Differential Regulation of the Expression of Two High-Aﬀinity Sulfate Transporters, SULTR1.1 and SULTR1.2, in Arabidopsis.

The molecular mechanisms regulating the initial uptake of inorganic sulfate in plants are still largely unknown. The current model for the regulation of sulfate uptake and assimilation attributes positive and negative regulatory roles to O -acetyl-serine ( O -acetyl-Ser) and glutathione, respectively. This model seems to suffer from exceptions and it has not yet been clearly validated whether intracellular O -acetyl-Ser and glutathione levels have impacts on regulation. The transcript level of the two high-afﬁnity sulfate transporters SULTR1.1 and SULTR1.2 responsible for sulfate uptake from the soil solution was compared to the intracellular contents of O- acetyl-Ser, glutathione, and sulfate in roots of plants submitted to a wide diversity of experimental conditions. SULTR1.1 and SULTR1.2 were differentially expressed and neither of the genes was regulated in accordance with the current model. The SULTR1.1 transcript level was mainly altered in response to the sulfur-related treatments. Split-root experiments show that the expression of SULTR1.1 is locally regulated in response to sulfate starvation. In contrast, accumulation of SULTR1.2 transcripts appeared to be mainly related to metabolic demand and is controlled by photoperiod. On the basis of the new molecular insights provided in this study, we suggest that the expression of the two transporters depends on different regulatory networks. We hypothesize that interplay between SULTR1.1 and SULTR1.2 transporters could be an important mechanism to regulate sulfate content in the roots.

Sulfur (S) is one of the most important macronutrients for plant growth and development. Higher plants acquire S predominantly in the form of anionic sulfate from the soil. In plastids, sulfate is reduced into sulfide, which then combines with O-acetyl-Ser to form Cys (Leustek et al., 2000). Subsequently, Cys is either converted into Met or directly incorporated into proteins or glutathione. S deficiency is a major abiotic stress that affects plant growth and crop productivity worldwide (McGrath et al., 1996). Plants mainly re-spond to S deficiency by regulating the expression of genes involved in the uptake and assimilation of sulfate and the remobilization of secondary metabolites (Maruyama-Nakashita et al., 2003. Numerous studies have shown that the regulation of sulfate transporters occurs predominantly at the mRNA level (Hawkesford et al., 1993;Smith et al., 1995Smith et al., , 1997Hawkesford and Wray, 2000;Takahashi et al., 1997Takahashi et al., , 2000Yoshimoto et al., 2002, Hopkins et al., 2005. Using different experimental conditions in various species, these studies led to the proposal of a general model describing the regulation of S uptake and assimilation . In this model, positive and negative regulatory roles are attributed to O-acetyl-Ser and glutathione, respectively. The validity of this model can, however, be questioned in the light of a number of experimental outcomes as described below. Glutathione is the major organic S-containing compound transported and stored in plants. It has an important role in regulating cellular S homeostasis and in controlling the redox cellular status (Noctor et al., 2002). Exogenous glutathione application causes a significant reduction in the expression of sulfate transporters and in the uptake of sulfate by roots (Herschbach and Rennenberg, 1994, Lappartient et al., 1999Vidmar et al., 1999). In rapeseed (Brassica napus) seedlings, split-root experiments showed that glutathione acts as a long-distance signal from the shoot to the root: the negative regulation of sulfate uptake resulting from the application of glutathione in one part of the root system was also demonstrated in the untreated part of the root (Lappartient and Touraine, 1996). However, such a negative effect of glutathione on the root sulfate uptake capacity was not observed in poplar after an exposure of aerial organs to H 2 S, although this treatment was shown to significantly increase leaf and root glutathione contents (Herschbach et al., 2000;Westerman et al., 2001). Also, the overexpression of the key enzyme for glutathione synthesis, g-glutamyl-Cys synthase, which resulted in the overaccumulation of glutathione, did not induce any down-regulation of the expression of sulfate transporters (Herschbach et al., 2000;Hartmann et al., 2004). Finally, increasing the glutathione content by providing O-acetyl-Ser to barley (Hordeum vulgare) plants did not result in a reduction in either the transcript accumulation of sulfate transporters or the sulfate uptake . In this context, the role of glutathione in down-regulating transporters involved in sulfate uptake needs further investigation.
O-Acetyl-Ser, a precursor of Cys synthesis in higher plants, has been proposed to be a positive regulator of the expression of genes involved in uptake and assimilation of sulfate in plants. Indeed, O-acetyl-Ser treatments induce the elevation of sulfate uptake rate in Lemna minor (Neuenschwander et al., 1991), barley , and potato (Solanum tuberosum; Hopkins et al., 2005). O-Acetyl-Ser treatments also result in an increase in the expression of all the genes involved in the sulfate assimilation pathway in Arabidopsis (Arabidopsis thaliana; Koprivova et al., 2000). The hypothesis that O-acetyl-Ser is a strong positive regulator of the S-deficiency response in plants is strengthened by the fact that in bacteria, O-acetyl-Ser plays an important role in S homeostasis: it has been proposed to stimulate the formation of a transcription initiation complex comprising the homotetrameric CysB protein and an RNA polymerase, which is active on several promoters of the Cys regulon (Kredich, 1992). In plants, some studies however lead to different conclusions. O-Acetyl-Ser treatment of pumpkin leaf discs did not result in an increase in the sulfate uptake rate (Rennenberg, 1983). In potato, the increase in the mRNA relative abundance of the StST1 sulfate transporter and the induction of sulfate uptake both precede the accumulation of O-acetyl-Ser (Hopkins et al., 2005). In addition, metabolite-to-gene network analysis showed that the level of expression of genes known to be induced by S deficiency is not correlated with O-acetyl-Ser accumulation (Hirai et al., 2005). Altogether, these observations suggest that O-acetyl-Ser cannot be the sole molecule sensed in response to S deficiency and that the main signal in the roots may be the depletion of intermediates of the sulfate assimilation pathway.
Concluding on the validity of the model proposed for the regulation of sulfate uptake transporters is not easy, mainly because the experiments performed to test the model used different plant species and different growth conditions. It is thus important (1) to extensively analyze to what extent the model is indeed valid and (2) to do the whole analysis in one species. In Arabidopsis, sulfate uptake at the root surface is suggested to be achieved by two high-affinity sulfate transporters (SULTR1.1 and SULTR1.2) that are coexpressed in root hairs and in root epidermal and cortical cells (Takahashi et al., 2000;Vidmar et al., 2000;Shibagaki et al., 2002;Yoshimoto et al., 2002Yoshimoto et al., , 2007El Kassis et al., 2007). SULTR1.2 transcripts are more abundant than SULTR1.1 ones (Fizames et al., 2004) and SULTR1.2 was shown to contribute to at least 80% of the sulfate uptake in standard growth conditions (Maruyama-Nakashita et al., 2003). Up to now, SULTR1.1 and SULTR1.2 were described to be tightly coregulated, although the magnitude of their up-or down-regulation is different in that SULTR1.1 transcripts show greater changes in their abundance (Yoshimoto et al., 2002;Maruyama-Nakashita et al., 2003, 2004a, 2004b. Exogenous supplies of O-acetyl-Ser or glutathione were shown to regulate SULTR1.1 and SULTR1.2 expression in agreement with the proposed model (Maruyama-Nakashita et al., 2004b). However, such treatment data do not constitute evidence for a direct role of O-acetyl-Ser or glutathione in the regulation of SULTR1.1 and SULTR1.2. In this study, we thus extensively tested whether SULTR1.1 and SULTR1.2 are indeed regulated in agreement with the proposed regulatory model. A standardized plant growth procedure was used to examine SULTR1.1 and SULTR1.2 gene expression in a wide variety of experimental conditions. Treatments known to have a pronounced effect on sulfate uptake, on S metabolite contents, or on the expression of sulfate transporters were used: S, nitrogen (N), or phosphorous (P) starvation, supplementation of the culture hydroponic solutions with either selenate, sugar, cadmium or sodium chloride, and photoperiod conditions according to the literature (Howden et al., 1995;Barroso et al., 1999;Lejay et al., 2003;Maruyama-Nakashita et al., 2004b;El kassis et al., 2007). By analyzing the root glutathione, O-acetyl-Ser and sulfate contents in conjunction with SULTR1.1 and SULTR1.2 transcript abundances, we show that the expression of these two transporters was not regulated in agreement with the current model. We propose that expression of SULTR1.1 and SULTR1.2 is regulated to a significant extent by distinct signaling pathways.

