Inter-organ signaling in plants: regulation of ATP sulfurylase and sulfate transporter genes expression in roots mediated by phloem-translocated compound


  • Anne G. Lappartient,

    1. Biochimie et Physiologie Moléculaire des Plantes, ENSA-M/INRA/CNRS URA 2133/Université Montpellier 2, 34060 Montpellier Cedex 1, France,
    Search for more papers by this author
    • These authors contributed equally to this work.

    • Present address: Rhône-Poulenc Agrochimie, CNRS UMR 41, BP 9163,

  • J. John Vidmar,

    1. Biochimie et Physiologie Moléculaire des Plantes, ENSA-M/INRA/CNRS URA 2133/Université Montpellier 2, 34060 Montpellier Cedex 1, France,
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Thomas Leustek,

    1. Center for Agricultural Molecular Biology, Rutgers University, New Brunswick, NJ 08903, USA, and
    Search for more papers by this author
  • Anthony D. M. Glass,

    1. Department of Botany, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada
    Search for more papers by this author
  • Bruno Touraine

    1. Biochimie et Physiologie Moléculaire des Plantes, ENSA-M/INRA/CNRS URA 2133/Université Montpellier 2, 34060 Montpellier Cedex 1, France,
    Search for more papers by this author

  • 14–20 rue Pierre Baizet, 69263 Lyon Cedex 09, France.

*For correspondence (fax +33 4675 25737; e-mail


Sulfate uptake and ATP sulfurylase activity in the roots of Arabidopsis thaliana and Brassica napus were enhanced by S deprivation and reduced following resupply of SO42–. Similar responses occurred in split-root experiments where only a portion of the root system was S-deprived, suggesting that the regulation involves inter-organ signaling. Phloem-translocated glutathione (GSH) was identified as the likely transducing molecule responsible for regulating SO42– uptake rate and ATP sulfurylase activity in roots. The regulatory role of GSH was confirmed by the finding that ATP sulfurylase activity was inhibited by supplying Cys except in the presence of buthionine sulfoximine, an inhibitor of GSH synthesis. In direct and remote (split-root) exposures, levels of protein detected by antibodies against the Arabidopsis APS3 ATP sulfurylase increased in the roots of A. thaliana and B. napus during S starvation, decreased after SO42– restoration, and declined after feeding GSH. RNA blot analysis revealed that the transcript level of APS1, which codes for ATP sulfurylase, was reduced by direct and remote GSH treatments. The abundance of AST68 (a gene encoding an SO42– transporter) was similarly affected by altered sulfur status. This report presents the first evidence for the regulation of root genes involved in nutrient acquisition and assimilation by a signal that is translocated from shoot to root.


The potential capacities for inorganic ion uptake by plant roots and their subsequent assimilation are high, but the actual rates of nutrient acquisition may be down-regulated according to plant demand rather than by ion concentrations in the rhizosphere. The demand-driven regulatory processes that are responsible for this control have been described at a physiological level. It is known, therefore, that these processes are ion-specific and involve signals originating from the shoot. Underlying mechanisms, however, remain obscure. In particular, the identities of the root targets for these inter-organ signaling processes are unknown. In this paper, we address this question with regards to S acquisition by roots.

