Arabidopsis thaliana NADPH oxidoreductase homologs confer tolerance of yeasts toward the thiol-oxidizing drug diamide.

To isolate new plant genes involved in the defense against oxidative stress, an Arabidopsis cDNA library in a yeast expression vector was transformed into a yeast strain deficient in the YAP1 gene, which encodes a b-Zip transcription factor and regulates general stress response in yeasts. Cells from ≈105 primary transformants were subjected to a tolerance screen toward the thiol-oxidizing drug diamide, which depletes the reduced glutathione in the cell. Four types of Arabidopsis cDNAs were isolated. Three of these cDNAs (P1, P2, and P4) belong to a plant ζ-crystallin family and P3 is an Arabidopsis homolog of isoflavonoid reductases. As such, all four isolated cDNAs are homologous to NADPH oxidoreductases. P1, P2, and P3 steady-state mRNAs accumulated rapidly in Arabidopsis plants under various oxidative stress conditions, such as treatment with paraquat, t-butylhydroperoxide, diamide, and menadione. The data suggested that proteins encoded by the isolated cDNAs play a distinct role in plant antioxidant defense and are possibly involved in NAD(P)/NAD(P)H homeostasis.

Aerobic organisms have evolved a number of enzymic and nonenzymic antioxidant defense mechanisms, which counteract the harmful effects of AOS and maintain the cellular steady-state of pro-oxidants and antioxidants (1,3). Nonenzymic antioxidants, such as ascorbic acid, glutathione, carotenoids, and ␣-tocopherol scavenge AOS directly through chemical mechanisms. The major enzymes of importance in O 2 .
Adverse environmental conditions often lead to the disruption of the pro-oxidant/antioxidant homeostasis. A disbalance in favor of pro-oxidants, potentially leading to cell damage, has been defined as oxidative stress (3). The activity of antioxidant defenses often correlates positively with aging or senescence, disease development, or ability to adapt to changing environment (4). Furthermore, hereditary defects in antioxidant protection have multiple pleiotropic effects both in prokaryotic and eukaryotic organisms. For example, mutations in the gene coding for the copper/zinc superoxide dismutase are associated with familiar amyotrophic lateral sclerosis in humans (6) and yeast mutants in mitochondrial manganese superoxide dismutase are hypersensitive to oxygen (7). The increase in activity of AOS-scavenging enzymes by gene engineering, for instance, prolongs the life span of Drosophila (8) and improves stress tolerance of plants (5).
The importance of the antioxidant defense mechanisms is mirrored by their complexity, and new antioxidant proteins are continuously described. The reversion of mutant phenotypes associated with defects in antioxidant defense by screening for extragenic suppressors or by gene overexpression has been found to be a suitable approach toward better understanding pro-oxidant/antioxidant homeostasis (9 -12).
Following such studies in yeasts, we became interested in examining whether the random overexpression of plant cDNAs in yeast cells would allow by-pass of phenotypes associated with a deficiency of antioxidant defense mechanisms. This approach allows the characterization of hitherto unidentified genes, the isolation of which would be not amenable by conventional methods.
The yeast mutant deficient in YAP1 was chosen in the present study. The YAP1 gene encodes a transcription activator of the b-Zip family of DNA-binding proteins and was originally isolated as a yeast homolog of mammalian jun (13). In later studies, however, the YAP1 gene was isolated by different research groups as a gene, which when overexpressed on multicopy plasmids, confers tolerance to different toxic compounds (14 -16). Direct expression analysis and DNA binding studies have shown that YAP1 plays an important role in stress-induced transcriptional activation of many yeast genes, and more specifically of antioxidant genes, including those encoding thioredoxin, TRX2 (17), and ␥-glutamylcysteine synthase, GSH1 (18). The levels of antioxidant defense enzymes, such as superoxide dismutase, glutathione peroxidase, and glucose-6-phosphate dehydrogenase are lower in yap1 mutants (16). The role of YAP1 in regulation of the antioxidant defense is reflected in the increased sensitivity of yap1 mutants to hydroperoxides, thiol oxidants, and redox-cycling drugs (17).
