Local and systemic regulation of sulfur homeostasis in roots of Arabidopsis thaliana


(e-mail hubberten@mpimp-golm.mpg.de).


Nutrients are limiting for plant growth and vigour. Hence, nutrient uptake and homeostasis must be adjusted to the needs of the plant according to developmental stages and environmental conditions. A split-root system was applied to analyse the systemic and local response of Arabidopsis thaliana to sulfur starvation. Arabidopsis thaliana plants in which only one root half was starved while the other root half was supplied with sulfate were analysed at the metabolic and transcriptional level. No systemic induction of sulfate uptake or expression of sulfate starvation marker genes was observed in split-roots sufficiently supplied with sulfate. Our data suggest that no activation of sulfur uptake takes part in sulfur-supplied root patches when the general sulfur status declines. When comparing roots of fully sulfate-starved plants with sulfate-starved split-root roots, expression of several potentially OAS responsive genes was attenuated in split-roots depending on the shoot sulfate status and the local root O-acetylserine concentration. In contrast, high-affinity sulfate transporters displayed similar expression in sulphate-starved split-roots and the corresponding controls. Feeding of 35SO42− to the shoot or to either part of a split-root system revealed that sulfate is the most prominent mobile sulfur-containing compound within the plant. Hence, we postulate a model whereby the soil sulfate availability regulates the sulfate uptake system of roots while the shoot sulfur status modulates the local O-acetylserine response in the root by passive ‘plant sulfur status-dependent’ transport of sulfate.


The availability of the macronutrient sulfur is crucial for plant growth and development, eventually determining plant vigour and yield. Sulfur is usually taken up as sulfate by the root system, and is then transported to the shoot via the xylem. In source leaves, it is reduced to sulfide through photosynthetic reduction equivalents. Sulfide and O-acetylserine (OAS) are the substrates for synthesis of cysteine, which is the precursor of metabolites containing reduced sulfur such as methionine and glutathione (Hoefgen and Nikiforova, 2008). Plants have developed mechanisms to counteract limitations in sulfur availability. The sulfate uptake and assimilation genes that are most consistently induced upon sulfur shortage are members of the sulfate transporter family (SULTR) 1, 2 and 4 and the adenosine 5′-phosphosulfate reductase (APR) family (Smith, 1980; Hawkesford et al., 1993; Gutierrez-Marcos et al., 1996; Takahashi et al., 1997; Kataoka et al., 2004a,b; Hoefgen and Nikiforova, 2008). Upon sulfate depletion, the content of OAS increases. OAS has been shown to directly influence the activity of APR and sulfate transporters, and to promote expression of several genes of the sulfur-reducing pathway (Neuenschwander et al., 1991; Smith et al., 1997; Koprivova et al., 2000; Kopriva et al., 2002; Hesse and Hoefgen, 2003; Hirai et al., 2003, 2005; Maruyama-Nakashita et al., 2004; Hubberten et al., 2012). Recently, Hubberten et al. (2012) provided strong evidence suggesting that OAS is a metabolite coordinating the expression of several genes that are strongly up-regulated upon sulfur starvation. In addition to several genes with unknown function, APR3, SULTR 1.1 and 1.2 displayed increased expression in response to elevated OAS content. The regulation of SULTR family 1 by OAS is controversial, being supported by a study by Maruyama-Nakashita et al. (2004) but questioned by Hopkins et al. (2005) and Rouached et al. (2008). In the latter, no correlation between SULTR expression and OAS content was observed, and thus OAS-independent signaling may play a role in regulating the expression of SULTR family 1. Other potential signals regulating sulfur assimilation are cysteine and glutathione. Both compounds decrease upon sulfur shortage and thus indicate the cellular sulfur status (Lappartient et al., 1999; Vauclare et al., 2002; Maruyama-Nakashita et al., 2004). Exogenously supplied GSH inhibits sulfur starvation-triggered activation of SULTR 1.1 and 1.2 and ATP sulfurylase (Lappartient et al., 1999; Vauclare et al., 2002; Maruyama-Nakashita et al., 2004). In contrast, no correlation between the expression of sulfate transporters and glutathione was found in studies by Buchner et al. (2004) and Rouached et al. (2008) in roots.

In addition to the regulation of sulfur homeostasis at the cellular level, understanding of the regulation of sulfur homeostasis at the plant level remains elusive. The shoot has been assumed to be the predominant site of sulfur reduction in the plant, although all enzymes of the sulfate reduction pathway have been shown to be present in roots, and some reduction capacity has been demonstrated (Pate, 1965; Hell, 1997; Lee and Leustek, 1998; Yonekura-Sakakibara et al., 1998; Heiss et al., 1999; Lappartient et al., 1999; Leustek and Saito, 1999; Saito, 2000). Thus, photosynthetically reduced sulfur must be transported within the plant to supply non-phototrophic sinks. The most prominent transport forms of reduced sulfur within the plant are glutathione (GSH) and S-methylmethionine (SMM) (Bourgis et al., 1999; Lappartient et al., 1999; Buchner et al., 2004; Durenkamp and De Kok, 2004). Lappartient et al. (1999) analysed systemic signals of sulfur starvation in Brassica napus using a split-root system, and observed an increase of APS1 protein in one split-root side when the other was exposed to limitations of sulfur, implying the existence of systemic signals between sulfur-starved and sulfur-supplied tissues. Further, GSH applied to one half of a split-root system was transported to the other root half, where it negatively affected the expression of APS3. As APS3 was also induced by sulfur starvation, it was proposed that the decrease of glutathione upon sulfur starvation coordinates this increase of expression. Moreover, transport of glutathione within the plant is supported by the work of Rouached et al. (2008) who observed increased glutathione levels in a sulfur-starved split-root in comparison to the −S control root when the other part of the root system was supplied with sulfate, implying that glutathione is transported between tissues.

