The glutathione-deficient, cadmium-sensitive mutant,cad2–1, ofArabidopsis thalianais deficient in γ-glutamylcysteine synthetase


  • Christopher S. Cobbett,

    1. Department of Genetics, The University of Melbourne, Parkville, 3052, Australia, and
    2. Laboratorium voor Genetica, Universiteit Gent, KL Ledeganckstraat 35, B-9000, Gent, Belgium
    Search for more papers by this author
  • Mike J. May,

    1. Laboratorium voor Genetica, Universiteit Gent, KL Ledeganckstraat 35, B-9000, Gent, Belgium
    Search for more papers by this author
    • Present address: Plant Genetic Systems NV, Plateaustraat 22, B-9000, Gent, Belgium.

  • Ross Howden,

    1. Department of Genetics, The University of Melbourne, Parkville, 3052, Australia, and
    Search for more papers by this author
  • Barbara Rolls

    1. Department of Genetics, The University of Melbourne, Parkville, 3052, Australia, and
    Search for more papers by this author
    • Present address: Department of Physiology, University of Queensland, St Lucia, 4072, Australia.

*For correspondence (fax +613 9344 5139;


This paper reports that the glutathione (GSH)-deficient mutant,cad2–1, of Arabidopsis is deficient in the first enzyme in the pathway of GSH biosynthesis, γ-glutamylcysteine synthetase (GCS). The mutant accumulates a substrate of GCS, cysteine, and is deficient in the product, γ-glutamylcysteine.In vitroenzyme assays showed that thecad2–1mutant has 40% of wild-type levels of GCS activity but is unchanged in the activity of the second enzyme in the pathway, GSH synthetase. TheCAD2locus maps to chromosome 4 and is tightly linked to a gene,GSHA, identified by a previously isolated cDNA. A genomic clone ofGSHAcomplements both the phenotypic and biochemical deficiencies of thecad2–1mutant. The nucleotide sequence of the gene has been determined and, in the mutant, this gene contains a 6 bp deletion within an exon. These data demonstrate that theCAD2gene encodes GCS. Thecad2–1mutation is close to the conserved cysteine which is believed to bind the substrate glutamate and the specific inhibitor L-buthionine-[S,R] sulfoximine (BSO). Both root growth and GCS activity of thecad2–1mutant was less sensitive than the wild-type to inhibition by BSO, indicating that the mutation may alter the affinity of the inhibitor binding site.


The tripeptide glutathione (GSH) is the most abundant of the free, low MW thiol compounds found in most living cells, including plants. GSH, γ-l-glutamyl-l-cysteinylglycine, or its analogue (homoglutathione, γ-l-glutamyl-l-cysteinyl-β-alanine, in some legumes) is ubiquitous in the plant kingdom and is proposed to play a significant role in a number of processes, particularly involving responses to stress (Alscher 1989;Marrs 1996;May et al. 1998;Noctor & Foyer 1998;Rea et al. 1998). One such role is the substrate for the biosynthesis of the heavy metal-binding peptides, phytochelatins (PCs). PCs [(γ-l-glutamyl-l-cysteinyl)n-glycine] are synthesised directly from GSH by the enzyme PC synthase. PCs form metal-thiolate bonds and the PC-metal complexes are sequestered to the vacuole (Rauser 1995). The essential role of PCs in metal detoxification in plants has been demonstrated by the isolation of mutants at the CAD1 locus in Arabidopsis which are deficient in PCs and in PC synthase activity, and which are sensitive to a range of heavy metals (Howden et al. 1995b).

In plants, GSH is synthesised enzymatically from its constituent amino acids via a two-step pathway similar to that found in other organisms. The first step, catalysed by γ-glutamylcysteine synthetase (GCS), forms γ-glutamylcysteine and the second, catalysed by GSH synthetase (GS), produces GSH through the addition of a glycine residue. Both enzymes require ATP for activity. cDNAs encoding enzymes of the GSH biosynthesis pathway have been isolated from Arabidopsis by the functional complementation of mutants of E. coli and S. cerevisiae deficient in these activities (May & Leaver 1994;Rawlins et al. 1995;Ullman et al. 1996; M.J. May et al. in preparation). Only a single gene encoding each enzyme has been isolated from Arabidopsis. In contrast, multiple genes appear to exist in both tomato (P. Goldsbrough, personal communication) and Brassica juncea (Schäfer et al. 1997).