In this paper, we present experiments aimed to isolate Arabidopsis cDNAs, of which the overexpression improves the survival of yeast cells deficient in YAP1 function in the presence of the thiol-oxidizing drug diamide (azodicarboxylic acid bis[dimethylamide]), which depletes intracellular glutathione in vivo (1) and oxidizes cysteine residues in proteins in vitro (19). The expression of antioxidant genes, such as the thioredoxin gene TRX2 in yeasts and manganese superoxide dismutase in Escherichia is strongly activated in diamide-treated cells (17,20). In yeast, diamide-stimulated up-regulation of the TRX2 gene is mediated through YAP1 activation (17). Furthermore, the YAP1 gene, when present on a high copy number plasmid, confers high tolerance of yeast cells to diamide (17). In Escherichia, the diamide-induced manganese superoxide dismutase gene activation is mediated through the superoxide anion sensor soxRS (20).
Escherichia coli strain XLI (Stratagene, La Jolla, CA) was used for plasmid manipulations and was grown on standard LB medium supplemented with antibiotics when required.
Plant Material-For DNA extraction and the analysis of tissuespecific gene expression, plants of Arabidopsis thaliana (L.) Heynh. ecotypes Columbia and Landsberg erecta were grown in soil at 23°C with 16 h light/8 h dark. For the restriction fragment length polymorphism (RFLP) mapping of P4 cDNA, plants of 98 inbred recombinant lines (23) were grown under the same conditions. For drug treatments, 3-week-old plants grown in vitro on a Murashige and Skoog medium (Life Technologies, Inc., Gaithersburg, MD) were used. Complete plants were collected and different compounds were infiltrated into the tissues under vacuum in Erlenmeyer flasks. Vacuum was released the moment air bubbles started appearing on the leaf surface. The following drugs were used: 5 ϫ 10 Ϫ6 M paraquat (methyl viologen), 5 ϫ 10 Ϫ6 M menadione, 1 mM diamide, 0.5 mM t-butylhydroperoxide (t-BOOH), 40 mM hydrogen peroxide, and 10 mM dithiothreitol. The infiltrated plants were transferred to Petri dishes and RNA was extracted at different time points. Plants infiltrated with paraquat were exposed to a 40-watt bulb from 10 cm distance to facilitate the light-dependent redox cycling of the drug.
Screening of Yeasts and Drug Tolerance Tests-An Arabidopsis cDNA library in the yeast expression vector pFL61 was kindly provided by Dr. M. Minet (24). The library was amplified in Escherichia as described (22). For the diamide resistance screening, WYT cells were transformed with the cDNA library and uracil prototrophic colonies were allowed to grow on minimal SD media plates for 4 days. Eleven independent transformations were made in parallel. Primary transformants were harvested from each plate and resuspended in 2 ml of 60% glycerol. Aliquots of cells were either directly plated onto diamide (1 and 1.5 mM) containing SD plates or frozen at Ϫ70°C for storage. Putative diamide-resistant colonies, which appeared after 4 -7 days of incubation at 30°C, were picked in 2 ml of SD medium and cultures were grown to stationary phase (36 h) on a gyratory shaker at 200 rpm, 30°C. Dilutions (10-and 50-fold) of cultures were spotted onto SD plates supplemented with diamide (0.5 and 1 mM) as 10-l aliquots. WYT(pFL61) was used as a control. Individual clones which grew at 1/50 dilution on 1 mM diamide were further analyzed. Plasmids from individually isolated yeast clones were rescued into Escherichia as described (22). To confirm the drug tolerance phenotype, WYT or DY strains were transformed with selected plasmids and at least two independent transformants were tested. To assay the drug tolerance, a semi-quantitative dilution test was used as described (25,26). Cultures of yeast cells in stationary or logarithmic phases were grown in SD medium at 30°C with shaking at 200 rpm. Cell suspensions were serially diluted in 0.3% NaCl and 10 l of dilutions were spotted on SD plates supplemented with drugs. After 3 days of incubation at 30°C, growth was assessed. Dilution series grown on plates without drugs were used to compare original cell density of different cultures. For the WYT strain, the following concentrations of drugs were used: 1 mM diamide, 0.2 and 0.3 mM t-BOOH, 0.2 mM cycloheximide, 100 M CdCl 2 . The diamide tolerance of DY strains was tested on plates with 1.5 mM diamide.
