Glutathione homoeostasis is critical to plant life and its adaptation to stress. The γ-glutamyl cycle of glutathione biosynthesis and degradation plays a pre-eminent role in glutathione homoeostasis. The genes encoding two enzymatic steps of glutathione degradation, the γ-glutamyl cyclotransferase (GGCT; acting on γ-glutamyl amino acids) and the Cys-Gly dipeptidase, have, however, lacked identification. We have investigated the family of GGCTs in Arabidopsis thaliana. We show through in vivo functional assays in yeast that all three members of the ChaC/GCG subfamily show significant activity towards glutathione but no detectable activity towards γ-glutamyl methionine. Biochemical characterization of the purified recombinant enzymes GGCT2;2 and GGCT2;3 further confirmed that they act specifically to degrade glutathione to yield 5-oxoproline and Cys-Gly peptide and show no significant activity towards γ-glutamyl cysteine. The Km for glutathione was 1.7 and 4.96 mM for GGCT2;2 and GGCT2;3 respectively and was physiologically relevant. Evaluation of representative members of other subfamilies indicates the absence of GGCTs from plants showing significant activity towards γ-glutamyl-amino acids as envisaged in the classical γ-glutamyl cycle. To identify the Cys-Gly peptidase, we evaluated leucine aminopeptidases (LAPs) as candidate enzymes. The cytosolic AtLAP1 (A. thaliana leucine aminopeptidase 1) and the putative chloroplastic AtLAP3 displayed activity towards Cys-Gly peptide through in vivo functional assays in yeast. Biochemical characterization of the in vitro purified hexameric AtLAP1 enzyme revealed a Km for Cys-Gly of 1.3 mM that was physiologically relevant and indicated that AtLAP1 represents a cytosolic Cys-Gly peptidase activity of A. thaliana. The studies provide new insights into the functioning of the γ-glutamyl cycle in plants.
Glutathione, γ-glutamyl cysteinyl glycine, the most abundant low-molecular-mass thiol in eukaryotic cells, plays many important functions in plants. The functions of glutathione in plants range from their requirement in growth and development to their involvement in the cellular response to environmental stresses [1,2].
Glutathione, as the principal redox buffer of the cell, protects the cellular components from oxidative damage and through its reactive thiol group, can also participate in thiol–disulfide exchange reactions that may be a key to linking the plant response to the redox state of the cell [1,3]. Glutathione is also involved in the transport of sulfur, and the detoxification of reactive oxygen species, xenobiotics (such as certain herbicides)  and heavy metals (such as cadmium and arsenic) . Low or impaired glutathione biosynthesis in plants leads to a defect in growth and development, whereas complete absence of glutathione from plants results in embryonic lethality [6,7]. Glutathione homoeostasis is therefore critical to plant life.
The γ-glutamyl cycle that describes the synthesis and degradation of glutathione plays a pre-eminent role in glutathione homoeostasis . The enzymatic activities corresponding to the different steps in this cycle have been assumed to be present in both mammals and plants. Glutathione degradation, according to this cycle, is initiated by γ-glutamyl transpeptidase (GGT), a membrane-bound extracellular enzyme in mammals (with the active site facing outside). In plants, there are three genes encoding GGTs that have been identified . GGT1 and GGT2  are localized on the apoplast for extracellular glutathione degradation, whereas GGT4 is localized on the vacuolar membrane facing the vacuolar lumen and can degrade glutathione and its conjugates in the vacuole . GGTs have been, for a long time, the only enzymes known to degrade glutathione in plants, but they act on non-cytosolic pools.
The enzymatic activity of GGT on glutathione has two possible reaction mechanisms. In the first pathway that has been clearly demonstrated in mammals and therefore thought to exist in plants, GGT hydrolyses glutathione along with an acceptor amino acid to yield the products γ-glutamyl amino acid and the Cys-Gly dipeptide. γ-Glutamyl amino acids are then cyclized by glutathione cyclotransferase to form 5-oxoproline, whereas the Cys-Gly peptide is cleaved to cysteine and glycine through a Cys-Gly peptidase. In the second pathway, GGT hydrolyses glutathione in the absence of an acceptor amino acid to yield glutamate and Cys-Gly peptide. The genes for both the γ-glutamyl cyclotransferase (GGCT) acting on γ-glutamyl amino acids and the Cys-Gly peptidase have not been identified in plants .
We have previously described two new pathways of cytosolic glutathione degradation that function independently of the classical GGT-dependent pathway . The first pathway, the Dug pathway, is fungal specific and in this pathway, glutathione degradation is initiated by the Dug2/Dug3 oligomer [12,13]. The second pathway involves the mouse ChaC1 protein homologues . These are GGCTs that are specific for glutathione. This pathway is also cytosolic and is conserved from bacteria to humans. Three homologues of the ChaC1 family exist in A. thaliana and recently one of the homologues, GGCT2;1 (γ-Glutamyl cyclotransferase ChaC2 like proteins), has been identified as being up-regulated during arsenic stress and was found to function in the degradation of glutathione and to a much lesser extent of the γ-glutamyl amino acid γ-glutamyl alanine . Both the deletion and the overexpression in plants led to the same phenotype of arsenic resistance. Whereas the deletion phenotype can be attributed to increased glutathione levels, the overexpression phenotype has been suggested to be a consequence of glutamate cycling. The roles of the other two ChaC homologues were, however, not investigated.