Design and Validity of the Experimental Setup
The main purpose of our experiments was to test the validity of the current model for the regulation of the expression of transporters involved in root sulfate uptake. This validity was assayed in the model plant Arabidopsis for the sulfate transporters SULTR1.1 and SULTR1.2. A common standardized system was used to assay a large set of different experimental conditions (see ''Materials and Methods''). For every sample, O-acetyl-Ser, glutathione, and sulfate contents and transcript abundance of SULTR1.1 and SULTR1.2 were systematically determined (Table I).
SULTR1.1 and SULTR1.2 mRNA accumulations were determined using quantitative PCR. Using this technique, the data must be normalized using control genes, expression of which remains invariant during the timescale of all the different experimental conditions assayed. Because no single reference gene has been previously described to have a stable expression level across all the treatments used in our study, we used four reference genes or gene pairs: the ACTIN2-ACTIN8 pair, the 60S ribosomal protein L23a gene, SCARECROW, and a region common to MAP3K-1 and MAP3K-2. These genes, already described to have invariant expression levels in different abiotic stress conditions, display different levels of relative expression (Charrier et al., 2002). Over the different treatments used in our experiments, the relative transcript levels of the four references commonly showed less than 2-fold variations (Supplemental Table S1). We then used the mean value of the expression levels of all the control genes or gene pairs to ensure maintaining of a stable and reliable reference point over 20 different conditions tested (Table I). In few cases, one of the treatments had a major impact on the mRNA accumulation of one of the reference genes. For example, the 1 mM glutathione treatment resulted in an approximately 5-times decrease in the accumulation of ACTIN2-8 mRNA relative to that of the other reference genes. In such a situation, the mRNA accumulation value for this reference gene was not taken into account for standardization.
Our results presented in Table I are in agreement with previously documented expression patterns of SULTR1.1 and SULTR1.2 (Maruyama-Nakashita et al., 2003). SULTR1.2 transcripts were indeed more abundant than SULTR1.1 transcripts (approximately 10 times in our experiments; Table I). Sulfate starvation, application of O-acetyl-Ser, selenate or cadmium all induced an increase in SULTR1.1 and/or SULTR1.2 mRNA levels, while the application of glutathione resulted in a strong reduction in the SULTR1.1 and/or SULTR1.2 expression (Table I). Our experimental framework is thus validated.
Is Glutathione Involved in the Regulation of the Expression of SULTR1.1 and SULTR1.2?
As already mentioned, glutathione has been proposed to be a negative regulator of sulfate uptake and of the gene expression of transporters involved in The 2DCt value expresses in a logarithmic scale the mRNA abundance of the gene of interest relative to the average mRNA abundance of the housekeeping genes evaluated for the same treatment. c The 2 2DDCt value expresses the mRNA abundance for the gene of interest in a given condition (treatment) relative to the average mRNA abundance for the same gene measured in the control treatment. A 2 2DDCt value of 4 means that the mRNA is four times more abundant in plants submitted to the considered treatment than in control plants.  sulfate uptake. To test this hypothesis, the root intracellular glutathione content was measured in plants submitted to the different treatments and related to the SULTR1.1 and SULTR1.2 mRNA accumulations. The treatments corresponding to the supplementation of the culture medium with glutathione had the greatest impact on the root glutathione content, raising it 2-to 3-fold when compared to the control (Table I). They also resulted in reduced expression of the sulfate transporters: reduction of SULTR1.2 mRNA accumulation was only significantly noticeable in plants treated with 0.5 mM glutathione, but SULTR1.1 mRNA accumulation was significantly reduced by a factor varying from 6 to 50 in plants treated with 0.1 to 1.0 mM glutathione, respectively (Table I). These results supported the hypothesis that glutathione is a negative regulator of expression of SULTR1.1 and SULTR1.2 genes. However, when the glutathione treatments were not considered, no correlation could be found between the intracellular content of the reduced form of glutathione (GSH) and the mRNA accumulations of either SULTR1.1 (R 2 , 0.01, P 5 0.685) or SULTR1.2 (R 2 5 0.1, P 5 0.171; Fig. 1). This lack of correlation is reliable since it was established from a wide range of root glutathione contents varying from 63 6 3 nmol.g 21 fresh weight (FW) to 239 6 19 nmol.g 21 FW in response to the Suc and the O-acetyl-Ser treatments, respectively. Our data thus do not support any role of the root intracellular glutathione on the expression of either SULTR1.1 or SULTR1.2.
Is O-Acetyl-Ser Involved in the Regulation of the Expression of SULTR.1.1 and SULTR1.2?
O-Acetyl-Ser has been proposed to be a positive regulator of sulfate uptake and of the genetic expression of transporters involved in sulfate uptake . To test this hypothesis, the root intracellular O-acetyl-Ser content was measured in plants submitted to different treatments (Table  I). Compared to the control treatment, the O-acetyl-Ser treatment resulted in an increase of over 100-fold in the root O-acetyl-Ser content. Interestingly, cadmium treatment also induced a significant overaccumulation of O-acetyl-Ser in roots, which corresponded to 8 to 10 times the O-acetyl-Ser content found in the control condition. A noteworthy increase of Ser acetyltransferase (SAT) gene expression from Arabidopsis has also been observed upon cadmium stress (Howarth et al., 2003). Apart from the O-acetyl-Ser and cadmium treatments, the O-acetyl-Ser content was evenly distributed within a 1 to 5 range, varying from 0.5 6 0.2 nmol g 21 FW measured in control condition at dawn to 2.6 6 0.2 nmol g 21 FW measured after 72 h of sulfate starvation. O-acetyl-Ser root contents were also high in response to the addition of glutathione or Suc and at the end of the light period (Table I).
The relationship between root O-acetyl-Ser contents and the mRNA accumulation of SULTR1.1 and SULTR1.2 was not as expected. The O-acetyl-Ser and cadmium treatments, which had the greatest impacts on the intracellular O-acetyl-Ser content, indeed induced approximately 2-to approximately 4-fold increases in mRNA accumulations of the two sulfate transporters (Table I). However, when the data obtained from plants submitted to the O-acetyl-Ser treatment were not taken into account, no correlation could be found between the intracellular O-acetyl-Ser content and the mRNA accumulations of either SULTR1.1 (R 2 5 0.02, P 5 0.519) or SULTR1.2 (R 2 5 0.06, P 5 0.27; Fig. 2). This conclusion, which is valid whether considering or not the cadmium treatment, is warranted by the fact that it has been obtained from the analysis of a wide range of O-acetyl-Ser contents. Our data are thus not in agreement with the hypothesis that intracellular O-acetyl-Ser plays a role in the gene expression of either SULTR1.1 or SULTR1.2.
Relationship between Expression of the SULTR1.1 and