DNA and RNA Work-RNA was extracted from Arabidopsis plants according to Shirzadegan et al. (27). Genomic DNA, extracted by the method of Mettler (28), was additionally purified by CsCl isopicnic centrifugation in the presence of 10 g ml Ϫ1 ethidium bromide. Gene copy number and possible RFLP were estimated by DNA gel blot analysis using Arabidopsis DNA digested with different restriction endonucleases according to standard procedures (22). Analysis of mRNA was carried out by glyoxal RNA gel blot analysis (22) with modifications according to Mironov et al. (29). For the DNA and RNA hybridizations on nylon membranes (Hybond N; Amersham, Aylesbury, United Kingdom) the buffer (0.25 M sodium phosphate, pH 7.2, 7% sodium dodecyl sulfate, 1% bovine serum albumin, 1 mM EDTA) of Church and Gilbert (30) was used. Radioactively labeled DNA probes were made by the random priming method according to the manufacturer's instructions (QuickPrime kit; Pharmacia, Uppsala, Sweden). The apx cDNA was isolated in our laboratory and was identical to the cDNA published by Kubo et al. (31).
Plasmid Construction-The expression cassettes containing cDNA clones designated P1 and P3 were excised from pFL61 as BglII fragments, which, after filling in DNA ends were ligated into the blunted SphI site of pGAD424 (Clontech, Palo Alto, CA). The shuttle vector pGAD424 has the LEU2 gene for selection in yeasts, which permitted the isolation of yeast strains with both pFL61 (URA3 selection) and pGAD234 based vectors.
Sequence Analysis-DNA sequencing was performed according to protocols provided by U. S. Biochemical Corp. Vector primers, GTTT-TCAAGTTCTTAGATGC and AGCGTAAAGGATGGGG, were used to obtain initial sequences. To complete the sequences on both strands, gene-specific primers were synthesized on an oligonucleotide synthesizer (Applied Biosystems, Foster City, CA). The DNA and amino acid sequence analyses were done with a software package by Genetics Computer Group Inc. (Madison, WI).

Isolation of Arabidopsis cDNAs Conferring Diamide Tolerance of yap1
Ϫ Yeast-Yeast cells lacking a functional YAP1 gene were transformed with an Arabidopsis cDNA library in the yeast expression vector pFL61 (24). Approximately 10 5 uracil prototrophic (URA3 ϩ ) colonies were selected in 11 independent transformation experiments. Forty thousand and 200,000 cells of pooled primary transformants were plated onto SD plates containing 1 and 1.5 mM diamide, respectively. Forty-eight colonies, which grew after 4 -7 days of selection, were tested for diamide tolerance, by spotting 10-l aliquots of diluted stationary phase cultures on plates containing 1 mM diamide; 23 clones showed confluent growth at a dilution of 1/50 on 1 mM diamide and were further analyzed in this study.
Partial DNA sequencing of the rescued plasmids was used to identify the clones. Four types of clones, designated P1, P2, P3, and P4 were found. Fourteen cDNA clones of eight different lengths contained the P1-type cDNA. The P2 (4 clones), P3 (3 clones), and P4 (2 clones) were represented each as two different length cDNAs. This shows the consistency in the recovery of Arabidopsis cDNAs in the described screen.
The longest cDNAs of each type were retransformed into the yap1 Ϫ mutant strain, WYT, and the diamide tolerance of two independent transformants for each cDNA was assessed in a semi-quantitative dilution series test, as shown in Fig. 1. WYT-(pFL61) cells from stationary phase culture are sensitive to 1 mM diamide and little growth was observed at the 1/10 dilution. Similar suppression of growth for strains containing P1, P2, P3, and P4 was observed only at a dilution of 1/250. Logarithmic phase WYT(pFL61) cells appeared to be more tolerant to diamide, although the strains containing the isolated cDNAs were still able to grow at the higher dilution (compare L lanes on Fig. 1). The isolated cDNAs were not able to improve the survival of yeast cells on plates supplemented with t-BOOH, cycloheximide, and cadmium (data not shown).
