Analysis of cytosolic and plastidic serine acetyltransferase mutants and subcellular metabolite distributions suggests interplay of the cellular compartments for cysteine biosynthesis in Arabidopsis


H. Hesse. Fax: +49 (0)331 567898247; e-mail:


In plants, the enzymes for cysteine synthesis serine acetyltransferase (SAT) and O-acetylserine-(thiol)-lyase (OASTL) are present in the cytosol, plastids and mitochondria. However, it is still not clearly resolved to what extent the different compartments are involved in cysteine biosynthesis and how compartmentation influences the regulation of this biosynthetic pathway. To address these questions, we analysed Arabidopsis thaliana T-DNA insertion mutants for cytosolic and plastidic SAT isoforms. In addition, the subcellular distribution of enzyme activities and metabolite concentrations implicated in cysteine and glutathione biosynthesis were revealed by non-aqueous fractionation (NAF). We demonstrate that cytosolic SERAT1.1 and plastidic SERAT2.1 do not contribute to cysteine biosynthesis to a major extent, but may function to overcome transport limitations of O-acetylserine (OAS) from mitochondria. Substantiated by predominantly cytosolic cysteine pools, considerable amounts of sulphide and presence of OAS in the cytosol, our results suggest that the cytosol is the principal site for cysteine biosynthesis. Subcellular metabolite analysis further indicated efficient transport of cysteine, γ-glutamylcysteine and glutathione between the compartments. With respect to regulation of cysteine biosynthesis, estimation of subcellular OAS and sulphide concentrations established that OAS is limiting for cysteine biosynthesis and that SAT is mainly present bound in the cysteine–synthase complex.


Cysteine biosynthesis is a central metabolic pathway to incorporate inorganic sulphur into organic molecules and is tightly linked to processes such as redox regulation and homeostasis, detoxification and defence (Kopriva 2006). In plants, cysteine is synthesized from serine, acetyl-coenzyme A (CoA) and sulphide by serine acetyltransferase (SAT, EC and O-acetylserine-(thiol)-lyase (OASTL, EC SAT generates the cysteine carbon skeleton, O-acetylserine (OAS) in the rate-limiting step of cysteine biosynthesis. OAS subsequently condensates with sulphide in the reaction catalysed by OASTL to form cysteine.

SAT and OASTL activity is present in the cytosol, plastids and mitochondria (Brunold & Suter 1982; Lunn et al. 1990; Ruffet et al. 1995; Warrilow & Hawkesford 1998; Hesse et al. 1999; Howarth et al. 2003; Kawashima et al. 2005), yet it is still not clearly resolved why the enzymes for cysteine biosynthesis are needed in all three cellular compartments. Although very recently tremendous progress has been made in understanding the contributions of the different Arabidopsis OASTL isoforms on cysteine biosynthesis using T-DNA insertion mutants (Heeg et al. 2008; Watanabe et al. 2008a) and mitochondrially localized SAT has been identified as the main provider for OAS (Haas et al. 2008; Heeg et al. 2008), only one recent study addresses the function and significance of the different SAT genes in Arabidopsis using a reverse genetics approach (Watanabe et al. 2008b). Moreover, subcellular concentrations of metabolites related to sulphur metabolism have not been established; thus, implications of metabolite concentrations on metabolite transport and consequences for the regulation of cysteine biosynthesis have so far remained speculative.

Five SAT isoforms exist in Arabidopsis thaliana. Cytosolic SERAT1.1, plastidic SERAT2.1 and mitochondrial SERAT2.2 are relatively high expressed in leaf and root tissue, whereas SERAT3.1 and SERAT3.2 are expressed at very low levels (; Howarth et al. 2003; Kawashima et al. 2005). Determination of in vitro substrate affinities of purified SAT revealed that SERAT3.1 and SERAT3.2 exhibit very low substrate affinities compared to SERAT1.1, SERAT2.1 and SERAT2.2 (Kawashima et al. 2005), suggesting that these isoenzymes may actually process different substrates in vivo. However, a study employing multiple knockout mutants for the different SAT isoforms revealed that all Arabidopsis SATs are functional and can, at least in part, complement for lacking SAT activity (Watanabe et al. 2008b). In line with the presence of multiple, differently localized SAT isoforms, the Arabidopsis genome encodes nine OASTL-like isoforms located in the cytosol, plastids and mitochondria (Warrilow & Hawkesford 1998; Dominguez-Solis et al. 2001; Wirtz & Hell 2006; Heeg et al. 2008; Watanabe et al. 2008a). OASTL assembles with SAT into the multimeric cysteine synthase complex (CSC). Assembly and disassembly of the complex modulates enzyme kinetics and is therefore believed to represent a key control point for the regulation of cysteine synthesis (Kredich & Tomkins 1966; Bogdanova & Hell 1997; Droux et al. 1998; Wirtz & Hell 2006). OASTL activity substantially decreases in the complex, whereas SAT exhibits slightly lower substrate affinities and displays positive cooperativity when it is bound in the complex. A high OASTL : SAT ratio has been considered important for efficient cysteine biosynthesis as a 100-fold excess of OASTL is required to achieve maximal OAS synthesis in vitro (Droux et al. 1998). Complex formation strongly depends on concentrations of OAS and sulphide, the two substrates for cysteine biosynthesis. High OAS concentrations promote CSC dissociation, whereas sulphide stabilizes the complex (Kredich, Becker & Tomkins 1969; Berkowitz et al. 2002; Wirtz & Hell 2006). This suggests that the CSC may play a central role for the regulation of cysteine biosynthesis in dependence of the sulphur status of the plant. Upon sulphur-limiting conditions, accumulating OAS is believed to dissociate the CSC leading to reduced SAT activity and, consequently, reduced OAS production and less acetyl-CoA consumption.

Apart from regulation through the CSC, SAT activity is modulated by feedback inhibition through cysteine. In A. thaliana, three SAT isoenzymes are cysteine sensitive (Noji et al. 1998; Wirtz & Hell 2003; Kawashima et al. 2005). Feedback inhibition of SAT is discussed as a mechanism to prevent accumulation of OAS under conditions in which sulphur is present in sufficient amounts (Saito 2000).

With respect to compartmentation of SAT and OASTL activity, chloroplastic SAT contributed 35% of total SAT activity in Spinacia oleraceae plants, and the residual 65% were assigned to the cytosol (Brunold & Suter 1982). In contrast, Smith (1972) showed that one-third of SAT activity was firmly bound to mitochondria in Phaseolus vulgaris. In Pisum sativum, mitochondrial SAT contributed with 75% mainly to total SAT activity, whereas 15% was located in the cytosol and 10% in chloroplasts (Ruffet et al. 1995). A study employing T-DNA mutants for the different Arabidopsis SAT isoforms revealed that mitochondrial sat exhibited significantly reduced activity in leaves, roots and stems, while cytosolic serat1.1was the only other sat mutant displaying reduced SAT activity in roots of Arabidopsis plants grown on sucrose-containing medium (Watanabe et al. 2008b). Similarly, transgenic Arabidopsis plants silenced for the mitochondrial SAT isoform showed dramatically reduced OAS levels and flux into cysteine and glutathione (Haas et al. 2008). In addition, SAT activity measurements in isolated mitochondria of heterotrophically grown Arabidopsis cell cultures revealed higher specific SAT activity in isolated mitochondria compared with crude plant extracts (Heeg et al. 2008). Thus, in Arabidopsis, the mitochondria are apparently the main site for cellular OAS synthesis. OASTL activity was mainly localized in the cytosol of S. oleraceae and Datura innoxia leaves, while chloroplasts exhibited lower OASTL activity and mitochondria even less (Lunn et al. 1990; Kuske et al. 1996). Similarly, studies using Arabidopsis T-DNA insertion lines for different compartment-specific OASTL isoforms revealed pronounced reductions for OASTL activity in leaves of mutants for the cytosolic and plastidic enzymes (Heeg et al. 2008; Watanabe et al. 2008a). As sulphate reduction exclusively takes place in plastids, cytosolic or mitochondrial cysteine biosynthesis is limited by diffusion or transport of sulphide.