An intermediate in glutathione degradation, and common to all the pathways of glutathione degradation, is the dipeptide Cys-Gly. The Cys-Gly dipeptide is cleaved by a Cys-Gly peptidase to yield the constituent amino acids. Peptidases with Cys-Gly activity have been described in mammals, yeast and bacteria [13,16–18] and the genes for the peptidases have been identified. In plants, Cys-Gly peptidases have also been biochemically demonstrated, but the genes encoding this activity or enzyme have, surprisingly, not yet been identified [11,19,20]. Among the cytosolic Cys-Gly peptidases of other organisms, Dug1p (M20 metallopeptidase family) homologues are present in mammals and yeast but are absent from plants . The M17 metallopeptidases, the leucine aminopeptidases (LAPs) which have been also shown to display significant Cys-Gly peptidase activity in mammals and pathogenic bacteria , in contrast, have homologues in plants. A. thaliana has been shown to contain LAPs that display activity towards a chromogenic substrate, leucine-p-nitroanilide (Leu-pNA); however, the activity towards other substrates has not been evaluated. The only study that has examined LAP activity towards different substrates was with the tomato LAP, LAP-A , where it was found that the enzyme showed very little activity towards Cys-Gly prompting the authors to argue against any role for these peptidases in glutathione or Cys-Gly degradation in plants, unlike the role proposed for these LAPs in mammals [23,24].
In the present study, we have sought to re-investigate these pathways and to identify candidate enzymes that might participate in these steps of glutathione degradation in the plant A. thaliana. We show through in vivo complementation studies of yeast mutants that all three ChaC homologues in A. thaliana (GGCT2;1, GGCT2;2 and GGCT2;3) function specifically in glutathione degradation and are thus functional GGCTs of the ChaC/GCG subfamily. Specific GGCTs for γ-glutamyl amino acids seemed to be absent from plants. We also demonstrate that AtLAP1 (A. thaliana leucine aminopeptidase 1), a known cytosolic LAP of A. thaliana can function as a CysGly peptidase by complementation of yeast mutants. These studies were followed up by in vitro characterization of the recombinant enzymes. The kinetic parameters of both the GGCTs and AtLAP1 reveal the physiological relevance in glutathione degradation in plants and together these studies succeed in delineating the pathway for glutathione degradation in plants.
Chemicals and reagents
All chemicals used in the present study were either of analytical or molecular biology grades and were obtained from commercial sources. Media components were purchased from Difco. Oligonucleotides were purchased from Sigma and IDT. Restriction enzymes, Vent DNA polymerase and other DNA-modifying enzymes were obtained from New England Biolabs. Gel extraction kits, plasmid miniprep columns and the Ni-NTA (Ni2+-nitrilotriacetate)–agarose resin were obtained from Qiagen. Ninhydrin reagents, glutamate, cysteine, glycine, glutathione, γ-Glu-Cys, γ-Glu-Ala, Cys-Gly dipeptide and Leu-pNA were procured from Sigma. γ-Glu-Met was from Bachem. The peptides Leu-Cys, Cys-Leu, Asp-Cys, Lys-Cys and Arg-Cys were synthesized on a PS-3 peptide synthesizer (Protein Technologies Inc.) at the Institute of Microbial Technology, Chandigarh, India. Masses of the purified peptides were confirmed by MALDI–TOF. Protein molecular mass markers were purchased from MBI Fermantas.
Strains, media and growth conditions
The Escherichia coli strain DH5α was used as a cloning host and E. coli strains BL21 and BL21 pLys2(DE3) were used as expression hosts. The Saccharomyces cerevisiae strain ABC1724 (MATα his3Δ1 leu2Δ0 lys2Δ0 met15Δ0 ura3Δ0 ecm38Δ::KanMX4 dug1-634) and ABC1723 (MATα his3Δ1 leu2Δ0 lys2Δ0 met15Δ0 ura3Δ0 ecm38Δ::KanMX4 dug3) were used for the genetic complementation assays to evaluate the functionality of the AtLAP and ChaC genes respectively. The S. cerevisiae strains were maintained on YPD [1% (w/v) yeast extract/2% (w/v) peptone/2% (w/v) glucose] medium. SD (synthetic defined) minimal medium contained yeast nitrogen base, ammonium sulfate and dextrose and was supplemented with methionine, histidine, leucine and lysine at concentrations of 80 mg/l. Glutathione and Cys-Gly dipeptide were used at concentrations of 300 and 100 μM respectively. Yeast transformation was carried out by the lithium acetate method .
The complete list of 14 predicted GGCTs containing the GGCT BtrG domain were extracted from the TAIR (the arabidopsis information resource) database. Additionally, we carried out BLAST analysis using representatives of the four different subfamilies of the GGCTs against the A. thaliana database. Reverse BLAST was further performed to investigate the possible orthologues. To identify the possible CysGly peptidase in A. thaliana, LAP protein sequence from Treponema denticola was used to search the A. thaliana database using BLAST. Multiple sequence alignment of the protein sequences was generated using ClustalW (http://www.ebi.ac.uk/Tools/msa/clustalw2/) using default parameters.
To establish a phylogenetic relationship among the A. thaliana GGCT fold proteins and other known members of GGCT proteins, we needed to ensure that a proper multiple sequence alignment of this diverse family was being fed into the phylogenetic analysis software. We first selected the four members of this family for which structures are known and carried out a structure alignment using the DeepAlign online tool [26,27]. We observed that the catalytic residues are conserved in the structural alignment. We then selected diverse homologues from each group and carried out a multiple sequence alignment of these protein sequences using T-Coffee . We validated the multiple sequence alignment by comparing it with the structural alignment conservation patterns. We found that catalytic residues also aligned well in the multiple sequence alignment output of T-Coffee. We then fed this multiple sequence alignment file into MEGA 6 software  to construct a Neighbour Joining (NJ) tree with 1000 bootstrapped data sets.