Several reports document that sulfur starvation leads to a several-fold increase in the rate of SO42–, but not NO3 and H2PO4–, uptake by roots ( Clarkson et al. 1983; Datko & Mudd 1984; Lee 1993). We previously showed that the rate of SO42– uptake and the activity of ATP sulfurylase, the enzyme catalyzing the first step of the S assimilation pathway, increased simultaneously in Brassica napus plants following SO42– withdrawal from culture solution, and then decreased after the restoration of SO42– supply ( Lappartient & Touraine 1996). These parallel patterns of de-represssion/repression were also evident in split root experiments where the roots were continuously supplied with SO42–, while other roots of the same plant where fed with an S-free medium. These results are critical because they demonstrate that the responses of ATP sulfurylase activity and SO42– uptake to S availability were not directly evoked by the composition of external solution supplied to roots, but rather were elicited by changes in plant nutritional status. The pattern of this regulatory system thus follows the general scheme postulated for other ions ( Clarkson & Lüttge 1991; Imsande & Touraine 1994; Touraine et al. 1994), whereby long-distance transport of a by-product of their assimilation represses their further absorption. Among the sulfur compounds that are accumulated and transported within the plants, we identified glutathione, a thiol known to inhibit SO42– uptake ( Herschbach & Rennenberg 1991; Herschbach & Rennenberg 1994; Rennenberg et al. 1989), as being this long-distance signal ( Lappartient & Touraine 1996). Recently, it has been shown that S starvation increased the abundance of mRNAs of genes encoding SO42– transporters ( Smith et al. 1997; Takahashi et al. 1997) and, to a lesser extent, the abundance of the APS2 (ASA1) ATP sulfurylase mRNA ( Logan et al. 1996). Together, these results suggest that the demand-driven regulation of ATP sulfurylase and SO42– uptake in roots is mediated via decreased abundances of the corresponding mRNAs in response to high levels of phloem-translocated GSH. Here, we present evidence that supports this hypothesis. This is the first report of the down-regulation of genes involved in mineral nutrition in roots by inter-organ signaling processes.


ATP sulfurylase activity and SO42– influx in the roots of Arabidopsis thaliana are regulated by demand-driven processes

Both SO42– influx and ATP sulfurylase activity increased in Arabidopsis thaliana plants upon the withdrawal of SO42– and declined sharply upon its restoration ( Fig. 1a). Feeding SO42– grown A. thaliana plants with 1 m m GSH diminished both activities ( Fig. 1b). Globally, these response patterns are essentially identical to those observed in B. napus ( Lappartient & Touraine 1996).

Figure 1.

Responses of SO42– influx and ATP sulfurylase activity in the roots of intact Arabidopsis thaliana to S starvation (a) and GSH supply (b).

Plants were grown on a complete nutrient solution containing 1.5 m m SO42– until the onset of treatment. Treatments (-S, transfer to S-free solution; +S, transfer to SO42– containing solution after a 3 day S-starvation period; GSH, 1 m m GSH added into the culture solution) were conducted for various times as indicated on the X-axis, so that analyses were always performed on 18-day-old plants. Data shown are the means of five replicates (independent batches of 20 plants), error bars represent SD.

It is noteworthy that, as previously reported for B. napus, these responses correlated negatively with internal concentrations of thiols, especially Cys and GSH, but not with those of SO42–. For example, whereas the concentrations of SO42–, Cys and GSH all decreased in the roots of S-starved A. thaliana, those of SO42 – grown plants treated with GSH for 2 h contained 2–3 times more Cys and GSH, but the same level of SO42– than controls (plants grown with SO42– and without GSH).

Expression of ATP sulfurylase protein in roots is affected by S status

In order to specifically address the question of inter-organ regulation recognized at the physiological level in B. napus, we performed split-root experiments which typically consisted of distributing the root systems of intact 21-day-old plants between two compartments in a 2:1 ratio. A modified culture solution (e.g. SO42 – free or 1 m m GSH added) was supplied to the larger compartment, whilst maintaining the smaller compartment, from which the proteins were extracted under control conditions (i.e. SO42– but not GSH). Using antibodies raised against the APS3 protein isolated from Arabidopsis thaliana ( Murillo & Leustek 1995), two bands were detected on all B. napus immunoblots ( Fig. 2a,b,d). The band of higher molecular mass corresponds to the predicted size of the purified polypeptide (50 kDa; Murillo & Leustek 1995). Because only this band was visible on A. thaliana immunoblots ( Fig. 2c,e), it must represent an ATP sulfurylase encoded by a gene of the APS1-APS3 family in B. napus roots, while the band of lower molecular mass (approximately 47 kDa) probably represents a cross-reacting protein. Whatever the identity of the 47 kDa protein, its steady state level correlated with that of the 50 kDa protein. In split root experiments, when S feeding was disrupted to a portion of the root system, the level of immunodetected protein in the remaining untreated portion of the root system increased several-fold within the first 24 h ( Fig. 2b). Conversely, it declined to a very low amount upon re-supply of SO42–. Identical responses were observed when the entire root system was exposed to the S-free solution ( Fig. 2a). When GSH was supplied, the APS3 polypeptide declined in abundance in the untreated portion of the root system ( Fig. 2d). Experiments repeated in A. thaliana revealed no differences from B. napus in response to either treatment ( Fig. 2c,e, respectively).