Although several studies have implied the existence of systemic signals regulating sulfur homeostasis at the whole-plant level, the identity of these signals and their modes of action requires further elucidation.


In order to investigate whether sulfur homeostasis in Arabidopsis thaliana is subject to systemic or local regulation, a split-root system was employed. Sulfate starvation was applied to one root half of the split-root system, and the metabolic and transcriptional responses of the shoot and the other root half were analysed (Figure 1). A glossary of the terms used to describe the individual tissues of the split-root plants is given in Figure 1. Three independent hydroponic split-root experiments were performed. Two experiments included 16 and 6 biological repetitions for split-root plants and control plants, respectively. In a third split-root experiment, the amount of sulfate supplied and the distribution of roots between the +S and −S split-root side was varied (Figure S1).

Figure 1.

 Design of the split-root system, and glossary of terms.
Overview of the cultivation of split-root plants and control plants. The terms used to describe the various tissues of split-root and control plants in the text are indicated.

Sulfur status of the root

Expression of −S marker genes in split-root experiments.

Several known sulfur starvation-induced genes were used as markers for the sulfur status of split-root plants in which one root half was starved for sulfate. Expression of the sulfate transporters SULTR 1.1, 1.2 (data not shown) and 2.1, the sulfur assimilation pathway genes APR3 and APS4, and known −S marker genes [SDI1 (At5g48850), LSU1 (At3g49580), ChaC-like (At5g26220) and SHM7 (At1g36370)] in the roots of split-root and control plants was analysed by quantitative RT-PCR (Figure 2a) (Maruyama-Nakashita et al., 2006; Howarth et al., 2009; Lewandowska et al., 2010; Hubberten et al., 2012). Expression of these genes was up-regulated 10-fold (SULTR 1.1), 3-fold (SULTR 1.2, data not shown), 6-fold (SULTR 2.1), 5-fold (APR3), 35-fold (LSU1, At3g49580), 25-fold (ChaC-like, At5g26220), 7-fold (SHM7, At1g36370) and 57-fold (SDI1, At5g48850), and down-regulated 3.4-fold (APS4) in the −S control roots in comparison to +S control roots.

Figure 2.

 Expression level of −S marker genes and mature miRNA395 in roots of split-root plants.
(a) The expression level of SULTR 1.1, 2.1, APR3, APS4, and the −S marker genes ChaC-like, SHM7, LSU1 and SDI1 in the roots of split-root and control plants is shown. (Experiment 1 is shown: control n=5, split-root n=14).
(b) Content of mature miRNA395 variants in roots of split-root and control plants (Experiment 1: mature miRNA395 ade: control n = 4, split-root +S n =9, split-root -S n=8; mature miRNA395 bcf: control n = 3, split-root +S n = 6, split-root -S n=8; Experiment 2: mature miRNA395 ade: control n = 4, split-root n = 8; mature miRNA395 cbf: control n = 4, split-root n = 8). Different lower-case and upper-case letters indicate values that are statistically significantly different (t-test with α ≤ 0.05) in experiments 1 and 2, respectively.

−S split-roots displayed a similar induction of expression of SULTR 1.1, 1.2 and 2.1 as −S control roots. In contrast, sulfur starvation marker genes such as APR3, ChaC-like, SHM7, SDI1 and LSU1 were less strongly up-regulated in −S split-roots compared to −S control roots (Figure 2a). As miRNA395 is thought to be a mediator of systemic regulation of sulfur homeostasis, we analysed the abundance of the two types of mature miRNA395 (Kawashima et al., 2009; Buhtz et al., 2010; Liang et al., 2010). Increases in mature miRNA395s ade or bcf of approximately 14–20- and 12–14-fold, respectively, were observed in−S control roots compared with +S control roots (Figure 2b). In comparison to the +S control root, the levels of transcripts of mature miRNA395 ade tended to increase in −S split-roots, although the induction was much lower than in −S control roots (Figure 2b). Further, +S control roots and −S and +S split-roots did not display any differences in the amount of mature miRNA bcf in either experiment. No significant increase in miRNA395 ade or bcf was observed in +S split-roots in comparison to +S control roots (Figure 2b).