Mutants affected in each of the two steps of GSH biosynthesis have been isolated in bacteria and yeasts (Apontoweil & Berends 1975;Glaeser et al. 1991;Grant & Dawes 1996;Mutoh & Hayashi 1988). In plants, a single GSH-deficient mutant has thus far been reported; the cad2–1 mutant of Arabidopsis, which has 15–30% of wild-type levels of GSH and is consequently PC-deficient and Cd-sensitive (Howden et al. 1995a). In this paper, we demonstrate that this mutant is both deficient in GCS activity and has a defect in the previously identified gene, GSHA (May & Leaver 1994).


The cad2–1 mutant is deficient in γ-glutamylcysteine synthetase activity

The GSH-deficiency observed in the cad2–1 mutant was most likely due to a deficiency in the GSH biosynthetic pathway. To explore this, the levels of cysteine, γ-glutamycysteine (GC) and GSH were measured in leaf tissue of wild-type and mutant plants grown in agar medium (Table 1). In the cad2–1 mutant, the levels of GC and GSH were about 30% and 45%, respectively, of those observed in the wild-type. In contrast, the level of cysteine in the mutant was approximately twofold that in the wild-type. Together, these data suggest that the mutant was affected in GSH biosynthesis and that because both an accumulation of cysteine and a decrease in GC was observed, it was more likely deficient in GCS rather than GS activity. This was confirmed by in vitro assays of the two biosynthetic activities which demonstrated that, while the level of GS activity was unchanged in the mutant, GCS activity was reduced to 40% of that in the wild-type (Table 1).

Table 1.  Biochemical characterisation of wild-type, cad2–1 and transgenic plants
(nmol g–1 DW)
(nmol g–1 DW)GSH
(μmol g–1 DW)
(nmol min–1 mg–1 protein)
  1. Metabolites and enzyme activities were measured in extracts of plants grown in agar medium as described in Experimental procedures. Percentage of wild-type is indicated in parentheses. pBI-GSHA-1 to −5 and pBI101–1 to −3 are independent transgenic derivatives of the cad2–1 mutant transformed with the plasmids indicated.

Wild-type 111 ± 9(100) 64 ± 8(100)2.95 ± 0.32(100)0.80 ± 0.05(100)0.69 ± 0.29
cad2–1 234 ± 3(209) 21 ± 1(33)1.32 ± 0.01(45)0.31 ± 0.02(40)0.67 ± 0.01
pBI-GSHA-1118 ± 4(106)210 ± 34(330)6.54 ± 0.37(222)2.57 ± 0.18(322)0.71 ± 0.10
-2134 ± 3(120)125 ± 1(197)5.22 ± 0.16(177)1.10 ± 0.01(138)0.82 ± 0.07(118)
-3118 ± 3(105)156 ± 6(244)4.80 ± 0.28(163)2.09 ± 0.40(263)1.04 ± 0.16(151)
-4131 ± 6(117)180 ± 5(283)5.54 ± 0.76(188)1.36 ± 0.18(170)0.72 ± 0.06(103)
-5110 ± 2(99)129 ± 1(203)5.47 ± 0.52(185)1.82 ± 0.26(228)0.71 ± 0.06(103)
pBI101-1224 ± 10(201) 26 ± 3(41)1.26 ± 0.55(43)0.30 ± 0.07(38)0.70 ± 0.010
-2249 ± 4(223) 21 ± 1(33)1.38 ± 0.94(47)0.28 ± 0.17(35)0.60 ± 0.08(88)
-3197 ± 9(176) 24 ± 3(38)1.32 ± 0.10(45)0.35 ± 0.01(44)0.70 ± 0.01(102)