Availability of cysteine in the different cellular compartments is crucial for downstream processes requiring cysteine, such as glutathione and protein biosynthesis. In plants, γ-glutamylcysteine synthesis by γ-glutamylcysteine-synthetase (GSH1) exclusively takes place in plastids and represents the rate-limiting step in glutathione biosynthesis (Wachter et al. 2005; Hothorn et al. 2006). Multiple transcript analysis of glutathione synthase (GSH2) indicates that transit peptide containing GSH2 transcript is present in low amounts; nevertheless, low amounts of plastidic GSH2 might be sufficient to promote chloroplastic glutathione synthesis (Wachter et al. 2005; Pasternak et al. 2008). However, glutathione uptake studies performed with wheat chloroplasts indicate that cytosolic glutathione can be efficiently taken up by a high and a low affinity system (Noctor et al. 2002).

To resolve to what extent the cytosol and plastids are involved in cysteine biosynthesis, we investigate here A. thaliana T-DNA insertion mutants for cytosolic SERAT1.1 and plastidic SERAT2.1. Non-aqueous fractionation techniques are further employed to address implications of compartmentalized enzyme activities and metabolite concentrations on the regulation of cysteine biosynthesis and to resolve transport processes between compartments.


Plant cultivation

T-DNA insertion mutants and Columbia-0 wild-type seeds were stratified in the dark for 2 d at 4 °C on soil or on plates containing one-half Murashige and Skoog (MS) media (Murashige & Skoog 1962). On soil, plants were grown for 2 weeks under short-day conditions (9/15 h light/dark cycle; 140 µmol m−2 s−1 photon flux density, 50% humidity, 21 °C) before being transferred to long-day greenhouse conditions (16/8 h light/dark cycle, 140 µmol m−2 s−1 photon flux density, 50% humidity, 21 °C). For hydroponic culture, plants were germinated and grown for 2 weeks on plates containing one-half MS media before being transferred to hydroponic culture boxes containing Hoagland medium (Arnon & Hoagland 1939). The hydroponic cultures were grown for an additional 3 weeks under long-day conditions before experiments were conducted. To avoid effects caused by use up of the medium, the Hoagland medium was replaced every 3 d.

Screening of T-DNA insertion mutants and crossing

T-DNA insertion lines for cytosolic serine acetyltransferase (csat) (SALK_142158) and plastidic serine acetyltransferase (psat) (SALK_120440) targeting SERAT1.1 and SERAT2.1, respectively, were selected from the SALK T-DNA collection (Alonso et al. 2003). Homozygous mutants were identified by genomic PCR. Primers used for genotyping are listed in Supporting Information Table S1. The dmsat double mutant was obtained from crosses of csat and psat single mutants. Homozygous plants for both mutations were identified by RT-PCR, and the F4 generation was selected for further analysis.

DNA isolation and Southern blot analysis

Total DNA was isolated from Arabidopsis leaves according to Dellaporta, Wood & Hicks (1983). DNA extracted from csat plants was SalI digested, and DNA extracted from psat mutants was digested with XbaI. Restriction enzymes did not cut in the gene targeted by the insertion. Southern blot was conducted according to Sambrook, Fritsch & Maniatis (1989) with gene and insertion-specific probes obtained by PCR. Primer sequences used to generate the probes are listed in Supporting Information Table S1.

Assay for SAT and OASTL enzyme activity

Soluble protein extracts were prepared using 100 mg frozen Arabidopsis leaf or root material or fractions of lyophilized powder from non-aqueous gradients and 500 µL extraction buffer [50 mm 4-(2-hydroxyethyl)-1-piperazine ethanesulphonic acid (HEPES)/KOH, pH 7.4, 5 mm MgCl2, 1 mm ethylenediaminetetraacetic acid (EDTA), 1 mm ethylene glycoltetraacetic acid (EGTA), 0.1% (v/v) Triton-X (Sigma-Aldrich, Munich, Germany), 10% (v/v) glycerol, 5 mm 1,4-bis-sulphanylbutane-2,3-diol (DTT), 2 mm benzamidine, 2 mmε-aminocapronic acid, 0.5 mm phenylmethane sulphonylfluoride (PMSF) and 1 g L−1 polyvinylpolypyrrolidone (PVPP)] modified from Geigenberger & Stitt (1993). After centrifugation, the supernatant was desalted and total protein quantified according to Bradford (1976). SAT activity was assayed via high-performance liquid chromatography (HPLC) by the determination of the OAS content (Lindroth & Mopper 1979; Kim et al. 1997). Ten microlitre protein extract was incubated in a 90 µL reaction mixture containing 50 mm Tris/HCl pH 7.5, 0.1 mm acetyl–coenzyme A and 1.25 mm EDTA. The reaction was started by adding 10 mm L-serine and was terminated after 15 min by the addition of 50 µL 0.1 N HCL. The mixture was centrifuged at 4 °C for 10 min at 13 000 g and the supernatant used for HPLC.

OASTL activity was determined by measuring cysteine formation according to Riemenschneider et al. (2005). Cysteine concentration was determined using the method described by Gaitonde (1967).

Metabolite analysis

For organic ion analysis, 20 mg of freshly ground frozen plant material or material from non-aqueous fractionation was homogenized in 200 µL 0.1 mm HCl. Samples were centrifuged for 5 min at 14 000 g and 4 °C. The supernatant was transferred to Ultrafree MC 5000 MC NMWL Filter Unit (Millipore, Schwalbach, Germany) and was centrifuged for 90 min at 5000 g and 4 °C. After filtration, 20 µL of the diluted sample was analysed by HPLC with conductivity detection facilitating a Dionex ICS-2000 system (Dionex, Idstein, Germany). Ions were eluted in a KOH gradient.

Individual soluble thiols were determined as the sum of their reduced and oxidized forms according to Hell & Bergmann (1990). Amino acids were determined as described by Kreft, Hoefgen & Hesse (2003). OAS was determined following a modified protocol from Kim et al. (1997). Fifty milligrams of freshly ground frozen plant tissue or material from non-aqueous fractionation was extracted for 20 min at 4 °C sequentially with 400 µL 80% (v/v), 400 µL 50% (v/v) and 200 µL 80% (v/v) aqueous ethanol (buffered with 2.5 mm HEPES/KOH, pH 6.2). Ethanol/water extracts were subjected to HPLC analysis using a Hyperclone C18 base-deactivated silica (BDS) column (Phenomenex, Aschaffenburg, Germany) connected to an HPLC system (Dionex). OAS was measured by pre-column online derivatization with ortho-phtalaldehyde in combination with fluorescence detection (Lindroth & Mopper 1979; Kim et al. 1997). OAS was eluted similarly to amino acids (Kreft et al. 2003), but at pH 6.2 and with 11% (v/v) tetrahydrofurane in 8.5 mm sodium phosphate buffer. OAS stability was tested by determining the recovery rate. More than 90% of OAS was recovered after 10 h at pH 6.2 and 3 °C. Absence of co-eluting compounds was tested with samples incubated with borate buffer at pH 10.7, at which OAS completely converts to N-acetylserine (NAS), which is not accessible for derivatization.

