The genome of Arabidopsis thaliana reveals that in this species the enzymes of glutathione biosynthesis, GSH1 and GSH2, are encoded by single genes. In silico analysis predicts proteins with putative plastidic transit peptides (TP) for both genes, but this has not been experimentally verified. Here we report a detailed analysis of the 5′ends of GSH1 and GSH2 mRNAs and demonstrate the subcellular targeting of the proteins encoded by different transcript types. GSH1 transcript analysis revealed two mRNA populations with short and long 5′-UTRs, respectively, both including the entire TP sequence. The ratio of long/total GSH1 transcripts was subject to developmental regulation. Transient transformation experiments with reporter gene fusions, bearing long or short 5′-UTRs, indicated an exclusive targeting of GSH1 to the plastids. Corroborating these results, endogenous and ectopically expressed GSH1 proteins were always present as a single polypeptide species with the size expected for correctly processed GSH1. Finally, the plastidic GSH1 localization was confirmed by immunocytochemistry. Similar to GSH1, multiple transcript populations were found for GSH2. However, here the prevalent shorter transcripts lacked a complete TP sequence. As expected, the large (but less abundant) transcript encoded a plastidic GSH2 protein, whereas GSH2 synthesized from the shorter transcript was targeted to the cytosol. The implications of the results for the compartmentation and regulation of GSH synthesis are discussed.
The biosynthesis of GSH occurs in two sequential ATP-dependent steps, catalyzed by γ-glutamylcysteine synthetase (GSH1; EC 126.96.36.199.) and glutathione synthetase (GSH2; EC 188.8.131.52.). Plant GSH1 and GSH2 cDNAs have been cloned and functionally expressed (May and Leaver, 1994; Ullmann et al., 1996; Wang and Oliver, 1996). Several studies have reported an upregulation of the expression of GSH1 and (to a lesser extent of) GSH2 genes during different stress regimes (Schäfer et al., 1998; Xiang and Oliver, 1998). Interestingly, in Arabidopsis thaliana suspension-cultured cells the transcriptional upregulation of GSH1 in response to stress treatments was not observed, whereas GSH1 activity was clearly induced, pointing to post-transcriptional regulation (May et al., 1998b). The underlying mechanism of this induction could be either a translational activation of GSH1 expression (Xiang and Bertrand, 2000), or a redox-based post-translational activation of the enzyme (Jez et al., 2004). Previous work on GSH1 and GSH2 expression has mainly focused on transcript amounts and/or enzyme activities. By contrast, little is known about GSH1 and GSH2 protein amounts and on the subcellular compartmentation of these enzymes (Noctor et al., 1998a,b).
As the need for cellular defense against ROS is not confined to a single compartment, GSH synthesis and intracellular transport have to meet the requirements of different organelles. Furthermore, GSH may perform stress-related metabolic functions in the cytosol and the chloroplast (see above). Thus, the question arises as to how the different subcellular compartments cover their specific demands for GSH, that is by full (or partial) synthesis or import via specific carriers. Earlier biochemical analysis of enzyme activity distribution had indicated that the enzymes for GSH biosynthesis may occur in the plastids and in the cytosol (Hell and Bergmann, 1988, 1990; Ruegsegger and Brunold, 1993). Recently, Moran et al. (2000) demonstrated for nodules of Glycine max, a tissue specialized for symbiotic N2-fixation, that GSH1 activity was exclusively confined to plastids and bacteroids, whereas for Vigna unguiculata significant GSH1 activity was also detected in cytosolic fractions. The same authors reported a predominantly cytosolic localization of GSH2 (and hGSH2) in nodules and leaf tissue (Moran et al., 2000). For Zea mays, Gomez et al. (2004) noted the absence of plastidic TPs in their isolated GSH1 and GSH2 cDNA clones. However, a database search revealed the presence of maize GSH1 and GSH2 transcripts (from the same genes) with putative TP sequences (accession numbers BM661181 and AJ579383).
Sequencing of the A. thaliana genome has revealed that in this species GSH1 and GSH2 are present as single genes (May and Leaver, 1994; The Arabidopsis Genome Initiative, 2000; Ullmann et al., 1996). Based on the A. thaliana GSH1 and GSH2 genomic sequences, both proteins are expected to include classical TPs for plastid localization, but the subcellular localization of the encoded GSH1 and GSH2 proteins has not yet been experimentally verified. In the present study, we have performed a detailed analysis of the 5′-ends of GSH1 and GSH2 transcripts in A. thaliana and their role for subcellular targeting of GSH1 and GSH2 proteins. Our results provide strong evidence for an exclusive plastidic localization of GSH1, whereas the major part of GSH2 transcripts encodes a cytosolic protein. Similar results were obtained with the allotetraploid species Brassica juncea. The implications for GSH synthesis and intracellular transport of GSH or its precursor EC will be discussed.