SULTR1.2 Transporters and Sulfate Accumulation in Roots
Internal SO 4 22 has been proposed as a signal responsible for the repression of sulfate uptake (Smith, 1975(Smith, , 1980Jensén and Kö nig, 1982;Datko and Mudd, 1984;Lass and Ullrich-Eberius, 1984). Particularly, sulfate transporter expression has been considered to correlate negatively to the sulfate content. The rela- Figure 1. Relationship between glutathione content and accumulation of SULTR1.1 and SULTR1.2 mRNA in Arabidopsis roots. The mRNA accumulation of SULTR1.1 (white circles) and SULTR1.2 (black diamonds) is plotted against the root glutathione content. Accumulation of SULTR1.1 and SULTR1.2 mRNA is expressed relative to that of the reference genes as 2DCt values. Each data point was obtained from the analysis of roots collected from a pool of six plants. For both SULTR1.1 and SULTR1.2, each symbol represents one of the different treatments used in this work and is the average of four to six biological repeats. The three glutathione treatments were not considered in this analysis. The two black lines represent linear regression lines corresponding to the least square adjustment of all the data obtained for either the SULTR1.1 or the SULTR1.2 gene; the corresponding Pearson's correlation coefficients (R 2 ) are reported.
tionship between sulfate transporter expression and sulfate content was then examined. As expected, 3 d after sulfate was withdrawn from the culture medium, the sulfate content in roots decreased to approximately 45% of the control value (8.7 6 0.8 mmol g 21 FW compared to 19.5 6 0.3 mmol g 21 FW; Table I). This decrease was negatively correlated to the increases in mRNA accumulations of both SULTR1.1 (4.5-fold) and SULTR1.2 (1.6-fold). The mechanism underlying this negative correlation is well known (Vidmar et al., 2000): when sulfate is withdrawn from the culture solution, the intracellular sulfate pool cannot be renewed; to overcome this sulfate shortage, the plant overexpresses transporters involved in sulfate uptake from the soil solution. When sulfate is present in a nonlimiting concentration in the culture medium, which is the case for all the treatments that we used apart from the sulfate deprivation, the situation may be completely different. Thus, for our further analysis of the relationship linking sulfate content and SULTR1.1 and SULTR1.2 mRNA accumulations in roots, we did not take into account the treatments in which sulfate was withdrawn from the culture medium.
Whatever the culture condition, the root sulfate content remained mostly stable (Table I). It was only markedly altered-actually lowered-in response to the O-acetyl-Ser, the glutathione, and the NaCl treat-ments. The root sulfate content thus displayed a 2.5 range of variation from 8.8 6 0.8 mmol g 21 FW (glutathione treatment) to 23 6 1 mmol g 21 FW (cadmium treatment). A positive correlation explaining 66% of the observed variability linked the SULTR1.1 mRNA accumulation level and the root sulfate content (R 2 5 0.66, P , 0.001; Fig. 3A). Our analysis of this correlation excludes that an increase in the intracellular sulfate content could be the cause of an increase in the expression of SULTR1.1, since such a mechanism would lead to an indefinite increase in sulfate uptake and accumulation. We instead consider that the increase in sulfate accumulation was the result of the increase in SULTR1.1 expression. In contrast, no  . Relationship between sulfate content and accumulation of SULTR1.1 and SULTR1.2 mRNA in Arabidopsis roots. The root sulfate content is plotted against the accumulation of SULTR1.1 (A) and SULTR1.2 (B) mRNA inferred from quantitative PCR analysis. Accumulation of SULTR1.1 and SULTR1.2 mRNA is expressed relative to that of the reference genes as 2DCt values. Each symbol represents one of the different treatments used in this work and is the average of four to six biological repeats. The sulfate-free treatments were not considered in this analysis. The two black lines represent linear regression lines corresponding to the least square adjustment of all the data obtained for either the SULTR1.1 or the SULTR1.2 gene; the corresponding Pearson's correlation coefficients (R 2 ) are reported. correlation was observed between the mRNA accumulation of SULTR1.2 and the root sulfate content (R 2 5 0.02, P 5 0.55; Fig. 3B). Thus, the variation in the root sulfate content appeared to be dependent on the level of SULTR1.1 mRNA accumulation but not on the level of SULTR1.2 mRNA accumulation. This is surprising since SULTR1.2 mRNAs were approximately 10 times more abundant than SULTR1.1 mRNAs (Fizames et al., 2004) and also since SULTR1.2 is known as the major root transporter involved in sulfate uptake from the soil solution (Maruyama-Nakashita et al., 2004b).