Cloning, expression and purification of Arabidopsis thaliana ChaC and LAP proteins
The open reading frames (ORFs) for the A. thaliana GGCT2;1, GGCT2;2, GGCT2;3, AT3G02910 and AIG2L (avirulence-induced gene 2-like protein) were obtained from TAIR. The ORFs were PCR-amplified and cloned into the yeast expression vector p416TEF at the BamHI and XhoI sites downstream of the translation elongation factor (TEF) promoter. For the purification of proteins, GGCT2;1 and GGCT2;3 were cloned in the pET23a expression vector at the NdeI and XhoI sites, whereas GGCT2;2 was cloned in the pET23d vector at NcoI and BamHI sites. All primers are listed in Table 1.
The A. thaliana LAP ORFs AT4G24200 (AtLAP1), AT4G30910 (AtLAP2) and AT4G30920 (AtLAP3) were obtained from TAIR. The AtLAP1, AtLAP2 and AtLAP3 genes were PCR amplified with primer sets corresponding to respective genes (Table 1) and subsequently cloned into the yeast expression vector p416TEF. The AtLAP2 and AtLAP3 genes were also expressed without their predicted N-terminal targeting signal by PCR amplification with EcoRIfw and XhoIrev primer sets. For expression and purification of the proteins, AtLAP1 was PCR- amplified with the appropriate primer set and cloned into the pET23a expression vector in the NdeI and XhoI restriction sites. N-terminal truncated AtLAP2 was haemagglutinin (HA)-tagged by using EcoRIFw and HA-XhoIrev primer and cloned in the pTEF416 expression vector.
The E. coli BL21 pLyS2(DE3) and BL21 host were used for expression and purification of C-terminal histidine-tagged proteins GGCT2;1, GGCT2;2, GGCT2;3 and AtLAP1 respectively. For in vitro purification, the cultures were grown up to A600=0.6 and were induced with 0.5 mM IPTG and further grown at 18°C for 18 h. Cells were harvested after 18 h by centrifugation at 2500 g for 5 min at 4°C following which they were lysed by sonication in the presence of 50 mM Tris/HCl (pH 8.0), 300 mM NaCl and 1 mM PMSF for 10 min with a 10 s on and 15 s off cycle. The lysates were centrifuged at 15000 g for 30 min at 4°C. The GGCT2;1 protein was solubilized from inclusion bodies using 6 M guanidium chloride and purified from Ni-NTA under denaturation conditions of 6 M guanidium chloride. The protein was refolded by dialysing against refolding buffer [50 mM Tris/HCl (pH 8.0), 300 mM NaCl and 0.5 M arginine]. For GGCT2;2 and GGCT2;3, the lysates were directly loaded on to an equilibrated Ni-NTA column and subsequently washed with wash buffer [50 mM Tris/HCl (pH 8.0), 300 mM NaCl, 1 mM PMSF and 20 mM imidazole]. Elution was performed in the presence of elution buffer carrying 250 mM imidazole and the samples were analysed on SDS/PAGE (10% and 14% gels). Proteins were dialysed using 50 mM Tris/HCl and 300 mM NaCl (pH 8.0) buffer at 4°C. Protein quantification was performed using Bradford reagent (Sigma) according to the manufacturer's instructions. BSA was used as a standard.
In vivo functional complementation assays in S. cerevisiae
The yeast strain ABC1723 or ABC1724 were transformed with control or test plasmids. Transformants were grown in minimal medium containing methionine and other amino acid supplements. The primary overnight culture was used to re-inoculate the secondary culture and incubated until A600=0.6. Equal numbers of cells were harvested, washed with water and suspended in sterile water to A600=0.2. Serial dilutions were prepared and 10 μl was spotted on to SD medium with or without glutathione, γ-Glu-Met or Cys-Gly dipeptide.
Gel filtration analysis of GGCT2; 2, GGCT2; 3 and AtLAP1 proteins
Gel filtration was carried out for further purification and to determine the oligomeric nature of the proteins. The active fractions eluted from the Ni-NTA column were loaded on to a size-exclusion gel filtration column Superdex 200 10/300 GL and gel filtration was performed using 50 mM Tris/HCl and 300 mM NaCl buffer (pH 8.0). Molecular mass determinations were made after calibration.
To study the GGCT activity of the proteins GGCT2;2 and GGCT2;3, 2.5 μg of purified recombinant proteins was incubated with 5 mM glutathione, in a 100 μl of reaction mixture that also contains 50 mM Tris/HCl buffer (pH 8.0). After 30 min of incubation at 37°C, the reaction was terminated by heating at 95°C for 5 min. Samples were centrifuged for 30 min to remove inactivated protein; 20 μl of reaction mixture was injected in C18 HPLC column (150 mm×4.6 mm). Acetonitrile (5%) in 2% perchloric acid was used as the mobile phase with a 1 ml/min flow rate at 25°C. Peaks of glutathione, Cys-Gly and 5-oxoproline were detected at 210 nm. Internal standards are used to confirm peaks.
To confirm the formation of 5-oxoproline in the reaction product we used LC–MS. The same reaction mixture as used for HPLC was used for LC–MS. The LC–MS was done for all of the reaction and product standards, i.e. glutathione, 5-oxoproline, Cys-Gly, cysteine, glycine and glutamate (Sigma–Aldrich). The reaction products of GGCT were analysed using LC–MS (TOF/QTOF mass spectrophotometer, Agilent Technologies G6550A) at the Institute of Microbial Technology. The LC–MS was carried out in ESI mode. The system was equipped with precolumn and reverse phase C18 column (3.5 μm). Acetonitrile (100%) gradient was run at a flow rate of 0.4 ml/min. The products were analysed using an intact mass detector. The mass range and scan rate were set to record m/z from 50 to 700/s.