GSH1 mRNAs: 5′-RACE and real-time PCR analysis reveal transcript heterogeneity and quantitative changes of transcript populations during plant development
The sequence of the AtGSH1 gene from A. thaliana predicts two putative TATA-boxes in the 5′-upstream region, an intron of 359 bp interrupting the 5′-UTR and a classical TP for plastidic localization. To experimentally verify the predicted differential transcription initiation, the 5′-regions of AtGSH1 transcripts were analyzed by sequencing 5′RACE amplification products. Two distinct populations of AtGSH1 transcripts were observed, their 5′UTRs having a length of 171 bp (5L, long; transcription start TS1) and 73 bp (5S, short; transcription start TS2) bases, respectively (Figure 1a,b). The start sites TS1 and TS2 are located 27 and 30 bp downstream of the respective TATA box, indicating that both TATA boxes are functional. In 5L and 5S transcripts, the intron localized in the 5′UTR was always correctly spliced out, and both transcripts included the complete predicted TP sequence. Initial observations from RACE analysis indicated that in roots 5S transcripts appeared to be more abundant, whereas leaves showed higher amounts of 5L transcripts (Figure 1a). To address the developmental regulation of AtGSH1 transcript pools, the amounts of long (5L) and total (5L + 5S) mRNAs were quantitatively determined by real-time PCR in different tissues (Figure 1c). Whereas the amounts of total AtGSH1 transcripts were rather similar in sink leaf, source leaf, seedling, root and suspension-cultured cells (being somewhat lower only in flowers), the relative abundance of long transcripts, as reflected in the ratio 5L/5L + 5S, varied greatly and pointed to a pronounced developmental regulation. The underlying regulatory mechanism – differential transcription rates and/or different transcript stabilities – remains to be investigated. The possible role of the different 5′UTRs for post-transcriptional regulation of AtGSH1 expression will be discussed below.
To evaluate whether the observed transcript heterogeneity is a unique feature of A. thaliana, we have extended our GSH1 transcript analysis to another species of the Brassicaceae family, B. juncea. For this allotetraploid species, three different GSH1 cDNAs were cloned (BjGSH1-1/2/3) and their 5′ends were analyzed by 5′RACE. For BjGSH1-1 and BjGSH1-3 partial, genomic clones were isolated, covering the N-terminal coding regions and extending up to about 400 bp into the promoter regions. In its 5′-region, the gene BjGSH1-1 shows the same organization as AtGSH1 (Figure 1b), with two predicted TATA boxes, an intron in the 5′UTR and a classical TP. As for AtGSH1, two transcript populations with different 5′UTR length (5L, 94 bp; 5S, 45 bp) were observed. Likewise, for BjGSH1-2 two transcript populations were found indicating a similar genomic structure (not shown). Only for BjGSH1-3 did 5′RACE analysis reveal a single major transcript population with a 5′UTR length of 95 bases, which in the genomic sequence is also interrupted by an intron; again, the 5′UTR is followed by a plastidic TP-encoding sequence (prediction by ChloroP; Emanuelsson et al., 1999, 2000).
Transcript analysis was based on the procedure of ‘RNA ligase mediated rapid amplification of cDNA ends’ (RLM-RACE; GeneRacerTM Kit; Invitrogen, Karlsruhe, Germany). This powerful technique eliminates truncated mRNAs, ensuring the exclusive amplification from full-length transcripts via the ligation of a synthetic RNA adapter to the 5′ end of cap-selected, full-length mRNA (Schaefer, 1995). Our transcript analysis of four different GSH1 genes from A. thaliana and B. juncea showed highly reproducible results in several independent reactions. Furthermore, the different transcript populations were confirmed by sequencing several independent clones. To investigate the possibility of differential splicing of GSH1 genes, additional RT-PCR reactions were performed using a further downstream reverse primer at position +358 to +381 (about 200 bp downstream of the 3′-end of the predicted transit peptide), together with sense primers located at the 5′-ends of long (5L, primer position −171 to −146) or short (5S, primer position −72 to −48) transcripts (see above). Again, the reactions did exclusively yield the amplification products predicted from the 5′RACE analysis (data not shown). Thus, alternative splicing of GSH1 transcripts leading to mRNAs without the TP-encoding sequence can be excluded.
Transformation of A. thaliana epidermal leaf and suspension-cultured cells with GSH1(TP)::RFP fusions: the predicted transit peptides target GSH1 exclusively to plastids
To determine the intracellular targeting of GSH1 protein, several reporter constructs were cloned for transient or stable transformation of A. thaliana, using particle bombardment and Agrobacterium tumefaciens, respectively, for gene delivery (Figure 2). First, the BjGSH1-1 full length ORF (or the BjGSH1-1-TP-encoding sequence) was fused with RFP to experimentally verify the predicted plastidic targeting by comparison with control constructs for plastids (RBCS-TP::GFP), mitochondria [serine hydroxymethyltransferase(SHMT)-TP::GFP] and peroxisomes (GFP::SKL; PTS1). Note that both BjGSH1-1 fusion constructs started at their 5′end with the predicted TP ATG start codon, thus lacking the 5′UTR sequence (for analysis of 5′UTR function see below). Transformed cells were analyzed by epifluorescence microscopy or CLSM. The results of these targeting experiments are presented in Figure 3. Both the BjGSH1-1 full-length ORF (Figure 3a) and the BjGSH1-1-TP (Figure 3c,d) delivered the C-terminally fused reporter protein RFP exclusively to plastids as shown by the RFP fluorescence in comparison with the GFP fluorescence of the plastidic control construct (Figure 3b). Furthermore, the results of a co-bombardment of BjGSH1-1-TP::RFP with the marker for mitochondria (Figure 3c) excluded this organelle as possible target for BjGSH1-1. Likewise, a peroxisomal targeting can be excluded for the following reasons: (i) the co-bombardment experiment with the peroxisomal marker (Figure 3d) provided no evidence for a functional (N-terminally located) PTS2, (ii) the only PTS2-like sequence motif near the N-termini of BjGSH1-1 and AtGSH1 (KRSKRGHQL) is compromised by the arginine residue in position 2 which inactivates a PTS2 sequence (Flynn et al. (1998) and literature cited therein), and (iii) the GSH1 sequences bear no classical PTS1 sequence at the C-terminus.