Sulphide was determined following a modified protocol from Vetter et al. (1989), Völkel & Grieshaber (1992) and Wohlgemuth, Taylor & Grieshaber (2000). Twenty milligrams of freshly ground frozen plant tissue or material from non-aqueous fractionation was extracted for 30 min in the dark with 160 µL extraction buffer [10 µL 3-bromomethyl-5-ethyl-2,6-imethyl-pyrazolo (1,2-α)pyrazol-1,7-dione (Calbiochem, Darmstadt, Germany); 100 µL 160 mm HEPES/16 mm EDTA, pH 8.0; 50 µL acetonitrile] and was stabilized with 100 µL 65 mm methanesulphonic acid. The extracts were centrifuged for 15 min at 14 000 g and 4 °C. The supernatant was diluted one to four with solvent solution (88% 0.25% acetic acid, 12% methanol) and was analysed with a Merck (Darmstadt, Germany)/Hitachi (Tokyo, Japan) HPLC system [AS-2000 autosampler; L-6200 intelligent pump; F-1050 fluorescence detector; RP select B (5 µm), LiChrospher 60, LiChroCART 125-4 chromatography column; D-7000 HPLC System Manager Version 2.1]. The column eluent was monitored by fluorescence detection (λex 380/λem 480). Mixed standards treated exactly as the sample supernatants were used as a reference for sulphide quantification. Spiking assays were performed with 20 µm authentic standard and confirmed the identity of the sulphide peak.

Non-aqueous fractionation (NAF)

Two technical replicates were made from each gradient. Six gradients were made from biologically independent plant material of wild type and two from biologically independent material of csat and psat mutants, respectively. dmsat mutants were analysed in two technical gradients. Sulphide levels were analysed in two technical gradients of each mutant and wild type, respectively.

The method described here is based on the original method for the determination of subcellular metabolite levels in leaves by Gerhardt & Heldt (1984). Frozen A. thaliana leaf material (8 g) from plants grown on soil was homogenized using a ball mill and was freeze-dried. The dry leaf powder was re-suspended in 20 mL of a tetrachlorethylene–heptane mixture 66:34 (v/v), density of 1.3 g cm−3, and was ultrasonicated for a total of 120 s, with 6 × 10 cycle, 65% power (Bandolin Sonoplus HD 200, MS 73/D; Bandolin, Berlin, Germany). After sonication, the suspension was poured through a nylon net (20 µm pore size), diluted threefold with heptane and centrifuged for 10 min at 4000 g. The sediment was re-suspended in 3 mL of a tetrachlorethylene–heptane mixture (1.3 g cm−3). The linear gradient (30 mL, 1.43–1.62 g cm−3) was prepared in polyallomer tubes (Beckmann, Munich, Germany) using a peristaltic pump (Econo Pump; Bio-Rad, Munich, Germany). Following sample loading, the gradients were centrifuged for 50 min at 5000 rpm and 25 °C (rotor AS4.13, ultracentrifuge Centrikon T-124; Kendro, Berlin, Germany). After centrifugation, fractions were taken from the top of the gradient and three times the sample volume heptane was added before centrifugation for 10 min at 4000 g. Subsequently, pellets were dissolved in 10 mL heptane. Ten 1 mL aliquots were transferred into 2 mL Eppendorf (Hamburg, Germany) tubes. After centrifugation for 10 min at 14 000 g, the supernatants were discarded and the pellets dried in a desiccator for 24 h. Aliquots were stored under vacuum until analysis. For all analysed metabolites and enzymes in the fractions of the gradients, the recovery rate was determined (Supporting Information Table S2).

Analysis of compartment-specific marker enzymes and metabolites

Uridine diphosphate (UDP)-glucose-pyrophosphorylase (UGPase, EC activity as cytosolic marker was determined according to Zrenner, Willmitzer & Sonnewald (1993). NADP-glyceraldehyde-3-phosphate dehydrogenase (GAPDH, EC activity as chloroplast marker was determined according to Stitt, Wirtz & Heldt (1983). According to Winter, Robinson & Heldt (1994), nitrate was used as vacuolar marker. Nitrate concentration was analysed by ion-exchange chromatography (IEC; see previous discussion). Marker measurements from two technical replicates of each gradient were averaged (Riens et al. 1991).

Sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and Western blotting

SDS–PAGE and Western blotting was conducted according to Sambrook et al. (1989). Western blots were blocked with skimmed milk and were probed with polyclonal primary antibody against adenosine 5′-phosphosulphate-reductase (APR, EC, OASTL (EC or cytochrome c oxidase. Primary antibodies were detected with anti-rabbit alkaline phosphatase-conjugated secondary antibodies. Blots were developed using CDP-Star chemoluminescense Kit (Roche, Indianapolis, IN, USA).

RNA extraction and RT-PCR

Total RNA was extracted with RNeasy Plant Mini Kit (Qiagen, Hilden, Germany) and was digested with TURBO DNA-free (Ambion, Huntingdon, UK). Absence of genomic DNA contamination was confirmed by PCR. Total RNA (5 µg) was subjected to cDNA synthesis using oligo(dT) primer and Superscript III (Invitrogen, Karlsruhe, Germany), according to the manufacturer's instructions.

The RT-PCR of SERAT1.1, SERAT2.1 and ACTIN cDNA was conducted in three repetitions with cDNA from biologically independent plant material of wild-type, csat, psat and dmsat plants. PCR reactions contained 1 µm Primer, Taq polymerase and buffer (Finnzymes, Espoo, Finland) and 5 mm deoxyribonucleotides. Primer sequences are listed in Supporting Information Table S1. After 5 min of initial denaturation at 92 °C, 30 cycles of 30 s at 92 °C, 30 s at 60 °C and 2 min at 74 °C were performed. Amplicons were made visible on ethidium bromide-stained agarose gels.

Data processing

Scanned gel and protein or DNA blot images were processed using Adobe Photoshop 7.0 software (Adobe Systems Inc., Mountain View, CA, USA). Pixel intensities were quantified using ImageJ ( software. Student's t-test was executed in Microsoft Excel. Descriptive statistics and graphical visualizations were computed with R. Hierarchical cluster analyses were performed as unweighted average linkage clustering on Euclidean distances (Mirkin 1996). Comparison between Euclidean distance calculations for seven gradients obtained from wild type and mutants and two wild-type gradients (Fig. 5b, Supporting Information Fig. S2) and comparison between Euclidean distance calculation and average abundance (Supporting Information Figs S1 & S2) demonstrate suitability of this method for evaluation of the NAF data. The Mantel test, and respective analysis of variance (anova), was performed as parametric Pearson correlation with 9999 bootstrap samples (Sokal & Rohlf 1995). The test of homogeneity was computed as described by Sokal & Rohlf (1995). The compartmental distribution was estimated according to Riens et al. (1991) by using a C version of the compartment calculation programme Bestfit (Steinhauser et al., unpublished data).

Figure 5.

Comparison of enzyme activity and metabolite distributions across non-aqueous gradients made from Arabidopsis leaf material. (a) Citrate synthase and serine acetyltransferase (SAT) activity in comparison to marker enzyme activities and concentrations for cytosol (UGPase), plastids [glyceraldehyde-3-phosphate dehydrogenase (GAPDH)] and vacuole (nitrate) in each fraction as percent of total. Mean values and SD from four biologically independent gradients made from wild-type leaf material were calculated for each fraction. Four fractions were taken from each gradient. Asterisks mark significant differences according to Student's t-test between citrate synthase and UGPase activities and between SAT and UGPase activities (*P < 0.05, **P < 0.01, ***P < 0.001). Significant differences between citrate synthase and SAT activities are indicated by the letter ‘a’ (aP < 0.05). Significant differences between citrate synthase activity and nitrate concentration and between SAT activity and nitrate concentration are indicated by the letter ‘b’ (bP < 0.05, bbP < 0.01, bbbP < 0.001). (b) Heat map visualization and cluster tree representation of the average distances among the compounds based on a total of seven individual non-aqueous gradients made from Arabidopsis wild-type, cytosolic serine acetyltransferase (csat), plastidic serine acetyltransferase (psat) and dmsat leaf material. Ten fractions were taken from each gradient. The heat map was generated by averaging the Euclidean distances, which are drawn in each cell, among all compounds or activities for each individual gradient on the basis of the percentage abundance in each fraction. The left top-side colour bar represents the colour code for the Euclidean distance in a range from 0 (very light blue) to 75 (dark blue) with a step width of 2.5 arbitrary units. Values above 50 are coloured dark blue. Assignment into the cluster groups was made according the average linkage cluster tree and is drawn as colour-coded row/column on the top and at the side: black = vacuole, green = chloroplast, light blue = cytosol #1, dark blue = cytosol #2. Cluster validity was tested by analysis of variance (anova) performed as parametric matrix correlation, that is, Mantel statistic (Sokal & Rohlf 1995), between the distance and the group matrix. The test revealed a statistically significant Pearson correlation of r = 0.4534 with a P < 2e-4 between the two matrices, evaluated by row and column permutation (bootstrap analysis with n = 9999 permutations). UGPase, UDP-glucose-pyrophosphorylase; OASTL, O-acetylserine-(thiol)-lyase; OAS, O-acetylserine; gEC, y-glutamylcysteine; glu, glutamate; asp, aspartate; cys, cysteine; gsh, glutathione; cp, chloroplast; vc, vacuole; cs, cytosol.