Content of sulfur-related metabolites in the split-root roots

The amount of sulfate, thiols and OAS in the roots of split-root plants was determined to identify metabolites correlated with the observed expression differences for −S-responsive genes (Figure 3). The amount of glutathione was reduced in −S control roots compared with the +S control root (67% glutathione), and a similar tendency was observed for cysteine (Figure 3b). Split-roots exhibited thiol levels similar to the corresponding controls. −S split-roots as well as −S control roots showed clearly decreased sulfate content compared to the +S control roots (Figure 3a). The sulfate content of the −S split-roots (47% of +S control) was elevated compared to the −S control roots in experiment 1 (28% of +S control) but did not show differences in experiment 2 (Figure 3a). +S split-roots and +S control roots displayed similar thiol and sulfate contents (Figure 3a,b). The OAS content was 4.7- and 9.2-fold elevated in the −S control roots in comparison to the +S control roots in experiments 1 and 2, respectively (Figure 3a). Moreover, the −S split-roots exhibited a 2.1- and 1.5-fold increase, respectively, compared with the OAS content of +S control roots. OAS levels in +S split-roots resembled those of the +S control roots (Figure 3a). A third split-root experiment confirmed the results of the experiments described above. Sulfate, OAS and glutathione levels behaved in a similar way as in the experiments described above. In split-root plants with an unequal distribution of split-root sides (<1:15 +S split-root to −S split-root), and thus with only minor access to sulfate, levels of all metabolites resembled the −S control plants. Split-root plants with more equally distributed split-roots (3:1–1:3 +S split-root to −S split-root) showed reduced OAS content in the −S split-root in comparison with the −S control (Figure S1).

Figure 3.

 Contents of sulfur-related metabolites in split-root experiments.
Content of sulfate (a, upper panel), OAS (a, lower panel), cysteine and glutathione (b) in roots of split-root and control plants ((a) Experiment 1 (sulfate): control n= 5, split-root n= 14 (OAS): control n= 6, split-root +S n= 6, split-root -S n= 7; Experiment 2 (sulfate and OAS): control n= 6, split-root n= 16; (b) experiment 1 is shown (cysteine and glutathione): control n= 5, split-root n= 14). Different lower-case and upper-case letters indicate values that are statistically significantly different (t-test with α ≤ 0.05) in experiments 1 and 2, respectively.

Sulfur status of the shoot

Indicative of the limitation of sulfur, −S control shoots displayed a reduction of the sulfate and glutathione contents in comparison with +S control shoots, while the cysteine level remained unaltered (Figure 4a,b). The limitation of sulfur to one root side in the split-root system leads to a reduction in the sulfate and glutathione content of split-root shoots. On average, the split-root shoots displayed approximately 60% of the sulfate content and 80% of the glutathione content of the +S control shoots (Figure 4a). The individual split-root shoots exhibited sulfate levels between −S and +S control shoots (Figure 4b). A third experiment confirmed the results of the first two experiments. Here, the shoot of split-root plants exhibited a sulfate level between that of the +S and −S controls. Split-root plants with unequally distributed split-root sides (<1:15 +S split-root to −S split-root) resembled the −S control plants (Figure S1).

Figure 4.

 Sulfur status of the split-root shoots does not affect the +S split-roots.
(a) Sulfate, glutathione and cysteine content of split-root shoots. Different letters indicate values that are statistically significantly different (control n = 5, split-root n = 14) (t-test with α ≤ 0.05).
(b) Expression of the −S marker gene SDI1 and the sulfate contents of single split-root shoots and control shoots ((SDI1 and sulfate): control n = 5).
(c) Expression of SDI1 and mature miRNA395 ade of the +S split-roots and control roots corresponding to split-root shoots (b) ((SDI1): control n = 5, (miRNA395 ade): control n = 4). ((a, b, c) Experiment 1 is shown).

To further characterize the sulfur status of the shoot, the expression of −S marker genes was analysed in split-root and control plants. The expression level of SDI1 (At5g48850), LSU1 (At3g49580) and APR3 was up-regulated in the split-root shoots in comparison with the +S control shoots (Figure 4b; only results for SDI1 are shown). A detailed analysis of the expression values for the individual split-root shoots demonstrated that SDI1 is up-regulated in some of the split-root shoots while others display +S control-like expression levels (Figure 4b). The sulfur starvation response displayed by some of the individual split-root shoots as indicated by increased expression of SDI1 is further supported by the simultaneous decrease in sulfate content, which is an additional indicator of the sulfur status of the respective shoot (Figure 4b). In lines 1, 21, 25, 31 and 39, the decrease of sulfate levels of the split-root shoots was most pronounced and directly correlated with the increase in expression of −S marker genes such as SDI1 (Figure 4b). The expression level of SDI1 and mature miRNA395 for the single +S split-roots in experiment 1 is shown in Figure 4b to allow comparison with the corresponding split-root shoots.

35SO42− treatment of split-root plants

To investigate the transport of sulfur-containing metabolites within the plant, we fed 35SO42− to one split-root of the split-root plants and analysed the distribution of 35S-labelling in the shoot and the split-roots. To investigate the influence of the sulfur status of the plant on sulfate uptake, two experiments were performed, varying the sulfur supply to the untreated split-root. While the 35S-treated split-root side was consistently supplied with sulfate, the other split-root side was exposed to medium containing sulfate (Figure 5a, lower panel) or medium without sulfate (Figure 5a, upper panel).