CAD2 corresponds to GSHA

The CAD2 locus was mapped to chromosome 4 (Fig. 1) and co-segregated with an RFLP detected by the cDNA clone, GSHA, which encodes GCS (see Experimental procedures;May & Leaver 1994), indicating GSHA was a likely candidate gene for CAD2. The nucleotide sequence (accession no. AF068299) of the corresponding region of a genomic clone, cAtECS4C3, (Fig. 2a) was determined. The CAD2 gene contains 15 introns ranging in size from 77 to 359 bp. A schematic illustration of the intron–exon structure of the gene is shown in Fig. 2(c). The consensus GT and AG sequences are found at the 5′ and 3′ splice sites, respectively, in every case. The exons range from 69 to 340 bp in size. In addition, the nucleotide sequence of the GSHA gene in the cad2–1 mutant contains a deletion of 6 bp in exon six (Fig. 2d). These 6 bp were present in the GSHA cDNA (May & Leaver 1994), in the genomic clone and in PCR products amplified from the Columbia wild-type ecotype. No other differences in nucleotide sequence between the wild-type clone and the cad2–1 mutant were observed, demonstrating that the deletion is the cad2–1 mutation and not a polymorphism.

Figure 1.

RFLP map position of the CAD2 locus on Chromosome 4.

The map distances in cM between CAD2 and various RFLP markers are shown. The complete designation for marker d104 is pCITd104.

Figure 2.

Structure of the CAD2 gene.

(a) Restriction map of cosmid clone cAtECS4C3 showing the positions of EcoRI sites (E) and the sizes (in kb) of EcoRI fragments. The arrow indicates the regions of the clone complementary to the cDNA and the direction of transcription.

(b) The approximately 6.6 kb NheI (N)-BglII (B) fragment ligated into pBI101 which complements the cad2–1 mutant.

(c) The intron/exon structure of the CAD2 gene. Sizes of exons (above) and introns (below) are shown in bp. The position of the cad2–1 mutation is indicated.

(d) The wild-type and cad2–1 nucleotide and derived amino acid sequences.

(e) The region of amino acid sequence around the conserved cysteine residue believed to be in the active site of the enzyme is aligned with the same region from the T. brucei GCS amino acid sequence. Residues absolutely conserved between the yeast, rat and T. brucei sequences are indicated by asterisks below the T. brucei sequence (Lueder & Phillips 1996). Residues conserved between the Arabidopsis and T. brucei sequences are indicated by asterisks between the two sequences. In the Arabidopsis sequence the amino acids affected by the cad2–1 mutation and the conserved cysteine residue are underlined.

To confirm further that the cad2–1 mutation is in the GSHA gene, a fragment spanning the GSHA gene (Fig. 2b) was ligated into the pBI101 vector (Clontech) to form pBI-GSHA which was transformed into cad2–1 plants. The pBI101 vector alone was also transformed into cad2–1 as a control. For pBI-GSHA and pBI101 a total of 11 and 9 kanamycin-resistant transgenic plants representing at least 5 and 3 independent events, respectively, were isolated. Among the progeny of cad2–1 plants transformed with the pBI-GSHA construct, kanamycin-resistant seedlings were uniformly cadmium-resistant. In contrast, among the progeny of cad2–1 plants transformed with pBI101, only cadmium-sensitive individuals were observed for both the unselected and kanamycin-resistant seedlings (data not shown).

From the progeny of each of five and three independent pBI-GSHA and pBI101 transformants, respectively, a line homozygous for the kanamycin-resistance transgene was isolated. These lines were assayed for cysteine, GC and GSH levels, and for GCS and GS activities along with the kanamycin-sensitive wild-type and cad2–1 mutant (Table 1). In each of the pBI-GSHA transformants, cysteine levels were decreased to approximately that in the wild-type, while the GC and GSH levels were increased to two- to threefold and 1.5- to 2.5-fold those in the wild-type, respectively. Similarly, the levels of GCS activity were 1.5- to threefold greater than in the wild-type, while the levels of GS activity remained unchanged. In contrast, the levels of metabolites and enzyme activities in the pBI101 transformants of cad2–1 were essentially the same as those in the cad2–1 parent. These data demonstrate clearly that the GSHA gene is able to complement the cad2–1 mutant at both the phenotypic and biochemical levels.