Expression of SULTR1.1 Is Locally Regulated in Response to the S Status
The existence of a shoot-to-root signal regulating the sulfate uptake activity has been suggested earlier in rapeseed seedlings (Lappartient and Touraine, 1996). To examine whether the expression of sulfate uptake transporters SULTR1.1 and SULTR1.2 is systemically regulated upon sulfate-deficient conditions or not, split-root experiments were carried out. In these experiments the root system was divided into two parts, and each part was individually submitted to either sulfate deficient (2S) or sufficient (1S) conditions. Split-roots submitted to 1S/2S condition were used as the control conditions of the split-roots submitted to 1S/1S and 2S/2S treatments (Fig. 4). Three days after the 2S treatment, sulfate, O-acetyl-Ser, and glutathione contents and mRNA accumulation for SULTR1.1 and SULTR1.2 have been measured in both 1S and 2S sides of all split-roots. Variations in O-acetyl-Ser and sulfate root contents in response to sulfate deficiency in split-roots were in agreement with results obtained in whole-root experiments. O-acetyl-Ser levels showed an apparent increase (approximately 30%) in all 2S root sides compared to 1S root sides (Fig. 4B). Sulfate content in 2S side of the split-roots decreased to approximately 53% of that of the 1S side (8.8 6 1.8 mmol g 21 FW compared to 16.5 6 1.2 mmol g 21 FW; Fig. 4A). Sulfate contents in the 1S or 2S sides were similar to that of 1S/1S or 2S/2S split-roots, respectively. This result indicates that sulfate content in S-deficient parts of the roots is not compensated by the S-sufficient parts. As expected, glutathione in 2S/2S (83.5 6 6.1 nmol g 21 FW) decreased compared to 1S/1S (100.4 6 7.0 nmol g 21 FW; Fig. 4A). However, unlike sulfate contents, no difference was observed for glutathione content in the 2S side (109.1 6 6.3 nmol g 21 FW) of the split-roots compared to 1S/1S control roots (Fig. 4A). These results suggest that glutathione content in roots is not only dependent on local root sulfate content but also determined by the availability of glutathione in other tissues of the plant.
As in the whole-root experiments (Table I), relative accumulation of SULTR1.2 mRNA in the 2S side (1.52 6 0.2) of split-roots was slightly induced upon sulfate starvation (compared to 1S side, 1.14 6 0.1; Fig.  4D). Interestingly, however, relative accumulation of Plants were grown hydroponically for 5 weeks, and then roots were split in two parts and maintained in complete nutrient solution for an additional week, before each part of split-roots was treated with either sulfate deficient (2S) or sufficient (1S) media. Each side of splitroots was separately harvested 3 d after the treatment, and accumulation of sulfate (black bars), glutathione (white bars; A), and O-acetyl-Ser (hatched bars; B) was determined. Abundance of SULTR1.1 (C) and SULTR1.2 (D) mRNA was normalized against their respective expression in 1S/1S control conditions. Individual measurements were obtained from the analysis of roots collected from a pool of five plants. Error bars correspond to SD; biological repeats (n 56). SULTR1.1 mRNA was strongly increased in the 2S side (3.24 6 0.57), while it remained unchanged in the 1S side (1.17 6 0.15) relative to 1S/1S control roots (1.0 6 0.31). The induction in the 2S side was nearly to the same extent as when both sides were starved for sulfate in 2S/2S control roots (Fig. 4C). The confined induction of SULTR1.1 expression in the 2S side is consistent with a local but not a systemic regulation in response to localized S starvation.
From this data it is obvious that the regulation of the expression of sulfate uptake transporters in Arabidopsis is dependent on local root sulfate content and is not mediated by GSH, as it has been suggested earlier (Lappartient and Touraine, 1996). These results are in agreement with a previous report in which an obvious role for the shoot-to-root signaling to regulate sulfate uptake and expression of the sulfate transporters in Brassica oleracea was not found (Buchner et al., 2004).

SULTR1.1 and SULTR1.2 Depend on Different Regulation Networks
The models currently for the genetic expression of sulfate uptake transporters seldom take into account that there are different genes encoding these transporters. In this study, analyses distinguishing the two high-affinity sulfate transporters SULTR1.1 and SULTR1.2 in Arabidopsis were performed. They showed that these transporters display similar alterations of the abundances of their transcripts in response to various treatments (Yoshimoto et al., 2002;Maruyama-Nakashita et al., 2003, 2004a, 2004b. SULTR1.1 and SULTR1.2 were thus considered as functionally redundant. When considering all our different experimental conditions, we observed a positive correlation explaining only approximately 26% of the observed variation between the mRNA abundances of SULTR1.1 and SULTR1.2 (R 2 5 0.26, P 5 0.001; Fig. 5). This positive correlation was due to data points corresponding to the three glutathione treatments only. When these treatments were not considered, no correlation could be obtained between the mRNA abundances of SULTR1.1 and SULTR1.2 (R 2 5 0.01, P 5 0.651; Fig.  5). The analysis of the results from the salt (100 mM NaCl) or the osmotic (200 mM mannitol) treatments gives a very good example of the lack of coregulation of the expressions of the two genes ( Fig. 6): SULTR1.2 transcripts were markedly overaccumulated in both conditions, while SULTR1.1 transcripts were rather underaccumulated in the same conditions. The validity of this conclusion was confirmed by analyzing hundreds of DNA chips through the Genevestigator ''Gene correlator'' tool (http://www.genevestigator. ethz.ch; Zimmermann et al., 2004); the Pearson's correlation coefficient (R 2 ) determined from the analysis of the mRNA accumulations of SULTR1.1 and SULTR1.2 was also only 0.23 (Supplemental Fig. S1). Altogether, these results suggest that the expressions of SULTR1.1 and SULTR1.2 are depending to a large extent on different regulation networks.
The contrasted mRNA accumulation profiles of SULTR1.1 and SULTR1.2 in response to the different culture treatments are presented in Figure 6. The order of magnitude of the between-treatment variation in mRNA accumulation was twice as high for SULTR1.1 as for SULTR1.2. Moreover, the transcript abundance of SULTR1.1 appeared less stable than that of SULTR1.2. Indeed, the three control treatments showed a marked variation between each other with respect to the mRNA accumulation of SULTR1.1 but not to that of SULTR1.2. Transcript accumulations of SULTR1.1 and SULTR1.2 were depending on different factors. The SULTR1.1 transcript level was mainly altered in response to the S demand: a marked increase in this level was only observed in response to S deficiency and supply of cadmium, while a significant decrease in this level was only noticed in response to the supply of glutathione. In contrast, the accumulation of SULTR1.2 transcripts appeared to be mainly related to a broader carbon (C), N, and S demand: the abundance of SULTR1.2 mRNAs was indeed significantly increased by the supply of Suc, O-acetyl-Ser, and markedly reduced by N deficiency and the supply of glutathione. A relationship between the SULTR1.2 mRNA abundance and a photosynthesis-driven C, N, and S demand was also noticeable when analyzing the diurnal cycle (Fig. 7). SULTR1.2 mRNA accumulation, which was the lowest at the end of the night, regularly increased during the daylight period to reach a maximum in the first part of the night. Such regulation was not observed for SULTR1.1 (Table I). It should be noticed that this latter result appears in contradiction with a previous report showing a diurnal change in SULTR1.1 mRNA (Lejay et al., 2003) similar to the one we observed for SULTR1.2. The discrepancy can be interpreted as follows. In the earlier study, mRNA abundance was determined by northern analysis using a SULTR1.1 probe that showed a high similarity with the SULTR1.2 sequence. Since SULTR1.2 mRNAs are approximately 10 times more abundant than SULTR1.1 ones, it is likely that the authors of this previous study actually analyzed the relative abundance of SULTR1.2 mRNAs. Altogether, these results suggest that the level of mRNA transcripts of SULTR1.2 depends on the general metabolic activity of the plant, while the level of mRNA transcripts of SULTR1.1 is more related to the sole S status of the plants. This functional conclusion can be related to the structural observation that SULTR1.1 and not SULTR1.2 possesses in its promoter a regulatory element involved in the response to sulfate starvation (Maruyama-Nakashita et al., 2005).

