild type ASTOL1 protein caused in three independent lines a 3-fold increase of OAS-TL activity when compared to WT (Supplementary Fig.聽15b). Neither loss-of-ASTOL1 function nor the overexpression of ASTOL1 mimicked the As(III) tolerance or the decreased growth phenotype of astol1 (Supplementary Fig.聽15c, d). These results suggest that the astol1 phenotypes are not caused by the loss of OAS-TL activity. This finding is consistent with the semi-dominant inheritance of the astol1 mutation. The results are also consistent with previous studies on Arabidopsis that knockout of each of the three major isoforms of OAS-TL did not impair growth25.Stabilized CSC leads to higher SAT activity in astol1 mutantIn plants, Cys biosynthesis is not limited by OAS-TL, but by the synthesis of its substrate OAS25,26,36. OAS is exclusively produced by SAT in plants. Remarkably, SATs are almost not regulated at the transcriptional level. The dominant mechanism for the regulation of SAT activity is the formation of the CSC complex. CSC formation significantly activates SAT and is controlled by the actual OAS supply22. We applied the well-characterized AtSAT5 protein as the bait protein to test whether Ser189Asn mutation in the mutant OsASTOL1S189N protein affects its association with SAT protein in an affinity column-based pull-down experiment29,37. Both OsASTOL1 and OsASTOL1S189N proteins interact with the His-tagged AtSAT5 protein (Supplementary Fig.聽16a, b). Similarly, AtOAS-TL A and AtOAS-TL AS102N bound to the column immobilized His-tagged AtSAT5 (Supplementary Fig.聽16c, d). Thus, mutation of the conserved Ser to Asn did not affect the formation of CSC.Next, we tested the ability of OAS to dissociate the recombinant CSCs consisting of His-SAT5 and the four candidate OAS-TLs. As expected, the CSC containing the wild type OsASTOL1 and the wild type AtOAS-TL A were dissociated by the application of OAS. In contrast, OAS failed to dissociate the CSC consisting of the OsASTOL1S189N and the AtOAS-TL AS102N protein (Fig.聽3a鈥揷, and Supplementary Fig.聽16e鈥揼). Since OAS-induced dissociation of the CSC is triggered by competition between OAS and the SAT C-terminus for binding to the active site of OAS-TL38,39, we quantified the affinity of wild type OsASTOL1 and the OsASTOL1S189N for OAS and the C-terminal fragment of SAT5 (AtSAT5C10) by microscale thermophoresis (MST). As expected, wild type ASTOL1 displayed high affinity towards OAS and AtSAT5C10 (Fig.聽3d, e and Supplementary Table聽3). The mutated OsASTOL1S189N physically interacted with the AtSAT5C10 fragment with a wild-type like affinity, but specifically lost the affinity towards OAS (Fig.聽3d, e and Supplementary Table聽3). Mutation of S102N in AtOAS-TL A resulted in the same specific loss of affinity towards OAS but did not affect the binding of AtSAT5C10 (Supplementary Fig.聽16h, i and Supplementary Table聽3). These biochemical properties of the mutated OAS-TLs provide a molecular explanation for the formation of an OAS-resistant CSC by the mutated OsASTOL1S189N and AtOAS-TL AS102N proteins and the catalytic inactivity of the both proteins.Fig. 3: astol1 mutant increases SAT activity by stabilizing cysteine synthase complex (CSC).a In vitro pull-down analysis of the dissociation effect of OAS on CSC. The fraction of different elution steps (lanes 1鈥?) from His-AtSAT5 protein (upper panel), His-AtSAT5 and mature OsASTOL1 protein (middle panel), and His-AtSAT5 and mature OsASTOL1S189N protein (bottom panel) was analyzed by SDS-PAGE and visualized by Coomassie Brilliant Blue staining. M marker; lane 1: crude extract of His-AtSAT5 (upper), OsASTOL1 (middle) and OsASTOL1S189N (bottom); lane 2: flow through of His-AtSAT5 (upper), OsASTOL1 (middle) and OsASTOL1S189N (bottom); lane 3: washing buffer (80鈥塵M imidazol); lane 4: OAS elution (10鈥塵M OAS鈥?鈥?0鈥塵M imidazol); lane 5: washing buffer (80鈥塵M imidazol); lane 6: final elution (400鈥塵M imidazol). The experiment was repeated three times and similar results were obtained. b The fraction of OAS elution and final elution was analyzed by SDS-PAGE and visualized by immunoblot, and a representative result of three repeated tests is shown. c The relative abundance of OsASTOL1 or OsASTOL1S189N in the OAS elution and the final elution was quantified by ImageJ software, compared to the corresponding AtSAT5 protein. Microscale thermophoresis (MST) analysis of in vitro binding of OsASTOL1 and OsASTOL1S189N protein with OAS (d) or AtSAT5C10 peptide (e). f Total SAT enzyme activity in whole soluble protein extracts of the shoots of 3-week-old WT and astol1(+/鈭?. g Relative expression levels of two rice SAT genes (Os01g0720700 and Os05g0533500) in the shoots of 3-week-old WT and astol1(+/鈭?, with OsHistone H3 as the internal reference. WT, wild type. astol1(+/鈭?, astol1 heterozygote. Data in c鈥?b>e are shown as mean鈥壜扁€塻.d., n鈥?鈥? technical replicates. Data in f鈥?b>g are shown as means鈥壜扁€塻.d., n鈥?鈥? (f) or 3 (g) biological replicates; each biological replicate represents an individual plant. Asterisks in c, f, g indicate significant difference by two-sided Student鈥檚 t test: **P鈥?lt;鈥?.01, ***P鈥?lt;鈥?.001.Full size imageTo test whether the endogenous SAT activity was stimulated by the presence of mutated OsASTOL1S189N protein in astol1, we compared SAT activities in crude plant protein extracts of WT and the astol1(+/鈭? mutant. The extractable SAT activity of astol1(+/鈭? was 2.8-fold higher than that in WT (Fig.聽3f), although transcript abundance of SAT genes was indistinguishable between astol1(+/鈭? and WT (Fig.聽3g). Taken together, these results imply that the Ser189Asn mutation in OsASTOL1S189N increases the stability of the CSC against the dissociation of OAS, resulting in significantly enhanced endogenous SAT activity in planta. astol1 mutant overaccumulates OAS and S metabolitesAs a result of the increased foliar SAT activity, the total steady-state level of the SAT product, OAS, Cys and most of the Cys-derived sulfur-metabolites tested were significantly higher in leaves of astol1(+/鈭? when compared to WT. Remarkably, the astol1 mutation-induced increase of OAS and sulfur-containing metabolites were also detected in roots of non-As(III)-stressed plants, suggesting that SAT activity is globally enhanced by the astol1 mutation (Fig.聽4a, b and Supplementary Table聽4). The astol1 mutation also caused decreased steady-state level of the SAT substrate, serine, in leaves and roots, suggesting a significant higher conversion of Ser into OAS in these organs (Fig.聽4a, b and Supplementary Table聽4).Fig. 4: astol1 mutant enhances S and Se uptake and phytochelatins-dependent As detoxification.a Metabolic pathway from serine to cysteine. b Relative fold change of S-related metabolites in astol1(+/鈭? compared to WT, visualized as the heat map (n鈥?鈥? biological replicates for sulfate, three biological replicates for all other metabolites; each biological replicate represents an individual plant). Abbreviations of metabolites: SO42鈭?/sup>sulfate, Ser serine, OAS O-acetylserine, Cys cysteine, Cyst cystathionine, GSH glutathione, Hcy homocysteine, Met methionine, SAH S-adenosylhomocysteine, SAM S-adenosylmethionine, MTA methylthioadenosine. c Relative expression levels of rice genes involved in sulfate uptake and reduction in the roots and shoots of 3-week-old WT and astol1(+/鈭?. Transcriptional induction of selected rice genes (OsSULTR1;1 (Os03g0195800) in roots, OsAPR1 (Os07g0509800) and OsSiR (Os05g0503300) in roots and shoots) was measured by Q-PCR, with OsHistone H3 as the internal reference gene. d S concentrations in roots and shoots of 5-week-old WT and astol1(+/鈭?. e Se concentrations in roots and shoots of 5-week-old WT and astol1(+/鈭? exposed to 2鈥壩糓 Se(VI) for 3 days. The concentrations of non-protein thiols in roots (f) and shoots (g) of 4-week-old WT and astol1(+/鈭? treated with 5鈥壩糓 As(III) for 3 days. DW, dry weight. FW, fresh weight. WT, wild type. astol1(+/鈭?, astol1 heterozygote. Data in c鈥?b>g are shown as means鈥壜扁€塻.d. n鈥?