The isolated Arabidopsis cDNAs were then tested for their ability to improve the diamide tolerance of the nearly isogenic to WYT, YAP1 strain DY. As shown in Fig. 2A, YAP1 strain DY showed better survival on 1.5 mM diamide, when it expressed P1, P2, P3, and P4 cDNAs. However, P1, P2, P3, and P4 clones were less efficient in rescuing cells from diamide toxicity as compared to the clone D3, which was isolated in a similar screening with a DY strain. 2 If the mechanisms of protection of yeast cells from diamide toxicity by the isolated clones were different, we may expect additive effects for the diamide tolerance, when several cDNAs are co-expressed. To test this hypothesis, we recloned the P1 and P3 expression cassettes into another yeast-Escherichia shuttle vector, which has a LEU2 gene as a selectable marker. Yeast strains containing the two plasmids were selected on SD medium plates lacking uracil and leucine. When assessed in a dilution test, none of the tested combinations showed strong additive effects, although slightly better growth on diamide plates may be seen for the strains, which combine P1 and P3, and P4 and P3 (Fig. 2B).
P1, P2, and P4 Are Members of a Gene Family-The longest P1 cDNA is 1211 bp and has an open reading frame (ORF) of 346 amino acids. Fig. 3A shows an RNA gel blot analysis using P1 as a probe, revealing a transcript of 1.24 kilobases. The longest P2 cDNA is 1206 bp, and RNA gel blot analysis showed a transcript of 1.25 kilobases (Fig. 3B). P2 cDNA has an ORF of 343 amino acids, which is 91% identical to the P1 ORF. Considering the lengths of the isolated P1 and P2 cDNAs, as compared to the estimated mRNA sizes, and homologies between the two cDNAs, we believe that nearly full-length cDNAs have been isolated. P1 and P2 have been previously described as expressed sequence tags and correspond to Arabidopsis cDNAs 97C1T7, 97B20T7, and TAGS137 (EMBL accession numbers T21718, T21706, and Z18002, respectively). Fig. 4 shows that a number of proteins identified in the EMBL data base have strong amino acid sequence similarity to P1 and P2. The best scores (Table I) were found to the rabbit protein AdRab-F, which is expressed only in the intestine of adult rabbits and not in newborn animals. Lower homology was found to -crystallins from guinea pigs, trypanosome Leishmania amazonensis, and to the plant cDNA TED2, originally isolated as a transcript up-regulated during in vitro xylem differentiation of zinnia mesophyll cells (Fig. 4). The actual function of these proteins in vivo is not well defined, although lens -crystallin have been shown to possess NADPH:quinone oxidoreductase activity in vitro (35).
The two isolated P4 cDNAs, of which the longest was 2175 bp, were different in one base pair at the 5Ј end and 12 bp at the 3Ј-nontranslated regions (not counting the polyadenylation tail). A 2.4-kilobase transcript was detected by RNA gel blot analysis using P4 as a probe (Fig. 3D). The P4-encoded polypeptide of 630 amino acids is preceded by an in-frame stop codon. Fig. 5 presents the sequence of the amino terminus of the P4 polypeptide. In this sequence, 250 amino acids are similar to those of a family of short-chain dehydrogenases. It shares, for example, 58.9% similarity and 38% identity with human NAD-FIG. 1. Diamide tolerance of yeast WYT cells. WYT strains containing either control plasmid pFL61 or pFL61 with Arabidopsis cDNAs P1, P2, P3, or P4 were tested for diamide tolerance in a semi-quantitative dilution test. Cultures of strains were grown to stationary or logarithmic phases, indicated as S and L to the right of the figures. Cell suspensions were serially diluted in 0.3% NaCl (fold of dilutions are indicated above the figures) and spotted as 10-l aliquots on plates without and with 1 mM diamide. Pictures of plates were taken after 3 days of incubation at 30°C. Two individual clones of yeast transformants with pFL61, pP1, pP2, pP3, and pP4 were analyzed.