Accession numbers

Accession numbers of genes analysed in this study are At5g56760 (SERAT1.1) and At1g55920 (SERAT2.1).


Molecular characterization of csat, psat and dmsat mutants

To reveal the importance of cytosolic and plastidic cysteine biosynthesis, two SAT T-DNA insertion mutants were characterized. The mutants carried the insertion in the gene for the plastidic SERAT2.1 (At1g55920) and the cytosolic SERAT1.1 (At5g56760) isoform, respectively. In addition, the two mutants were crossed and the resulting double knockout mutant (dmsat) was investigated. To confirm single insertion of T-DNA into the genome, Southern blot analysis was conducted with plants homozygous for each single mutant (Fig. 1a). DNA blot analysis using gene- and T-DNA-specific probes, respectively, revealed that the insertion was located exclusively in the target genes (Fig. 1a). The csat mutant apparently carried a triple T-DNA insertion in the first exon (Fig. 1b). In the psat mutant, only one T-DNA insertion in the SERAT2.1 gene was identified. RT-PCR analysis established that T-DNA insertion had led to a complete disruption of the respective SAT gene, as endogenous SERAT1.1 and SERAT2.1 gene expression was completely abolished in csat and psat mutants (Fig. 1c). In the double mutant, neither transcript accumulated to detectable levels, which confirmed the identity of the double mutant.

Figure 1.

Molecular characterization of serine acetyltransferase (SAT) T-DNA insertion mutants. (a) DNA blot hybridization of wild-type (WT), csat and psat DNA with a gene- and a T-DNA-specific probe. (b) Localization of T-DNA insertions in csat and psat mutant alleles. Arrowheads indicate the position of T-DNA insertion. Exons are marked as black boxes; white boxes depict untranslated regions. Locations and directions of primers used for RT-PCR are indicated by black arrowheads. (c) RT-PCR analysis of sat mutant and wild-type plants. Gene-specific primers were used to amplify the SERAT2.1 and SERAT1.1 transcript. csat, cytosolic serine acetyltransferase; psat, plastidic serine acetyltransferase.

SAT and OASTL activity in wild-type and sat mutant plants

During the growth period under standardized greenhouse conditions, none of the mutants revealed a visible morphological phenotype or growth retardation when compared to wild type. To examine the direct consequences of the knockout of the compartment-specific SAT isoforms, total activities of SAT and OASTL were investigated. In leaves, total sat activity was significantly reduced to 71% of wild-type activity in csat plants. In roots, csat and dmsat plants displayed 68 and 80% of wild-type SAT activity (Fig. 2a,b). No significant alteration in total SAT activity was detected in psat plants. Compared to wild type, OASTL activity was not altered in the mutants (Fig. 2c,d).

Figure 2.

Serine acetyltransferase (SAT) and O-acetylserine-(thiol)-lyase (OASTL) activity in mutant and wild-type plants. (a,b) SAT and (c,d) OASTL activity was analysed in leaves (a,c) and roots (b,d) of sat mutant lines and wild type. Mean value and SD of four biological replicates for each mutant and the wild type are presented. Asterisks mark significant differences between wild type and mutant (*P < 0.05).

Analysis of metabolites related to cysteine biosynthesis

Concentrations of metabolites required for and dependent on cysteine biosynthesis were measured in the different sat mutants and wild type. OAS and cysteine content significantly decreased in leaves and roots of csat and dmsat mutants compared to wild type (Fig. 3a–d). While in leaves glutathione levels remained unchanged in all mutants and γ-glutamylcysteine content was reduced only in leaves of dmsat plants compared to the wild type, γ-glutamylcysteine and glutathione levels were significantly reduced in the csat mutant in root tissue (Fig. 3e–h). Sulphate measurements exhibited a significant decrease of this metabolite in csat leaves; in roots, sulphate levels remained unaltered in all three mutants compared to wild type (Fig. 3i,j). Sulphide levels were strongly reduced in the psat mutant, while they were only slightly decreased in the csat mutant relative to wild type. The dmsat mutant displayed no change in sulphide content versus wild type (Fig. 3k). Levels of methionine, a sulphur containing amino acid derived from cysteine, did not display significant differences in leaves of sat mutant plants compared to the wild type (data not shown).

Figure 3.

Metabolic analysis of sat mutants. Mean value and SD for O-acetylserine (OAS) (a,b), cysteine (c,d), γ-glutamylcysteine (e,f), glutathione (g,h), sulphate (i,j) and sulphide (k) concentrations in leaves (a,c,e,g,i,k) and in roots (b,d,f,h,j) of three to four biological replicates for each mutant and wild type. Asterisks mark significant differences between wild type and mutant (*P < 0.05, **P < 0.01). FW, fresh weight.

Subcellular distribution of enzyme activities and metabolites in sat mutant and wild-type plants

To further investigate the compartmentation of cysteine biosynthesis NAF was employed. NAF bases on separation of proteins and metabolites from specific compartments within the cell aggregating upon lyophilization via density gradients in a non-aqueous environment. As the non-aqueous environment diminishes catabolic and anabolic reactions, NAF is specifically suited to determine steady-state metabolite levels within the different subcellular compartments (Stitt et al. 1983; Gerhardt & Heldt 1984; Riens et al. 1991; Farréet al. 2001; Tiessen et al. 2002).