Figure 2.

Levels of the anti-APS3 immunodetected polypeptide in the roots of Brassica napus and Arabidopsis thaliana in response to S-starvation or GSH-supply.

Plants were grown hydroponically in complete nutrient media containing 1.5 m m (A. thaliana) or 1 m m (B. napus) SO42–. Roots were harvested from 18-day-old (A. thaliana) or 21-day-old (B. napus) plants, following periods of treatments as indicated, and proteins were extracted. Total proteins were electrophoretically separated and transferred to nitrocellulose membranes, and ATP sulfurylase polypeptide was detected with antibodies raised against APS3.

(a) B. napus, SO42– deprivation for 0–72 h (time as indicated) and SO42– re-feeding (72/24). (b) B. napus, similar experiment as in Fig. 2(a), except that approximately two-thirds of the root system of each plant were deprived of S for the various periods, and that proteins were extracted from the remaining roots, which were continuously fed with SO42– (split-root experiments). (c) A. thaliana, SO42– deprivation for 0–72 h (time as indicated) and SO42– re-feeding (72/24). (d) B. napus, GSH supply for 0–24 h, split-root experiments (see Fig. 2b). (e) A. thaliana, GSH supply for 0–24 h.

Accumulations of ATP sulfurylase and SO42– transporter mRNAs in roots are affected by the S status

Using APS1, a cDNA isolated from A. thaliana sharing an 86% identity with APS3 ( Leustek et al. 1994 ; Murillo & Leustek 1995), we were able to detect ATP sulfurylase transcript in all samples from roots of B. napus analyzed. Its abundance declined in response to GSH treatment, either in whole-root (not shown) or split-root plants ( Fig. 3a). Similar patterns were obtained when treating whole roots of A. thaliana with 1 m m GSH ( Fig. 3b).

Figure 3.

Effects of SO42–, GSH or Cys supply on the content of Cys and GSH, and on the level of APS1 and AST68 transcripts in roots from Brassica napus and Arabidopsis thaliana.

Plants grown as in Fig. 2. Thiol values represent the average of four replicates (error bars, SD).

(a) GSH content and APS1 transcript level in roots of B. napus supplied with 1 m m GSH, split root experiment (see text). (b) GSH content and APS1 transcript level in roots of A. thaliana supplied with 1 m m GSH. (c) GSH content and AST68 transcript level in roots of A. thaliana starved of SO42– (-S), and starved then re-supplied with SO42– (+S). (d) Cys and GSH contents, and APS1 transcript level in the roots of A. thaliana grown on normal SO42 – containing culture solution (control), to which 1 m m BSO, 1 m m Cys or both were eventually added for 24 h, as indicated above the bars. (e) Cys and GSH contents in the roots of A. thaliana starved of SO42– for 48 h prior to the onset of treatments. Treatments indicated above the bars were supplied at 1 m m concentration for 24 h. (f) APS1 and AST68 transcripts level in the roots of A. thaliana plants treated as in Fig. 3(e).