FIG. 2. Diamide tolerance of DY strains and analysis of putative additive effects of P clones.
A, the DY strain was transformed with pFL61, pP1, pP2, pP3, and pP4. The diamide tolerance of stationary phase cells was tested as described in the legend to Fig. 1. As an additional control, the DY strain containing the pD3 cDNA isolated in a different screening 2 was used. B, to analyze the possible additive effects of P clones for diamide tolerance, expression cassettes containing P1 and P3 cDNA were recloned in another vector pGAD424, which provide LEU2 selection in yeasts. Double transformants of WYT yeasts were selected on plates without uracil and leucine. The diamide tolerance of stationary phase cells was tested as described in the legend to Fig. 1. dependent 15-hydroxyprostaglandin dehydrogenase, 53% similarity and 28% identity with NADP-dependent glucose dehydrogenase from Bacillus megaterium, and 50% similarity and 30% identity with the Arabidopsis 3-oxoacyl (acyl carrier protein) reductase. The amino acid residues, which are important for the enzymic activity of short-chain dehydrogenases (39,40), are well conserved in the P4 polypeptide (Fig. 5).
The carboxyl-terminal part of the P4 polypeptide, starting from the 284th amino acid, shows similarity with the P1 polypeptide (Table I). The highest similarity can be found to the L. amazonensis -crystallin, which is followed by the TED2 cDNA, the guinea pig lens -crystallin, and the rabbit protein AdRab-F ( Fig. 4 and Table I). Therefore, P4 is most probably a two-domain protein. Specific peptide sequences for targeting to different cell compartments could not be identified in P1, P2, or P4, suggesting that the proteins reside in the cytosol.
The gene copy numbers for P1, P2, and P4 were estimated on DNA gel blots of Arabidopsis genomic DNA digested with DraI, SacI, EcoRV, PstI, and HindIII. DNA extracted from two different land races of Arabidopsis was compared to find RFLP. The pattern of bands hybridizing with the P1 and P2 probes is highly similar, as predicted from the high DNA sequence identity of these two cDNAs (Fig. 3, A and B). Since single hybridizing bands of the same molecular weight were detected with P1 and P2 probes in DNA digested with SacI and PstI, these two genes are likely to be tightly linked on the chromosome. Additionally, weakly hybridizing bands were also detected that may indicate the presence of divergent genes in the Arabidopsis genome. The P4 gene is apparently unique in the Arabidopsis genome, because a simple hybridization pattern was observed (Fig. 3D). The P4 locus is polymorphic and an RFLP polymorphism was found, with four out of the five used restriction endonucleases tested, between Arabidopsis ecotypes Columbia and Landsberg erecta. Using the RFLP polymorphism obtained with DraI and DNA extracted from 98 Arabidopsis inbred recombinant lines (23), the P4 gene was mapped in the middle of chromosome 1 close to the RFLP marker m213 (mapping data are available from the laboratory of Dr. C. Dean, John Innes Institute, Norwich, UK).
The highest levels of P1, P2, and P4 steady-state mRNAs were detected in leaves and less in stems. Weak expression occurred in flowers and very weak expression in roots of Arabidopsis plants (Fig. 3, A, B, and D). From the sequence comparison, it was clear that the cDNAs P1, P2, and P4 corresponded to a rather divergent family of plant proteins, which were tentatively named together with zinnia TED2, plant -crystallins.