Assay of γ-glutamyl cyclotransferase activity on glutathione, γ-Glu-Cys using the Dug1p-coupled assay
GGCT activity towards glutathione or γ-Glu-Cys as substrates was carried out using Dug1p-coupled assay . This assay is based on the fact that degradation of glutathione by ChaC protein will release Cys-Gly and 5-oxoproline and has been described earlier . The Cys-Gly so released will be further acted on by the Cys-Gly dipeptidase Dug1p. The cleavage of Cys-Gly will lead to formation of cysteine and glycine. In the case of γ-Glu-Cys, the enzyme would directly release cysteine. The cysteine so formed will be detected by the sensitive acidic ninhydrin detection method . In this assay, 1 μg of GGCT2;2, 5 μg of GGCT2;3 and 10 μg of GGCT2;1 were used. The free cysteine generated was measured by the acidic ninhydrin method . For kinetic experiments, concentrations of glutathione ranging from 0.2 to 15 mM were incubated with GGCT2;2 and GGCT2;3 in 50 μl of reaction mixture and reactions were performed as described above. The above method can detect as little as 50 nmol of cysteine. Kinetic parameters were calculated using Michaelis–Menten equation kinetics in GraphPad Prism 5 software.
Assay of γ-glutamyl cyclotransferase activity on γ-Glu-Ala and glutathione using the 5-oxoprolinase-coupled assay
Activity against γ-Glu-Ala (and also glutathione as a control) was assayed using the 5-oxoprolinase-coupled assay essentially as described earlier  using 5 mM substrate and 2.5 μg of the proteins. This assay was based on the indirect measurement of 5-oxoproline released during the reaction by converting it into glutamate with the action of yeast recombinant 5-oxoprolinase. The final glutamate measurements were done using an Amplex Red kit. The activities were calculated using glutamate standard curve. The histograms were created using GraphPad Prism 5 software.
Cys-Gly peptidase activity and dipeptide activity assay for AtLAP1 protein
The Cys-Gly peptidase activity assay as well as the dipeptidase activity assay against other cysteine-containing dipeptides (Cys-Leu, Leu-Cys, Gly-Cys, Arg-Cys, Lys-Cys and Asp-Cys) was based on the estimation of released cysteine from the substrate by the action of enzyme . Cysteine estimation was modified from the acid ninhydrin method developed by Gaitonde . The standard reaction mixture in total reaction volume of 150 μl contained 100 ng of purified enzyme, 100 μM MnCl2, 10 mM Tris/HCl (pH 8.0), 50 mM NaCl and substrate (Cys-Gly) was added in the range 0.05–8 mM. The reaction was initiated by addition of protein and incubated at 37°C for 10–30 min. The reaction was terminated by addition of 5% of trichloroacetic acid (TCA). Each sample was centrifuged at 12500 g for 2 min and the supernatant was used to estimate the cysteine released in the reaction. To 125 μl of supernatant, 125 μl of acetic acid and 125 μl of acidic ninhydrin reagent (125 mg of ninhydrin reagent was dissolved in 3 ml of acetic acid and 2 ml of concentrated HCl) were added and samples were boiled in a water bath for 10 min. To 300 μl of samples, 1000 μl of absolute ethanol was added and absorbance was taken at 560 nm. The control reaction was also run against each concentration of substrate and subtracted. A standard graph for cysteine was plotted and used for the calculation of free cysteine in the enzyme reaction. The relative activity of AtLAP1 protein was determined with hexameric fraction of protein. Purified protein (100 ng) was incubated with 2 mM Cys-Gly in the presence of either Zn2+ or Mn2+ (250 μM) at 37°C for 30 min and released cysteine was measured by ninhydrin as described above. The relative activity of AtLAP1 for different dipeptides was assayed with 100 ng of protein and 2 mM substrate in presence of 100 μM MnCl2 at 37°C for 30 min and released cysteine was measured by ninhydrin as described above. The activities against all dipeptides were calculated using a cysteine standard curve and were plotted using GraphPad Prism 5 software.
Protein electrophoresis and Western blotting
To study the heterologus expression of AtLAP2 in S. cerevisiae, the N-terminal truncated AtLAP2 was HA-tagged at the C-terminus and cloned in a pTEF416 shuttle vector. The protein was expressed in yeast and total crude cell extracts were prepared as described previously . Equal amount of protein samples (20 μg) were resolved by SDS/PAGE (12% gel) and electroblotted on to Hybond ECL nitrocellulose membrane (GE Healthcare) in a mini trans-blot apparatus (Bio-Rad Laboratories) at 110 V for 1 h using Tris-glycine transfer buffer at 4°C. After blocking the membrane for 1 h at room temperature in 5% dried skimmed milk in tris-buffered saline and tween 20 (TBST), it was probed with mouse monoclonal anti-HA primary antibody (Cell Signaling Technology) at a dilution of 1:3000 in TBST for 4 h at room temperature. After four 10 min washing steps of the blot with TBST, the membrane blot was incubated for 1 h in horse anti-mouse (horseradish peroxidase-conjugated) antibody at a dilution of 1:2500 in TBST at room temperature. The signal was detected with an ECL Plus Western detection kit as per the manufacturer's instruction.
Leu-pNA peptidase activity assay for AtLAP1 protein
For metal-dependency studies, purified hexameric AtLAP1 protein was incubated with 2 mM Leu-pNA substrate in the presence of either Zn2+ or Mn2+ at 37°C for 30 min and measurement was carried out at 405 nm. The activity in the presence of Mn2+ was considered as 100% and the relative activity of Zn2+ was calculated.