Figure 5.

35SO42− labelling of split-root plants.
Determination of 35SO42−-labelling in shoots and roots of split-root plants after feeding one split-root side with 35SO42− for 2–72 h.
(a) Upper panel: No sulfate was supplied to the untreated split-root (= 4). Lower panel: sulfate was supplied to the untreated split-root (= 4).
(b) Radiogram of a split-root plant fed with 35SO42− for 72 h on the left +S split-root side (circled).

Radioactivity was detected in the treated root after 2 h incubation (Figure 5a). After 6 h, radioactive label was present in the shoot, and, to a much smaller extent, in the untreated parts of the split-root. The radioactivity constantly increased over time in all tissues, including the untreated split-root. After 72 h, the strongest labelling was displayed by shoots while the untreated split-root exhibited 10-fold less labelling than the treated split-root (Figure 5a,b). Sulfate supply to the untreated split-root (+S or −S) did not influence the uptake and distribution of the 35S label within the plant (Figure 5a). Split-root plants treated with 35SO42− for 72 h were exposed to a phosphorimager plate. In situ accumulation of label was detected in the shoot and the unlabelled split-root side.

Determination of 35SO42−-labelled compounds in split-root roots and shoots

The results described above confirmed transport of 35S within the split-root system. Characterization of the transported sulfur-containing metabolites was achieved by investigating labelled metabolites using a modified amino acid TLC method (adapted from Bettelheim and Landsberg, 2009) (Figure S2). +S split-roots were incubated for 72 h with 35SO42− as described above, and labelling of the −S split-root was analysed. Labelling of sulfate, glutathione and several unidentified substances was detected in the untreated split-root (Figure 6a,b). To further prove which metabolite is transported within the plant, we treated shoots of A. thaliana wild-type and adenosine-5′-phosphosulfate kinase 1 and 2 double knockout mutants (APK-KO) (Mugford et al., 2009) grown in the presence and absence of the glutathione synthesis inhibitor buthionine sulfoximine (BSO) with 35SO42−, and investigated labelling of shoots and roots. APK-KO mutants are impaired in their ability to produce 3’-phosphoadenosine-5-phosphosulfate, which is the substrate of sulfotransferase reactions. Consequently, APK-KO mutants are generally limited in sulfated compounds (Mugford et al., 2009). Sulfate labelling was detected in roots of all treatments. No labelling of glutathione was detected in roots of BSO-treated plants, though labelling of glutathione in the shoot was detected. Labelling was detected in more nonpolar bands in the roots. Interestingly, these bands did not overlap between the APK mutant and wild-type, demonstrating that none of these metabolites is the source of the detected sulfate in the root. Thus, although several bands could not be characterized, the experiment clearly shows that sulfate itself is transported to the root from the shoot. Similarly to the roots, large differences in the labelling pattern of APK mutant and wild-type were also detected in the shoots.

Figure 6.

 Compound determination for 35SO42− labelling experiments.
(a) +S split-roots (= 4) were fed with 35SO42 for 72 h. Reduced extracts (10 mm DTT) of shoots and roots of split-root plants were separated by TLC (Figure S2).
(b) Shoots of wild-type and APK double knockout mutants (= 4) grown in the presence and absence of BSO were treated with 35SO42−. Shoot and root tissue was harvested after 72 h. Extracts were reduced by adding 10 mm DTT, and labelling of metabolites was analysed by TLC.


Nutrients taken up from the soil by the root must be distributed within the whole plant system. To analyse the signals and mechanisms that enable the plant to adjust the uptake of sulfate by the root to the needs of sulfur of the shoot, we investigated A. thaliana plants in which sulfate starvation was applied to one root half in a split-root approach. Upon sulfate starvation of one split-root side, we did not observe any activation of sulfur-responsive genes or metabolites on the +S side of the split-root system (Figures 2 and 3). A similar observation was reported by Rouached et al. (2008), who also did not observe activation of SULTR 1.1 and 1.2 in the sulfate-supplied split-root side upon depletion of sulfate on the other split-root side. Additionally, we showed that a systemic activation is also absent for APR3 and other highly sulfur-responsive genes such as ChaC-like (At5g26220), SHM7 (At1g36370), SDI1 (At5g48850) and LSU1 (At3g49580) (Nikiforova et al., 2003; Maruyama-Nakashita et al., 2006; Howarth et al., 2009; Lewandowska et al., 2010; Hubberten et al., 2012). Moreover, sulfur starvation of one split-root side did not systemically induce uptake and distribution of 35S-labelling within the plant when the other split-root side was fed with 35SO42−, in comparison with split-root plants where both split-roots were supplied with sulfate (Figure 5a). Our results indicate that the root does not signal information on the local sulfate availability to the rest of the plant.