The distribution of the different compartments throughout non-aqueous gradients of wild-type and sat mutant leaf material was determined using compartment-specific markers (Fig. 4). The abundance of these marker enzyme activities, protein levels and metabolite concentrations in each of the 10 collected fractions reflects the enrichment of the respective compartments in each fraction. Thus, the distribution of the marker represents the distribution of the specific compartment throughout the gradient. As the separation of subcellular compartments is not complete with this method (Stitt et al. 1983; Gerhardt & Heldt 1984; Riens et al. 1991), two criteria have to be met to designate different subcellular compartments. Firstly, it is crucial that each marker is more abundant in at least one fraction compared to the other markers, and second, the distribution of the markers across the gradient has to be sufficiently distinct from each other. As markers for the plastids, cytosol, mitochondria and vacuole, respectively, served NADP-GAPDH (Gerhardt & Heldt 1984; Riens et al. 1991) and UGPase activity (Oparka et al. 1992), cytochrome c oxidase levels (Farréet al. 2001) and nitrate (Kaiser et al. 1989; Winter, Robinson & Heldt 1993; Martinoia, Massonneau & Frangne 2000) concentration. The plastidic marker GAPDH was enriched in the lighter fractions 7–10, while the vacuolar marker nitrate was clearly enriched in the densest fraction (Fig. 4a,b). Localization of the plastidic compartment across the gradients was confirmed by immunoblot analysis for S-adenylyl-sulphate reductase (APS-reductase, EC abundance within the different fractions (Fig. 4d,e). The cytosolic marker UGPase was relatively equally distributed throughout the gradients with a slight enrichment in the densest fraction, but, compared to GAPDH and nitrate, was clearly enriched in the middle region of the gradients (Fig. 4a,b). The mitochondrial marker cytochrome c oxidase was detected throughout the gradient with enrichment in the more dense fractions (Fig. 4c). To confirm mitochondrial localization between the cytosolic and vacuolar compartments, citrate synthase activity was measured in four gradients of biologically independent Arabidopsis Columbia-0 leaf material. Four fractions were collected per gradient (Fig. 5a). Citrate synthase activity measurements established that despite the clearly distinct distribution of mitochondria and cytosol as well as mitochondria and vacuole, the mitochondrial marker was in no fraction superior compared to the other markers. Thus, unfortunately, and similar as reported previously for many plant species (Gerhardt & Heldt 1984; Farréet al. 2001; Tiessen et al. 2002; Fettke et al. 2005), we were not able to unambiguously resolve the mitochondrial compartment via NAF. In consequence, as the mitochondrial markers had intermediate abundances between cytosolic and vacuolar markers in all fractions, we cannot differentiate between partially cytosolic and vacuolar or mitochondrial distributions of a given metabolite by comparing its abundance in each fraction with the marker distributions throughout the gradients. Nevertheless, the achieved separation is equivalent to other published examples of NAF (see e.g. Gerhardt & Heldt 1984; Dancer, Neuhaus & Stitt 1990; Farréet al. 2001; Tiessen et al. 2002, Fettke et al. 2005; Riewe et al. 2008). The gradients prepared from Arabidopsis leaf material clearly resolved the three compartments, cytosol, plastids and vacuole, although it has to be noted that mitochondria were contained in the cytosolic and vacuolar compartments (Figs 4a,b & 5a). Thus, compartments assigned to a vacuole and cytosol in this study in fact represent vacuolar and partially mitochondrial and cytosolic and partially mitochondrial compartments. With respect to variability in the determination of enzyme activities and metabolite concentrations, averages and SDs of marker enzyme activities and marker metabolite abundances in each fraction are depicted for gradients made from wild-type, csat, psat and dmsat mutants of two biologically independent experiments (Fig. 4a,b), and a third experiment with four gradients of biologically independent plant material from Arabidopsis wild-type leaves (Fig. 5a). Marker distributions across the gradients were comparable for all three experiments. Moreover, significant similarities in the range of 0.828 ≤ r ≤ 0.962 (P < 1e-04) between all gradients could be confirmed by Mantel test (Sokal & Rohlf 1995; Supporting Information Table S3). Test of homogeneity (Sokal & Rohlf 1995) confirmed no significant differences among the obtained correlations (P = 0.8754). Taken together, these analyses on the comparability of the different gradients established that we were able to reliably determine the subcellular localization of sulphur metabolism-related metabolites and enzyme activities.

Figure 4.

Distribution of compartment-specific markers in non-aqueous gradients made from Arabidopsis thaliana leaves. The distribution of marker enzyme activities and the marker metabolite nitrate across gradients made from wild-type, cytosolic serine acetyltransferase (csat), plastidic serine acetyltransferase (psat) and dmsat leaf material in two biologically independent experiments is shown in (a) experiment 1 and (b) experiment 2. Mean values and SD of marker enzyme activity or concentration, respectively, in each fraction is depicted as percent from total. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as marker for the chloroplasts; UDP-glucose-pyrophosphorylase (UGPase) is a marker for the cytosol, and nitrate was used as a marker for the vacuole. Significant differences between vacuolar and cytosolic marker and plastidic and cytosolic marker concentrations and activities, respectively, in each fraction are indicated by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001). Letters mark significant differences between plastidic and vacuolar markers in each fraction (aP < 0.05, aaP < 0.01, aaaP < 0.001). (c) Distribution of mitochondria across the gradients was determined by Western blot detecting cytochrome c oxidase abundance in each fraction. A representative protein blot is shown. (d) Western blot detecting APS-reductase abundance in each of the fractions taken from one representative gradient to confirm the distribution of plastids throughout the gradient and (e) Western blot detecting APS-reductase abundance in fraction 1, 10 and a comparable volume of leaf extract before fractionation.

To determine to which percentage an enzyme activity or metabolite of interest was localized in the vacuole, cytosol and plastids, the abundance of the respective enzyme activity or metabolite was measured in each fraction of a gradient and the distribution throughout the gradient fitted using the approach described by Riens et al. (1991). The Bestfit algorithm uses the compartment-specific marker distributions to calculate theoretical abundances in each fraction in 1% intervals from 0 to 100% for each compartment and yields the subcellular distribution, which most closely relates to the experimental results. A measure for quality of the fit for each compound is given in Supporting Information Table S4.

Applying the Bestfit approach to determine the subcellular distribution of SAT activity yielded a cytosolic and vacuolar localization for wild type and the different SAT mutants (Fig. 6a). No SAT activity was found in the plastidic compartment. However, as none of the Arabidopsis SAT isoenzymes is localized in the vacuole (Noji et al. 1998; Howarth et al. 2003; Kawashima et al. 2005), a vacuolar localization of SAT activity can be excluded. The distribution of wild-type SAT activity and the distribution of the mitochondrial marker citrate synthase across the gradients strongly resembled each other as revealed by Student's t-test (Fig. 5a). Together with the significant differences compared to cytosolic and vacuolar marker distribution, this implies that most SAT activity is located in mitochondria. For OASTL activity, the Bestfit approach revealed distribution between the cytosol and plastids for all mutants and wild type (Fig. 6b). Comparison of Euclidean distances calculated for OASTL activity and abundance of glutamate and aspartate, two metabolites that have previously been shown to be localized to relatively equal amounts in the cytosol and plastids (Gerhardt & Heldt 1984; Riens et al. 1991; Winter, Lohaus & Heldt 1992; Winter et al. 1993, 1994; Leidreiter et al. 1995; Tilsner et al. 2005), revealed strong similarity (Fig. 5b). Unlike as for SAT activity, no clear conclusion about mitochondrial localization can be drawn for metabolites. However, the subcellular distribution of OAS strongly resembled the distribution of SAT activity (Figs 5b & 6c). Sulphide was mainly located in plastids and to a minor extent in the cytosol and vacuole (Fig. 6d). For cysteine, the Bestfit approach revealed a predominantly cytosolic localization, as more than 90% of the cellular cysteine was assigned to this compartment in all lines analysed (Fig. 6e). Comparing the Euclidean distances for this metabolite across the gradient with the Euclidean distances for the markers, a strong similarity between cysteine and the cytosolic marker UGPase could be observed (Fig. 5b). Bestfit calculations for γ-glutamylcysteine revealed distribution between all three compartments; however, γ-glutamylcysteine was enriched in the cytosol compared to vacuole and plastids (Fig. 6f). Glutathione was relatively equally distributed between all three compartments (Fig. 6g). Calculation of Euclidean distances for this metabolite revealed similarity to OAS and SAT activity (Fig. 5b). Interestingly, the compounds contained in cluster cytosol #2 have been described to occur in substantial amounts in mitochondria (Ruffet et al. 1995; Zechmann, Zellnig & Müller 2005; Heeg et al. 2008). The determination of subcellular sulphate pools revealed that sulphate is predominantly localized in the vacuole (Figs 5b & 6h). Low amounts of sulphate could be detected in the cytosol and sulphate was below detection limit in plastids.

Figure 6.

Subcellular distribution of cysteine synthesis-related enzyme activities and metabolite concentrations in serine acetyltransferase (SAT) T-DNA insertion mutants compared to wild type. Percent distribution of SAT activity (a), O-acetylserine-(thiol)-lyase (OASTL) activity (b), O-acetylserine (OAS) (c), sulphide (d), cysteine (e), γ-glutamylcysteine (f), glutathione (g) and sulphate (h) concentration between vacuolar, cytosolic and plastidic compartments. Mitochondria were contained in the vacuolar and cytosolic compartments. Enzyme activities and metabolite concentrations measured in each of 10 fractions taken from non-aqueous gradients made from leaf material of wild-type, csat, psat and dmsat plants were assigned to the respective compartments by applying the Bestfit algorithm (Riens et al. 1991), which compares the distribution of the respective metabolite or enzyme activity with the distribution of each marker across the respective gradient. Two biologically independent gradients of wild-type, csat and psat mutants were analysed, and mean values of the percent distribution of enzyme activities and metabolite abundances between the different compartments are depicted. For dmsat (white bars) and sulphide, only one gradient was analysed. csat, cytosolic serine acetyltransferase; psat, plastidic serine acetyltransferase.