鈥? (c, d), 4 (e) or 5 (f, g) biological replicates; each biological replicate represents an individual plant. Asterisks in b鈥?b>g indicate significant differences by two-sided Student鈥檚 t test: *P鈥?lt;鈥?.05, **P鈥?lt;鈥?.01, ***P鈥?lt;鈥?.001.Full size imageOAS acts as the signal of sulfur deficiency and induces the expression of genes involved in S uptake and assimilation21,40,41. The transcript levels of several genes involved in sulfate uptake and reduction, including root plasma membrane-localized high-affinity sulfate transporter OsSULTR1;1, were substantially upregulated in astol1 when compared with WT (Fig.聽4c). Consistently, the astol1(+/鈭? mutant accumulated more sulfate, total S and Se (Fig.聽4b, d, e); the latter is a chemical analog of S and is also taken up by SULTRs. Moreover, knockout of Osastol1 in the astol1 background reduced the shoot S concentration to the WT level, whereas knockout of wild-type OsASTOL1 did not significantly affect shoot S concentration (Supplementary Figs.聽6f and 15e), further supporting that the specific mutation in Osastol1 leads to the mutant phenotypes in astol1. Taken together, these results demonstrate that substantial accumulation of OAS leads to increased sulfate uptake and assimilation into Cys in the astol1 mutant.To explain the OsASTOL1S189N-induced gain of As(III) tolerance, we quantified the impact of As(III) on thiol compounds in the roots and shoots of WT and astol1(+/鈭?. Upon 5鈥壜礛 As(III) treatment, the steady-state levels of Cys, and the As(III) chelators GSH, PC2, PC3, and PC4 were all significantly increased in the roots of astol1(+/鈭? when compared with WT (Fig.聽4f), demonstrating the enhanced capacity of the astol1 plants for complexation of As(III). The levels of Cys and PC2 in the shoots of astol1(+/鈭? were also increased significantly (Fig.聽4g), although the levels of both thiols were lower than those in the roots. Furthermore, the transcript levels of OAS-TL and SAT genes in shoots and roots of WT and astol1(+/鈭? showed no significant change when treated with As(III) (Supplementary Fig.聽17a鈥揹), suggesting that these genes are not transcriptionally induced by As(III) stress.OsASTOL1S189N reduces As and increases Se in rice grainTo evaluate the biotechnological capability of the astol1 mutation for improvement of the nutritional value of rice grain, we quantified the yield and accumulation of As, S, and Se in the grain of rice plants grown on paddy fields with background levels of As and Se in the soils. Grain As concentration of astol1(+/鈭? was 29% and 33% lower than that of WT, respectively, in the two field experiments (Fig.聽5a and Supplementary Fig.聽18a). In contrast, astol1(+/鈭? accumulated 77% and 107% higher Se concentration in rice grain than WT, respectively, in the two field experiments (Fig.聽5c and Supplementary Fig.聽18b). We also determined As concentrations in different shoot tissues of astol1(+/鈭? and WT sampled at plant maturity from the two field sites. The astol1(+/鈭? mutant contained significantly higher As concentrations in node I and internode than WT at one site, but a significantly lower As concentration in flag leaf blade at the other site (Supplementary Fig.聽19a, b).Fig. 5: Mutation allele of OsASTOL1 reduces As and increases S and Se concentrations in rice grain. Grain As (a), S (b), and Se (c) concentrations of WT, astol1(+/鈭?, and three independent OsASTOL1聽overexpression lines in the astol1(+/鈭? or astol1(+/+) background. Plants were grown in a paddy field of Nanjing (12.5鈥塵g鈥塳g鈭? As and 0.61鈥塵g鈥塳g鈭? Se in the soil). d鈥?b>f Yield components of WT, astol1(+/鈭?, and three overexpression lines in astol1(+/鈭? or astol1(+/+) background. After harvest, filled spikelets (d), 1000-grain weight (e), and grain yield (f) were recorded. DW, dry weight. WT, wild type. astol1(+/鈭?, astol1 heterozygote. astol1(+/+), astol1 homozygote. Data in a鈥?b>f are shown as means鈥壜扁€塻.d., n鈥?鈥? (a鈥?b>c) or 6 (d鈥?b>f) biological replicates; each biological replicate represents an individual plant. Different letters in a鈥?b>f indicate significant differences (P鈥?lt;鈥?.05) using one-way ANOVA followed by Tukey鈥檚 test.Full size imageTo overcome the growth inhibition in astol1(+/+), we overexpressed wild-type OsASTOL1 in astol1(+/鈭? mutant and obtained overexpression lines in the WT, astol1(+/鈭?, or astol1(+/+) backgrounds among the segregated T1 plants. The overexpression lines in the astol1(+/+) background rescued the early leaf death phenotype of astol1(+/+) and also improved the growth of astol1(+/鈭? under low light conditions (Supplementary Fig.聽20a鈥揻). Notably, the overexpression lines in both the backgrounds of astol1(+/鈭? and astol1(+/+) still accumulated higher S concentrations in the shoots and higher As concentrations in the roots than WT (Supplementary Fig.聽20g鈥搇). Moreover, when grown in a paddy field, overexpression of wild-type OsASTOL1 in both the astol1(+/鈭? and astol1(+/+) backgrounds restored their growth and agronomic traits to the levels of WT (Fig.聽5d, e and Supplementary Fig.聽21a鈥揺). Furthermore, the overexpression lines in both backgrounds of astol1(+/鈭? and astol1(+/+), like astol1(+/鈭?, had significantly lower grain As concentrations than WT, with decreases of 32鈥?9% and 23鈥?0%, respectively (Fig.聽5a). These lines accumulated significantly more S in the grain than WT (Fig.聽5b). In addition, the overexpression lines in the backgrounds of astol1(+/鈭? and astol1(+/+) accumulated 95鈥?21% and 15鈥?8% higher Se concentration in the grain than WT, respectively (Fig.聽5c). These results indicate that the Ser189Asn mutation in OsASTOL1S189N in rice enriches S and Se and reduces As accumulation in rice grain, while the growth inhibition in astol1(+/+) can be overcome by balancing the mutant and wild-type OsASTOL1 protein.DiscussionRice accumulates about 10-times higher As in the grain than other cereal crops due to the mobilization of As(III) in submerged paddy soils and the high capacity of rice roots for uptake of As(III)3,42,43. Consequently, many Asian populations are exposed to highly levels of dietary As. On the other hand, around 75% of the rice grains produced globally have insufficient levels of Se to meet human鈥檚 requirement44. In this study, we show that Ser189Asn mutation in OsASTOL1S189N serves as a molecular switch in the sulfur assimilation pathway to produce multiple desirable consequences, including increased As tolerance, decreased As accumulation in the grain and increased Se accumulation in the grain. These phenotypes can be explained by improved uptake and assimilation of sulfate and selenate, and the enhanced synthesis of thiol compounds. This reinforced capacity to synthesize PCs allows for improved complexation of As11,12,13,14, facilitates the vacuolar sequestration of As-thiol complexes and, consequently, reduces As mobility to rice grain15,16,17. Increased synthesis of thiol compounds is more pronounced in the roots than in the shoots of the mutant, which is consistent with roots being the main tissue for As sequestration. A radioactive 73As tracer experiment showed that approximately 90% of the arsenite taken up by rice roots is sequestered in the roots45. Furthermore, enhanced GSH and PC synthesis should also improve the resistance towards heavy metals like Cd and Hg as well as some biotic stresses18,46,47.We show that Ser189Asn mutation in OsASTOL1S189N results in a total loss of the OAS-TL activity but a gain-of-function of increased SAT activity, and it is the latter that explains the observed phenotypes of the astol1 mutant, as summarized in our model (Fig.