The decreased abundance of APS1 transcript was observed in roots from SO42 – grown A. thaliana plants that were treated with 1 m m Cys for 24 h ( Fig. 3d). These plants, however, accumulated Cys and GSH, presumably resulting from the metabolism of Cys to GSH. To distinguish between the effects of Cys and GSH, we used BSO to inhibit γ-glutamylcysteine synthetase. This enzyme catalyzes the phosphorylation of BSO, whose product is tightly bound to the enzyme and is thus irreversibly inhibited ( Griffith 1982). The S-alkyl moiety of BSO that binds at the acceptor amino acid site of the enzyme confers a high specificity for γ-glutamylcysteine synthetase ( Griffith & Meister 1979). Simultaneous treatment with BSO and Cys blocked the accumulation of GSH in Arabidopsis roots ( Fig. 3d). In addition, BSO blocked the Cys-induced decrease in APS1 transcript ( Fig. 3d), and the Cys-induced decline in ATP sulfurylase activity ( Lappartient & Touraine 1996).

In order to further discriminate between the effects of Cys and GSH in root tissues, BSO experiments have been performed in 48 h S-starved plants. The impact of various treatments (the addition of BSO and Cys, separately or together, or SO42– or GSH for 24 h) on internal Cys and GSH levels are given in Fig. 3(e). As shown in Fig. 3(f), the abundance of APS1 mRNA declined in response to all the treatments that increased the GSH content (SO42–, Cys, GSH), but remained essentially unchanged in plants treated with BSO (alone or with Cys) that had higher Cys levels, but not higher GSH levels.

The transcript level of a root sulfate transporter (AST68) was investigated. It has been reported that the abundance of AST68 transcript in Arabidopsis grown on an agar medium is strongly induced by sulfur starvation ( Takahashi et al. 1997 ). In the roots of hydroponically grown Arabidopsis, AST68 mRNAs were barely detectable when plants were continuously fed SO42–. Conversely, the AST68 transcript level dramatically increased upon S-starvation, and sharply declined after the restoration of SO42– to S-starved plants ( Fig. 3c). Due to the low level of transcript in the roots of SO42 – grown plants, it was impossible to determine the effect of GSH feeding on the abundance of AST68 mRNA. However, in plants starved of S for 48 h, the abundance of AST68 transcript declined in response to both GSH and Cys feeding and BSO blocked effect of Cys ( Fig. 3f).


Observed modifications of metabolism or transport in roots when nutrient solution is changed may involve both local effects in this organ and responses to altered nutritional demand of the whole plant. To discriminate between these processes, it is necessary to perform split-root experiments where the observed responses necessarily involve increased or decreased rates in root–shoot cycling of some compounds in the xylem and phloem. Depending on the treatment imposed, these regulatory signals may originate either from the shoot (as a consequence of changes in metabolic rates in leaves due to changed nutrient availability), or from the treated roots. In whichever organ the signal originates, it enters the untreated roots exclusively via the phloem pathway. 6(5)carboxyfluorescein is a fluorescent compound that can diffuse in the apoplasm, but remains strictly confined to the symplasm once it has crossed the plasma membrane ( Grignon et al. 1989; Grignon et al. 1992). When this dye is supplied to one part of the root system, it is transported to other roots where it is strictly localized in the phloem tissue (N. Grignon et al. unpublished results).

Conducting split-root experiments in Brassica napus, we have demonstrated that the regulation of ATP sulfurylase protein and mRNA underlies the responses of root ATP sulfurylase activity to S deprivation ( Lappartient & Touraine 1996). The finding that A. thaliana, from which the specific nucleotidic and proteic reagents were obtained ( Leustek et al. 1994; Murillo & Leustek 1995), responds similarly indicates the conservation of the response in both species. Physiological studies showed that ATP sulfurylase activity and SO42– influx in roots displayed remarkably parallel responses to S starvation in A. thaliana ( Fig. 1) as in B. napus ( Lappartient & Touraine 1996). Furthermore, since the up-regulation of SO42– uptake was visible in membrane vesicles isolated from S-starved B. napus plants ( Hawkesford et al. 1993), the response of SO42– uptake to S deprivation is likely due to changes in the levels of transporter proteins. Consistent with this hypothesis, mRNAs that hybridize with AST68 in Arabidopsis roots were up-regulated in S-starved plants and down-regulated when SO42– was restored ( Fig. 3c). Accumulation of transcripts coding for high affinity SO42– transporters in roots of S-starved Stylosanthes hamata ( Smith et al. 1995) and Hordeum vulgare ( Smith et al. 1997; Vidmar et al. 1997) have also been reported. Therefore, demand-driven regulation of ATP sulfurylase activity and SO42– uptake, involving the inter-organ signaling process between shoot and root, is likely to operate in both Arabidopsis thaliana and Brassica napus, and to trigger repression/de-repression of gene expression in roots.