As the needs for sulfate of the shoot may be assumed to be communicated to the uptake system of the root, we investigated the sulfur status of the shoot. Shoots that have established a sulfur starvation response should, if existing, release a signal to activate uptake in the roots. With respect to sulfate levels and −S marker gene expression (SDI1), the split-root shoots exhibited all intermediate states between starved and supplied controls (Figure 4a,b).

The data suggest that the −S response, including activation of expression of −S-responsive genes, is activated when the sulfate content falls below a certain threshold. Although several split-root plants showed increased expression of −S -responsive genes (SDI1) and decreased sulfate content in the shoot (Figure 4b), none of these plants displayed induced expression of SULTR or other −S marker genes in the +S split-root (Figure 4b), indicating that no systemic activation of sulfur response from shoot to root occurred. In agreement with these results, the level of miRNA395, which has been discussed as a systemic signal (Kawashima et al., 2009; Buhtz et al., 2010; Liang et al., 2010), was not increased in the +S split-roots, indicating that miRNA395 was not transported from the shoot to the root (Figure 4b). Further, no miRNA395-induced cleavage of SULTR 2.1 and APS4 transcripts was observed in +S split-roots in comparison with +S control roots (Figure 2a) (Kawashima et al., 2009). Upon exposure of +S split-roots to sulfate concentrations between 50 and 400 μm, no differences in the sulfate content of the shoots were observed when the root system deviated equally between split-root sides, showing that low- and high-affinity systems were able to recruit sulfate efficiently over a wide range of sulfate concentrations (Figure S1). Only when the split-root system was unequally arranged, with few +S split-roots, did the shoots display sulfate levels similar to the −S control (Figure S1). Further, taking into account that, in all experiments, the sulfate content of equally distributed split-root shoots was between that of the +S and −S controls, the implication is that the sulfur status of the shoot is determined by the ratio of sulfur-supplied to sulfur-starved roots and the total amount of roots, and not by activation of sulfate uptake in sulfur-supplied roots by the shoot sulfur status (Figure 4b and Figure S1). It may be concluded that activation of the high-affinity uptake system is a purely local response when sulfate availability drops below a critical level. Although systemic signals have been demonstrated for nitrate, phosphate and iron, we did not find any indication in this experimental system for signals from a starved shoot or locally starved root activating the sulfate uptake system or other parts of the sulfur starvation response in a sulfur supplied root. (Drew and Saker, 1975; Burns, 1991; Laine et al., 1995; Grusak and Pezeshgi, 1996; Schmidt, 1996; Tillard et al., 1998; Schikora and Schmidt, 2001; Aung et al., 2006; Bari et al., 2006; Lin et al., 2008; Pant et al., 2008; Vert et al., 2009).

SULTR 1.1 and 1.2 and SULTR 2.1 expression displayed similar increases upon sulfur starvation in −S split-roots and −S control roots. In contrast, differences were observed with respect to the transcript levels of other −S-induced genes (APR3, APS4, ChaC-like, SHM7, SDI1 and LSU1) (Figure 2a) as well as mature miRNA395 (Figure 2b). As the local environment for −S control roots and −S split-roots was identical, a metabolite or signal delivered from the split-root shoot or the sulfur-supplied split-root must be assumed to be responsible for repression of the activation of −S marker genes in −S split-roots. Thus, we analysed sulfur-related metabolites such as OAS, thiol and sulfate in the various tissues of split-root plants. No differences between the concentrations of cysteine and glutathione between −S split-roots and −S control roots were observed (Figures 4a and S1). Sulfate levels in the −S split-roots were only marginally increased compared to the −S control roots, and this increase was not consistent within all experiments (Figures 3a and S1). Most strikingly, OAS levels in the −S split-root side were not as strongly increased as in the roots of −S control plants, but instead resembled the levels for the +S split-roots and +S controls (Figures 3a and S1). The role of OAS as a mediator of the sulfur starvation response remains a topic of debate (Smith et al., 1997; Koprivova et al., 2000; Kopriva et al., 2002; Hesse et al., 2003; Hirai et al., 2003; Maruyama-Nakashita et al., 2004; Ohkama-Ohtsu et al., 2007; Hubberten et al., 2012). Recently, Hubberten et al. (2012) provided strong evidence suggesting that OAS is a direct regulator of six genes including APR3, SDI1, LSU1, ChaC-like and SHM7. In accordance with this data, the differences in OAS content perfectly explain the expression pattern of the −S-responsive genes APR3, SDI1 (At5g48850), LSU1 (At3g49580), SHM7 (At1g36370) and ChaC-like (At5g26220) in −S split-roots and the roots of −S controls. Further, it remains contentious whether induced expression of these genes in −S split-roots is only due to the slight increase in the OAS content, or whether basal induction of the genes by other signals also occurs.

Remarkably, SULTR 1.1, 1.2 and 2.1 are similarly induced in −S split-roots and −S control roots, indicating that SULTR 1 and 2 expression is regulated independently of the OAS content or is more sensitive in its response to minor changes in OAS levels. Hubberten et al. (2012) observed a significant up-regulation of SULTR 1.1 and 1.2 in roots of plants that, as a result of inducible over-expression of a serine acetyltransferase (SERAT) gene, accumulated OAS within 3 h of treatment.