聽6). The Ser189Asn mutation in OsASTOL1S189N specifically impairs its affinity towards the substrate OAS (Fig.聽3d), but surprisingly does not impact its physical interaction with the SAT5C10 peptide (Fig.聽3e). In fact, the CSC complex formed from OsASTOL1S189N and SAT becomes much more resistant to the dissociation of OAS (Fig.聽3a鈥揷), which may also result from the loss of binding of OAS to OsASTOL1S189N. The increased stability of the CSC complex leads to a significantly higher SAT activity in the astol1 mutants, because SAT is substantially activated in the CSC complex due to release from cysteine inhibition48,49. Because production of OAS is the rate-limiting step, increased OAS drives Cys biosynthesis catalyzed by other isoforms of OAS-TL either in the chloroplast or in the cytoplasm, as proposed in the model (Fig.聽6). Moreover, OAS is a signaling molecule that can induce the expression of genes involved in S uptake and assimilation21,40,41. Indeed, several genes in the sulfate uptake and reduction pathway聽were significantly upregulated in the astol1 mutants (Fig.聽4c). In addition to sulfate, many sulfur-containing metabolites are also increased in the astol1(+/鈭? mutant, indicating that sulfate reduction and assimilation are enhanced (Fig.聽4b). Increased biosynthesis of OAS, Cys, and other elevated thiol compounds in the astol1 mutants causes a higher demand for carbon. Consequently, norm light conditions in paddy fields helped the mutants to overcome the imposed challenge for the enhanced synthesis of the diverse carbon and sulfur-containing metabolites. The relatively poor growth of the mutants under low light intensity in growth rooms might be also partially explained by a misbalance of enhanced level of sulfur-containing defense compounds in the artificial absence of stresses, e.g., high light and pathogens, which demand sulfur-containing defense compounds. However, in open paddy fields the potentially defense-primed astol1(+/鈭? mutants performed as well as the wild type, since the wild type was also challenged with environmental cues inducing sulfur-defense metabolite synthesis.Fig. 6: Proposed working model for the regulation of OsASTOL1 in sulfur metabolism in rice.Schematic overview of enzymatic reactions (black arrows) and metabolites involved in the sulfur metabolism and its regulatory mechanism. In wild type rice, OAS is produced by cysteine synthase complex (CSC), which results in two effects: (i) induces expression of genes in sulfate uptake and assimilation (blue arrow) to enhance the supply of sulfide; (ii) dissociates the cysteine synthase complex (solid red line) to inactivate SAT. These effects control the flux of S towards cysteine synthesis in vivo. However, in the astol1 mutant, the mutated OsASTOL1S189N protein is difficult to be dissociated from the CSC by OAS (dotted red line), which results in SAT being constitutively active and producing more OAS. Increased OAS level enhances sulfate/selenate uptake and assimilatory reduction and arsenic detoxification (bigger size of black and blue arrows).Full size imageOur model (Fig.聽6) also explains the dosage effect of the mutant allele, as well as the alleviation effect on the mutant growth phenotypes when wild-type OsASTOL1 was overexpressed in the astol1 mutants (Fig.聽5 and Supplementary Fig.聽20), resulting from the balance in the competition between the mutant and wild-type OsASTOL1 for SAT. Taken together, the specific Ser189Asn mutation in OsASTOL1S189N enables it to act as a regulatory protein of Cys biosynthesis, even though the mutant protein is inactive as an OAS-TL enzyme. It is remarkable that a single amino acid mutation in OAS-TL can trigger such an overwhelming impact on the uptake and assimilation of the macronutrient sulfur, albeit plastidic OAS-TLs is dispensable25 (Supplementary Fig.聽15c鈥揺). The full conservation of Ser189 in all known plant OAS-TLs suggests that the herein identified mechanism to trigger S and Se accumulation in grains will also be applicable in other plant species. As a proof of concept, we generated an OsASTOL1S189N-like mutation in cytosolic AtOAS-TL A (AtOAS-TLAS102N), which, indeed, caused the identical biochemical properties as observed in the plastid-localized OsASTOL1S189N protein of rice (Fig.聽2e and Supplementary Fig.聽16e鈥搃).In the reference plant Arabidopsis thaliana, cysteine synthesis takes place in all subcellular compartments that are capable of protein biogenesis, which is the dominant sink for cysteine21. Remarkably, these subcellular compartments are also equipped with a CSC, which is supposed to regulate the OAS precursor biogenesis by controlling the SAT activity in these compartments. Since 80% of total SAT activity is localized in mitochondria50,51, the mitochondrial CSC is supposed to efficiently regulate cellular OAS synthesis in plants49. Accordingly, downregulation of mitochondrial SAT3 impairs growth of Arabidopsis52. However, expression of SAT in the cytosol of Arabidopsis and Nicotiana tabacum also resulted in enhanced cysteine formation, demonstrating that enhanced OAS production in these compartments also trigger net cysteine synthesis rate53,54. In particular, the plastid-localized CSC is supposed to integrate environmental cues to coordinate cysteine biosynthesis for GSH production to maintain redox homeostasis upon diverse stresses55,56,57. In contrast to OAS biosynthesis, the S-precursor for cysteine biosynthesis is exclusively generated in plastids in a photosynthesis-dependent manner21. The uncoupling of the OAS-sensing function of OAS-TL from its regulatory role inside the CSC stimulated sulfate uptake from the soil and its reductive assimilation in plastids significantly. It remains to be shown if the CSCs in other subcellular compartments contribute to the coordination of S-flux into cysteine with translation demand or have distinct features (e.g., coordination with supply of other macronutrients, or coordination with stress responses). The OsASTOL1S189N mutation provides a versatile tool to dissect these potential regulatory functions of the individual CSCs in each of these compartments by allowing specific engineering of CSCs at the subcellular level.In conclusion, our study demonstrates that S assimilation can be effectively controlled by engineering the sensitivity of the CSC against dissociation by OAS, highlighting the physiological importance of the regulatory function of the CSC. Triggering of S assimilation in rice has multiple beneficial impacts on the food security (sustainable growth) and the nutritional value of rice since it enhances the tolerance of rice towards arsenic and heavy metals, and positively affects Se聽content but limits As content of grains. The key to optimization of the beneficial traits is to adjust the balance between the wild type like and the OsASTOL1S189N-like OAS-TL protein, as too much-mutated protein had a deleterious effect on plant growth possibly through disturbance in the balance between C, N, and S metabolism. This goal can be achieved by expression of wild-type OsASTOL1 in the astol1(+/+) mutant as demonstrated in our study (Fig.