Our earlier work in B. napus ( Lappartient & Touraine 1996; Lappartient & Touraine 1997) suggests that the control of root ATP sulfurylase activity and SO42– uptake is mediated by phloem-translocated GSH. Consistent with this hypothesis, the abundance of transcripts in sulfur-starved plants was increased, and feeding plants with GSH resulted in a decline in transcript levels ( Fig. 3a). Because the GSH supply resulted in increased accumulation of both Cys and GSH, both compounds might be responsible for these effects. Phloem sap analyses in split-root GSH-fed plants, however, indicated that GSH rather than Cys is the phloem-translocated signal acting in B. napus ( Lappartient & Touraine 1996). Here we present further evidence in favor of this hypothesis in A. thaliana. In SO42– grown plants as well as in S-starved plants, the repression of APS1 mRNA accumulation by external Cys was relieved by BSO ( Fig. 3d,f, respectively). The expression of AST68 in S-starved Arabidopsis followed the same pattern ( Fig. 3f). Because the level of expression of this gene was so low in unstarved plants, it has been impossible to study its response to the supply of GSH in plants adequately fed with SO42–. It is likely that several SO42– transporters co-exist in the plasma membrane of root cells, and the gene coding for one of these may be expressed at a higher level than AST68 and hence be more useful for investigating the response of SO42– uptake to GSH. Nevertheless, our results clearly show that GSH rather than Cys acts as a repressor of the SO42– transporter and ATP sulfurylase genes in roots. In Cys or GSH treated plants, it is possible that Cys oxidization releases SO42– in the roots, as demonstrated in cultured tobacco cells ( Harrington & Smith 1980). However, this would not explain the results reported in this paper. First, no increase in the concentration of SO42– in root tissues was measured in GSH-treated A. thaliana (see Results) or B. napus plants for either split-root or whole-root experiments ( Lappartient & Touraine 1996). Second, analyses of phloem exudates showed that SO42– is a minor component of sieve sap in B. napus compared to GSH, and that it does not vary in concentration in response to a change in the SO42– supply (II in Lappartient & Touraine 1996) or GSH feeding (data not shown). Third, if SO42– was produced in Cys treatment, this production would be even higher in plants treated with both Cys and BSO, which would not support an effect of internal SO42– on the APS1 and AST68 transcript level since BSO relieved the inhibition of their accumulation by Cys.

Although it remains to be determined whether the down-regulation of ATP sulfurylase and SO42– transporter transcripts is related to the repression of gene expression or to an altered rate of mRNAs decay, we conclude that a regulation of the amount of these transcripts is operating in root cells in response to phloem-translocated GSH. GSH, an end-product of the sulfate assimilation pathway thus appears to serve as a signal responsible for integrating S status at the whole plant level. It is known that glutamine, a by-product of the nitrogen assimilation pathway, depresses nitrate reductase mRNA levels in tobacco ( Deng et al. 1991) and corn ( Li et al. 1995), but no report of the response of gene expression in roots to changes in phloem translocation has been published yet. Although glutathione is a ubiquitous compound that regulates redox potential in all living organisms, the demand-driven regulation of ATP sulfurylase activity and SO42– uptake appears to be distinct from the oxidative stress response ( Lappartient & Touraine 1997). The site of biosynthesis of this transducer may be in the leaves or roots of S-fed plants, and the pathway for inter-organ communication is the phloem. The molecule which ultimately interacts with gene expression and the transduction pathway in the root cells remain to be elucidated. Whatever the precise events occurring within the root cells, the operation of the signaling process described here enables plants to respond with great flexibility to fluctuations of S availability in the root environment. This behaviour may well be a general feature of the regulation of nutrient acquisition and metabolism ( Clarkson & Lüttge 1991; Glass 1990; Imsande & Touraine 1994; Touraine et al. 1994). If so, it would be an important component of the plasticity of plant responses to a fluctuating environment.