Calculation of subcellular metabolite concentrations

Assessment of the subcellular metabolite concentrations was achieved by normalizing the relative abundances of metabolites in the different subcellular compartments to the total content of the respective metabolite in leaf tissue and to the average organelle volumes published for spinach, barley, potato and rape (Winter et al. 1992, 1993, 1994; Leidreiter et al. 1995; Tilsner et al. 2005). Table 1 summarizes the subcellular metabolite concentrations for wild-type plants determined from two gradients of biologically independent plant material. Subcellular metabolite concentrations estimated with this method were well comparable with those published for mesophyll cells of different plant species. The cytosolic glutathione concentration assessed here (Table 1) was in the range of that determined by Meyer, May & Fricker (2001), who measured 2.7–3.2 mm glutathione in the cytosol of A. thaliana mesophyll cells. To further validate our data, we estimated the subcellular glutamate concentrations to be approximately 17.5 mm in plastids, 21 mm in the cytosol and 0.3 mm in vacuoles (Table 1). These values are in good agreement with the organellar concentrations of glutamate determined for S. oleraceae and Brassica napus leaf cells (Riens et al. 1991; Tilsner et al. 2005), where glutamate occurred in concentrations of 14 and 16 mm, respectively, in plastids; 21 and 14 mm, respectively, in the cytosol; and 0.5 and 1.9 mm, respectively, in vacuoles. Assessment of subcellular concentrations of sulphur metabolism-related metabolites revealed that OAS was, with an estimated value of approximately 40 µm, more than ten times higher concentrated in the cytosol compared with the two other compartments. Sulphide concentrations of 125 µm were estimated in plastids, but interestingly, the cytosol also contained 55 µm sulphide. Very low sulphide concentrations were found in the vacuolar compartment. Sulphide and OAS concentrations in the cytosol were in a similar range. Cysteine concentrations in the cytosol were estimated to be over 300 µm, whereas the other two compartments only contained below 10 µm cysteine. With regard to metabolites relying on cysteine for biosynthesis, γ-glutamylcysteine was shown to be present in concentrations of approximately 150 and 300 µm in the cytosol and about 90 and 120 µm in plastids. Low γ-glutamylcysteine concentrations were calculated for the vacuolar compartment. Though more variable, subcellular glutathione concentrations in the cytosol and plastids were in the range of 3 mm and, with below 1 mm, were much lower in the vacuole. Sulphate concentrations were calculated to be higher than 8 mm in the vacuolar compartment and lower than 4 mm in the cytosol.

Table 1.  Compartment-specific metabolite concentration in Arabidopsis leaves
MetaboliteTotal tissue content (nmol g fresh weight−1)Compartment-specific concentration (mm)
  1. The compartment-specific concentration was estimated from total tissue content, the percent distribution between the different subcellular compartments obtained after non-aqueous fractionation of wild-type leaf material and the average organellar volumes published for spinach, barley, potato and rape (Winter et al. 1992, 1993, 1994; Leidreiter et al. 1995; Tilsner et al. 2005). The two values for each compartment correspond to estimated concentrations calculated with the percent organellar distributions obtained from two independent gradients. n.d., not determined.

O-acetylserine2.5 ± 0.230.002 | 0.0010.047 | 0.0310.005 | 0.001
Sulphide13.2 ± 2.370.005 | n.d.0.055 | n.d.0.125 | n.d.
Cysteine17.4 ± 1.880.003 | 0.0020.332 | 0.3050.009 | 0.001
γ-Glutamylcysteine19.5 ± 2.120.008 | 0.0010.297 | 0.1540.124 | 0.086
Glutathione733.1 ± 122.040.733 | 0.6133.518 | 3.0123.176 | 2.544
Sulphate5242.2 ± 260.998.737 | 8.3873.823 | 1.276–|–
Glutamate2262.7 ± 133.820.269 | 0.56523.758 | 20.81619.610 | 16.483


The cytosolic and plastidic serine acetyltransferase isoforms SERAT1.1 and SERAT2.1 do not contribute to cysteine biosynthesis to a major extent

While the cysteine biosynthesis pathway has been elucidated in plants, its regulation through compartmentation and the efficiency of cysteine biosynthesis in the different cellular organelles is still not clearly resolved. The plastids have long been assumed to largely contribute to cysteine biosynthesis as sulphur reduction exclusively takes place in this compartment (Leustek et al. 2000; Saito 2000; Droux 2004; Kopriva 2006). Furthermore OASTL activity was shown to be highest in the cytosol and plastids of spinach leaves, compartments where the SAT to OASTL ratio supports efficient cysteine synthesis (Lunn et al. 1990; Ruffet, Droux & Douce 1994). Very recently, this has been confirmed for Arabidopsis with OASTL T-DNA insertion mutants (Heeg et al. 2008; Watanabe et al. 2008a). Here we complement these data by analysing T-DNA insertion mutants for the plastidic (SERAT2.1) and cytosolic (SERAT1.1) SAT enzymes for their specific impact on cysteine biosynthesis. In addition, analysis of the subcellular distribution of enzyme activities and metabolites related to cysteine biosynthesis in the mutants and wild type should give insight into the regulation of cysteine biosynthesis and help to reveal to what extent the different compartments contribute to this biosynthetic pathway.

T-DNA insertion resulted exclusively in disruption of SERAT1.1 and SERAT2.1 genes, respectively, and consequently in the absence of SERAT1.1 and SERAT2.1 specific transcripts in csat, psat and dmsat mutant plants. In line with the lack of visible phenotypic differences between sat mutant and wild-type plants, high residual SAT activity was observed for all three mutants compared to wild type. Quantitative RT-PCR demonstrated that residual SAT activity does not result from transcriptional up-regulation of other SAT isoenzymes in the mutants (data not shown). Therefore, it is likely that a different SAT activity, presumably the mitochondrial SERAT2.2, which has been shown to exhibit better kinetic parameters compared with SERAT3.1 and SERAT3.2 (Kawashima et al. 2005), predominantly contributes to OAS formation. In agreement with this observation, it was demonstrated that mitochondrial SAT activity constitutes more than 75% of total SAT activity in plants (Ruffet et al. 1995; Heeg et al. 2008), and specific down-regulation of mitochondrial SAT through expression of artificial micro RNA led to severely retarded growth, strongly reduced OAS levels and impaired incorporation of radioactively labelled serine into cysteine and glutathione (Haas et al. 2008). A very recent study employing knockout plants for all members of the Arabisopsis SAT gene family as well indicated that mitochondrial SAT predominantly contributes to cellular OAS formation as T-DNA insertion mutants for this enzyme displayed strongly reduced SAT activity and OAS levels in leaves and roots of 2-week-old seedlings grown on sucrose-containing medium (Watanabe et al. 2008b). Presumably due to the different growth conditions, the authors did not observe significantly reduced SAT activity, OAS and thiol levels in the mutants for cytosolic SERAT1.1; however, quadruple mutants, in which only SERAT1.1 or SERAT2.2 genes were retained, grew normally. This suggested that SERAT1.1 also plays an important role for OAS formation in vivo (Watanabe et al. 2008b). The significant decrease of OAS and cysteine concentrations in csat and dmsat mutants observed here provides further evidence that SERAT1.1 is involved in OAS and cysteine biosynthesis to a certain extent. In contrast, plastidic OAS synthesis appears to be of marginal importance for cysteine biosynthesis in A. thaliana as enzyme activities and metabolite levels related to cysteine biosynthesis exhibited no significant modulations in the mutants for the sole plastidic SAT isoform SERAT2.1 (the results presented here; Watanabe et al. 2008b). In agreement with these observations, plastidic OASTL does also not appear to be of substantial importance for cysteine biosynthesis in A. thaliana (Heeg et al. 2008; Watanabe et al. 2008a). A possible role for SERAT1.1 and SERAT2.1 may be to overcome transport limitations of OAS from mitochondria. As similar tendencies were observed for the extent to which metabolite concentrations were affected in leaf and root material when the different mutants were compared to each other and to wild type, the different SAT enzymes appear to fulfil similar functions in leaves and roots. Analysis of sulphide levels hinted to a function of SERAT1.1 and SERAT2.1 in the regulation of sulphate reduction. Similarly, it has been hypothesized that sulphate activation and reduction may be reduced in Arabidopsis plants silenced for the mitochondrial SAT enzyme (Haas et al. 2008). The absence of OAS produced by compartment-specific isoforms in the mutants could have resulted in reduced sulphate assimilation. That OAS positively regulates sulphate assimilation has been demonstrated for bacteria and plants (Neuenschwander, Suter & Brunold 1991; Kredich 1992; Koprivova et al. 2000; Tsakraklides et al. 2002). However, dmsat mutants did not display reduced sulphide levels compared to wild-type plants, indicating that signalling leading to down-regulation of sulphate reduction in the single mutants had been abrogated in the double mutant. Sulphate levels were reduced only in leaves of csat plants implying a role of SERAT1.1 in the regulation of sulphate uptake. Interestingly, decreased sulphate levels were also observed in mutants lacking the cytosolic OASTL enzyme (Heeg et al. 2008). Thus, disturbing cytosolic cysteine synthesis seems to interfere with sulphate uptake.