聽5), or by the expression of mutant Osastol1 in wild-type background driven by a suitable promoter.MethodsPlant materials and growth conditionsRice (Oryza sativa cv. Kasalath) seeds were mutagenized by treating them with 0.8% ethyl methanesulfonate (EMS) for 12鈥塰. M2 seeds obtained from self-pollinated M1 plants were used for the screening of arsenite (As(III)) tolerant mutants based on root elongation at 20鈥壜礛 As(III). The M3 generation of the candidate astol1 mutant was used to verify the As(III) tolerant phenotype. In brief, the root length of the seedlings after germination was recorded as R0. Seedlings were transferred to deionized water for 24鈥塰 and the root length was recorded as R1. Seedlings were then transferred to deionized water or deionized water containing 20鈥壜礛 As(III) for 48鈥塰 and root length was recorded as R2. The relative root elongation (RRE) was calculated as (R2-R1)/(R1-R0)鈥壝椻€?00. Similarly, the RRE of astol1 was calculated when treated with 8鈥壜礛 arsenate (As(V)).For hydroponic experiments, rice plants were grown in 陆 Kimura B nutrient solution in a growth chamber with 65% relative humidity under 12鈥塰/12鈥塰 light/dark (light intensity ~鈥?00鈥壩糾ol m鈭? s鈭?) cycle at 28鈥壜癈/25鈥壜癈 day/night temperatures. The composition of half-strength Kimura B nutrient solution was: 0.09鈥塵M KH2PO4, 0.27鈥塵M MgSO4, 0.18鈥塵M (NH4)2SO4, 0.09鈥塵M KNO3, 0.18鈥塵M Ca(NO3)2, 3鈥壩糓 H3BO3, 0.5鈥壩糓 MnCl2, 1鈥壩糓 (NH4)6Mo7O24, 0.4鈥壩糓 ZnSO4, 0.2鈥壩糓 CuSO4, and 20鈥壩糓 Fe(III)-EDTA. The nutrient solution was renewed every 3 days. For field experiments, plants were grown in paddy fields in the experimental farm of Nanjing Agricultural University in Nanjing (Jiangsu province, June-October; soil As content 12.5鈥塵g鈥塳g鈭? and soil Se content 0.61鈥塵g鈥塳g鈭?) or Lingshui (Hainan province, January-May; soil As content 3.1鈥塵g鈥塳g鈭? and soil Se content 0.11鈥塵g鈥塳g鈭?).MutMap-based gene cloningTo clone the causal gene responsible for the astol1 mutant phenotype, the astol1 mutant as paternal was backcrossed with wild type (Kasalath) to generate F2 progeny. The F2 plants which showed tolerance to 20鈥壜礛 As(III) and the early leaf death phenotype were selected for DNA extraction for whole-genome resequencing using the MutMap method32. In brief, DNA was extracted from 42 F2 plants showing the As(III)-tolerant and early leaf death phenotype and 10 wild-type plants (Kasalath) and mixed in an equal ratio to prepare pooled genomic DNA of astol1 mutant and wild-type, respectively. Each of the two pooled DNA samples was used for the preparation of libraries for Illumina sequencing according to the protocol for the Paired-End DNA Sample Prep kit (Illumina). The DNA library of each pool was subjected to the whole-genome sequencing using an Illumina Hiseq4000 sequencer (100脳 and 30脳 coverage for astol1 and wild-type, respectively) and 150-bp paired-end reads were generated. After removing the adapter sequences and low-quality reads, clean reads were mapped to the reference genome sequence (Os-Nipponbare-Reference-RGAP7, MSU) using BWA software58, followed by SNP-calling using GATK software59. Causative variants were calculated using the SNP-index method32 and candidate SNPs were identified. The candidate SNPs were confirmed by PCR sequencing. The primers used are listed in Supplementary Table聽1.Plasmid construction and plant transformationTo generate the pUbi-OsASTOL1 and pUbi-Osastol1 constructs, the open reading frames (ORFs) of OsASTOL1 and Osastol1 were amplified from cDNA of cv. Kasalath and astol1 homozygous mutant, respectively, and then cloned into the KpnI/SpeI sites of the pTCK303 vector.To generate a pCRISPR-OsASTOL1 gene-editing construct with a single target site, a 20鈥塨p target sequence from the second exon of OsASTOL1 (5鈥?AATGGGTGAGACCATCGCCA-3鈥? was designed and then cloned into pOs-Cas9 vector according to Sun et al.60.To generate a pCRISPR-OsASTOL1 gene-editing construct with double target sites, specific target sequences of OsASTOL1 were blasted by CRISPR-GE (http://skl.scau.edu.cn/home/) and two target sequences (5鈥?AATGGGTGAGACCATCGCCA-3鈥?and 5鈥?TTGGCTCAATCAGCACACTC-3鈥? were selected and cloned into single guide RNA (sgRNA) expression cassettes by overlapping PCR, producing pU3-OsASTOL1T1-sgRNA and pU6a-OsASTOL1T2-sgRNA fragments, respectively. The two fragments were sequentially cloned into the BsaI site of the pYLCRISPR/Cas9-MH vector.To generate a pOsASTOL1-GUS construct, a 2.7-kb DNA fragment of the OsASTOL1 upstream sequence before ATG was PCR amplified from the Kasalath genomic DNA and subcloned into the PacI/AscI sites of pS1aG-3 vector61 to replace the original HvPht1;1 promoter before the GUS (尾-glucuronidase) reporter gene.To generate OsASTOL1-YFP and Osastol1-YFP constructs, the ORFs of OsASTOL1 and Osastol1 without the stop codon were cloned into the pEarleyGate 101(C-YFP-HA) vector using Gateway technology. In brief, the ORFs of OsASTOL1 and Osastol1 were PCR-amplified using attB adapter-containing primers and then the fragments were cloned into pDONR vector using BP Clonase (Invitrogen). OsASTOL1-pDONR and Osastol1-pDONR were subcloned into pEarleyGate 101(C-YFP-HA) vector using LR Clonase (Invitrogen).Because the full-length OsASTOL1 protein was expressed with a very low efficiency in E. coli likely due to the presence of the signal peptide in the N terminus, the signal peptide (87 amino acid residues) of OsASTOL1 was deleted before expression in E. coli. To generate the OsASTOL1-His, Osastol1-His, AtOAS-TL A-His and AtOAS-TL AS102N-His for in vitro enzyme assays and for protein binding analysis, the ORFs of OsASTOL1 without the first 261鈥塨p sequence in the N terminus, Osastol1 without the first 261鈥塨p sequence in the N terminus, AtOAS-TL A and AtOAS-TL AS102N (site-directed mutation AtOAS-TL A) without the stop codon were amplified and cloned into the NdeI/XhoI sites of pET-29a vector, respectively. Constructs of different alleles at the 189th site of OsASTOL1-His were generated by site-directed mutagenesis.For in vitro pull-down assays, the construct of pET-28a-His-AtSAT5 was used62. To generate the OsASTOL1, Osastol1, AtOAS-TL A and AtOAS-TL AS102N constructs, the ORFs of OsASTOL1 (without the first 261鈥塨p sequence in the N terminus), Osastol1 (without the first 261鈥塨p sequence in the N terminus), AtOAS-TL A, and AtOAS-TL AS102N with the stop codon were PCR-amplified and cloned into the NdeI/SacI sites of pET-32a vector, respectively. To generate the S-tag-OsASTOL1, S-tag-Osastol1, S-tag-AtOAS-TL A, and S-tag-AtOAS-TL AS102N constructs, the ORFs of OsASTOL1 (without the first 261鈥塨p sequence in the N terminus), Osastol1 (without the first 261鈥塨p sequence in the N terminus), AtOAS-TL A, and AtOAS-TL AS102N with the stop codon were PCR-amplified and cloned into the NcoI/SacI sites of pET-29a vector, respectively.All constructs were verified by DNA sequencing analysis. The primers used are listed in Supplementary Table聽1. pTCK303-Ubi-OsASTOL1 (for overexpression), pOs-Cas9-OsASTOL1 (for knock-out), and pTCK303-Ubi-Osastol1 (for complementation test) were transformed into wild-type Kasalath. pTCK303-Ubi-OsASTOL1 and pYLCRISPR/Cas9-MH-OsASTOL1 were transformed into Osastol1 heterozygous mutant for complementation test. pOsASTOL1-GUS vector was transformed into cv. Nipponbare for tissue localization assay. All plasmids were transformed into rice plants using Agrobacterium-mediated transformation63.Hydroponic experimentsFor uptake experiments, 4-week-old rice seedlings of wild-type Kasalath and astol1 heterozygous mutant were exposed to half strength Kimura B nutrient solution containing 5鈥壜礛 As(III), 5鈥壜礛 As(V), or 2鈥壜礛 Se(VI) for 3 days.For complementation test, 5-week-old rice seedlings of wild-type Kasalath, astol1 heterozygous and homozygous mutants, knockout lines in different background genotypes were exposed to half strength Kimura B nutrient solution containing 5鈥壜礛 As(III) for 3 days. The background genotypes of rice plants were identified by a derived cleaved amplified polymorphic sequence (dCAPS) marker. The PCR products were digested with DdeI. The primers used are listed in