In accordance with our split-root data, Rouached et al. (2008) reported the independence of SULTR1 expression from the OAS content, thus suggesting regulation of SULTR family by OAS and other OAS-independent signals, such as sulfate content. The behaviour of glutathione and OAS content in the split-roots is in contrast to the observations of Rouached et al. (2008) who performed a comparable split-root experiment. Here, the OAS levels of +S and −S split-roots were similar to the corresponding controls, and glutathione levels on the −S split-root side matched +S control levels. The different results obtained between the split-root experiments may be due to the different set-up of the split-root systems. While Rouached et al. (2008) used a hydroponic split-root system in which the root system of adult plants was separated, potentially leading to crosswise distribution of physically linked roots, the split-root system used in this study was generated by cutting the primary root tip, inducing growth of two independent primary roots, and thus ensuring the mutual independence of both root parts.

The decreased OAS levels in −S split-roots in comparison to −S control roots indicate that a metabolite is transported to the starved split-root side leading to consumption of OAS and eventually producing cysteine. 35SO42− feeding experiments suggest that sulfate is transported in the plant from shoots to roots. When feeding 35SO42− to one split-root side of the split-root system, a 35S-derived signal was detected in the shoot and the unlabelled split-root. This transport appears to be independent of the sulfur supply of the untreated split-root (+S or −S media). A more detailed analysis revealed that a 35S signal is emitted from sulfate, glutathione and several as yet unidentified bands on the untreated split-root side (Figures 6 and S2). Experiments in which shoots of wild-type and APK 1/2 double knockout mutants grown on media with and without BSO (an inhibitor of glutathione synthesis) were fed with 35SO42 indicated that sulfate itself is transported from the shoot to the root where it is further incorporated into downstream metabolites. 35S-labelled sulfate was detected in the roots of all treatments (Figure 6b). Glutathione was not detectable in roots treated with BSO, although 35S-glutathione was clearly visible in the corresponding shoots. Hence, our data question the transport of glutathione or any other form of reduced sulfur from the shoot to the root, and instead provide evidence that reduction of sulfate in the root itself is the dominant source of reduced sulfur present in the root. Further labelling of the unidentified bands did not overlap between wild-type and the APK double knockout mutant, thus excluding these metabolites as potential sulfur transport forms between shoot and root (Figure 6a,b). Sulfate transported from the shoot to the root may explain the observed differences in the OAS content in the −S split-root. In the root, sulfate derived from the shoot is reduced to sulfide, and is then available for reaction with OAS to form cysteine, thus reducing the OAS level and consequently attenuating the OAS response locally in the −S split-root.

As cysteine and glutathione levels in the −S split-roots match those of the −S controls (Figure 3b), it may be assumed that reduced sulfur (e.g. cysteine) is immediately consumed for synthesis of other compounds, for example proteins. It thus must be assumed that a low rate of reduction that is not able to completely compensate for the loss of thiols due to the local absence of sulfate is sufficient to decrease OAS levels significantly.

In turn, the plant sulfur status-dependent delivery of sulfate to roots controls OAS levels as an effective mechanism by which the plant avoids activation of the starvation response when the overall plant sulfur status is sufficient even though the local root environment is low in available sulfate (Figure 7).

Figure 7.

 Model of local and systemic regulation of root sulfur homeostasis.
(a) Under conditions of sufficient sulfate availability, sulfate is taken up by the sulfate uptake system.
(b) At low sulfate concentrations, the sulfate uptake system, including SULTR family 1 and 2, is induced to mobilize sulfate to allow local consumption and export to other plant parts. This induction is independent of OAS or the systemic sulfate status.
(c) When soil sulfate levels become too low, local cellular sulfate and consequently thiol levels decrease in the respective root patch. Eventually, the decrease in thiols leads to retardation of lateral root growth. As the shoot sulfate status is high, small amounts of sulfate are transported to the root and are sufficient to decrease root OAS levels and ensure minimal flux to thiols, preventing an OAS-induced sulfur starvation response.
(d) When the whole plant sulfur content decreases, sulfate transport from shoot to root declines, and thus OAS levels in the root increase, activating the full OAS/−S response. Sulfur recovery mechanisms are induced to ensure remobilization of sulfur to guarantee survival of the plant.
The two sulfate symbols are used to illustrate sulfate taken up by the root and sulfate supplied by the shoot, but do not imply the existence of separate pools in the root. Dashed arrows indicate strong limitation of flux.