P3 Encodes a Putative Isoflavonoid Reductase-The longest P3 cDNA was 1116 bp in length and RNA gel blot analysis showed a transcript of 1.2 kilobase (Fig. 3C). The P3-homolo- FIG. 3. Analyses of gene copy number and tissue-specific expression. Genomic DNA of Arabidopsis was extracted from two different isolates, Landsberg erecta (L) and Columbia (C). DNA samples (1 g) were digested with DraI, SacI, EcoRV, PstI, and HindIII, electrophoresed on 0.8% agarose gel, and blotted to nylon membranes. After hybridization with 32 P-labeled DNA probes for P1 (A), P2 (B), P3 (C), and P4 (D) membranes were exposed to x-ray film. 35 S-Labeled HindIII DNA fragments of phage were used as a molecular weight markers (MW), their positions on the membranes are indicated to the left, numbers correspond to molecular masses expressed in thousands of base pairs. For the determination of sizes and tissue distribution of the mRNAs, total RNA was extracted from roots (R), leaves (L), stems (S), and flowers (F) of Arabidopsis plants. RNA gel blots were as described under "Materials and Methods." gous cDNAs from Arabidopsis are known as expressed sequence tags ATTS4709, ATTS4648, and 7315 (EMBL data bank accession numbers F13584, Z48456, and T44052, respectively). Clone P3 contains one continuous ORF, encoding a polypeptide of 310 amino acids with high similarity to isoflavonoid reductases (IFRs) from different plant species. Fig. 6 shows that the P3 polypeptide has 77% similarity (63% identity) with tobacco cDNA A662 and 70% similarity (55% identity) to alfalfa and chickpea IFRs. Single hybridizing bands were detected on DNA gel blots when P3 was used as a probe, indicating that P3 is a single-copy gene in Arabidopsis (Fig.  3C). Significant differences were found in expression in different plant organs. Expression was very high in stems, weaker in flowers, and more weak in leaves and roots (Fig. 3C).
Response to Oxidative Stress Conditions-The isolated cDNAs render yeast more resistant to diamide. Our hypothesis is that in plants these genes play a role in oxidative stress tolerance. To test this, the levels of the respective mRNAs in plants exposed to oxidative stress were analyzed by RNA gel blot analysis. As a positive control, the mRNA levels of the cytosolic ascorbate peroxidase (apx) were analyzed in the same RNA samples. Ascorbate peroxidases are major H 2 O 2 -scavenging enzymes in chloroplasts and cytosol (44). The apx mRNA levels and enzyme activities have been shown to increase in plants exposed to stress conditions, such as drought, salt stress, and treatment with paraquat, which result in a prooxidant state in cells (45).
Oxidative stress was imposed by the following treatments: the hydroperoxides, hydrogen peroxide (H 2 O 2 ), and t-BOOH were used to induce direct oxidation of cell constituents. Treatment with diamide was used to induce pro-oxidant conditions through depletion of intracellular glutathione. The herbicide paraquat (methyl viologen) and the quinone, menadione, are redox-cycling drugs. In the cell they serve as efficient electron acceptors from electron transport chains. Reduced paraquat and semi-quinones are readily oxidized by oxygen, leading to the formation of O 2 . anions. Then, the cycle of reduction-oxidation is repeated, a process called redox cycling (1). Furthermore, potent antioxidants such as reduced glutathione, dithiothreitol (DTT), and N-acetylcysteine, may cause the activation of plant oxidative stress defense genes (46). Therefore, we analyzed the expression of the isolated genes after the infiltration of plants with DTT. The chemicals mentioned above were applied by infiltration into Arabidopsis plants. Although infiltration by itself caused activation of apx, P1, P2, and P4 genes, this method was used to eliminate problems of drug uptake through roots or epidermis.
As shown in Fig. 7, apx mRNA levels steadily increased up to 8 h after infiltration of water. The strongest response was observed in response to paraquat treatment; accumulation of mRNA was detected after 15 min of treatment and reached a plateau after 30 min, which lasted at least 8 h. Transiently higher steady-state RNA levels (compared to control) were observed in plants treated with t-BOOH, diamide, and menadione, with the highest mRNA levels approximately 1 h after infiltration. The apx gene was also induced by DTT and levels of mRNA detected in H 2 O-infiltrated plants after 8 h were already observed after 15 min following DTT infiltration. No differences were observed between control and H 2 O 2 -infiltrated plants.