Supplementary Table聽1.To evaluate the effects of OsASTOL1 overexpression or knockout, 5-week-old rice seedlings of wild-type Kasalath, astol1 heterozygous mutant, knockout lines, overexpression lines, and null lines in different background genotypes were exposed to half strength Kimura B nutrient solution containing 5鈥壜礛 As(III) for 3 days. The background of transgenic plants was genotyped by a derived cleaved amplified polymorphic sequence (dCAPS) marker (DdeI) targeting the mutated site of Osastol1 in astol1. The primers used are listed in Supplementary Table聽1.Elemental analysisRoots and shoots of rice plants in hydroponic experiments were separated and sampled at the end of the treatments. Plants grown in field experiments were separated into different tissues. Plant samples were dried at 65鈥壜癈 for 3 days, ground to fine powder and then digested with HClO4/HNO3 (15/85, v/v) in a heating block. Brown rice samples were digested with high-purity HNO3 in a microwave oven. The concentrations of As, Se, and other mineral elements were determined using inductively coupled plasma mass spectrometry (ICP-MS, NexION 300X, PerkinElmer). Blanks and certified reference materials (rice flour NIST1568b from National Institute of Standards and Technology, Gaithersburg, MD, USA and spinach leaves GBW10015 from Institute of Geophysics and Geochemical Exploration, Langfang, China) were included for quality control in the analysis.Analysis of metabolitesNon-protein thiols (glutathione and phytochelatins) were quantified using a HPLC-Fluorescence method according to Minocha et al.64. In brief, rice roots were homogenized with a thiol extraction solution (0.1% trifluoroacetic acid (TFA), 6.3鈥塵M diethylenetriamine-pentacetate acid (DTPA), 10鈥塵M tris(2-carboxyethyl)phosphine (TCEP)). After centrifugation, the supernatant was derivatized with monobromobimane (mBBr). The thiols were separated by reversed-phase HPLC and quantified with standard curve by collecting the fluorescence signal under 380鈥塶m excitation and 470鈥塶m emission.Metabolites involved in sulfur metabolism were analyzed by Metware Biotech Co., Ltd (Wuhan, China). In brief, roots and shoots of 5-week-old wild-type rice and astol1 heterozygous mutant were frozen in liquid nitrogen and freeze dried. Three biological replicates were included for each line with six plants for each biological replicate. The metabolites from freeze-dried samples were extracted overnight by 80% ethanol and qualitatively analyzed using a liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) system according to Chen et al.65. The metabolites were quantified using standard curves. For the measurement of sulfate, roots and shoots of rice plants were extracted with deionized water at 4鈥壜癈 for 1鈥塰 and heated at 95鈥壜癈 for 15鈥塵in. After centrifugation and purified through 0.22-碌m Millipore filters, the supernatant was quantified on a Dionex ICS-600 ion chromatography system.RNA extraction and quantitative real-time PCRTotal RNA was extracted from different plant tissues using a RNeasy plant extraction kit (Bioteke) following the manufacturer鈥檚 protocol. First-strand cDNA was synthesized using 2鈥壜礸 RNA using oligo dT(23) primer according to the manufacturer鈥檚 instructions of the HiScript Q Select RT SuperMix for quantitative PCR kit (Vazyme). Quantitative real-time PCR was performed on a CFX96 Real-Time PCR Detection System (Bio-Rad) using an SYBR Green Master Mix kit (Vazyme) according to the manufacturer鈥檚 instructions. The rice OsHistone H3 gene was used as the internal reference. The expression level of each gene was calculated as 2-螖Ct relative to the internal reference. The primers used are listed in Supplementary Table聽1.Tissue and subcellular localization of OsASTOL1To investigate the expression pattern of OsASTOL1, the GUS reporter activity was analyzed by histochemical staining. Roots and leaves of pOsASTOL1-GUS transgenic plants at the seedling stage were collected and incubated with a GUS staining solution for 2鈥塰 at 37鈥壜癈. Leaves were decolorized by 90% acetone overnight. The expression levels of OsASTOL1 in the roots and leaves of cv. Kasalath and in various tissues/organs of cv. Nipponbare at different growth stages were quantified by quantitative real-time PCR. The samples were collected according to the method described by Sun et al.60.To investigate the subcellular localization of OsASTOL1, the pEarleyGate 101-OsASTOL1-YFP and pEarleyGate 101-Osastol1-YFP constructs were transformed into rice protoplasts using polyethylene glycol (PEG)-mediated transformation61, or in tobacco (Nicotiana benthamiana Domin) leaf epidermal cells using Agrobacterium-mediated transformation66. The pEarleyGate 104 (N-YFP) vector containing a 35鈥塖::YFP construct was used as a control. Protoplasts isolated from rice or tobacco were observed using a confocal laser scanning microscope (Zeiss). YFP fluorescence was observed at 525鈥塶m for emission and 510鈥塶m for excitation, and chlorophyll autofluorescence was detected at 575鈥?40鈥塶m for emission and at 546鈥塶m for excitation.Immunoblotting analysisFor immunological detection of OsASTOL1 in rice, total soluble proteins were extracted from 200鈥塵g leaf sample with 1鈥塵l plant protein extraction buffer (CW0885, CWBIO). Proteins were denatured by heating for 10鈥塵in in a boiling-water bath and separated on 12% SDS-PAGE. Subsequently, proteins were transferred onto a polyvinylidene difluoride (PVDF) membrane and immunoblots were probed with the following antibodies in the TBST solution (100鈥塵M Tris-HCl, 150鈥塵M NaCl, 0.1%(v/v) Tween 20, pH 7.4) with 5% nonfat dry milk. A rabbit anti-OAS-TL A antibody (against full-length Arabidopsis cytosolic OAS-TL A, 1:5000 dilution) and a rabbit anti-RbcL antibody (AS03037, Agrisera, 1:10000 dilution) were used as primary antibodies and HPR-conjugated goat anti-rabbit IgG (AT0097, CMCTAG, 1:10000 dilution) was used as secondary antibody. Chemiluminescence was performed using ECL Western Blotting Substrate (IF6747, CMCTAG) according to manufacturer鈥檚 instructions.To produce different variants of AtOAS-TL A and OsASTOL1, the plasmids pET-29a-AtOAS-TLA-His, pET-29a-AtOAS-TLAS102N-His, pET-29a-OsASTOL1-His, pET-29a-Osastol1-His, and different allele of pET-29a-OsASTOL1-His were transformed into E. coli strain BL21 (DE3). The bacterial cells were cultured at 37鈥壜癈 until the OD600 reached 0.5, and then induced by 0.3鈥塵M isopropyl-尾-D-thiogalactopyranoside (IPTG) at 16鈥壜癈 for 16鈥塰. Cells were collected and sonicated by an ultrasonic cell crusher. After centrifugation and heating, denatured proteins in the supernatant were separated by 12% SDS-PAGE. Western blot analysis was performed using a mouse anti-His-tag antibody (AF5060, Beyotime,1:2000 dilution). HPR-conjugated goat anti-mouse IgG (AT0098, CMCTAG, 1:10000 dilution) was used as the secondary antibody.Analysis of OAS-TL and SAT enzyme activityTo determine the OAS-TL enzyme activity, total soluble proteins were extracted from 200鈥塵g rice leaves with 1鈥塵l plant protein extraction buffer (CW0885, CWBIO). Recombinant OAS-TL proteins were expressed in E. coli after induction by 0.3鈥塵M isopropyl-尾-D-thiogalactopyranoside (IPTG) at 16鈥壜癈 for 16鈥塰. Cells were sonicated and filtered through 0.45-碌m Millipore filters. The supernatant (crude extract protein) was added to Ni-NTA packed columns (Sangon Biotech Co. Ltd, Shanghai) and recombinant proteins were eluted with a PBS buffer (20鈥塵M Na3PO4, 100鈥塵M NaCl, pH 7.4) containing 150鈥塵M imidazole. All proteins were quantified with a BCA Protein Assay Kit (CW0014S, CWBIO). The OAS-TL enzyme activity was determined at 25鈥壜癈 in a total volume of 100鈥壜祃 containing 50鈥塵M HEPES/KOH (pH 7.5), 5鈥塵M Na2S, 10鈥塵M OAS, 5鈥塵M DTT and 1-2鈥壜祃 of crude extract proteins according to Heeg et al.25. Recombinant proteins were diluted 1:100 before the enzyme activity assay.To determine the SAT enzyme activity, total soluble proteins were extracted from 200鈥塵g rice leaves with 1鈥塵l plant protein extraction buffer (CW0885, CWBIO). The SAT activity53 was assayed as described before, but with some modification.