As an integrated model of plant sulfate homeostasis, the local sulfur starvation response in the root may be divided into mechanisms that appear to be induced locally and independently of the sulfur status of the plant, such as sulfate transporter induction, and mechanisms that are attenuated if significant amounts of sulfate are delivered from the shoot or other parts of the root system (Figure 7). The local induction of sulfate transporters seems reasonable, keeping uptake of sulfate active even at low sulfate concentrations in the soil (Figure 7a,b). In accordance with the data in our study, Hubberten et al. (2009) demonstrated that local sulfur reduction is essential for lateral root growth in A. thaliana. This indicates that local sulfate uptake must be maintained at low sulfate concentrations to enable synthesis of thiols and hence lateral root growth (Figure 7a,b). OAS-dependent mechanisms systemically integrate the plant sulfur status, and are only induced if the general plant sulfur status is severely reduced (Figure 7c,d). Genes whose expression was correlated with OAS content in this study perfectly overlap with the OAS-responsive genes identified previously (Hubberten et al., 2012). However, except for APR3, the function of the other OAS-responsive genes remains unclear. Future studies investigating the role of the OAS-correlated genes will help to finally understand why their expression is attenuated in sulfur-limited roots when the sulfate status in general is sufficient. It may be assumed that plants have evolved mechanisms to avoid energetically cost-intensive processes in locally sulfur-starved roots while the overall plant sulfur status is sufficient, and it may be concluded that the OAS response as part of the sulfur starvation response has too severe an impact on metabolism or development, demanding strict regulation and only being activated when the general sulfur status is dramatically decreased.

Experimental procedures

Plant material and cultivation

Arabidopsis thaliana plants of the Columbia-0 ecotype were used in all experiments. Arabidopsis thaliana plants were grown under standard conditions (140 μmol m−2 sec−1, 50% relative humidity, 21°C) under a 16/8 h light/dark cycle. In the third hydroponic split-root experiment, A. thaliana plants were grown under a 12/12 h light/dark cycle.

Split-root agar plates

Split-root agar plates were prepared using Arabidopsis medium (half-strength MS medium; Murashige and Skoog, 1962) with or without 1% sucrose and 1% agarose. Slices of 0.5–3 mm width were removed to generate separated patches. The separated patches were supplemented with MgSO4 or the corresponding amount of MgCl2 to prepare sulfur-supplied and sulfur-depleted patches. Root tips of 3–4-day-old A. thaliana seedlings were cut with a scalpel and grown for 10 more days, developing two primary roots. Each primary root was transferred to one segment of the split-root agar plate. One split-root side was treated with 20 μl of 0.5 μm35SO42− (10 μCi) solution. Radiograms of plants were generated by exposing whole plants to phosphoimager plates for 3–5 days. Phosphoimager plates were read using a Thypoon Trio+ variable mode imager (GE Healthcare Life Sciences, www.gelifesciences.com). To exclude passive transport of 35SO42− within the separated agar plates, a control experiment was performed in which the untreated split roots were separated from the shoot (Figure S3a,b). In addition, a second control was applied in which control plants were grown in the unlabelled agar segment of a split-root labelling experiment (Figure S3a,b). The control experiments clearly show that only marginal amounts of 35SO42− were transported passively between separated agar plates.

Wild-type and APK mutant grown on agar plates

Arabidopsis thaliana wild-type and adenosine-5′-phosphosulfate kinase (APK) 1 and 2 double knockout mutants (APK-KO) were grown for 14 days on Arabidopsis medium with or without 1% sucrose and 1% agarose (Mugford et al., 2009), and transferred to plates with or without 2.5 mm buthionine sulfoximine (BSO). After 2 days, 10 μl of 1 μm35SO42− (10 μCi) solution was applied carefully to single leaves. Shoot and root material was harvested 3 days after the start of the treatment. A control experiment was performed in which the shoot–root connection of plants was interrupted at the start of labelling (Figure S3c). Only marginal amounts of 35SO42− were transported passively within the medium.

Split-root hydroponics

Split-root and control plants were generated as described above and placed in a hydroponic system containing Hoagland medium (Hoagland, 1919). The hydroponic plants were grown for an additional 12 days before starving them for 5 days. Plants were grown at MgSO4 concentrations of 500 μm in 500 ml pots. For −S treatments, the medium was supplemented with equal amounts of MgCl2.

RNA extraction

RNA was extracted using an RNeasy plant mini kit (Qiagen, http://www.qiagen.com) according to the manufacturer’s instructions. Total RNA was subsequently digested using Turbo DNAfree™ DNase (Ambion, http://www.invitrogen.com) according to the manufacturer’s instructions. Absence of genomic DNA contamination was confirmed by RT-PCR using intron-specific primers (Table S1).

cDNA synthesis for analysis of mRNA

After DNase digestion, 1 μg total RNA was subjected to cDNA synthesis using Superscript III (Invitrogen, http://www.invitrogen.com). The quality of the cDNA was assessed by quantitative PCR using primers amplifying sequences located at the 3’ and 5’ ends of glyceraldehyde-3-phosphate dehydrogenase mRNA. Only cDNA exhibiting a ratio between the relative amount of 3’ and 5’ amplicon of <2 was used for further analysis.

cDNA synthesis for analysis of mature miRNA395

Total RNA (1 μg) was mixed with 1 μl 10 mm dNTPs, 1 μl of 2.5 μm stem-loop reverse-transcription primer specific for miRNA ade or cdf, and RNase-free H2O to make up a final volume of 36.5 μl, and heated at 65°C for 5 min and immediately chilled on ice. Then 10 μl 5 × first-strand buffer, 2 μl 0.1 m DTT, 0.5 μl RNase inhibitor and 1 μl MultiScribe reverse transcriptase (Applied Biosystems, http://www.appliedbiosystems.com) were added. The reaction mixture was incubated for 30 min at 16°C, followed by 1 h at 42°C and 5 min at 85°C, and then held at 4°C. Primer sequences are shown in Table S1.