To corroborate the plastidic targeting in differentiated cells, the BjGSH1-1-TP::RFP construct was delivered to leaf epidermal cells of A. thaliana (Figure 3e,f) via particle bombardment. Again, RFP fluorescence was exclusively detected in plastids, as demonstrated by comparison with chlorophyll autofluorescence. The same result was obtained with stable A. thaliana transformants expressing a BjGSH1-1::RFP fusion (Figure 3g,h). In summary, the results confirmed that the TP of BjGSH1-1 targets the mature protein exclusively to plastids with no indication of targeting to any other cellular compartment. Identical results were observed for AtGSH1 (see below).
Different length of GSH1 transcript 5′UTRs does not affect plastidic targeting of GSH1 protein
For AtGSH1, BjGSH1-1 and BjGSH1-2, two transcript populations with different 5′UTR length (5L,5S) were observed (see Figure 1b). Therefore, the possible influence of the 5′UTR on translation initiation at the predicted TP start codon was addressed. AtGSH1TP5L and AtGSH1TP5S sequences, encompassing the long or short 5′UTR with the entire TP sequence, were fused with RFP and GFP, respectively (see Figure 2). Co-bombardment of AtGSH1TP5S::GFP and AtGSH1TP5L::RFP revealed identical targeting to plastids (Figure 4a,b). This was independently confirmed by co-bombardment of AtGSH1TP5L::RFP with the plastidic control construct (Figure 4c,d). Therefore, under these experimental conditions the length of the GSH1-5′UTR does not affect translation initiation at the predicted ATG start codon.
Due to the cloning strategies, transcripts resulting from the AtGSH1TP5S::GFP and AtGSH1TP5L::RFP constructs may contain additional 5′UTR sequence (up to 41 bp from the 35S promoter and 6/19 bp from multiple cloning site; prediction of transcription start site according to Odell et al., 1985), which potentially might affect the secondary structure of the 5′UTR. In addition, it cannot be excluded that the expression of the above constructs under the control of the 35S promoter could titrate out a potential 5′UTR-binding factor. However, in a large number of independent transient expression experiments with widely differing levels of reporter protein expression we never observed a cytosolic targeting. Independent confirmation of the exclusive presence of a single plastid-localized GSH1 protein was obtained by immunological analysis (see below).
Evidence for exclusive plastidic localization of endogenous GSH1 protein
To independently verify the exclusive plastidic localization of endogenous GSH1, several experimental approaches were undertaken: (i) immunocytochemical localization of GSH1 protein, (ii) immunoblot analysis of different subcellular fractions, and (iii) analysis of ectopically expressed GSH1 protein provided with a C-terminal his-tag.
In B. juncea leaf sections, strong and specific immunosignals were confined to the chloroplasts (Figure 5). A faint background staining was observed in the cytosol of mesophyll and guard cells, however, comparison with the control indicates that this background signal is non-specific. Identical results were obtained with leaf sections from Cd-treated plants (data not shown). Thus, the results from immunocytological analysis corroborated the plastidic localization of GSH1 protein in mesophyll and guard cells.
Immunoblot analysis of subcellular fractions. To exclude the possibility that, as a consequence of dual targeting (Chew et al., 2003), GSH1 protein could also be imported into mitochondria, purified chloroplasts and mitochondria from B. juncea leaves were analyzed by immunoblot for the presence of GSH1 protein and the mitochondrial marker alternative oxidase. As expected from the reporter gene experiments, GSH1 protein was detected in the chloroplast but not in the mitochondrial fraction (Figure 6, top panel). Note that in whole cell extracts and purified chloroplast of B. juncea, the GSH1 signal sometimes appeared as a doublet of almost identical size (difference <1 kDa), most likely due to the presence of at least three GSH1 genes (see above).
Quantitative analysis of the distribution of GSH1 protein in plastids versus cytosol by comparing different fractions after differential centrifugation gave ambiguous results for the following reasons. During the extraction procedure, a substantial amount of plastids is broken, releasing plastid proteins. Furthermore, the relative degree of protein loss may vary for different proteins due to their different sizes and/or plastid-internal binding and/or compartmentation. As expected for an N-terminally processed plastidic AtGSH1 protein, a single GSH1 immunosignal of 51 kDa was detected in whole cell extracts and purified intact chloroplasts (data not shown), corresponding exactly to the theoretical size after removal of the TP sequence (50.9 kDa; prediction of TP according to Target P (Emanuelsson et al., 1999, 2000). Note that a translational start at the second or third ATG codon would have yielded GSH1 proteins of 54.7 and 52.9 kDa, respectively, which would have been clearly distinguished as different proteins.
Analysis of ectopically expressed GSH1 protein. As the possible presence of a minor amount of cytosolically localized, unprocessed GSH1 protein was difficult to exclude (see above), we decided to express a full-length BjGSH1-1-His6 construct in A. thaliana. The ectopically expressed BjGSH1-1 protein could be easily recovered and highly enriched. Total protein extracts and the affinity-purified BjGSH1-1-His6 protein were analyzed by SDS-PAGE (Figure 6, bottom panel). A single protein species was detected by immunoblot analysis, the size of which exactly matched the predicted Mr for the BjGSH1-1-His6 protein after removal of the transit peptide. From the immunoblot detection limit we could infer that the amount of N-terminally unprocessed BjGSH1-1-His6 protein was definitely below 1%.