聽 The SAT activity was first assayed at 25鈥壜癈 in a total volume of 100鈥壜祃 containing 50鈥塵M HEPES/KOH (pH 7.5), 10鈥塵M serine, 1鈥塵M acetyl-CoA, 5鈥塵M DTT and 70鈥壜祃 protein extract. After reaction for 30鈥塵in, 50鈥壜祃 20% trichloroacetic acid (TCA) was added and the content was centrifuged. The pH of the supernatant was adjusted to 7.4 with 40鈥壜祃 2鈥塎 Tris, to which 50鈥壜祃 50鈥塵M Na2S and 4 U purified AtOAS-TL A were added to ensure an excess of OAS-TL activity. The reaction mixture was incubated at 25鈥壜癈 for 20鈥塵in to allow OAS produced by SAT to be transformed to cysteine. The production of cysteine was determined with an acid-ninhydrin solution.Functional analysis of OAS-TL in E. coli NK3 strainTo investigate whether OsASTOL1 functions as an OAS-TL enzyme, the empty vector pET-29a, pET-29a-OsASTOL1-His, and pET-29a-Osastol1-His were transformed into the cysteine auxotroph E. coli NK3 strain. pET-29a-AtOAS-TL A-His was used as a positive control. All bacterial cells were grown on M9 minimal medium (0.2鈥塯鈥塋鈭? leucine, 0.2鈥塯鈥塋鈭? tryptophan, 50鈥壜礸鈥塵l鈭? kanamycin, 1鈥塵M IPTG, and 1.5% agar) with or without 0.5鈥塵M cysteine.Size exclusion chromatography analysisUntagged OsASTOL1 and OsASTOL1S189N were expressed and purified according to Wirtz et al.26. Molecular weights of OsASTOL1 and OsASTOL1S189N were determined by size exclusion chromatography (SEC) using a FPLCTM system. Briefly, 1鈥塵l of OsASTOL1 or OsASTOL1S189N (1鈥塵g) was injected into the sample loop and was eluted through the HiLoadTM 16/60 Superdex 200 prep grade column (120鈥塵l, GE Healthcare) in buffer (20鈥塵M Tris-HCl, 150鈥塵M NaCl, pH 7.5) at 1鈥塵l/min flow rate. Calibration of the Superdex 200 column was performed with the high and low molecular weight gel filtration calibration kits (GE Healthcare) according to the manufacturer鈥檚 instructions. The molecular weights of OsASTOL1 and OsASTOL1S189N聽were calculated from the calibration curve with the measured elution volumes.Sequence alignment and visualization of protein structureMultiple sequence alignments of OAS-TL proteins were conducted using the ClustalW method in DNAMAN software. Visualization of protein structure and site-directed mutation of Arabidopsis cytosol OAS-TL A (PDB ID: 1z7w) were performed by PyMOL software.In vitro pull-down assayTo investigate the interactions of OsASTOL1, OsASTOL1S189N, AtOAS-TL A or AtOAS-TL AS102N with AtSAT5 and the dissociation effect by OAS, a revised in vitro pull-down assay was conducted. In brief, soluble proteins were extracted from cells harboring the pET-28a-His-AtSAT5 plasmid and added to a NiCl2 containing HiTrapTM Chelating column (GE Healthcare). The column was washed with Buffer W (50鈥塵M Tris-HCl, pH 8.0, 250鈥塵M NaCl, 80鈥塵M Imidazol), followed by washing with Buffer W containing 10鈥塵M OAS (Solar Bio) to remove the bacterial OAS-TL. The crude extract containing untagged plant OAS-TL proteins was added to the above column containing the His-tagged AtSAT5 protein. The untagged OAS-TL protein was specifically eluted with Buffer W containing 10鈥塵M OAS. After washing with Buffer W, the His-tagged AtSAT5 was eluted with Buffer E (50鈥塵M Tris-HCl, pH 8.0, 250鈥塵M NaCl, 400鈥塵M Imidazol). The fractions obtained from different steps of elution were analyzed by SDS-PAGE followed by Coomassie staining.For immunological detection of OAS-TL and SAT proteins, the His-AtSAT5 and S-tagged OAS-TL proteins were used to perform the same steps as described above but with a different Ni-NTA packed columns (Sangon Biotech Co. Ltd, Shanghai) and a new washing buffer (50鈥塵M Tris-HCl, pH 8.0, 250鈥塵M NaCl, 20鈥塵M Imidazol). A rabbit anti-S-tag antibody (101290-T38, Sino Bio, 1:5000 dilution) and a mouse anti-His-tag antibody (AF5060, Beyotime,1:2000 dilution) were used as the primary antibodies.Microscale thermophoresis analysisTen C-terminal residues38 of AtSAT5 (AT5G56760, sequence: FISEWSDYII), required for binding with OAS-TL, was synthesized by GenScript Biotech (Nanjing). Recombinant His-tagged OAS-TL proteins were labeled with RED-tris-NTA 2nd Generation Dye for 30鈥塵in. For the binding assay, 100鈥壜礛 labeled His-tagged OAS-TL protein was incubated with a serial dilution of the ligand (OAS or AtSAT5C10 peptide) in the PBS buffer (pH 7.6) containing 0.05% Tween 20. Microscale thermophoresis (MST) analysis was performed on a Monolith NT.115 instrument (Nano Temper, Germany) using auto-detected LED power (5鈥?0%) and MST power (Medium or High). The data were evaluated using the MO Affinity Analysis software.Reporting summaryFurther information on research design is available in the聽Nature Research Reporting Summary linked to this article. Data supporting the findings of this work are available within the paper and its Supplementary Information files. A reporting summary for this article is available as a Supplementary Information file. The genetic materials generated and analyzed during the current study are available from the corresponding author upon request. The source data underlying Fig.聽1b, d, g鈥搃, 2b鈥揺, 3, 4b鈥揼, 5, as well as Supplementary Figs.聽1a, 2b鈥揹, 3b, 4鈥?a data-track=\"click\" data-track-label=\"link\" data-track-action=\"supplementary material anchor\" href=\"/articles/s41467-021-21282-5#MOESM4\">6d鈥揻, 10a, 12b, 14, 15b鈥揺, 16鈥?a data-track=\"click\" data-track-label=\"link\" data-track-action=\"supplementary material anchor\" href=\"/articles/s41467-021-21282-5#MOESM4\">20d鈥搇, 21c鈥揺, and Supplementary Table聽4 are provided as a Source Data file. References1.Stone, R. Food safety. Arsenic and paddy rice: a neglected cancer risk? Science 321, 184鈥?85 (2008).CAS聽 PubMed聽 Article聽Google Scholar聽 2.Meharg, A. A. et al. Geographical variation in total and inorganic arsenic content of polished (white) rice. Environ. Sci. Technol. 43, 1612鈥?617 (2009).ADS聽 CAS聽 PubMed聽 Article聽Google Scholar聽 3.Zhao, F. J., McGrath, S. P. Meharg, A. A. Arsenic as a food chain contaminant: mechanisms of plant uptake and metabolism and mitigation strategies. Annu Rev. Plant Biol. 61, 535鈥?59 (2010).CAS聽 PubMed聽 Article聽Google Scholar聽 4.Li, G., Sun, G. X., Williams, P. N., Nunes, L. Zhu, Y. G. Inorganic arsenic in Chinese food and its cancer risk. Environ. Int. 37, 1219鈥?225 (2011).CAS聽 PubMed聽 Article聽Google Scholar聽 5.Williams, P. N. et al. Occurrence and partitioning of cadmium, arsenic and lead in mine impacted paddy rice: Hunan, China. Environ. Sci. Technol. 43, 637鈥?42 (2009).ADS聽 CAS聽 PubMed聽 Article聽Google Scholar聽 6.Williams, P. N. et al. Increase in rice grain arsenic for regions of Bangladesh irrigating paddies with elevated arsenic in groundwaters. Environ. Sci. Technol. 40, 4903鈥?908 (2006).ADS聽 CAS聽 PubMed聽 Article聽Google Scholar聽 7.Panaullah, G. M. et al. Arsenic toxicity to rice (Oryza sativa L.) in Bangladesh. Plant Soil 317, 31鈥?9 (2009).CAS聽 Article聽Google Scholar聽 8.Huhmann, B. L. et al. Field study of rice yield diminished by soil arsenic in Bangladesh. Environ. Sci. Technol. 51, 11553鈥?1560 (2017).ADS聽 CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 9.Muehe, E. M., Wang, T., Kerl, C. F., Planer-Friedrich, B. Fendorf, S. Rice production threatened by coupled stresses of climate and soil arsenic. Nat. Commun. 10, 4985 (2019).ADS聽 PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 10.Ha, S. B. et al. Phytochelatin synthase genes from Arabidopsis and the yeast Schizosaccharomyces pombe. Plant Cell 11, 1153鈥?164 (1999).