The compartmentation of cysteine biosynthesis

The use of mutants or transgenic plants for cysteine biosynthetic enzymes is limited as it provides only indirect evidence for the interplay of the different compartments for cysteine biosynthesis through overall metabolite measurements. Subcellular metabolite analysis complements this approach by allowing direct evaluation of the impact of the different compartments on sulphur metabolism through assessment of organelle steady-state metabolite concentrations. The only method available today for analysis of subcellular metabolite concentrations is NAF as, in contrast to other methods such as organelle purification or immunolabelling after chemical fixation of leaf tissue, it efficiently quenches metabolism directly after sampling through freezing and subsequently through the non-aqueous environment in which the separation of organelles is achieved. Comparison of the distribution of SAT activity with that of the mitochondrial marker citrate synthase throughout non-aqueous gradients of wild-type leaf material suggests that most SAT activity is located in mitochondria in A. thaliana. This observation is consistent with the finding that csat, psat and dmsat mutants exhibited high residual overall SAT activity and confirms recent results obtained by Heeg et al. (2008), who showed that specific SAT activity in the mitochondria of heterotrophically grown Arabidopsis cell cultures was higher compared with crude plant extract. Very recently, Haas et al. (2008) and Watanabe et al. (2008b) were able to demonstrate that plants with decreased mitochondrial SAT activity display strongly reduced OAS levels and reduced flux into cysteine and glutathione (Haas et al. 2008). As csat plants still displayed approximately 80% of wild-type SAT activity and no SAT activity was detected in plastids, the cytosol will presumably constitute the residual 20% of SAT activity. Co-distribution of OAS content and SAT activity in the non-aqueous gradients further suggests largest amounts of OAS in the mitochondria, fewer cytosolic and little plastidic OAS. In contrast, OASTL activity was localized in the cytosol and plastids, hence similarly distributed as in P. sativum and D. innoxia (Lunn et al. 1990; Kuske et al. 1996). In agreement with these observations, Heeg et al. (2008) and Watanabe et al. (2008a) demonstrated that T-DNA insertion mutants for cytosolic and plastidic OASTL displayed reduced activity relative to wild type and mitochondrial OASTL mutant. Transgenic potato plants silenced for cytosolic and plastidic OASTL, respectively, also exhibited reduced OASTL activity with a more pronounced effect being associated with silencing of the cytosolic enzyme (Riemenschneider et al. 2005). The distribution of cysteine across the gradients strongly resembled the distribution of the cytosolic marker leading to a predominantly cytosolic localization of this metabolite in the Bestfit approach. In agreement with this finding, analysis of cysteine content in different compartment-specific OASTL mutants revealed reduced cysteine levels in leaves only for mutants of the cytosolic OASTL enzyme (Heeg et al. 2008; Watanabe et al. 2008a). Moreover, among the different OASTL mutants, only the mutant lacking the major cytosolic enzyme exhibited reduced incorporation capacity for radioactively labelled sulphate (Heeg et al. 2008). Thus, given that cysteine is predominantly formed in the cytosol, OAS has to be transported from the mitochondria into the cytosol for efficient cysteine biosynthesis. In plants, the de novo sulphate reduction takes place in the plastids (Leustek et al. 2000; Kopriva et al. 2001; Kopriva 2006). Consistent with this, a large part of cellular sulphide was localized in plastids. Nevertheless, considerable amounts of sulphide were distributed to the vacuolar and cytosolic compartments providing experimental proof for earlier postulations of cytosolically localized sulphide (Riemenschneider et al. 2005; Wirtz & Hell 2007; Heeg et al. 2008; Watanabe et al. 2008a,b). The presence of sulphide in compartments different from plastids requires transport across membranes. It has been speculated that H2S reaches the cytosol via diffusion through the chloroplast envelope membrane (Wirtz & Hell 2007). However, the chloroplast stroma reaches a pH of 8.5 under illumination (Heldt et al. 1973; Laisk & Oja 1988; Wu & Berkowitz 1992), conditions at which 95% of sulphide would be present in the charged HS- form. Therefore, sulphide is likely transported across the chloroplast envelope membrane. Vacuolar localized sulphide might result from degradation of thiol-containing compounds in the vacuole, similar to that described by Grzam et al. (2006, 2007). As the vacuolar and cytosolic compartments cannot be separated from the mitochondria via NAF (Gerhardt & Heldt 1984; Farréet al. 2001; Tiessen et al. 2002), low concentrations of sulphide within mitochondria cannot be excluded. However, the presence of sulphide within mitochondria is unlikely, as already nano-molar concentrations of sulphide in this compartment would lead to inhibition of cytochrome c oxidase similar to inhibition of this enzyme by cyanide (Smith, Kruszyna & Smith 1977; Nicholson et al. 1998; Dorman et al. 2002; Truong et al. 2006). Taken together, the need to avoid accumulation of toxic sulphide concentrations in mitochondria, the absence of detectable SAT activity in plastids, the presence of sulphide in the cytosol and the predominantly cytosolic localization of cysteine demonstrate that the cytosol represents the main site for cysteine biosynthesis in A. thaliana. Predominantly mitochondrial localized SAT activity and only little contributions of SERAT1.1 to total SAT activity and OAS levels, however, necessitate OAS transport into the cytosol. These findings are in agreement with results obtained by Heeg et al. (2008) and Watanabe et al. (2008a) showing that cytosolic OASTL dominantly contributes to cysteine biosynthesis, whereas plastidic OASTL only plays a minor role.