The γ-glutamyl cyclotransferases family of Arabidopsis thaliana consists of 14 proteins that fall into three distinct subfamilies, but lack any members that fall into the GGCT subfamily
To identify whether a gene encoding GGCT that acts on γ-glutamyl amino acids exists in plants, we first examined the TAIR database for GGCTs. A total of 14 GGCTs from the Arabidopsis database were identified with the GGCT BtrG domain. The known subfamilies include GGCT that acts on γ-glutamyl amino acids, GGACT that acts on γ-gamma glutamyl amine amino acids, Butirosin biosynthesis protein (BtrG) that acts on γ-glutamyl-butirosin and ChaC/GCG that acts on glutathione. Sequences of known members of these subfamilies were retrieved and were used as query sequences for BLAST analysis against the Arabidopsis protein database. Three members fell into the ChaC/GCG (γ- glutamyl cyclotransferase acting on glutathione) family, two members belonged to the GGACT family and no members fell into either the GGCT family or the BtrG family, whereas nine members fell into a new group of GGCTs that seem to have members only in plants, but do not as yet have an identified substrate. We labelled this new family GGCT-PS (plant-specific). The different members of each subfamily are similar to each other; however, across subfamilies all of the members are characterized by a GGCT BtrG fold and thus have structural but not sequence similarity. In order to establish phylogenetic relationships among these A. thaliana GGCT fold proteins and other known members of the GGCT family, we selected the diverse members of each subfamily known and built a phylogenetic NJ-based tree using MEGA 6 software with 1000 bootstrapped data sets as described in the Experimental section. The phylogenetic tree confirmed five branches that were supported by bootstrap percentages (Figure 1). From this analysis it is apparent that A. thaliana lacks any member in GGCT and BtrG subfamily (Table 2).
NJ tree of GGCT fold proteins
|Similarity (among subfamily) (%)||71-69||57-65||Nil||Nil||88-66|
|Similarity (among subfamily) (%)||71-69||57-65||Nil||Nil||88-66|
In the absence of any member in the GGCT family that acts on γ-glutamyl amino acids, we initially examined a member of the undefined family for activity against γ-glu-tamyl amino acids. We cloned AIG2L, a member of GGCT-PS subfamily whose crystal structure is already known  in the yeast expression vector p416TEF but did not observe any growth on γ-Glu-Met (Supplementary Figure S1). We have also examined a member of the GGACT subfamily AT3G02910 for the ability to grow on γ-Glu-Met in the yeast assay but, again, found no growth suggesting no activity of this enzyme towards γ-Glu-Met unlike GGCT (Supplementary Figure S1).
GGCT2;1, GGCT2;2 and GGCT2;3, the A. thaliana homologues of the mouse ChaC1/yeast GCG1 protein, function in GSH degradation
The mouse ChaC1 protein and its yeast homologue GCG1 (ORF YER163c) are GGCTs that specifically act on glutathione and can initiate cytosolic glutathione degradation to yield 5-oxoproline and Cys-Gly. A. thaliana has three homologues, At5G26220 (GGCT2;1), At4G31290 (GGCT2;2) and At1G44790 (GGCT2;3), that encode proteins of 216, 227 and 199 amino acids respectively. These proteins show more than 50% similarity to the previously characterized mouse ChaC1 proteins and when aligned with ChaC1 through multiple sequence alignments revealed conserved motifs and presence of catalytic glutamate residues (results not shown). Sequence comparisons revealed that all three ChaC proteins of A. thaliana proteins showed greater similarity to mouse ChaC2 than to mouse ChaC1. Further analysis (using reverse BLAST analysis) confirmed their greater similarity to ChaC2 and thus we have named the second and third ORFs GGCT2;2 and GGCT2;3 to indicate this greater similarity to ChaC2.
GGCT2;1 was shown to be a cytosolic enzyme using a GFP-tagged protein . However, the localization of GGCT2;2 and GGCT2;3 is not known. Using localization prediction tools we attempted to predict localization of GGCT2;2 and GGCT2;3. Both are likely to be cytosolic as they lack any signal peptide and give no significant score with transmembrane prediction tools (results not shown). All three ORFs were cloned into a yeast expression vector p416TEF. These constructs were then transformed into a S. cerevisiae met15∆ ecm38∆ dug3∆ strain. This strain is an organic sulfur auxotroph due to met15Δ and is deficient in glutathione utilization due to a both ecm38Δ (which encodes γ-GGT) and dug3Δ (an essential component of the alternative DUG pathway) . This strain background was used for carrying out in vivo assays to evaluate the glutathione-degrading ability of the encoded proteins. When the clones were transformed into this strain and glutathione used as the sole source of sulfur we observed that cells expressing these proteins could grow on glutathione in a manner similar to yeast expressing mouse ChaC1 (Figure 2A). We were also interested to identify whether these ChaC-like proteins of A. thaliana have any activity towards γ-glutamyl amino acids. We first examined the activity towards γ-Glu-Met. It was observed that the mouse GGCT (C7orf24) which is a known GGCT shows activity towards γ-Glu-Met (used as positive control) but neither GGACT (which is specific for γ-glutamyl-amine) nor GGCT2;1, GGCT2;2 or GGCT2;3 displayed any activity towards γ-Glu-Met (Figure 2B).
In vivo growth assay of A. thaliana GGCT2 activity towards glutathione in S. cerevisiae
In vitro purification and characterization of recombinant GGCT2;1, GGCT2;2 and GGCT2;3 from E. coli reveal them to be γ-glutamyl cyclotransferases specifically acting on glutathione
GGCT2;1 was recently purified and shown to be able to act on glutathione as well as γ-Glu-Ala with a preferential activity towards glutathione . In vivo studies with GGCT2;2 and GGCT2;3 showed significant activity towards glutathione but not to γ-glutamyl-containing dipeptides. To evaluate more rigorously the ability of GGCT2;2 and GGCT2;3 to act on glutathione and γ-glutamyl amino acids, these proteins were cloned with a histidine-tag in an E. coli expression vector and the proteins were purified from E. coli using Ni-NTA affinity chromatography as described in the Experimental section. Eluants from the Ni–NTA chromatography were further run through Superdex 200GL size-exclusion chromatography columns. The proteins obtained were more than 95% pure as seen on Coomassie Blue-stained SDS/PAGE and were of the expected size, i.e. 26 and 22 kDa (Supplementary Figure S2A). GGCT2;2 has a tendency to aggregate since a significant fraction of the protein also showed a higher-molecular-mass peak at the void volume along with the monomeric peak. GGCT2;3, however, gives a predominantly single peak at 22 kDa that corresponds to the monomeric protein and a small peak at void volume (Supplementary Figures S2B and S2C).