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 11.Song, W. Y. et al. Arsenic tolerance in Arabidopsis is mediated by two ABCC-type phytochelatin transporters. Proc. Natl Acad. Sci. USA 107, 21187鈥?1192 (2010).ADS聽 CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 12.Song, W. Y. et al. A rice ABC transporter, OsABCC1, reduces arsenic accumulation in the grain. Proc. Natl Acad. Sci. USA 111, 15699鈥?5704 (2014).ADS聽 CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 13.Yang, J. et al. OsCLT1, a CRT-like transporter 1, is required for glutathione homeostasis and arsenic tolerance in rice. New Phytol. 211, 658鈥?70 (2016).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 14.Clemens, S. Ma, J. F. Toxic heavy metal and metalloid accumulation in crop plants and foods. Annu Rev. Plant Biol. 67, 489鈥?12 (2016).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 15.Hayashi, S. et al. Phytochelatin synthase OsPCS1 plays a crucial role in reducing arsenic levels in rice grains. Plant J. 91, 840鈥?48 (2017).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 16.Deng, F. et al. Engineering rice with lower grain arsenic. Plant Biotechnol. J. 16, 1691鈥?699 (2018).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 17.Batista, B. L. et al. Identification and quantification of phytochelatins in roots of rice to long-term exposure: evidence of individual role on arsenic accumulation and translocation. J. Exp. Bot. 65, 1467鈥?479 (2014).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 18.Cobbett, C. S., May, M. J., Howden, R. Rolls, B. The glutathione-deficient, cadmium-sensitive mutant, cad2-1, of Arabidopsis thaliana is deficient in 纬-glutamylcysteine synthetase. Plant J. 16, 73鈥?8 (1998).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 19.Pasternak, M. et al. Restricting glutathione biosynthesis to the cytosol is sufficient for normal plant development. Plant J. 53, 999鈥?012 (2008).CAS聽 PubMed聽 Article聽Google Scholar聽 20.Khan, M. S. et al. Sulfite reductase defines a newly discovered bottleneck for assimilatory sulfate reduction and is essential for growth and development in Arabidopsis thaliana. Plant Cell 22, 1216鈥?231 (2010).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 21.Takahashi, H., Kopriva, S., Giordano, M., Saito, K. Hell, R. Sulfur assimilation in photosynthetic organisms: molecular functions and regulations of transporters and assimilatory enzymes. Annu Rev. Plant Biol. 62, 157鈥?84 (2011).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 22.Hell, R. Wirtz, M. Molecular biology, biochemistry and cellular physiology of cysteine metabolism in Arabidopsis thaliana. Arabidopsis Book 9, e0154 (2011).PubMed聽 PubMed Central聽 Article聽Google Scholar聽 23.Shirzadian-Khorramabad, R. et al. A mutation in the cytosolic O-acetylserine (thiol) lyase induces a genome-dependent early leaf death phenotype in Arabidopsis. BMC Plant Biol. 10, 80 (2010).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 24.Lopez-Martin, M. C., Becana, M., Romero, L. C. Gotor, C. Knocking out cytosolic cysteine synthesis compromises the antioxidant capacity of the cytosol to maintain discrete concentrations of hydrogen peroxide in Arabidopsis. Plant Physiol. 147, 562鈥?72 (2008).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 25.Heeg, C. et al. Analysis of the Arabidopsis O-acetylserine(thiol)lyase gene family demonstrates compartment-specific differences in the regulation of cysteine synthesis. Plant Cell 20, 168鈥?85 (2008).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 26.Wirtz, M., Droux, M. Hell, R. O-acetylserine (thiol) lyase: an enigmatic enzyme of plant cysteine biosynthesis revisited in Arabidopsis thaliana. J. Exp. Bot. 55, 1785鈥?798 (2004).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 27.Wang, C. et al. OASTL-A1 functions as a cytosolic cysteine synthase and affects arsenic tolerance in rice. J. Exp. Bot. 71, 3678鈥?689 (2020).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 28.Wirtz, M., Berkowitz, O., Droux, M. Hell, R. The cysteine synthase complex from plants. Mitochondrial serine acetyltransferase from Arabidopsis thaliana carries a bifunctional domain for catalysis and protein-protein interaction. Eur. J. Biochem 268, 686鈥?93 (2001).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 29.Droux, M., Ruffet, M. L., Douce, R. Job, D. Interactions between serine acetyltransferase and O-acetylserine (thiol) lyase in higher plants鈥搒tructural and kinetic properties of the free and bound enzymes. Eur. J. Biochem. 255, 235鈥?45 (1998).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 30.Combs, G. F. Jr. Selenium in global food systems. Br. J. Nutr. 85, 517鈥?47 (2001).CAS聽 PubMed聽 Article聽Google Scholar聽 31.Rayman, M. P. Food-chain selenium and human health: emphasis on intake. Br. J. Nutr. 100, 254鈥?68 (2008).CAS聽 PubMed聽 Article聽Google Scholar聽 32.Abe, A. et al. Genome sequencing reveals agronomically important loci in rice using MutMap. Nat. Biotechnol. 30, 174鈥?78 (2012).CAS聽 PubMed聽 Article聽Google Scholar聽 33.Emanuelsson, O., Nielsen, H. von Heijne, G. ChloroP, a neural network-based method for predicting chloroplast transit peptides and their cleavage sites. Protein Sci. 8, 978鈥?84 (1999).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 34.Hell, R., Bork, C., Bogdanova, N., Frolov, I. Hauschild, R. Isolation and characterization of two cDNAs encoding for compartment specific isoforms of O-acetylserine (thiol) lyase from Arabidopsis thaliana. FEBS Lett. 351, 257鈥?62 (1994).CAS聽 PubMed聽 Article聽Google Scholar聽 35.Bonner, E. R., Cahoon, R. E., Knapke, S. M. Jez, J. M. Molecular basis of cysteine biosynthesis in plants: structural and functional analysis of O-acetylserine sulfhydrylase from Arabidopsis thaliana. J. Biol. Chem. 280, 38803鈥?8813 (2005).CAS聽 PubMed聽 Article聽Google Scholar聽 36.Saito, K., Kurosawa, M., Tatsuguchi, K., Takagi, Y. Murakoshi, I. Modulation of cysteine biosynthesis in chloroplasts of transgenic tobacco overexpressing cysteine synthase [O-acetylserine(thiol)-lyase]. Plant Physiol 106, 887鈥?95 (1994).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 37.Wirtz, M., Heeg, C., Samami, A. A., Ruppert, T. Hell, R. Enzymes of cysteine synthesis show extensive and conserved modifications patterns that include N-alpha-terminal acetylation. Amino Acids 39, 1077鈥?086 (2010).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 38.Francois, J. A., Kumaran, S. Jez, J. M. Structural basis for interaction of O-acetylserine sulfhydrylase and serine acetyltransferase in the Arabidopsis cysteine synthase complex. Plant Cell 18, 3647鈥?655 (2006).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 39.Huang, B., Vetting, M. W. Roderick, S. L. The active site of O-acetylserine sulfhydrylase is the anchor point for bienzyme complex formation with serine acetyltransferase. J. Bacteriol. 187, 3201鈥?205 (2005).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 40.Hubberten, H. M. et al. Additional role of O-acetylserine as a sulfur status-independent regulator during plant growth. Plant J. 70, 666鈥?77 (2012).CAS聽 PubMed聽 Article聽Google Scholar聽 41.Hopkins, L. et al. O-acetylserine and the regulation of expression of genes encoding components for sulfate uptake and assimilation in potato. Plant Physiol. 138, 433鈥?40 (2005).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 42.Williams, P. N. et al. Greatly enhanced arsenic shoot assimilation in rice leads to elevated grain levels compared to wheat and barley. Environ. Sci. Technol. 