Increased expression of GSH1 protein in response to Cd exposure does not yield additional GSH1 polypeptides
During Cd exposure, the formation of PCn catalyzed by phytochelatin synthase leads to a high demand for GSH in the cytosol and to the induction of GSH1 expression (Haag-Kerwer et al., 1999; Heiss et al., 2003; Schäfer et al., 1998). Therefore, it was of interest to probe for possible additional GSH1 polypeptides in Cd-treated plants. Hydroponically grown 6-week-old B. juncea plants were exposed to 25 μm Cd2+ for up to 144 h and the expression of GSH1 protein in leaves and roots was determined by SDS-PAGE immunoblot analysis (Figure 7). As expected from previous work (Schäfer et al., 1998), the amount of GSH1 protein increased after about 72 h in leaves and after less than 48 h in roots. However, a single GSH1 band was observed throughout the Cd exposure period with no indication of additional GSH1 proteins. Thus, the committing step for GSH synthesis remains confined to the plastids even in the presence of a high demand for GSH in the cytosol. A preliminary analysis by real-time PCR revealed that in response to Cd treatment the ratio of long to total AtGSH1 transcripts showed a dynamic response, the implications of which are yet unknown (data not shown).
Transcript heterogeneity for AtGSH2: prevalent short transcripts encode a cytosolic GSH2 protein, whereas only a minor amount of long transcripts encodes a plastidic GSH2 protein
Similar to AtGSH1 and BjGSH1 genes, the analysis of the AtGSH2 gene predicted an N-terminal TP, most likely representing a plastidic targeting sequence (Figure 8, upper left panel). The only putative TATA box was located about 80 bp upstream of the 5′end of the longest AtGSH2 transcripts, making a functional role rather unlikely. Transcript analysis of AtGSH2 delivered a complex picture. Again, 5′RACE revealed considerable transcript heterogeneity, but, in marked contrast to AtGSH1, most transcripts did not include the complete TP sequence (Figure 8, upper left panel). Only the longest transcripts (transcript 1) encompassed the entire TP, whereas transcripts 2–5 lacked varying parts of it. Quantitative analysis of A. thaliana seedlings by real-time PCR revealed that only 8.6 ± 0.7% of total transcripts contained the entire TP sequence. Therefore, it was expected that, except for transcript 1, the major part of AtGSH2 mRNAs would be translated, starting from the second ATG, into a cytosolic GSH2 protein. It is noteworthy that the 5′ends of GSH2 transcripts 3–5 determined by the 5′-RACE procedure exactly match with the 5′ends of previously published A. thaliana GSH2 cDNA clones (Figure 8, upper right panel).
To verify the functionality of the predicted TP present in the long AtGSH2 transcripts, in vivo targeting experiments were performed with AtGSH2TPS::GFP (transcript 5) and AtGSH2TPL::RFP (transcript 1) fusions. Both constructs were co-bombarded into A. thaliana suspension-cultured cells (Figure 8a,b). Alternatively, the AtGSH2TPL::RFP fusion was co-bombarded with a plastidic marker (Figure 8c,d). As expected, the AtGSH2TPS::GFP fusion protein showed cytosolic localization (Figure 8a), whereas the AtGSH2TPL::RFP fusion protein was strictly confined to the plastidic compartment (Figure 8b,c). The latter observation confirms that in vivo the shorter AtGSH2 transcripts are the result of differential transcription initiation in the 5′upstream context of the AtGSH2 gene (RLM-RACE yields products from capped transcripts only; see above), rather than representing processing products of the long transcripts. Otherwise, the transient transformation with the AtGSH2TPL::RFP construct should have yielded plastidic and cytosolic fusion proteins. In addition to the reporter gene fusions described above, the same experiments performed with gene constructs that encoded the corresponding full-length GSH2 proteins yielded identical results (data not shown).
The genome sequence of A. thaliana indicated that the enzymes of GSH synthesis, GSH1 and GSH2, are both encoded by single genes, and TP have been predicted for both proteins, suggesting a plastidic (and/or mitochondrial) targeting. However, many studies have shown that such predictions may be prone to errors. Moreover, differential transcription initiation and/or splicing may affect the N-terminal structure of the mature protein. As conflicting results have been reported for the subcellular localization of GSH1 and GSH2 in different plant species, we have readdressed this topic in the model plant A. thaliana. In this report, we demonstrate the exclusive plastidic localization of GSH1 protein in leaf cells and heterotrophic suspension-cultured cells from A. thaliana and B. juncea. The cumulative evidence from (i) transcript analysis (see Figure 1), (ii) in vivo targeting studies with different GSH1-reporter gene fusion proteins (see Figures 3 and 4), (iii) immunocytochemical localization of GSH1 protein in the chloroplasts (Figure 5), (iv) immunological detection of GSH1 protein in purified organelles (Figure 6), and (v) analysis of ectopically expressed His-tagged GSH1 protein (Figure 6), demonstrates that in the investigated species GSH1 is entirely a plastidic protein. In contrast, the second enzyme of GSH synthesis, GSH2, appears to be largely confined to the cytosol, but a low amount of long GSH2 transcripts may assure some GSH2 delivery to the plastids. This unexpected result has a direct bearing on the regulation of GSH-requiring processes in other compartments. Considering the absence of a substantial GSH efflux from chloroplasts (see below), our data strongly suggest that plastid-derived γ-glutamylcysteine (EC) is the precursor of GSH in other compartments.
Based on observations from in vivo labeling of GSH with monochlorobimane, Meyer and Fricker (2002) postulated that in A. thaliana suspension-cultured cells the artificial depletion of the cytosolic GSH pool does not cause a substantial GSH efflux from the chloroplast. The authors assumed that reduced sulfur exits the chloroplast as HS−, cysteine or EC, or a combination of these compounds. The observed absence of significant GSH efflux from chloroplasts to the cytosol was corroborated by a recent study on GSH distribution in poplar leaves (Hartmann et al., 2003). Based on these results, an exclusive plastidic localization of GSH1 would require an export of EC, especially in periods of high demand for GSH in the cytosol and other compartments. Preliminary data from a GSH transport study with wheat chloroplasts indicated that two uptake systems for GSH may exist in the plastid envelope, and the authors speculate that the high-affinity component may represent an active uptake component (Noctor et al., 2002). However, several questions remain as yet unresolved: (i) are these plastid envelope GSH transport systems in vivo unidirectional, (ii) are GSH and GSSG equally transported, and (iii) do these uptake mechanisms operate as uniporters or do they represent exchange reactions?