To determine the kinetic parameters of the activity of GGCT2;2 and GGCT2;3 towards glutathione we used the previously described Dug1p-coupled assay for ChaC family proteins . HPLC studies confirmed that the products of the reaction are 5-oxproline and Cys-Gly (Supplementary Figure S3). The formation of 5-oxoproline was further confirmed by LC–MS (Supplementary Figure S4) as described in the Experimental section. The monomeric fractions of GGCT2;2 and GGCT2;3 were assayed as described in the Experimental section. Both GGCT2;2 and GGCT2;3 showed Michaelis–Menten kinetics towards glutathione (Figures 3A and 3B respectively; Table 3). With GGCT2;2, we did observe some inhibition at high substrate (GSH) concentrations although in sufficient inhibition to fit the substrate inhibition equation (results not shown). Neither GGCT2;1, GGCT2;2 nor GGCT2;3 showed any significant activity towards γ-Glu-Cys (Figure 4A). GGCT2;1 has been shown to display some activity towards γ-Glu-Ala  and we therefore evaluated GGCT2;2 and GGCT2;3 activity towards the same substrate using the 5-oxoprolinase-coupled assay (see the Experimental section). Although we could detect activity against γ-Glu-Ala, it was significantly less than that seen against glutathione (Figure 4B).
Michaelis–Menten plot of GGCT2;2 and GGCT2;3 for glutathione
Comparative in vitro activity of GGCT2;1, GGCT2;2 and GGCT2;3 towards GSH and γ-Glu-Cys
|Enzyme||Substrate||kcat (min−1)||Km (mM)||kcat/Km (min−1·mM−1)|
|Enzyme||Substrate||kcat (min−1)||Km (mM)||kcat/Km (min−1·mM−1)|
AtLAP1, the cytosolic leucine aminopeptidase of Arabidopsis thaliana can complement the yeast dug1 mutant that is deficient in Cys-Gly peptidase
As homologues of yeast Dug1p (or mammalian CNDP2, cytosolic nonspecific dipeptidase) Cys-Gly peptidases are absent from A. thaliana and all other plants, we focused our attention on members of the LAP M17 family. A. thaliana has three LAPs, AtLAP1 (AT2g24200), AtLAP2 (AT4g30920) and AtLAP3 (AT4g30910), with 80–90% sequence identity. AtLAP1 and AtLAP2 peptidases have been shown to have activity against Leu-pNA but have not been examined for Cys-Gly peptidase activity. AtLAP1 has been suggested to be cytosolic and this is also supported by complete cytosolic proteome analysis [33,34], whereas AtLAP2 and AtLAP3 have N-terminal targeting sequences that suggest that they are organellar. Using prediction tools (SUBA3) both AtLAP2 and AtlAP3 are predicted to be chloroplast-localized, and complete chloroplast proteome analysis has confirmed the presence of AtLAP2 at least in the chloroplast . The AtLAP1, AtLAP2 and AtLAP3 genes were cloned in the yeast expression vector p416TEF. The initial assay for functionality was carried out by in vivo complementation of the S. cerevisiae strain ABC1724 (MATα his3Δ1 leu2Δ0 lys2Δ0 met15Δ0 ura3Δ0 ecm38Δ::KanMX4 dug1-634). This strain is an organic sulfur auxotroph but fails to grow on glutathione as it also carries a deletion in the S. cerevisiae Cys-Gly peptidase (Dug1) gene. The Dug1 peptidase is required for the utilization of Cys-Gly dipeptide (released from glutathione). In order to grow, it is essential to release cysteine from the dipeptide. If the different AtLAPs have Cys-Gly dipeptidase activity they would complement the dug1-deficient mutant by releasing the cysteine from Cys-Gly and permitting growth on medium lacking sulfur. We transformed the AtLAP1, AtLAP2 and AtLAP3 (cloned downstream of the yeast TEF promoter in p416TEF) and observed complementation of the dug1 mutants on glutathione medium by both AtLAP1 and AtLAP3 suggesting that AtLAP1 and AtLAP3 could function in vivo as an efficient Cys-Gly peptidase (Figure 5A). The inability of AtLAP2 to complement the dug1 defect could be a result of the extra N-terminal targeting sequence that might cause the protein to misfold. We therefore expressed both AtLAP2 and AtLAP3 without the N-terminal targeting sequences. It was observed that AtLAP2Δ1–55 that lacked the N-terminal targeting sequence was not able to complement the dug1 defect (even though the expression of the HA-tagged protein could be detected in yeast extracts; results not shown). In contrast, AtLAP3Δ1–56 without the N-terminal plastid targeting sequence was complementing the defect. The growth of these transformants was also checked on Cys-Gly dipeptide as a substrate and in the present study too we had observed that AtLAP1 as well as AtLAP3 or AtLAP3Δ1–56 (with or without the targeting sequence) was complementing the dug1 defect. Thus AtLAP1 and AtLAP3 are peptidases with activity towards Cys-Gly dipeptide (Figure 5B).