41, 6854鈥?859 (2007).ADS聽 CAS聽 PubMed聽 Article聽Google Scholar聽 43.Ma, J. F. et al. Transporters of arsenite in rice and their role in arsenic accumulation in rice grain. Proc. Natl Acad. Sci. USA 105, 9931鈥?935 (2008).ADS聽 CAS聽 PubMed聽 Article聽Google Scholar聽 44.Williams, P. N. et al. Selenium characterization in the global rice supply chain. Environ. Sci. Technol. 43, 6024鈥?030 (2009).ADS聽 CAS聽 PubMed聽 Article聽Google Scholar聽 45.Zhao, F. J., Stroud, J. L., Khan, M. A. McGrath, S. P. Arsenic translocation in rice investigated using radioactive As-73 tracer. Plant Soil 350, 413鈥?20 (2012).CAS聽 Article聽Google Scholar聽 46.Schiavon, M., Pilon-Smits, E. A., Wirtz, M., Hell, R. Malagoli, M. Interactions between chromium and sulfur metabolism in Brassica juncea. J. Environ. Qual. 37, 1536鈥?545 (2008).CAS聽 PubMed聽 Article聽Google Scholar聽 47.Parisy, V. et al. Identification of PAD2 as a gamma-glutamylcysteine synthetase highlights the importance of glutathione in disease resistance of Arabidopsis. Plant J. 49, 159鈥?72 (2007).CAS聽 PubMed聽 Article聽Google Scholar聽 48.Wirtz, M. et al. Mitochondrial cysteine synthase complex regulates O-acetylserine biosynthesis in plants. J. Biol. Chem. 287, 27941鈥?7947 (2012).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 49.Wirtz, M. et al. Structure and function of the hetero-oligomeric cysteine synthase complex in plants. J. Biol. Chem. 285, 32810鈥?2817 (2010).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 50.Watanabe, M. et al. Comparative genomics and reverse genetics analysis reveal indispensable functions of the serine acetyltransferase gene family in Arabidopsis. Plant Cell 20, 2484鈥?496 (2008).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 51.Ruffet, M. L., Lebrun, M., Droux, M. Douce, R. Subcellular-distribution of serine acetyltransferase from Pisum sativum and characterization of an Arabidopsis thaliana putative cytosolic isoform. Eur. J. Biochem. 227, 500鈥?09 (1995).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 52.Haas, F. H. et al. Mitochondrial serine acetyltransferase functions as a pacemaker of cysteine synthesis in plant cells. Plant Physiol. 148, 1055鈥?067 (2008).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 53.Wirtz, M. Hell, R. Dominant-negative modification reveals the regulatory function of the multimeric cysteine synthase protein complex in transgenic tobacco. Plant Cell 19, 625鈥?39 (2007).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 54.Noji, M. Saito, K. Molecular and biochemical analysis of serine acetyltransferase and cysteine synthase towards sulfur metabolic engineering in plants. Amino acids 22, 231鈥?43 (2002).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 55.Muller, S. M. et al. The redox-sensitive module of cyclophilin 20-3, 2-cysteine peroxiredoxin and cysteine synthase integrates sulfur metabolism and oxylipin signaling in the high light acclimation response. Plant J. 91, 995鈥?014 (2017).PubMed聽 Article聽 CAS聽 PubMed Central聽Google Scholar聽 56.Park, S. W. et al. Cyclophilin 20-3 relays a 12-oxo-phytodienoic acid signal during stress responsive regulation of cellular redox homeostasis. Proc. Natl Acad. Sci. USA 110, 9559鈥?564 (2013).ADS聽 CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 57.Dominguez-Solis, J. R. et al. A cyclophilin links redox and light signals to cysteine biosynthesis and stress responses in chloroplasts. Proc. Natl Acad. Sci. USA 106, 1292鈥?292 (2009).CAS聽Google Scholar聽 58.Li, H. Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754鈥?760 (2009).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 59.McKenna, A. et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 20, 1297鈥?303 (2010).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 60.Sun, S. K. et al. Decreasing arsenic accumulation in rice by overexpressing OsNIP1;1 and OsNIP3;3 through disrupting arsenite radial transport in roots. New Phytol. 219, 641鈥?53 (2018).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 61.Tang, Z. et al. Knockdown of a rice stelar nitrate transporter alters long-distance translocation but not root influx. Plant Physiol. 160, 2052鈥?063 (2012).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 62.Birke, H., Hildebrandt, T. M., Wirtz, M. Hell, R. Sulfide detoxification in plant mitochondria. Methods Enzymol. 555, 271鈥?86 (2015).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 63.Hiei, Y., Komari, T. Kubo, T. Transformation of rice mediated by Agrobacterium tumefaciens. Plant Mol. Biol. 35, 205鈥?18 (1997).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 64.Minocha, R., Thangavel, P., Dhankher, O. P. Long, S. Separation and quantification of monothiols and phytochelatins from a wide variety of cell cultures and tissues of trees and other plants using high performance liquid chromatography. J. Chromatogr. A 1207, 72鈥?3 (2008).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 65.Chen, W. et al. A novel integrated method for large-scale detection, identification, and quantification of widely targeted metabolites: application in the study of rice metabolomics. Mol. Plant 6, 1769鈥?780 (2013).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 66.Sparkes, I. A., Runions, J., Kearns, A. Hawes, C. Rapid, transient expression of fluorescent fusion proteins in tobacco plants and generation of stably transformed plants. Nat. Protoc. 1, 2019鈥?025 (2006).CAS聽 PubMed聽 Article聽 PubMed Central聽Google Scholar聽 Download referencesAcknowledgementsThis work was supported by the grants from the Natural Science Foundation of China (31520103914), the Ministry of Science and Technology Key R D program (2016YFD0100704), the Innovative Research Team Development Plan of the Ministry of Education of China (grant no. IRT_17R56), the Fundamental Research Funds for the Central Universities (grant no. KYT201802)聽and German Research Foundation聽(DFG, WI3560/1-2, WI3560/6-1聽and HE1848/15-2, HE1848/19-1).Author informationAffiliationsState Key Laboratory of Crop Genetics and Germplasm Enhancement, College of Resources and Environmental Sciences, Nanjing Agricultural University, Nanjing, 210095, ChinaSheng-Kai Sun,聽Xuejie Xu,聽Zhong Tang,聽Zhu Tang,聽Xin-Yuan Huang聽 聽Fang-Jie ZhaoCentre for Organismal Studies (COS), Heidelberg University, 69120, Heidelberg, GermanyMarkus Wirtz聽 聽R眉diger HellAuthorsSheng-Kai SunView author publicationsYou can also search for this author in PubMed聽Google ScholarXuejie XuView author publicationsYou can also search for this author in PubMed聽Google ScholarZhong TangView author publicationsYou can also search for this author in PubMed聽Google ScholarZhu TangView author publicationsYou can also search for this author in PubMed聽Google ScholarXin-Yuan HuangView author publicationsYou can also search for this author in PubMed聽Google ScholarMarkus WirtzView author publicationsYou can also search for this author in PubMed聽Google ScholarR眉diger HellView author publicationsYou can also search for this author in PubMed聽Google ScholarFang-Jie ZhaoView author publicationsYou can also search for this author in PubMed聽Google ScholarContributionsF.J.Z. and S.K.S designed the research. S.K.S., X.X., Zhong T., and Zhu T. performed experiments. S.K.S., X.Y.H., and M.W. analyzed data. S.K.S., F.J.Z., M.W., and R.H. wrote and revised the manuscript with inputs of other authors.Corresponding authorCorrespondence to Fang-Jie Zhao.Ethics declarations Competing interests The authors declare no competing interests. Additional informationPeer review information Nature Communications thanks Satoru Ishikawa and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer review reports are available.Publisher鈥檚 note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary information CommentsBy submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Sign up for the Nature Briefing newsletter 鈥?what matters in science, free to your inbox daily.