DISCUSSION
The molecular mechanisms underlying the regulation of sulfate uptake by roots remain largely unknown. The current model attributes positive and negative regulatory roles to O-acetyl-Ser and glutathione, respectively . However, the model is questioned in the light of a number of experimental outcomes. Data are still lacking to demonstrate that O-acetyl-Ser and glutathione are indeed molecular signals regulating either sulfate uptake or the gene expression of the transporters responsible for it. Several studies showed that regulation of sulfate transporters occurs predominantly at the mRNA level (Hawkesford et al., 1993;Smith et al., 1995Smith et al., , 1997Hawkesford and Wray, 2000;Takahashi et al., 2000;Yoshimoto et al., 2002, Hopkins et al., 2005. Our objective was to test the current model by analyzing the mRNA accumulation of the two Arabidopsis high-affinity sulfate transporters that are involved Figure 6. Effect of individual treatments on the accumulation of SULTR1.1 and SULTR1.2 mRNA. Accumulation of SULTR1.1 and SULTR1.2 mRNA is expressed relative to that of the reference genes as 2DCt values. The treatments to which the plants were submitted are detailed in ''Materials and Methods''. Every data point was obtained from the analysis of roots collected from a pool of six plants. Error bars correspond to SD; biological repeats (4 # n # 6).
in sulfate uptake from the soil solution, SULTR1.1 and SULTR1.2. This study used two strategies that have not been previously utilized to investigate SULTR gene expression. First, a common standardized system was used to assay a large set of different experimental conditions, including conditions that have not been tested before for their effect on the expression of sulfate transporters. Second, the transcript levels of SULTR1.1 and SULTR1.2 were systematically compared to each other but also to the root sulfate, glutathione, and O-acetyl-Ser contents. Numerous studies have shown that glutathione application to plants results in a repression of the expression of sulfate transporters with no exception (Brunold and Schmidt, 1978;Hawkesford and Smith, 1997;Smith et al., 1997;Lappartient et al., 1999;Vidmar et al., 1999Vidmar et al., , 2000Vauclare et al., 2002;Maruyama-Nakashita et al., 2004b). Our data are in full agreement: exogenous applications of glutathione indeed resulted in a reduction in the transcript accumulation of SULTR1.1 and SULTR1.2. However, given the many aspects of cellular metabolism that the glutathione is engaged in (Noctor et al., 2002), such feeding data do not constitute evidence for a direct role in the regulation of SULTR1.1 and SULTR1.2. Indeed, the analysis of our whole data set indicates that there is no correlation between the root glutathione contents and the mRNA levels of either SULTR1.1 or SULTR1.2 (Fig. 1).
Such a lack of relationship between increased levels of glutathione and decreased accumulation of sulfate transporter transcripts in roots has already been documented in barley , and in maize (Zea mays; Nocito et al., 2006). Our results give a broader experimental basis to this observation.
The down-regulation of the expression of sulfate transporters in response to glutathione has actually only been demonstrated following exogenous applications of glutathione. One explanation could be that plants seem to be able to distinguish between the glutathione pool synthesized in the cell and glutathione taken up by the roots from the nutrient solution, and to respond differently in the two situations (Hartmann et al., 2004). Another explanation could be that our glutathione measurements gave access to the average of different cell pool contents that would mask a great heterogeneity in the distribution of glutathione within the root or even within the cells. Intercompartmental variations in glutathione content and redox status were indeed suggested to be crucial in signaling (Noctor et al., 2002). These two explanations might merge: an exogenous application of glutathione perceived at the root epidermis may down-regulate SULTR1.1 and SULTR1.2 since these transporters are expressed in the outer cell layers of the roots, whereas increases in endogenous glutathione contents in inner cell layers resulting from different treatments would not. To our knowledge, it is still impossible to analyze the subcellular distribution of glutathione, and thus to analyze more deeply the effect of glutathione on the expression of SULTR1.1 and SULTR1.2. The above-mentioned explanations still remain insufficient to interpret the whole set of data. Indeed, the N shortage and the 200 mM mannitol treatments both resulted in a similar glutathione accumulation but in a greater than 4-fold variation in SULTR1.2 mRNA accumulation between each other (Table I). Similar discrepancies can be found through the pairwise analysis of many different treatments. It seems unlikely that glutathione would be systematically distributed in different cells or subcellular compartments in response to the different treatments.
Glutathione and sulfate are both translocated within the sieve sap (Rennenberg et al., 1979), although glutathione is the major form of long-distance transport of reduced S in plants (Rennenberg et al., 1979;Herschbach et al., 2000). Interestingly, split-root experiments revealed that glutathione content in the 2S side of the split-roots, unlike sulfate content, remains high, and is comparable to the glutathione content of 1S/1S control roots (Fig. 4A). Two possible explanations may account for the observed high levels of glutathione in 2S side of the split-roots. First, the sulfate translocated from S-sufficient parts of the plant and the metabolized sulfate in the 2S side of roots could contribute to maintain the high levels of glutathione in the 2S side of the root. However, because there is no compensation of sulfate content in S-deficient parts of the root by the S-sufficient parts (Fig. 4A), such Roots of 6-week-old plants grown under an 8-h-day/16-h-night cycle from germination were collected just before switching to day, and 2, 5, and 8 h thereafter. Additionally samples were collected 5 and 10 h after switching back to night. Accumulation of SULTR1.2 mRNA is expressed relative to that of the reference genes as 2DCt values. Every data point was obtained from the analysis of roots collected from a pool of six plants. Error bars correspond to SD; biological repeats (n 56). an explanation cannot be deemed entirely convincing. An alternative explanation would be that the glutathione supply from the S-sufficient parts of the plant could maintain high glutathione levels in the 2S side of the roots. Our results suggest that the glutathione content in roots is not only dependent on local root sulfate contents but also determined by the availability of glutathione in other tissues of the plant. Furthermore, glutathione has been postulated as a systemic regulator; however, using split-root systems no evidence was found for a systemic regulation of the expression of SULTR1.1 and SULTR1.2 in response to sulfate starvation (Fig. 4). These results are in full agreement with a previous report in which the significance of nonprotein thiols as signal in the shoot/root regulation of sulfate uptake and expression of the sulfate transporters in B. oleracea appears to be limited (Buchner et al., 2004). Taken together, our results strongly argue against a relay model in which glutathione would be a significant intermediary in the repression of the expression of either SULTR1.1 or SULTR1.2, even if it may be a component of this regulation.
Role of O-Acetyl-Ser in the Regulation of SULTR1.1 and SULTR1.2 O-Acetyl-Ser has been proposed to be a positive regulator of the expression of sulfate transporters . This proposal was mainly supported by the strong parallelism found between the increase in O-acetyl-Ser content and the transcript abundance of sulfate transporters in both O-acetyl-Ser feeding and S-depletion experiments Kim et al., 1999). Considering similar experimental conditions, our results are in agreement with these earlier observations (Table I). However, the analysis of our whole set of experimental conditions clearly showed that there is no correlation between the root O-acetyl-Ser content and the mRNA levels of either SULTR1.1 or SULTR1.2 (Fig. 2). This result suggests that O-acetyl-Ser does not play a major role in the regulation of these two genes. This conclusion is corroborated by the recent finding that O-acetyl-Ser accumulation in S-starved potato plants was preceded by the increase in both the StST1 sulfate transporter mRNA level and the induction of sulfate uptake (Hopkins et al., 2005). As already detailed in the case of glutathione, it is not obvious to understand why exogenous application of O-acetyl-Ser results in an increase of the transcript level of either SULTR1.1 or SULTR1.2 while changes in endogenous O-acetyl-Ser contents are not correlated to changes in the transcript level of these two genes. Differences in subcellular distributions of O-acetyl-Ser may be part of the explanation. There is, however, a specific feature of O-acetyl-Ser that may help to interpret our results. O-acetyl-Ser is synthesized from Ser and acetyl-coenzyme A by the Cys synthase protein complex. This complex is also involved in the synthesis of Cys from O-acetyl-Ser and sulfide. Importantly, dissociation and stabilization of the Cys synthase complex is achieved by O-acetyl-Ser and to a lesser extent Cys on one side, and sulfide on the other side, respectively . When the complex is formed, which is mainly dependent on the availability of sulfide, O-acetyl-Ser is metabolized to Cys. However, when sulfide is lacking or when there would be an excess of O-acetyl-Ser compared to sulfide, part of O-acetyl-Ser would not be metabolized. The current hypothesis is that in that case, free O-acetyl-Ser would play a role in up-regulating the expression of sulfate transporters to ultimately increase the sulfide pool. In that context, it would not be the root O-acetyl-Ser content alone that would be the critical variable regulating the expression of sulfate uptake transporters, but the resultant relative contents of O-acetyl-Ser, sulfide, and maybe Cys.