Functional complementation assay of AtLAP1, AtLAP2, AtLAP3, AtLAP2Δ1-55 and AtLAP3Δ1-56 in S. cerevisiae dug1-deficient mutants for growth on glutathione and Cys-Gly
In vitro purification and kinetic characterization of AtLAP1 reveals that it has significant activity towards Cys-Gly dipeptide
Owing to our interest in the cytosolic pathways, we decided to focus on the AtLAP1 protein. AtLAP1 has been previously purified from the plant cytosol and shown to be a hexameric protein that can act on the model substrate LeupNA [33,36]. However, the activity against natural substrates such as the Cys-Gly dipeptide was never evaluated. In order to understand the biochemical behaviour of AtLAP1 protein towards Cys-Gly, we purified the recombinant histidine-tagged protein from E. coli. The protein showed a single band of approximately 54 kDa on SDS/PAGE analysis (Supplementary Figure S5A). After Ni–NTA chromatography we carried out a gel filtration on Superdex 200 10/300 GL and found that, as reported, the AtLAP1 protein exists predominantly as a hexameric form although some small monomeric fractions were also observed (Supplementary Figures S5B and S5C). We examined both fractions for activity; however, the monomeric fraction showed very little activity against the model substrate Leu-pNA as well as Cys-Gly. As this is a metallopeptidase, its activity was checked towards both Cys-Gly and Leu-pNA in the presence of Zn2+ and Mn2+. We observed significantly higher activity of AtLAP1 towards both substrates in the presence of Mn2+ as compared with Zn2+. AtLAP1 showed approximately 25% activity for Cys-Gly in the presence of Zn2+ as compared with Mn2+ (results not shown) and about 4% activity for Leu-pNA in the presence of Zn2+ as compared with Mn2+ (results not shown). To evaluate the substrate specificity of the enzyme, we assayed the activity against several different dipeptides that included Arg-Cys, Lys-Cys, Asp-Cys, Leu-Cys and Cys-Leu. (As the assay was based on cysteine release, the peptides had one of the residues as cysteine.) Highest activity was observed for Cys-Gly (27.02 μmol/min/mg), the activity towards Leu-Cys and Cys-Leu was 19.4 and 20.3 μmol/min/mg, whereas negligible activity (less than 1.5 μmol/min/mg) was observed for Arg-Cys, Lys-Cys or Asp-Cys (Figure 6). The kinetic parameters were also determined towards Cys-Gly, Cys-Leu and Leu-Cys as described in the Experimental section using Mn2+ as the metal (Figure 7; Table 3). The activity towards Cys-Gly and comparison with other substrates suggested that the enzyme could function as a Cys-Gly peptidase. Furthermore, as glutathione concentrations in the cell are known to range from 1 to 10 mM, the activity towards Cys-Gly seemed physiologically relevant.
AtLAP1 activity towards different dipeptides
Michaelis–Menten plot of AtLAP1 for Cys-Gly, Cys-Leu and Leu-Cys substrates
The studies described in the present report throw light on several different aspects of glutathione degradation in plants leading to new insights into the γ-glutamyl cycle that is critical for understanding glutathione homoeostasis in plants.
First, the studies with the three ChaC/GCG members of A. thaliana reveal a class of enzymes that act preferentially on glutathione to degrade it to 5-oxoproline and Cys-Gly dipeptide. The members show a similarity of about 70% between them, but all of them behaved similarly in showing high specificity towards glutathione and with an affinity that is physiologically relevant. With the description of this family of three new enzymes involved in glutathione degradation, together with the GGTs, plants have a total of six known enzymes involved in glutathione degradation. Although the GGTs clearly have roles in glutathione degradation outside the cytosol (in the vacuole or the apoplast) the ChaC family appears to be cytosolic. The reasons for this multiplicity of enzymes in the cytosolic degradation of glutathione is unclear but seems to underline a larger role for glutathione in nutritional homoeostasis since glutathione upon degradation can provide three key metabolites for biosynthetic pathways: glutamate, cysteine and glycine. Although the regulation of GGCT2;1 is known to some extent be induced by arsenite  and with sulfur stress , very little is known about the regulation of the other GGCTs acting on glutathione.
Studies with the yeast S. cerevisiae have revealed that the half-life of glutathione under conditions of sulfur starvation is about 30 min. More surprisingly, however, was the observation that even under standard growth conditions (unstressed) there was still significant turnover with a half-life of glutathione of about 90 min . The reasons for the turnover of glutathione even under unstressed conditions are still not clear and such continuous turnover is yet to be established in other organisms. Whereas the kinetics of the plant ChaC proteins have revealed that the Km for glutathione is physiologically relevant, the catalytic efficiency of these proteins was lower than that of the mammalian homologue expressed under ER stress conditions , but similar to the yeast enzyme . It is possible therefore that in plants these enzymes may be involved in the turnover of glutathione both under stress conditions (GGCT2;1), but possibly also in the slow, but continuous, turnover of cytosolic glutathione under unstressed conditions.
Secondly, the investigations on the GGCTs reveals that plants, like yeasts, do not carry any significant activity towards γ-glutamyl amino acids and are thus essentially lacking in the classical GGCT. GGCT2;1, GGCT2;2 and GGCT2;3 all display some limited activity towards γ-Glu-Ala; however, the predominant activity was towards glutathione. Furthermore, we could not find any significant activity towards γ-Glu-Met by all three proteins through our in vivo assay or towards γ-Glu-Cys in the in vitro assays. We have also investigated members of a novel family of GGCTs that we refer to as GGCT-PS. This is a family of nine members that shows no sequence similarity to other families. However cloning and expressing a representative member from this family has not indicated any activity towards either glutathione or γ-Glu-Met. The absence of any significant GGCT activity towards γ-glutamyl amino acids similar to the mammalian GGCT encoded by C7Orf24  seems to indicate that the generation of γ-Glu-amino acids in plant cells might be minimal. This has been observed by Ohkama-Ohtsu et al. , and in the study by these workers, glutathione was proposed to be the major contributor to 5-oxoproline formation . Degradation by the ChaC family yields 5-oxoproline and Cys-Gly directly, whereas degradation by the GGT family is thought to yield either glutamate (if only hydrolysis is involved) or γ-glutamyl amino acid (if transpeptidation with amino acid is involved) and being located at the membrane these have a role in amino acid transport. In mammals, it is thought that amino acid transport occurs with the latter reaction. In plants, it seems likely that the principal product is glutamate and not γ-glutamyl amino acid. This has also been suggested in mammals [40,41], but the absence of γ-glutamyl amino acid-dependent cyclotransferases seems to lead greater credence to this hypothesis at least in plants and yeasts where this activity is minimally present. The role of the γ-glutamyl cycle in the movement of glutathione therefore needs to be viewed differently in plants with the findings that have emerged from these studies (Figure 8).