Based on our observations, and accounting for the absence of a substantial transport of plastidic GSH to the cytosol (see above), we postulate that the dipeptide formed from the primary S- and N-assimilation products, cysteine and glutamate, is transported from the plastids to the cytosol where it acts as the precursor for cytosolic GSH synthesis. It may be speculated that EC also becomes hydrolyzed in the cytosol into its amino acids, thereby providing a cysteine export route from the chloroplast. Note that the mechanism of cysteine export from chloroplasts still remains largely unknown.
GSH2 enzyme activities have been detected in plastids and in the cytosol (Hell and Bergmann, 1988), however, in most studies GSH2 activities were considerably higher in the cytosol compared with chloroplasts (Moran et al., 2000; Noctor et al., 2002). RML-RACE analysis of GSH2 transcripts and reporter gene studies have revealed that only a minor amount of mRNAs included a functional plastidic TP, whereas the prevalent shorter transcripts encoded a cytosolic protein (Figure 8). Likewise, two types of transcripts, encoding most probably plastidic and cytosolic proteins, were detected for GSH2 in the legume Lotus japonicus (Matamoros et al., 2003). It is noteworthy that in the recent analysis of the proteome of A. thaliana chloroplasts GSH1 but no GSH2 protein was detected (Kleffmann et al., 2004), whereas in the proteomes of mitochondria (Millar et al., 2001) and peroxisomes (Fukao et al., 2002) both proteins appeared to be absent. From these observations it follows that, except for plastids, cytosolic GSH has to be imported into the other GSH-containing compartments (Jiménez et al., 1997, 1998). Recently, the first plant GSH transporters have been cloned and functionally characterized (Bogs et al., 2003; Zhang et al., 2004), but their subcellular localization remains to be investigated.
Whether the different compartmentation of GSH1 and GSH2 in plants is related to the evolutionary origins of both genes is a matter of speculation. When compared with all other non-plant eukaryotic organisms, plant GSH1 sequences form a clearly separated clade, whereas plant GSH2 genes group with all other eukaryotic GSH2 sequences (Copley and Dhillon, 2002). These authors propose that plant GSH1 originally may have evolved in a cyanobacterium.
The experiments performed in our study have focused mainly on the localization of GSH1 (and GSH2) in non-stressed cells. Several previous studies have shown that expression of GSH1 is strongly induced in response to a number of stress treatments, including salt exposure (Mittova et al., 2003) and chilling (Gomez et al., 2004). However, the possible effects of these stress treatments on the compartmentation of GSH1 and GSH2 proteins have not yet been analyzed. Interestingly, in maize the amount of GSH1 transcripts and protein was exclusively induced in the bundle sheath cells but not in the mesophyll cells (Gomez et al., 2004), adding another spatial dimension to the dynamic control of GSH synthesis after stress exposure. While it cannot a priori be excluded that during periods of high GSH demand in the cytosol, additional unknown mechanisms may lead to the formation of cytosolic GSH1 protein, our analysis of GSH1 protein expression in response to Cd-treatment did not indicate the formation of additional GSH1 species (see Figure 7).
The occurrence of distinct GSH1 transcript pools with different 5′UTR length remains an interesting observation at this point (Figure 1), the functional significance of which is currently explored. The strong developmental modulation of the 5L/5L + 5S transcript ratio points to a possible involvement in the regulation of GSH1 expression. Furthermore, preliminary data revealed that the transcript ratio was also modulated in response to various stress treatments (Cd exposure, abscisic acid, H2O2; data not shown). As both transcripts start in canonical distance to corresponding TATA boxes, and as the two TATA boxes are separated by a stretch of 100 bp, it appears likely that two partially overlapping – and differentially regulated – promoters direct the differential transcription of both mRNAs.
A search for proteins binding to the 5′UTR of GSH1 transcripts, using an RNA-protein hybrid screen, has yielded a candidate protein factor involved in cellular redox control which is currently being characterized to define its possible role for translational control of GSH1 expression (A. Wachter, unpublished data). In a preliminary report, Xiang and Bertrand (2000) described the binding of a redox-sensitive protein component to the 5′UTR from GSH1 transcripts, however, no detailed experimental data were provided. May et al. (1998b) have previously presented evidence for post-transcriptional regulation of GSH1 expression in A. thaliana cells under stress exposure. While circumstantial evidence supports a stress-related post-transcriptional regulation of GSH1 expression, the underlying mechanisms remain largely unknown. As stated in the Introduction section, the post-transcriptional regulation may involve a translational control (Xiang and Bertrand, 2000) and/or a post-translational redox-mediated activation of the GSH1 enzyme (Jez et al., 2004).
We conclude that in cells of A. thaliana and B. juncea, GSH1 is exclusively localized in the plastids whereas GSH2, albeit also present in the chloroplasts, is, to a large extent, a cytosolic protein. Thus, the compartmentation of GSH synthesis functionally links the different cellular compartments and may provide a platform for intracellular redox signaling. Future work will address the possible role of the 5′-UTR for translational control of GSH1 expression and the mechanisms of intracellular transport of both EC and GSH.
Arabidopsis thaliana, ecotype Columbia, was grown under greenhouse conditions (approximately 8-h light period). For reporter gene studies, heterotrophic A. thaliana suspension-cultured cells were cultivated in the dark at 25°C in MS medium (Trezzini et al., 1993). Indian mustard (B. juncea [L.] Czern. Vittasso) was grown on sand as previously described (Schäfer et al., 1998). For Cd treatment, 5-week-old sand-grown plants were transferred to hydroponic culture and treated with 25 μm Cd(NO3)2 after 1 week adaptation according to Schäfer et al. (1998). Plant tissues for protein and RNA extraction were immediately frozen in liquid nitrogen and stored at −80°C.