Relationship between Expression of SULTR1.1 and SULTR1.2 and Sulfate Content in Roots
Our work aimed at characterizing the regulation of SULTR1.1 and SULTR1.2 at the mRNA accumulation level and not at analyzing the role of these transporters in sulfate uptake, and we concluded that the root sulfate content did not play any role in the regulation of either SULTR1.1 or SULTR1.2 gene expression when sulfate supply from the medium is not limited. However, our analysis surprisingly revealed that the root sulfate content showed a strong positive correlation with mRNA accumulation of the less expressed of the sulfate transporters SULTR1.1 and not with mRNA accumulation of the more expressed SULTR1.2. The contradiction might be only apparent. First, considering that the regulation of SULTR1.2 might be dependent on the metabolic demand, it should be reasonable to consider that the sulfate taken up following the increase in SULTR1.2 expression is rapidly reduced and metabolized and thus not accumulated. The fact that SULTR1.1 transcripts are approximately 10 times less abundant than those of SULTR1.2 should not be readily interpreted to mean SULTR1.1 has a secondary role. SULTR1.1 transcript and protein accumulation increases under sulfate starvation (Yoshimoto et al., 2007). The increase in SULTR1.1 protein levels, following the increase in its transcript levels, could partially provide an explanation for its contribution in accumulation of sulfate in roots. It is also possible that a synergistic functional interaction between SULTR1.1 and SULTR1.2 could explain why an increase in SULTR1.1 expression had a major impact on root sulfate accumulation. Consistent with this idea, two other sulfate transporters, SULTR2.1 and SULTR3.5, have been shown to synergistically interplay to increase the sulfate transport capacity in yeast (Kataoka et al., 2004). In plants, hetero-oligomerization of different subunits appears to be a typical feature to alter transport properties as exemplified in the case of Suc transporters (Reinders et al., 2002), ammonium transporters (Ludewig et al., 2003), and channels (e.g. K 1 channels; Obrdlik et al., 2004). From these considerations, we hypothesize, that interplay between SULTR1.1 and SULTR1.2 transporters could be an important mechanism to regulate sulfate content in the roots.
Regulation of the Expression of SULTR1.1 and SULTR1.2 Involves Partially Independent Signaling Pathways SULTR1.1 and SULTR1.2 are the two high-affinity sulfate transporters that are expressed in the same set of cells at the root-soil interface and contribute to sulfate uptake from the soil solution (Takahashi et al., 2000;Vidmar et al., 2000;Shibagaki et al., 2002;Yoshimoto et al., 2002;El Kassis et al., 2007). They have up to now been considered as equivalent but not equal (Maruyama-Nakashita et al., 2003). The concept of equivalence was established on the basis that these two transporters showed similar biochemical characteristics and similar regulations of expression (Yoshimoto et al., 2002;Maruyama-Nakashita et al., 2003, 2004a, 2004b. But these two transporters were not considered as equal, since SULTR1.1 plays a minor role in comparison to SULTR1.2; SULTR1.1 mRNAs are approximately 10 times less abundant than SULTR1.2 ones and SULTR1.1 contributes to less than 30% of the sulfate uptake (Shibagaki et al., 2002;Yoshimoto et al., 2002;El Kassis et al., 2007). Our results are not in agreement with the concept of equivalence, at least as far as regulation of expression is considered. SULTR1.1 and SULTR1.2 were shown to be coregulated only to a limited extent (r 2 ; 0.26) and this weak correlation between their levels of expression was only attributable to the glutathione treatments (Fig. 5). Data extracted from the available sets of Arabidopsis microarray data revealed a similar weak correlation (r 2 ; 0.25) between SULTR1.1 and SULTR1.2 mRNA accumulations (Supplemental Fig.  S1; Zimmerman et al., 2004). Thus, regulation of the expression of SULTR1.1 and SULTR1.2 involves to a significant extent independent signaling pathways.
Our analysis suggests a specialization in the regulation of these two transporters. The abundance of SULTR1.1 mRNA transcripts was mainly altered in response to S-related treatments, whereas the level of SULTR1.2 mRNA transcripts was mainly altered by changes in the C, N, and S demands (Fig. 8). Interestingly, we found that SULTR1.1 expression is locally regulated in response to sulfate starvation treatment, whereas such a regulation is not observed for SULTR1.2 as demonstrated by results obtained in split-root experiments (Fig. 4, C and D). The identification of an S specificity of the SULTR1.1 regulation is consistent with the presence of a S-responsive cisregulatory element (SURE element) in the SULTR1.1 promoter but not in the SULTR1.2 promoter (Maruyama-Nakashita et al., 2005). In Arabidopsis, this element has been identified in promoters of different S-responsive genes and was shown to be necessary and sufficient for mediating a negative regulation by sulfate, Cys, or glutathione in the SULTR1.1 promoter (Maruyama-Nakashita et al., 2005). The differential regulation of SULTR1.1 and SULTR1.2 is further strengthened by the recent identification of a trans-regulatory element, the EIL3 transcription factor, which strongly alters the transcription of SULTR1.2 but has a limited effect on SULTR1.1 (Maruyama-Nakashita et al., 2006). These results support our analysis that SULTR1.1 and SULTR1.2 are subjected to some extent to different regulatory networks.
Our proposal that SULTR1.2 is transcriptionally regulated by the metabolic demand is mainly supported by our data showing that the expression of SULTR1.2 is diurnally regulated and is also dependant on the S, N, and C availabilities (Table I; Fig. 7). In this context, sulfate uptake would appear to be coordinated with sulfate assimilation since regulation of the expression of adenosine 5-phosphosulfate reductase, the key enzyme of sulfate assimilation pathway, has also been shown to be dependent from the S, N, and C demands . Sulfate uptake from the soil solution was shown to be essentially ensured by the SULTR1.2 sulfate transporter (Maruyama-Nakashita et al., 2003). It would actually not be surprising that the transporter that plays a major role in sulfate uptake is regulated depending on the metabolic demand. All together, these arguments suggest that the metabolic demand plays a major role in the regulation of SULTR1.2 expression.
Similar to the case of SULTR1.2, other transporters involved in the uptake of essential macronutrients such as N (e.g. NRT2.1), are also transcriptionally regulated depending on the S, N, and C demands (Lejay et al., 2003). This is reminiscent of data showing that sulfate assimilation is well coordinated with nitrate and C assimilation . Interestingly, recent data have confirmed close connections between C, N, and S utilization and cytokinin in plants, pointing out to novel metabolic and transport processes that might be under the influence of cytokinin (Ohkama et al., 2002;Maruyama-Nakashita et al., 2004c;Brenner et al., 2005;Sakakibara et al., 2006). Cytokinins have been reported to signal the plant N, S, sugar, or P status (Sakakibara et al., 1999;Martín et al., 2000;Gessler et al., 2004;Franco-Zorrilla et al., 2005). Also, synthesis of cytokinins depends on the availability of different macronutrients: cytokinin contents are increased by nitrate supply (Takei et al., 2001;Collier et al., 2003), and decreased by N or P starvation (Salama and Wareing, 1979;Horgan and Wareing, 1980). Finally, applications of exogenous cytokinin were shown to repress the expression of transporters involved in sulfate, nitrate, or phosphate uptake and to reduce the net nitrate uptake (Martín et al., 2000;Maruyama-Nakashita et al., 2004b;Brenner et al., 2005;Dluzniewska et al., 2006). It has been well demonstrated that cytokinins significantly down-regulate SULTR1.1 and SULTR1.2 with major impact on the latter; however, it should be noted that this negative effect is independent of the S limitation response (Maruyama-Nakashita et al., 2004c). Taken together, cytokinins, as common integrative mediators in signaling of plant nutritional status, could take part in regulation of SULTR1.2 expression in response to metabolic demand.
In conclusion, this work is the first extensive comparison of the effect of a wide set of culture conditions on accumulation of SULTR1.1 and SULTR1.2 transcripts. The expression of SULTR1.1 and SULTR1.2 was shown not to be regulated consistently to the current model. Our results also show for the first time that the regulation of these two high-affinity sulfate transporters depend to a significant extent on different signaling pathways, and raise the possibility of an interplay between SULTR1.1 and SULTR1.2 to regulate the sulfate content in the roots. These observations indicate that the regulation of the root sulfate uptake is more complex than previously thought and pave the way for further work reconsidering the nature of signal(s) involved in the regulation of the expression of SULTR1.1 and SULTR1.2. Determining additional level of regulation acting on the sulfate transporters will be required to fully appreciate the regulation mechanisms for sulfate transport in plants.