The mutant GCS activity is less sensitive to inhibition by BSO

GCS catalyses the reaction of l-glutamate with MgATP to form γ-glutamylphosphate as an enzyme-bound intermediate. Reaction of γ-glutamylphosphate with the α-amino group of l-cysteine completes the catalytic cycle (Griffith 1982). Buthionine sulfoximine (BSO) closely approximates the structure of the γ-glutamylphosphate cysteine adduct and binds tightly but non-covalently to a catalytically active cysteine near the active site (Griffith 1982). This cysteine residue is conserved in the amino acid sequences derived from all GCS genes cloned to date (Lueder & Phillips 1996), including Arabidopsis. In view of the proximity of the cad2–1 mutation to this residue (Fig. 2e), it was possible that the mutant GCS in cad2–1 may have altered affinity for BSO. In support of this, while the root growth of both wild-type and cad2–1 seedlings was inhibited in the presence of 1 mm BSO, the degree of inhibition was greater for the wild-type than for the mutant (Fig. 3). No difference in root growth between cad2–1 and wild-type seedlings was observed in the absence of BSO. Furthermore, the sensitivity of the enzyme to inhibition by BSO in vitro was tested as described in Experimental procedures. These data show that the activity of the mutant GCS is less sensitive to inhibition than the wild-type enzyme (Table 2).

Figure 3.

Growth of wild type and cad2–1 mutants seedlings in the presence of 1 mm BSO.

Seeds of cad2–1 (a) and the wild-type (b) were germinated on solidified mineral salts medium supplemented with 1 mm BSO and photographed after 5 days. Bar = 1 cm.

Table 2.  Inhibition of mutant and wild-type GCS activity by BSO
GCS activity (nmol min–1 mg–1 protein)
BSO added
  1. Data are the mean ± SE of three separate determinations.

nil0.90 ± 0.030.31 ± 0.02
0.1 mm0.52 ± 0.050.30 ± 0.03
1.0 mm0.10 ± 0.010.18 ± 0.01


This work demonstrates that the GSHA gene encoding the enzyme γ-glutamylcysteine synthetase is mutated in the cad2–1 mutant. We have shown that the effects on the intermediates of the GSH biosynthetic pathway are consistent with such a defect and that the mutant is deficent in GCS activity in vitro. The GSHA gene, previously identified as complementing both E. coli (May & Leaver 1994) and yeast (unpublished data) mutants deficient in GCS, is closely linked to the CAD2 locus. Both the Cd-sensitive and biochemical phenotypes of the cad2–1 mutant can be complemented by the GSHA gene and there is a small deletion mutation in the GSHA gene in the cad2–1 mutant. Together, these observations support the conclusion that the CAD2 gene encodes GCS activity.

The most apparent effect of inhibition of GSH biosynthesis by BSO is the inhibition of root growth. Consistent with this, a second, more extreme, cad2 mutant which exhibits a root growth, seedling lethal phenotype has recently been identified (R. Wilson, personal communication). That the cad2–1 mutant does not exhibit any phenotype in the absence of Cd suggests that the residual synthesis of GSH (30–45%) (Howden et al. 1995b; this study) is sufficient for growth and metabolism. The existence of a more extreme cad2 allele indicates that the cad2–1 mutant enzyme retains partial activity and supports the idea that, in Arabidopsis, GCS activity is encoded by a single gene. Other circumstantial evidence also suggests that there may be a single gene: the same GCS gene has been identified by functional complementation of either E. coli or yeast in at least three separate experiments, and there appear to be no cross-hybridising sequences in the Arabidopsis genome.