5′-RACE analysis of GSH1 and GSH2 transcripts
The 5′-regions of transcripts were cloned by the RML-RACE technique using the GeneRacerTM Kit (Invitrogen) for amplification of full-length cDNAs. Starting from root or leaf total RNA, all steps were performed according to the manufacturer's protocol. Total RNA was isolated according to Logemann et al. (1987). For each GSH1/GSH2 gene, two specific primers were designed to perform nested PCR following the initial PCR reaction (for all primer sequences see Table 1). Gene specific primers for AtGSH1 were P1 and P2 (nested). To probe for further downstream located transcription start sites, additional primers, P3 and P4, were used; however, no evidence was obtained for transcripts other than those reported in the Results section. For B. juncea, two different cDNA clones, BjGSH1-1 (AJ563921) and BjGSH1-2 (AJ563922), were previously isolated from a B. juncea root cDNA library (Schäfer et al., 1998). A third cDNA clone, BjGSH1-3 (AJ563923), was amplified by RACE from B. juncea leaf cDNA with primer P5. Gene-specific primers for the amplification of 5′-regions of full-length BjGSH1 cDNAs were P6 and P7 for BjGSH1-1, P5 and P7 for BjGSH1-2, and P8 and P9 for BjGSH1-3. The 5′-region of AtGSH2 was amplified with primers P10 and P11. All amplification products were cloned into pGEM-T (Promega, Mannheim, Germany) and sequenced from both ends (SEQLAB, Göttingen, Germany), using vector-specific primers. DNA analysis was performed using Clone Manager 5.2 (Sci Ed Central, Cary, NC, USA) and GeneBank databases using the blast program (Altschul et al., 1997).
Table 1. Primer sequences (for further details see Experimental procedures). Note that primers used for cloning reporter gene constructs contained restriction sites (in bold face)
Isolation of the BjGSH1-1 5′-flanking genomic region by inverse PCR
To isolate the 5′-flanking region of the known BjGSH1-1 cDNA sequence, inverse PCR was performed as described by Triglia et al. (1988) with BclI-, DraI- and EcoRV-restricted genomic DNA as template. In total, 1241 bp of the genomic DNA upstream of the translational start of BjGSH1-1 was isolated from two consecutive inverse PCR runs with primers P12 and P13 for the first round, and P14 and P15 for the second round (AJ564376). Following the same strategy, a genomic clone with about 400 bp of promoter sequence was obtained for BjGSH1-3 (AJ564377).
Transcript quantitation by real-time PCR
Total RNA was extracted from various tissues of A. thaliana WT plants using the RNeasy Plant Mini Kit (Qiagen, Hilden, Germany), following the manufacturer's instructions. To eliminate residual genomic DNA present in the preparation, the samples were treated with RNase free DNaseI (Promega), and the RNA was subsequently bound to RNeasy Spin columns (Qiagen) for purification. After elution with RNase free water, 2 μg of RNA was transcribed into first strand cDNA using the Omniscript RT Kit from Qiagen with an oligo dT primer. Real-time PCR was performed using the Platinum Taq-DNA Polymerase (Invitrogen) and SYBR-Green as fluorescent reporter in the Biorad iCycler (Biorad, Munich, Germany). Primers for the coding region of AtGSH1 were P16 and P17. The 5′UTR of long GSH1 transcripts was amplified with primers P18 and P19. Primer sequences for actin (Act2/8) were reported previously (Ha et al., 1999). A serial dilution of cDNA derived from sink leaves was used as standard curve to calculate amplification efficiency for AtGSH1 and actin primers. Each reaction was performed in triplicates, and specificity of amplification products was confirmed by melting curve and gel electrophoresis analysis. Relative abundance of total AtGSH1 transcripts and long AtGSH1 transcripts was calculated and normalized with respect to Act2/8 mRNA according to the method of Muller et al. (2002). The results were expressed relative to sink leaf given the arbitrary value of 1 for both total and ratio long/total transcripts.
Plasmids for reporter gene studies
All constructs for transient transformation were cloned into the vector pFF19 (Timmermans et al., 1990) for expression under the control of an enhanced 35S promoter. For cloning of pFF19-GFP, EGFP cDNA was amplified from the plasmid pEGFP-N1 (Clontech, Palo Alto, CA, USA) using primers P20 and P21, and, after digestion with PstI and SphI, was cloned into pFF19. For the peroxisomal control construct pFF19-GFP::PTS1, EGFP was amplified from pEGFP-N1 with primers P20 and P22, thereby introducing a peroxisomal targeting sequence [PTS1 (SKL)] at the C-terminus of GFP, and inserted into PstI and SphI sites of pFF19. For the plastidic control construct pFF19-RBCS-TP::GFP, a fragment of 197 bp, containing the TP of the small subunit of RUBISCO from A. thaliana, was released from plasmid pBinAR218 (a gift from Rita Zrenner, MPIMP, Golm) by KpnI and SalI and cloned into KpnI/PstI sites of pFF19-GFP. pBinAR218 was generated by amplifying the TP-encoding sequence from A. thaliana cDNA with primers P23 and P24, the product being digested with Asp718 and SalI and ligated into appropriate sites of pBinAR (Höfgen and Willmitzer, 1992). For the mitochondrial construct, pFF19-SHMT-TP::GFP, the TP-encoding sequence of serinhydroxymethyl transferase, was amplified by PCR with primers P25 and P26, using plasmid pSHMT3.6.2 as template (a gift from Rüdiger Hell, HIP, Heidelberg) and, after digestion with SacI and XbaI, cloned into appropriate sites of pFF19GFP.