Plant Growth Conditions
The Arabidopsis (Arabidopsis thaliana) plants used in all experiments were obtained from the Nottingham Arabidopsis Stock Centre (Columbia  ecotype; reference N60000). Plants were grown hydroponically under nonsterile conditions. Seeds were germinated directly on top of pierced Eppendorf tubes filled with sand and placed on rafts floating in a nonsterile hydroponic solution, as described previously by Lejay et al. (1999). Plants were germinated and grown for 1 week in tap water then in the following nutrient solution: 0.5 mM KNO 3 ; 1 mM MgSO 4 ; 1 mM KH 2 PO 4 ; 0.25 mM Ca(NO 3 ) 2 ; l00 mM NaFeEDTA; 30 mM H 3 BO 3 ; l0 mM MnCl 2 ; l mM CuCl 2 ; 1 mM ZnCl 2 ; 0.1 mM (NH 4 ) 6 Mo 7 O 24 ; and 50 mM KCl. The nutrient solution was renewed every 4 d and on the day before the beginning of every treatment. Plants were grown in a growth chamber under the following environmental conditions: light/dark cycle of 8 h/16 h, light intensity of 250 mmolÁm 22 Ás 21 , temperature of 24/20°C, and RH of 75%. The treatments were applied to 6-week-old plants. For the control condition, plants were kept in culture medium with the abovementioned composition. For the sulfate-(2S), phosphate-(2P), or N-free (2N) treatments, plants were transferred into nutrient solutions of similar composition as described above, except that 1 mM MgSO 4 , 1 mM KH 2 PO 4 , or 0.5 mM KNO 3 and 0.25 mM Ca(NO 3 ) 2 were replaced by 1 mM MgCl 2 , 1 mM KCl, or 0.5 mM KCl and 0.25 mM CaCl 2 , respectively. Plants were grown in media lacking S for either 24 or 72 h, and in media lacking N or P for 24 h only. Other treatments consisted of growing the plants in the above-mentioned nutrient solution supplemented with 0.1, 0.5 or 1 mM glutathione, 0.2 mM O-acetyl-Ser, 20 mM cadmium (CdCl 2 ), 0.5 mM selenate (Na 2 SeO 2 ), 100 mM NaCl, or 200 mM mannitol for 24 h. A Suc treatment was performed by adding 30 mM Suc to the nutrient solution immediately after light was switched off; the plants were then grown for 4 h in this solution, as previously described in Lejay et al. (2003). A treatment with 30 mM mannitol was used as an osmotic control for the Suc treatment. At the end of the treatments, roots were cut, immediately frozen in liquid N, and stored at 280°C.
The split-root experiments were essentially performed as previously described by Gansel et al. (2001). Briefly, plants were grown hydroponically on complete media as described above. After 2 weeks, seedlings were removed so that only one seedling per tube was maintained. The root system of 5-week-old plants was gently separated into two approximately equal parts, and each part was transferred to separate containers supplied with 1 mM MgSO 4 solution and allowed to adapt for 3 d to split-root conditions. After washing the roots with double deionized water, the localized supply treatments were initiated by supplying each part of the root with either a sulfate-free (2S) or complete nutrient solution (1S). The side of the root system containing the main root was treated with 1S or 2S in an equal number of replicates. After 3 d of treatment, root metabolite (glutathione, O-acetyl-Ser, and sulfate) contents were expressed on an FW basis.