Efficiency of cysteine biosynthesis is restricted by OAS levels

Assessment of the distribution of metabolites between different subcellular compartments allowed estimation of metabolite concentrations in the vacuole, cytosol and plastids. Unfortunately, NAF does generally not allow sufficient separation of mitochondria; thus, metabolite concentrations in this compartment could not be evaluated, and metabolite concentrations in the vacuole and cytosol are overestimated if a metabolite occurs predominantly in mitochondria. However, as the mitochondrial volume is small compared to the volumes of cytosol and vacuole, the error introduced by the presence of mitochondria in the vacuolar and cytosolic compartments is expected to be small. The calculated cytosolic and plastidic OAS concentrations were essentially below the Km values (0.3 and 0.6 mm; Wirtz, Droux & Hell 2004) of the cytosolic and plastidic OASTL isoenzyme, but in the range promoting binding of OAS by OASTL (Wirtz et al. 2004). In contrast, the estimated sulphide concentration in plastids and the cytosol was considerably higher than the Km values of the respective OASTL isoenzymes (3.2 and 5.6 µm; Wirtz & Hell 2006). So far, only one study reports on cellular sulphide concentrations in plants (Papenbrock et al. 2007). Depending on the cell layer, the authors were able to measure sulphide concentrations between 1 and 15 µm using an H2S microelectrode. Total leaf tissue sulphide levels determined in our study normalized to the total cellular volume yield, with 18.5 µm, comparable concentrations. The determination of subcellular OAS and sulphide concentrations holds important implications for OAS rather than sulphide being the limiting factor for cysteine biosynthesis in the cytosol and plastids. Studies on plants overexpressing SAT, OASTL or Pseudomonas aeruginosa APR corroborate this finding by demonstrating that increased supply with OAS leads to enhanced cysteine biosynthesis (Saito et al. 1994; Harms et al. 2000; Tsakraklides et al. 2002; Sirko, Blaszczyk & Liszewska 2004).

Consequences of subcellular metabolite concentrations for the CSC

Cysteine biosynthesis appears to be controlled by regulation of the CSC through metabolites (Kredich et al. 1969; Droux et al. 1998; Wirtz et al. 2001). Metabolite-driven dissociation and association of the complex was intensively studied in vitro, and the equilibrium constants for OAS (58–77 µm) and sulphide (approximately 50 µm) were resolved (Berkowitz et al. 2002; Wirtz & Hell 2006). Estimation of the cytosolic OAS concentration under non-stress conditions revealed that it is below the equilibrium constant for OAS; therefore, OAS levels in the cytosol should not be sufficient to dissociate the CSC. The concentration of sulphide in the cytosol was in the range of the association constant of the CSC. Consequently, the cytosolic sulphide and OAS concentrations rather promote association than dissociation of the complex and therefore support maximum efficiency of cytosolic OAS synthesis under non-stressed conditions. In plastids, the concentrations of OAS and sulphide also support association of the CSC. However, the fact that no SAT activity could be assigned to plastids argues against efficient OAS synthesis within this compartment, which is consistent with the results obtained from the analysis of psat mutant plants.

SERAT1.1 appears not to be completely inhibited despite high cytosolic cysteine concentrations

Feedback inhibition of cytosolic SERAT1.1 by cysteine is thought to represent a mechanism to regulate cysteine synthesis in A. thaliana (Noji et al. 1998). The high cytosolic cysteine concentration revealed in this study by far exceeds the Ki value of SERAT1.1 determined in vitro by Noji et al. (1998), consequently implying complete inactivity of this enzyme. However, that SAT activity as well as OAS and cysteine levels were significantly reduced in the csat mutant indicates that feedback inhibition of SERAT1.1 is not complete at the conditions applied in this study. In Glycine max, feedback sensitivity of SAT is abolished after phosphorylation (Liu et al. 2006). Yet, the phosphorylation site regulating feedback sensitivity in G. max is not conserved in A. thaliana. Although we cannot exclude post-translational modifications to regulate the feedback sensitivity of Arabidopsis SERAT1.1, we suggest that the cysteine sensitivity of SAT may be regulated by its assembly into the CSC. Here we demonstrate that cellular metabolite concentrations promote assembly of the CSC and propose that SAT present in the CSC is rendered insensitive to feedback inhibition. Consistent with this theory, increased thiol levels were obtained by heterologous expression of cysteine-sensitive SAT from Escherichia coli and A. thaliana in different plant species (Blaszczyk, Brodzik & Sirko 1999; Harms et al. 2000; Wirtz & Hell 2003; Sirko et al. 2004).

OAS and thiols are presumably efficiently transported between the compartments

In agreement with recent observations of Haas et al. (2008), Heeg et al. (2008) and Watanabe et al. (2008a,b) we have shown above that mitochondria appear to have a large impact on the cellular OAS biosynthesis, whereas cysteine is mainly synthesized in the cytosol. The finding that cytosolic cysteine biosynthesis apparently still efficiently takes place in mutants for cytosolic and plastidic SAT isoforms suggests that OAS is exported from the mitochondria into the cytosol. Cytosolic cysteine synthesis further requires import of cysteine into plastids, where γ-glutamylcysteine synthesis exclusively takes place (Wachter et al. 2005; Hothorn et al. 2006; Pasternak et al. 2008). Despite missing direct evidence, transport of OAS and cysteine between the compartments has been assumed in previous studies to explain the lack of phenotypic consequences of compartment-specific knockout of OASTL enzymes (Riemenschneider et al. 2005; Heeg et al. 2008; Watanabe et al. 2008a). The finding that γ-glutamylcysteine is predominantly synthesized in plastids (Wachter et al. 2005; Pasternak et al. 2008), but that cysteine concentration is low in this compartment indicates rapid incorporation of cysteine into γ-glutamylcysteine. For efficient glutathione synthesis in the cytosol (Wachter et al. 2005; Pasternak et al. 2008), γ-glutamylcysteine has to be transported there. Our finding that the γ-glutamylcysteine concentration was higher in the cytosol compared with plastids supports the hypothesis that γ-glutamylcysteine is efficiently transported against the concentration gradient. In agreement with results obtained by Meyer et al. (2001), we determined the cytosolic glutathione concentration to be approximately 3 mm. As the major part of glutathione is presumably synthesized in the cytosol (Wachter et al. 2005), the high plastidic glutathione concentration measured here indicates glutathione import into chloroplasts. Supporting this, glutathione transport studies with isolated wheat chloroplasts have led to evidence for two glutathione transport systems (Noctor et al. 2002). Thus, although glutathione biosynthesis will take place in plastids to some extent, glutathione is likely to be imported into plastids from the cytosol.

To summarize our findings, we present here a model describing the impact of the different subcellular compartments on cysteine biosynthesis in A. thaliana (Fig. 7).

Figure 7.

A model for cysteine biosynthesis in Arabidopsis. The model describes the contribution of the different subcellular compartments on cysteine biosynthesis. O-acetylserine (OAS) is mainly synthesized in mitochondria by SERAT2.2 and in part in the cytosol by SERAT1.1 (1). Only very little OAS is made in plastids. OAS is present bound to O-acetylserine-(thiol)-lyase (OASTL) (2), with the main OASTL pool being localized in the cytosol, where cysteine biosynthesis primarily takes place. Free OAS is capable to induce expression of genes involved in sulphate uptake and reduction. OAS-loaded OASTL traps sulphide for cysteine synthesis. Sulphide gets into the cytoplasm via transporter-mediated export from the chloroplasts (3). Cysteine for γ-glutamylcysteine synthesis is derived in part from the cytoplasm (4,5). γ-Glutamylcysteine is transported from the chloroplasts into the cytosol for glutathione synthesis (6,7). In addition to glutathione synthesis in the chloroplasts (7), glutathione might be imported into the chloroplasts from the cytosol (8). cOASTL, pOASTL, mOASTL, cytosolic, plastidic and mitochondrial O-acetylserine-(thiol)-lyase (OASTL) isoforms, respectively; γ-EC, γ-glutamylcysteine; GSH, glutathione.


We are grateful to Dr Dirk Buessis [Max-Planck-Institute of Molecular Plant Physiology(MPI-MP), Golm Germany] for help during reprogramming the Bestfit programme (Riens et al. 1991) for larger datasets. Further, we would like to thank Dr Stanislav Kopriva [John Innes Centre (JIC Norwich, UK)] and Dr Alisdair Fernie (MPI-MP Golm, Germany) for providing APR and cytochrome c oxidase antibodies, respectively. This work was supported by a grant of the German-Israelian Program (DIP), and by grants of the Ministerio de Educación y Ciencia (BIO2004-00784) and Junta de Andalucía (CVI-273), Spain.