Quantitative RT-PCR

PCR was performed in an optical 384-well plate using an ABI PRISM® 7900 HT sequence detection system (Applied Biosystems). Reactions contained 5 μl 2 × SYBR® Green Master Mix reagent (Applied Biosystems), 1 μl of fivefold diluted shoot cDNA or 2.5-fold diluted root cDNA, and 200 nm of each gene-specific primer. The total reaction volume was 10 μl. The thermal profile used was 50°C for 2 min, 95°C for 10 min, and 40 cycles of 95°C for 15 sec and 60°C for 1 min. sds 2.0 software (Applied Biosystems) was used for data analysis. The CT values for genes were normalized to the CT values of ubiquitin. PCR efficiencies were calculated using the linregpcr 7.0 program (Ramakers et al., 2003). All primer sequences were designed using the criteria described by Czechowski et al. (2004). Primer sequences are shown in Table S1.

HPLC of thiol compounds

First 200 μl of 0.1 M HCl containing 10 mg polyvinylpolypyrrolidone was added to 20 mg plant material. The supernatants were buffered to basic conditions by adding 1.6 volumes of 0.25 m 2-(cyclohexylamino)ethanesulfonic acid/NaOH, pH 9.4, and reduced by adding DTT to a concentration of 2 mm. Thiols were derivatized by applying monobromobimane to a concentration of 0.625 mm (Hell and Bergmann, 1988). HPLC of the samples was performed using a Hypersil ODS C18 column (Thermo Scientific, http://www.thermoscientific.com). Samples were eluted using increasing concentrations of methanol in an acetic acid/methanol mixture. The column eluent was monitored by fluorescence detection (λex 380 nm/λem 480 nm).


OAS content was determined using a modified amino acid analysis protocol (Scheible et al., 1997) described in detail by Krueger et al. (2009).

High-performance ion exchange chromatography

Ground frozen plant material (20 mg) was homogenized in 200 μl of 0.1 mm HCl. Samples were centrifuged at 14 000 g for 5 min at 4°C. The supernatant was transferred to an Ultrafree MC 5000 MC NMWL filter unit (Millipore, http://www.millipore.com) and centrifuged at 5000 g for 90 min at 4°C. Samples were diluted 1:20 and analysed by high-performance anion exchange chromatography with conductivity detection on a Dionex ICS-2000 system (Dionex, http://www.dionex.com). Elution of the ions was conducted using a KOH gradient.

TLC of 35S-labelled compounds

Shoots and roots of 35SO42−-treated plants were harvested and extracted 1:20 w/v in 80% ethanol (15 min at room temperature on a shaker at 1400 rpm) and 1:20 w/v in 50% ethanol. Extracts were reduced by adding DTT to a concentration of 10 mm, and 20 μl aliquots of extract were spotted onto TLC plates (SI250-PA, Mallinckrodt Baker, http://www.jtbaker.nl). Separation was performed in a saturated TLC chamber with a mobile phase consisting of n-butanol/acetic acid/H2O (4:1:1). Phosphoimager plates were developed for 1–5 days and were read using a Thypoon Trio+ variable mode imager (Sartorius).

Quantification of 35S-labelled compounds

Whole roots or shoots were extracted in 200 μl of 0.1 m HCl for 1 h. Upon addition of 4.8 ml scintillation cocktail, the radioactivity was counted indirectly by fluorescence measurement in a LS6500 multipurpose scintillation counter (Beckman Coulter, http://www.beckmancoulter.de).

Statistical analysis

Statistically significant differences between the groups were tested by anova using a standard Bonferroni-adjusted significance level of  0.01 (Table S2) in MultipleExperiment Viewer (MeV) (Saeed et al., 2003). If anova indicated differences between the dataset, a crosswise comparison of the single groups using a t-test was performed. Significance levels are presented in Table S2. All error bars in the figures indicate standard errors.


We wish to thank Dr Marc Lepetit and Dr Alain Gojon (both Institut National de la Recherche Agronomique/SubAgro, Montpellier, France) for their help setting up the split-root system, and Dr Bikram Datt Pant (Max Planck Institut für Molekulare Pflanzenphysiologie, Golm, Germany), for providing primers for miRNA395 analysis. We wish to thank Franziska Brückner for excellent technical support, and Dr Stanislav Kopriva (John Innes Centre, Norwich, UK) for providing the APK double knockout mutants. Dr Gareth Catchpole (Metanomics, Berlin, Germany) is gratefully acknowledged for proofreading of this manuscript. Financial support for this project from the French–German Research Network Program (P2R; Deutsche Forschungsgemeinschaft HO1916/3-2) and the Max-Planck Society is gratefully acknowledged.