Experimental procedures

Plant material

Arabidopsis thaliana plants (ecotype Columbia) were grown hydroponically in sterile vessels as described previously ( Touraine & Glass 1997), and experiments were performed on 18-day-old plants. Brassica napus plants (cv Drakkar) were grown hydroponically under non-sterile conditions as described by Lappartient & Touraine (1996), and experiments were performed on 21-day-old plants. Sulfate-free solutions were identical to basic nutrient solutions, except that CaCl2 and MgCl2 were substituted for CaSO4 and MgSO4, respectively. For glutathione, cysteine and BSO treatments, basic nutrient solutions were supplemented with 1 m m GSH, 1 m m Cys and 1 m m BSO, respectively.

SO42– and thiol determinations

SO42– was extracted in 0.1 N HCl for 1 h (20 ml per g fw) and assayed according to a turbidimetric method ( Tabatabai & Bremmer 1970). Thiols were extracted and determined by reverse-phase HPLC after reduction with DTT and derivatization with monobromobimane (Calbiochem) as described previously ( Lappartient & Touraine 1996). Using this methodology, individual thiols were determined as the sum of their reduced and oxidized forms ( Lappartient & Touraine 1997).

Influx measurements and enzyme assay

The rate of 35S accumulation in the roots of Arabidopsis was measured over 5 min exposure to labeled nutrient solution (specific radioactivity 5.2 Mbq mmol–1) in culture vessels. Roots were rinsed for 3 min in unlabeled culture solution and digested in 0.1 N HCl for 1 h (20 ml per g fw). Radioactivity was determined by liquid scintillation counting (460-C Tri-Carb, Packard).

ATP sulfurylase activity was determined from crude extracts of root tissues according to the molybdolysis assay, as described by Lappartient & Touraine (1996).

Western blots

Protein concentration in crude extracts of root tissues was determined according to Schaffner & Weissmann (1973). Aliquots of these extracts containing 5 μg of total protein were separated by SDS-PAGE on 10% polyacrylamide gel followed by electrophoretic blotting onto nitrocellulose membrane (Sartorius). The level of ATP sulfurylase protein was tested using a polyclonal serum raised in rabbit against APS3 protein overexpressed in E. coli ( Murillo & Leustek 1995). The antibody was specifically recognized using the ECL kit (Amersham) according to the manufacturer’s instructions.

RNA extraction and Northern blots

Total RNAs isolated from roots by guanidine HCl extraction followed by LiCl precipitation ( Logeman et al. 1987 ) were quantified spectrophotometrically. RNAs (20 μg per lane) were separated on 1.2% agarose/formaldehyde gel and transferred to Hybond N+ membrane (Amersham), and fixed by UV cross-linking. Membranes were hybridized to radiolabeled APS1 (GenBank accession number U05218), AST68 (GenBank accession number AB003591), or EF1α ( Axelos et al. 1989 ) as a control, probes for 18 h at 47°C in SSCX5, 1% sarkosyl, 50% (v/v) formamide, 10% (wt/vol) dextran sulfate, and denatured salmon sperm DNA. The 32P-labeled cDNA probes were synthesized by random priming (prime-a-gene kit, Promega). Membranes were washed in 0.1×SSc, 0.1% SDS at 47°C, and scanned on a phosphoimager (Molecular Dynamics). Quantification was performed using Image-Quant (Molecular Dynamics).


This work was supported in part by a fellowship from the Ministère de l’Enseignement Supérieur et de la Recherche to A.G.L. and from the Ministère des Affaires Etrangères to J.J.V.