P1 and P2 genes responded similarly, although, considering the high sequence homology, the detected hybridization signals might be derived from P1 and P2 mRNA together. Like apx, P1 and P2 mRNAs were induced by water infiltration. Rapid in-TABLE I Similarity between -crystallins Names of sequences: L.c., AdRab-F from rabbit intestine; L.a., -crystallin-like protein from trypanosome L. amazonensis; zeta, guinea pig -crystallin; TED2, zinnia TED2 cDNA. For references, see Fig. 4. For P4, only the C-terminal part was used. The protein identity and similarity was calculated by the Genetics Computer Group (GCG, Madison, WI) program (version 7.3).  4. The P1, P2, and P4 cDNA are homologous to the -crystallin-like proteins. The amino acid sequences of the ORF encoded by the P1, P2, and P4 cDNAs were aligned to the sequences of protein AdRab-F from rabbit intestine (L.c.) (Boll and Mantei, PIR accession number S47093), -crystallin-like protein from trypanosome L. amazonensis (L.a.) (32), guinea pig -crystallin (zeta) (33), and zinnia cDNA TED2 (TED2) (34). The amino acids, which are identical in P1 ORF compared with other amino acid sequences are indicated by dots and boxed. In addition, amino acids identical to P4 ORF are bold and shaded. Dashes represent gaps introduced for the better alignment. duction of these genes was found after menadione treatment, with detectable increases over the control after 15 min and a plateau of expression after 30 min. The response to another redox-cycling drug methyl viologen (paraquat) was detectable after 15 min and the accumulation of P1 and P2 mRNAs after 8 h was the highest detected (Ϸ10-fold). The accumulation of P1 and P2 mRNAs in t-BOOH-treated and diamide-treated plants was faster as compared to the control, however, expression was lower after 8 h. Hydrogen peroxide slightly induced P1 and P2 genes after 8 h. DTT repressed accumulation of P1 and P2 mRNAs.
The P3 mRNA levels were unresponsive to water infiltration, whereas expression weakened in t-BOOH-treated and DTTtreated plants. The P3 gene was slowly induced by paraquat, hydrogen peroxide, diamide, and an Ϸ3-fold increase in mRNA levels is detected after 8 h. Menadione did not affect P3 mRNA levels.
The P4 gene was induced in the control infiltration with water. H 2 O 2 , t-BOOH, and DTT prevented this water-induced mRNA accumulation, whereas diamide, menadione, and methyl viologen had little effect on P4 gene expression.
The same membranes were also hybridized with a tobacco cDNA probe of the chlorophyll a/b-binding protein (cab). All of treatments lowered the cab gene expression (data not shown).

DISCUSSION
In this work we have isolated four Arabidopsis cDNAs which improve the tolerance of yeast cells to the thiol oxidizing drug diamide. Plant -crystallins and IFR-like cDNAs were consistently recovered in our screening for diamide tolerance of the yap1 Ϫ yeast mutant. Furthermore, we have found that the cDNAs isolated also improved the survival of YAP1 yeast strain, DY which has a normal response to oxidative stress. The mechanism of protection is apparently rather specific for diamide. Efforts to find cross-tolerance to other drugs were not successful, however, the choice of stresses as well as possible growth conditions were not exhaustively tested in this study. This observation is not in disagreement with the data obtained by multicopy plasmid screening in yeasts. Significant cross-tolerance was usually observed when genes encoding regulatory proteins, like YAP1 or YAP2 were overexpressed (17,47,48). The tolerance provided by enzymes is rather limited to the particular stress agent (17,48).
Although we cannot entirely exclude the direct detoxification of diamide by -crystallins or IFR, our expression analysis suggest that the isolated cDNAs encode enzymes with a distinct role in stress response. The genes encoding P1 and P2 were rapidly induced by drugs which induce redox cycling, and in general their expression is very similar to the expression of the well characterized plant antioxidant protein, ascorbate peroxidase (45).