Although the amino acid sequence of the Arabidopsis GCS enzyme is not clearly related to the enzymes from other organisms (May & Leaver 1994), evidence for the conservation of a specific region of the amino acid sequence of GCS from a number of organisms, including Arabidopsis, has been presented (Lueder & Phillips 1996) (Fig. 2e). Of particular interest is a conserved Cys residue important for the catalytic activity of the protein, notably in the binding of Glu and BSO (Griffith 1982). The consequences of the cad2–1 mutation on enzyme activity are of considerable interest since the mutation lies 12 amino acids on the N-terminal side of the conserved Cys residue (Fig. 2e). It is likely that the lowered activity of GCS in cad2–1 is due to an alteration in the conformation of the active site as a result of the nearby mutation. While the mutant enzyme remains subject to inhibition by BSO, it appears to be less susceptible than the wild-type, supporting the idea that the mutation influences the affinity of the enzyme for BSO and, by inference, Glu. We are currently evaluating whether the mutation in cad2–1 GCS affects the Km for the substrates and whether the enzyme has lowered affinity for other inhibitors.

Experimental procedures

Plant growth conditions

Plants were grown for 3 weeks under sterile conditions in 150 mm Petri dishes on 100 ml of nutrient medium solidified with 0.6% (w/v) agarose at a light intensity of 100 μmol m–2 s–1 with a photoperiod of 16 h. The nutrient solution was supplemented with 1% (w/v) sucrose (pH 5.8) (Estelle & Somerville 1987). Greenhouse-grown plants were grown for 3 weeks on a mixture of potting compost/vermiculite (1:1) at a light intensity of 100 μmol m–2 s–1 with a photoperiod of 16 h and were watered weekly with the nutrient solution described.

Quantitation of cysteine, GC and GSH

Extraction and determination of acid-soluble thiols from leaf extracts was essentially as described by Rüeggseger & Brunold (1992). Triplicate samples were assayed.

Enzyme activity assays

The activities of GCS and GS were determined by postreaction derivatisation of thiols with monobromobimane and separation of the thiol-bimane conjugate by HPLC coupled to a fluorescence detector essentially as described by Rüeggseger & Brunold (1992). Leaf tissue from 3-week-old plants grown in the greenhouse or under sterile conditions was harvested, flash-frozen in liquid nitrogen and stored at –70°C. Three grams of frozen tissue was ground in ice-cold extraction buffer (100 mm Tris–HCl, 10 mm MgCl2, 5 mm EDTA, pH 8) at a tissue-to-buffer ratio of 1:4. Debris was removed by centrifugation. The supernatant was desalted by Sephadex G-25 chromatography using the extraction buffer for elution. For each tissue sample, desalted extracts were divided into 250 μl aliquots and stored at –70°C until needed. For the measurement of GCS activity, 250 μl of the desalted extract was incubated at 37°C for 45 min in a total volume of 500 μl containing 100 mm HEPES-NaOH (pH 8), 40 mm MgCl2, 30 mm L-Glu, 0.8 mm L-Cys, 0.4 mm DTT, 5 mm ATP, 5 mm phospho-enol pyruvate and 5 units of pyruvate kinase. When testing for inhibition by BSO, 5 mm ATP and the appropriate concentration were added to the desalted extract 10 min prior to the addition of the other components. At the end of the reaction, 50 μl of the reaction mix was added to 200 μl of 50 mm CHES, pH 9 and was derivatised for 15 min with 15 μl of 15 mm monobromobimane (Molecular Probes) in acetonitrile. The reaction was stopped with 700 μl of 5% (v/v) acetic acid and protein sedimented by centrifugation for 10 min at top speed in an eppendorf centrifuge. For the measurement of GS activity, 250 μl of the desalted extract was incubated at 37°C for 45 min in a total volume of 500 μl containing 90 mm Tris–HCl (pH 8.4), 20 mm MgCl2, 45 mm KCl, 2 mm Gly, 0.5 mmγGluCys, 4 mm DTT, 7 mm ATP, 5 mm phospho-enol pyruvate and 5 units of pyruvate kinase. At the end of the reaction, 25 μl of the reaction-mix was added to 200 μl of 50 mm CHES, pH 8.4 and was derivatised for 15 min with 20 μl of 15 mm monobromobimane (Molecular Probes) in acetonitrile. The reaction was stopped with 1 ml of 5% (v/v) acetic acid and protein sedimented by centrifugation for 10 min at top speed in an eppendorf centrifuge and the supernatant stored at –70°C until needed. For quantification of the reaction products, 50 μl samples of the supernatant from the derivatisation reactions were manually injected on a Vydac C18 column (Alltech) attached to a Vista 5500 HPLC (Varian), eluted isocratically in a mixture of 95% buffer A and 5% buffer B (Buffer A is 5% (v/v) methanol, 0.25% (v/v) acetic acid, pH 3.9; Buffer B is 90% (v/v) methanol, 0.25% (v/v) acetic acid), and detected using a Waters 470 scanning fluorescence detector by excitation at 380 nm and emission at 480 nm. Peak areas were quantified against a standard curve. Protein was measured by the method of Lowry with modifications by Peterson (1979). Duplicate samples were assayed.