Real-Time Quantitative Reverse-Transcription PCR
Total RNA was extracted from frozen root tissues using the Plant RNeasy extraction kit (QIAGEN). Any residual genomic DNA was eliminated using an RNAse-free DNAse I (QIAGEN). Total RNA was quantified with a UV spectrophotometer. Two micrograms of total RNA were reverse transcribed using the SuperscriptIII RT kit (Life Technologies/Gibco-BRL). Complementary DNA (cDNA) was kept at 220°C until analyzed. Real-time reversetranscription PCR was performed with an ABI Prism 7700 sequence detection system, using the SYBR green PCR master mix kit (Applied Biosystems). Reactions were performed in MicroAmp 96-well plates (Applied Biosystems) covered with optical adhesive covers (Applied Biosystems). PCR reactions were performed in a final volume of 25 mL containing 300 nM each of the forward and reverse primers, 12.5 mL of the SYBR green master mix, and 5 mL of a 1:50 cDNA dilution. All PCR reactions were performed in triplicate. Samples were submitted to 50°C for 2 min, then to 95°C for 10 min, and finally to 40 cycles of 95°C for 15 s followed by 60°C for 1 min, except for SULTR1.1 and SULTR1.2 for which the annealing/extension temperature was adjusted to 65°C. Data were analyzed using the Sequence Detection System software version 1.7 (Applied Biosystems). The specificity of the amplified PCR products was systematically confirmed by a melting curve analysis using the ABI PRISM Dissociation Analysis software (Applied Biosystems). Performing real-time PCR on RNA samples that were not subjected to cDNA synthesis checked the absence of genomic DNA contamination. We also checked that reactions performed without template did not result in any PCR product.
Quantification of the relative transcripts levels was performed using the comparative C T method (Livak and Schmittgen, 2001). For every data point, the threshold cycle (C T ) value was the average of the C T values obtained from the triplicate PCR analysis. For every treatment, four to six biological replicates were performed, leading to four to six corresponding samples. For every sample, the relative gene expression of each of the SULTR genes DC T,SULTR was expressed following normalization against the average of the C T values obtained for the genes used for standardization: DC T,SULTR 5 C T,SULTR 2 (average C T,standards ). When a treatment of interest (TOI) was compared to a reference treatment (RT), for instance for pairwise comparisons in chart bars, the relative expression of a SULTR gene was expressed as a DDCt value calculated as follows: DDCt 5 DC T,TOI 2 DC T,RT . The fold change in relative gene expression was determined as 2 2DDCT .

Statistical Analysis
For statistical analysis of quantitative data, the Analysis Toolpak add-in program for Excel 2004 for Macintosh (Microsoft) was used to calculate the square of Pearson product-moment correlation coefficient (R 2 ) and Student's t tests to establish if the correlation coefficient is significantly different from zero. For all the t-test analyses the difference is considered statistically significant with a probability of P , 0.05.

Glutathione Measurements
The glutathione extraction protocol was adapted from Creissen et al. (1999) and glutathione was analyzed using monobromobimane (Newton et al., 1981). Briefly, weighed root material (approximately 20 mg) was extracted with 3 volumes of 0.1 M HCl during 15 min at 4°C. The insoluble residues were removed by centrifuging at 20,000g for 15 min at 4°C. Extracts were neutralized by adding an equal volume of 0.1 M NaOH. Aliquots of 50 mL were subjected to derivation by adding 35 mL of water, 10 mL of 1 M Tris-HCl, pH 8, and 5 mL of 10 mM monobromobimane (Interchim). Samples were incubated for 15 min at 37°C before the addition of 45 mL of pure acetic acid. Separation of monobromobimane-derivatized thiols was achieved by reversed-phase HPLC (Nucleodur C18 column, 250 3 4.6 mm; Macherey-Nagel). The compounds were eluted at a flow rate of 0.8 mL/min using the following twophase acetonitrile gradient in 0.1% (v/v) trifluoroacetic acid: acetonitrile concentration was raised from 0% to 10% (v/v) in 15 min, then from 10% to 99.9% in 2 min. Fluorescence of glutathione derivatives was measured with a ProStar Fluorescence detector (Varian; excitation 385 nm; emission 484 nm). Identification and quantification of glutathione were performed by comparison of retention times and peak areas with samples containing known amounts of pure synthetic glutathione (G4251; Sigma-Aldrich) and analyzed using the same procedure.

Sulfate Measurements
Weighed fresh roots were grounded into powder in liquid N and extracted in water by incubation for 30 min at 70°C. The extract was centrifuged at 20,000g for 30 min, and the supernatant filtered through a 0.45-mm filter unit. Ion concentrations were determined by high-pressure ionic chromatography (ICS-2500 apparatus; Dionex) using the AS11 anion exchanging column (Dionex) and a NaOH gradient. NaOH concentration was raised from 3.5 to 5 mM in 3 min, then linearly from 5 to 35.4 mM in 8 min. Identification and quantification of sulfate were performed by comparison of the retention times and peak areas-integrated using the Chromeleon software (Dionex)-with the standards.

O-Acetyl-Ser Measurements
Metabolites were extracted with 3 volumes of 0.1 M HCl using 0.02 g FW of root material that was grinded in liquid N to a fine powder. Cell debris were sedimented by centrifuging at 16,800g for 15 min at 4°C. The resulting supernatants were used for metabolite analysis after a short-term storage at 280°C. O-Acetyl-Ser was quantified using the AccQ-Tag fluorescence dye (Waters) after separation by reversed-phase HPLC on a Nova-Pak C18 3.9 3 150 mm column (pore size 4 mm) according to Wirtz et al. (2006). Chromatograms were recorded with a Jasco 920 fluorescent detector at 250-nm excitation wavelength and 395-nm emission wavelength and processed with the Millenium 32 software.

Supplemental Data
The following materials are available in the online version of this article.