The original interest in -crystallin stems from the fact that a mutation in the -crystallin gene is associated with formation of an autosomal dominant congenital cataract in guinea pigs (49). Since cataract development is thought to result from oxidative damage of lens proteins (1), it was logical to presume that high levels of -crystallin is a species-specific adaptation to control the formation of active oxygen species. It was further found that detoxification of hydrogen peroxide is more efficient in lens of guinea pigs compared to rat lens, which have only small amounts of -crystallin (49). The expression of -crystallins is also induced in rat lens under oxidative stress conditions (50).
-Crystallin specifically binds NADP(H) and is a novel NAD-PH:quinone oxidoreductase (35,49). Based on biochemical data, Rao and Zigler (49) hypothesized that one of the important functions of -crystallins is an initiation of NADPH/NADP redox cycling, which in turn is coupled to the up-regulation of the hexose monophosphate shunt. Hexose monophosphate shunt is the principal source of NADPH in cells and one of the components of antioxidant defense, because NADPH is needed for the reduction of oxidized glutathione and ascorbic acid, whose formation greatly increases under oxidative stress (51)(52)(53).
The P3 gene was slower in response to paraquat and H 2 O 2 and was not induced by water infiltration. Considering, the importance of H 2 O 2 metabolism in processes of cell wall crosslinking and lignin biosynthesis (54) and possible xylem-specific expression of the P3 gene, its product may play a role in AOS homeostasis in specific cell types. The P3 clone is highly homologous to isoflavonoid reductases. IFRs are believed to be specifically involved in the biosynthesis of isoflavonoid phytoalexins in Leguminosae (55). However, IFR cDNA homologs have been isolated from several species, which do not synthe-  (36), Arabidopsis 3-oxoacyl (acyl carrier protein) reductase (oxoacyl) (37), and glucose dehydrogenase from B. megaterium (glucose) (38). The amino acids, which are identical in P4 and other short-chain dehydrogenases, are indicated by dots and boxed. The amino-terminal part of the oxoacyl dehydrogenase is a chloroplast transit peptide. The amino acid residues, important for the enzymic activity of the short-chain dehydrogenases are indicated with lozenges. The arrowhead above the P4 sequence shows the beginning of the P4 domain with homology to the -crystallins. To improve alignment, gaps indicated by dashes have been introduced. size isoflavonoid phytoalexins and the tobacco IFR-like protein expressed in E. coli did not show any enzymatic activity (41,56). The IFR activity of legume proteins suggests that a whole IFR-like family are oxidoreductases utilizing NAD(P)H as a co-factor and the use of isoflavonoids as electron acceptors is likely to be an acquired enzymatic activity in legumes.
The expression of the P4 gene was only elevated in control infiltration with water; furthermore, its expression was suppressed by diamide and t-BOOH. Apparently, P4 protein, which is composed of two domains has a rather specific function in plant cells. Although it was isolated in our screening most likely due to the presence of the carboxyl-terminal -crystallin domain, it also possesses an amino-terminal domain which shows homology to the short-chain dehydrogenases. The arrangement of the two domains in the putative P4 protein is quite interesting. If enzymatically active, the amino-terminal dehydrogenase domain could, through the oxidation of a substrate molecule, generate NADPH, from which electrons may then be transferred to another acceptor (quinones?) by the -crystallin domain. It is tempting to propose that, short -crystallins encoded by the P1 and P2 may in fact compete with P4 for the interaction with some plant protein(s), and in this way are lowering P4 activity.
In this work, we have identified 4 new Arabidopsis cDNAs, which, when expressed under control of a yeast promoter on a multicopy plasmid provide better tolerance of yeasts to thiol oxidizing drug diamide. It might not be coincidental that plant -crystallin homologs were isolated by such a screen. Loss of -crystallin activity in guinea pigs leads to cataract formation and this effect can be mimicked by diamide (57). The putative NAD(P)H oxidoreductase activities of proteins encoded by the isolated cDNAs suggest the importance of NAD(P)H/NAD(P) homeostasis in plant cell antioxidant defense. The use of other drugs and chemicals in similar functional screening of plant cDNA libraries may provide an important tool for the identification of plant genes involved in stress response and defense, which would not be amenable to isolation by conventional screening.