For cloning of pBjGSH1-1-TP::RFP, plasmid pBK-CMV[ECS1-1] (Schäfer et al., 1998) was first digested with AflII and PflMI, yielding an internal BjGSH1-1 1155 bp fragment (positions 189–1344 of BjGSH1-1), which was discarded, and the remaining plasmid fragment, which was used for ligation with the following two fragments: (i) an 81-bp fragment (positions 197–278 of BjGSH1-1) amplified with primers P27 and P28 from pBK-CMV[ECS1-1], and digested with AflII and AscI; (ii) the DsRED cDNA, which was amplified from the vector pDSRED (Clontech) with primers P29 and P30, and digested with AscI and PflMI. All three fragments were ligated resulting in a pBK-CMV-vector which contained 277 bp of 5′ region of BjGSH1-1, representing the TP in a translational fusion with DsRED. This sequence was amplified with primers P31 and P32, and, after restriction, was cloned into KpnI and PstI sites of pFF19, yielding the plasmid pBjGSH1-1-TP::RFP.
To clone pFF19RFP, the DsRED cDNA was amplified from vector pDSRED with primers P33 and P32, and, after digestion, inserted between the KpnI and PstI sites of pFF19. For cloning pBjGSH1-1::RFP, the full-length cDNA of Bj GSH1-1 (1542 bp) was amplified with primers P34 and P35 to introduce XhoI and KpnI sites and to mutate the stop codon. This fragment was cleaved with XhoI and KpnI and inserted into the appropriate sites of pFF19RFP, resulting in pBjGSH1-1::RFP. Note that both BjGSH1-1 constructs do not contain 5′UTR sequence of BjGSH1-1.
To investigate a possible role of the 5′UTR of AtGSH1 transcripts, the TP-encoding sequence of AtGSH1 with either short (5S) or long (5L) 5′UTR (transcript populations identified by RLM-RACE, see above) were fused to GFP or DsRED, respectively. Briefly, 5S plus TP-encoding sequence (301 bp) was amplified using primers P36 and P37, digested with SacI and XbaI, and cloned into pFF19GFP, whereas 5L plus TP-encoding sequence (399 bp) was amplified with primers P37 and P38, digested with XhoI and KpnI and cloned into pFF19RFP.
For the targeting analysis of proteins encoded by AtGSH2 transcripts, the sequence extending from the 5′-transcript end to the 3′-end of the predicted plastidic TP was cloned in front of a reporter gene. For the construct based on the long transcript, a 226-bp fragment of AtGSH2 was amplified with primers P39 and P40, digested with NruI and KpnI and cloned into appropriate sites of pFF19RFP. For the corresponding construct of the short transcripts, an 83-bp fragment of AtGSH2 was amplified with primers P40 and P41, digested with NruI and KpnI and cloned into vector pFF19GFP.
For stable transformation of A. thaliana with a BjGSH1-1::RFP fusion, the BjGSH1-1::RFP encoding sequence was released from plasmid pFF19BjGSH1-1::RFP by XhoI and cloned into the SalI site of pBinAR (Höfgen and Willmitzer, 1992).
Transient transformation of suspension-cultured cells and leaf epidermal cells
Arabidopsis thaliana suspension-cultured cells were filtrated onto sterile filters, which were then placed on petri dishes containing solid MS cell culture medium, supplemented with 22 g l−1 mannitol and 22 g l−1 sorbitol. Cells were incubated for 3 h before particle bombardment. The gene gun procedure has been previously described (Lehr et al., 1999). Each plate was bombarded two to three times with 0.75 μg of plasmid on wolfram particles. After bombardment, cells were incubated at 25°C in the dark for 24–48 h. For gene delivery to leaf tissue, 3-week-old A. thaliana seedlings grown on petri dishes were bombarded.
Reporter protein localization by epifluorescence and confocal laser scanning microscopy
For microscopical analysis, cell walls of transiently transformed suspension-cultured cells were partially digested by incubation in 0.1% MES buffer with 3.8% CaCl2, 9% mannitol, 2% cellulase, 0.1% driselase, and 1% pectinase for 6 h. Cells were embedded in glycerin on a slide and analyzed by epifluorescence microscopy with different filter sets (excitation 30-BP540 nm and emission 585 nm long pass for DSRED; excitation 450–490 nm and emission 515 nm longpass for GFP). Pictures were taken with a digital camera using analySIS (Soft Imaging System). Transformed suspension-cultures cells were also analyzed by confocal scanning microscopy (DMRBE; Leica, Bensheim, Germany) with the following filter settings: for DSRED excitation 568 nm and emission 590 nm long pass, for GFP excitation 488 nm and emission 515 nm longpass. For confocal analysis of transient transformed A. thaliana leaves, LSM410 (Zeiss, Jena, Germany) was used with the following filters: for DSRED excitation 543 nm and emission bandpass 575–640 nm, for chlorophyll autofluorescence excitation 488 nm and emission 645–700 nm. For confocal analysis of stable transformed A. thaliana plants, SP2AOBS (Leica) was used with the following filters: for DSRED excitation 543 nm and emission bandpass 562–611 nm, for chlorophyll autofluorescence excitation 488 nm and emission 630–716 nm. All images were edited with Adobe Photoshop 5.5.