RFLP mapping

Thirty-six F3 families derived from a cross between the cad2–1 mutant (Columbia ecotype) and the Landsberg erecta ecotype were scored for Cd-resistance. Genomic DNA was extracted from each F3 family using the method of Dellaporta et al. (1983). Lambda clones corresponding to the RFLP markers (Chang et al. 1988) indicated in Fig. 1 were obtained from D. Smyth. The GSHA cDNA clone detected an EcoRI RFLP between the Columbia and Landsberg ecotypes. DNA from these clones was labelled with 32P-dATP using a nick-translation procedure essentially as described by Sambrook et al. (1989). Southern transfer was essentially as described by Sambrook et al. (1989). Southern filters were prehybridised at 42°C (50% formamide, 5× SSPE, 5× Denhardt’s solution, 1% w/v dried milk powder, 0.1% w/v SDS). The filters were hybridised (50% formamide, 5× SSPE, 3× Denhardt’s solution, 0.1% w/v SDS 9% w/v dextran sulphate, 100 mg ml–1 salmon sperm DNA) at 42C with each DNA probe and washed in 2× SSPE (20 min), 2× SSPE/0.1% SDS (20 min) and 0.1× SSPE (10 min), at 65°C.

Nucleotide sequencing of the CAD2 gene

A genomic clone, cAtECS4C3, hybridising to the GSHA cDNA was isolated from a cosmid library (Olszewski et al. 1988) and the nucleotide sequence of the wild-type CAD2 gene was determined from subclones derived from cAtECS4C3. The cad2–1 allele was amplified in an approximately 5 kb fragment of DNA using genomic DNA from the cad2–1 mutant and the ELONGASE Enzyme Mix (GibcoBRL), according to the manufacturer’s instructions. Nucleotide sequencing reactions were performed by using the double-stranded templates and the dye-terminator cycle-sequencing AmpliTaq kit (ABI) with a series of primers specific for the amplified product spaced at 300–400 bp intervals. Sequence products were resolved on a 373 DNA sequencer (ABI). The difference observed between the wild-type and mutant sequences was determined from both strands in two independent PCR products.

Transformation of the cad2–1 mutant

An approximately 6.6 kb NheI-BglII fragment from cAtECS4C3 was ligated between the XbaI and BamHI sites of pBI101 (Clontech) and transformed into Agrobacterium tumefaciens strain GV3101/pMP90 by electroporation. This construct and the parent plasmid, pBI101, were transformed into cad2–1 plants by the vacuum infiltration procedure (Bechtold et al. 1993). Transformant seedlings were selected in the presence of 50 μg ml–1 kanamycin.


We thank Teva Vernoux for critical reading of the manuscript. M.J.M. is indebted to EMBO for a Post Doctoral fellowship. R.H. was supported by a University of Melbourne Scholarship. C.S.C. is supported by a grant from the Australian Research Council.

GenBank database accession number AF068299 (CAD2 genomic sequence).