Schematic representation of the γ-glutamyl cycle in plants in the absence of significant activity towards γ-glutamyl amino acids
Thirdly, we have been able to identify the enzymes responsible for Cys-Gly peptidase activity in plants. The Cys-Gly peptidases are the other enzymes of the γ-glutamyl cycle that have been known in plants for many years, but have surprisingly remained unidentified. Although the AtLAP1 peptidase has been previously well characterized as a cytosolic peptidase, its ability to cleave Cys-Gly peptidase was never examined. LAP-A of tomato has been the only LAP of plants that has been extensively characterized in terms of substrate specificity. The LAP-A enzyme of tomato was found to have only minimal activity towards Cys-Gly. The detailed kinetics towards Cys-Gly, however, was never investigated, but tomato LAP-A had the highest affinity for Arg-Leu and Leu-Leu among the dipeptides and the highest kcat towards Leu-Leu. The Km of 1.4 mM towards Arg-Leu is quite close to what we have observed as the Km of AtLAP1 for Cys-Gly . AtLAP1, however, did not show any significant activity towards either Arg-Cys or Lys-Cys (analogues of Arg-Leu) and also showed maximal activity towards Cys-Gly indicating that the two enzymes are quite distinct. The relative activities towards the six dipeptides further suggest that the natural substrate of AtLAP1 could well be Cys-Gly in vivo. When one compares the affinity of the Cys-Gly peptidases of other organisms towards Cys-Gly, the mammalian CNDP2 (Km=0.6 mM) and yeast Dug1p (Km=0.4 mM) Cys-Gly peptidases which belong to a different family of M20 metallopeptidases , as well as the LAP of T. denticola (Km=0.48 mM)  is reasonably comparable to what has been observed with the AtLAP1 (Km=1.3 mM) and is within the range that is likely to be physiologically significant given the high levels of glutathione in living systems. Cys-Gly, in addition to being an intermediate in glutathione degradation, also has important redox properties . It can contribute to the redox environment of the cell and excessive accumulation has been shown to be toxic in yeast cells, possibly owing to its pro-oxidant nature [44,45]. Its cleavage to the constituent amino acids is therefore critical to the cell, first for removing the metabolite from the cytoplasm and secondly for releasing the constituent amino acids for use by the cell. The Cys-Gly released by the cytosolic GGCTs are likely to be substrates for the AtLAP1 protein which can function as a Cys-Gly peptidase. But what about the other Cys-Gly formed outside the cytosol and in other compartments generated by the GGT enzymes? One possibility is the presence of still uncharacterized membrane-associated peptidases found in these environments . A second possibility is that these peptides are transported by the di- and tri-peptide transporters into the cytosol. A. thaliana has multiple peptide transporters that might be participating in their transport . Among the two putative plastid LAPs, AtLAP3 has activity against Cys-Gly as well and could also play some role in these pathways. But AtLAP2 (and its N-terminal deletion mutant) did not show any activity against Cys-Gly. This could indicate different substrate specificity, alternatively, the possibility still exists that the native organellar requirement could be a prerequisite for its activity. Determining the role of these enzymes is also important in the light of the fact that glutathione synthesis occurs in the plastid. Interestingly, AtLAP2 and AtLAP3 are overlapping ORFs on the same strand. Both AtLAP1 and AtLAP2/AtLAP3 have been shown to be induced upon pathogenic attack and stress response.
Glutathione degradation is critical to glutathione homoeostasis, which in turn is critical for plant life and its adaptation to stress. One hopes therefore that with the identification of these important players in glutathione homoeostasis that have led to fresh insights into the pathways, a better understanding of glutathione homoeostasis and its role in both redox and nutritional homoeostasis in plants can now be achieved.
Shailesh Kumar purified and performed enzymatic characterization of AtLAP1. Amandeep Kaur purified and characterized the GGCT enzymes. Banani Chattopadhyay initiated the work with the plant Cys-Gly peptidases and cloned and initially characterized the different AtLAP genes. Anand Bachhawat supervized and integrated the different components and all authors contributed in the writing and preparation of the manuscript.
We thank Tejaswani Modi and Sayali Chougale for their help in the cloning of GGCT2;3 and AtLAP genes. We thank Dr Ashish and Sameer K Nath of Institute of Microbial Technology, Chandigarh, for custom synthesizing the dipeptides. We thank Shashi B. Pandit for help with the phylogenetic analysis. We thank Dr Akhilesh Kumar for helpful discussions.
This work was supported by the Department of Science and Technology and the Department of Biotechnology, Government of India [grant numbers SB/SO/BB/017/2014 and BT/PR11656/BRB/10/681/2008 (to A.K.B.)]; the Council of Scientific and Industrial Research [(to S.K.)]; the Department of Biotechnology, Government of India [(to B.C.)]; and the Department of Science and Technology JC Bose National Fellowship from the Government of India [grant number SR/S2/JCB-98/2011 (to A.K.B.)].
avirulence-induced gene 2-like protein
A. thaliana leucine aminopeptidase
γ-glutamyl amine cyclotransferase
open reading frame
TBS with Tween 20
These authors contributed equally to the study.