Stable A. tumefaciens-mediated transformation of A. thaliana by floral dip
Arabidopsis thaliana plants were transformed by the floral dip method according to Clough and Bent (1998). After transformation, seeds were screened on solid MS medium containing 0.8% agar and 50 μg ml−1 kanamycin under sterile conditions. After 2 weeks, transformants were transferred to soil and screened by PCR with insert-specific primers (P32, P33) and by immunoblot analysis for expression of the transgene (BjGSH1-1::RFP fusion protein).
Expression of recombinant BjGSH1-1 protein in Escherichia coli and antiserum production
To express the full-length BjGSH1-1 protein in E. coli, the corresponding cDNA was amplified from the GSH1-1 cDNA clone with primers P44 and P45. The amplification product was digested with BamHI and PspAI and ligated into the corresponding sites of the pQE30 vector (Qiagen). The pQE-construct was cotransformed with the plasmid pUBS520 (containing the dnaY insert and the lacIq gene; Brinkmann et al., 1989) into DH5α cells. For induction, a 1:100 dilution in LB medium of an overnight culture was grown at 37°C to an OD600nm of 0.5–0.7 and then induced with 1 mm IPTG. After induction, bacteria were transferred to 30°C and incubated for an additional h. Cells were lysed using an 8 m urea buffer and the recombinant His-tagged protein was purified by Ni2+ affinity chromatography (Qiagen) according to the manufacturer's protocol. The purified protein (approximately 1.2 mg) was used to raise a polyclonal antiserum in rabbits (BioGenes, Berlin, Germany).
Ectopic expression of recombinant, his-tagged BjGSH1-1 protein in A. thaliana
For stable overexpression of His-tagged BjGSH1-1 in A. thaliana, the coding sequence of BjGSH1-1 was amplified with primers P42 and P43 to introduce restriction sites and a 6xHIS tag at the C-terminus of BjGSH1-1. The PCR product was digested with BamHI and SalI and cloned into appropriate sites of pBinAR. All constructs were confirmed by sequencing (SEQLAB). For the isolation of His-tagged BjGSH1-1 protein from transformant plants, 10 g of leaf tissue was extracted with lysis buffer and purified under native conditions by Ni2+ affinity chromatography (Qiagen) according to the manufacturer's protocol. Different fractions of the purification procedure were analyzed by immunoblot.
For total protein extraction, 100 mg of deep-frozen plant material was ground in liquid nitrogen, vortexed with 375 μl extraction buffer (100 mm HEPES pH 7.1, 250 mm sorbitol, 10 mm MgCl2, 10 mm KCl, 2 mm DTT and 1 mm PMSF), and centrifuged for 20 min at 15 000 g, and, subsequently, 30 min at 500 000 g, at 4°C. Protein concentration of the supernatant was determined using the Bradford (1976) assay.
Immunoblotting was performed using the semidry procedure and a 1:10 000 dilution of GSH1-1-antiserum. From leaf and root tissue, 20 and 10 μg total protein, respectively, were loaded on a discontinuous 9% SDS-polyacrylamide gel. Proteins were transferred to a PVDF membrane (Immobilon; Millipore, Billerica, MA, USA) at 3.5 mA cm−2 and 15 V for 45 min. After blocking with 5% low fat milk powder in TBST for 1 h, the membrane was incubated with the primary antiserum in a 1:10.000 dilution in 5% BSA (in TBS) at 4°C for 12 h. Immunoblots were developed with anti-rabbit IgG-horse radish peroxidase conjugate (Sigma) and subjected to enhanced chemiluminescence detection (Super Signal West Dura; Pierce, Rockford, IL, USA), according to the manufacturer's protocol.
Youngest leaves from 6 week-old B. juncea plants were cut into small pieces (5 × 7 mm), degassed and fixed in PBS containing 4% paraformaldehyde, pH 7, under vacuum for 5 h. After overnight incubation at 4°C, samples were washed twice with PBS for 1 h. Leaf samples were dehydrated in an ethanol series and after a xylol substitution series (XEM-200; Vogel, Giessen, Germany) embedded in paraffin (Paraplast Plus; Sigma-Aldrich, Seelze, Germany). For immunocytochemistry, 4 μm sections were prepared (microtom RM2125RT; Leica) and transferred to slides.
For immunodetection of GSH1 protein, the ExtrAvidin® Peroxidase Staining Kit No. EXTRA-3 (Sigma) was used according to the manual. After removal of paraffin and a rehydration step, slides were incubated for 5 min in PBS, 0.1% trypsin for antigen unmasking. The GSH1-antiserum was used in a dilution of 1:200. The biotinylated goat anti-rabbit IgG and ExtrAvidin-Peroxidase were both diluted 1:25. This kit was combined with NovaRED (Vektor, Burlingame, CA, USA) as substrate. For blocking, 2 μl of GSH1-antiserum was incubated with 10 μg of recombinant BjGSH1-1 protein in 100 μl PBS for 1 h at 37°C and for 2 h at 4°C. After incubation the blocking reaction was centrifuged for 10 min at 15 000 g, and the supernatant was used in a final dilution of 1:200. The same procedure was carried out for antiserum without GSH1 protein.
We gratefully acknowledge critical reading of this manuscript and valuable suggestions by Phil Mullineaux (Essex, UK). Access to CLSM equipment was kindly provided by the MPI Ladenburg. The plasmid p218 was a gift from Rita Zrenner (MPIMP, Golm). Furthermore, we acknowledge the help of Oliver Berkowitz (RSBS, Canberra) and Andreas Meyer (HIP, Heidelberg) for CLSM analysis of transient BjGSH1-1TP::RFP expression in A. thaliana leaf epidermis and BjGSH1-1::RFP expression in transgenic A. thaliana plants, respectively. We thank Marcus Koch (HIP, Heidelberg) for helpful discussions on the evolution of B. juncea. This work was supported by a DFG grant to TR (FOR 383).