AtRAD1, a plant homologue of human and yeast nucleotide excision repair endonucleases, is involved in dark repair of UV damages and recombination

Authors


*For correspondence (fax +32 2 776 4676; e-mail bruno.tinland@ea.monsanto.com).
†Present address: Department of Biochemistry and Molecular Biology, Facultat de Químiques, Martí i Franquès 1, Universitat de Barcelona, E-08028 Barcelona, Spain.
‡Present address: Institute of Molecular Biology and Biochemistry, Michurin St. 86, 480012 Almaty, Kazakhstan.

Summary

Plants are unique in the obligatory nature of their exposure to sunlight and consequently to ultraviolet (UV) irradiation. However, our understanding of plant DNA repair processes lags far behind the current knowledge of repair mechanisms in microbes, yeast and mammals, especially concerning the universally conserved and versatile dark repair pathway called nucleotide excision repair (NER). Here we report the isolation and functional characterization of Arabidopsis thaliana AtRAD1, which encodes the plant homologue of Saccharomyces cerevisiae RAD1, Schizosaccharomyces pombe RAD16 and human XPF, endonucleolytic enzymes involved in DNA repair and recombination processes. Our results indicate that AtRAD1 is involved in the excision of UV-induced damages, and allow us to assign, for the first time in plants, the dark repair of such DNA lesions to NER. The low efficiency of this repair mechanism, coupled to the fact that AtRAD1 is ubiquitously expressed including tissues that are not accessible to UV light, suggests that plant NER has other roles. Possible ‘UV-independent’ functions of NER are discussed with respect to features that are particular to plants.

Introduction

Plants are unique in their requirement of sunlight for photosynthesis, and due to their sessile lifestyle they are exposed to the harmful effects of solar UV radiation. Direct incidence of UV light inflicts oxidative lesions and cross-links on DNA, namely cyclobutane pyrimidine dimers (CPDs) and (6-4)-photoproducts, but it also induces significant changes in plant growth and morphological and biochemical alterations ( Jansen et al. 1998 ; Vonarx et al. 1998 ).

Through evolution plants have acquired two main protective strategies to cope with UV effects. Firstly damage avoidance which provided by shielding through the production of UV-absorbing compounds, epicuticular waxes and cuticular structures ( Bornman et al. 1997 ). Secondly, excision or direct reversion by photoreactivation of UV-induced DNA lesions.

Photoreactivation has been demonstrated from bacteria to higher eukaryotes. By photoreactivation, UV-induced pyrimidine dimers are restored to their native base form in a light-dependent, error-free and economic manner mediated by the action of photolyases, which are structurally and functionally conserved ( Yasui & Eker 1998). Excision repair strategies have been characterized in Escherichia coli, yeast and mammalian systems, but their presence in plants has not yet been extensively described. Excision is accomplished through specific base excision repair enzymes or by a more versatile pathway termed nucleotide excision repair (NER). NER components recognize a broad spectrum of DNA damage, from chemical alterations to photochemical lesions such as CPDs and (6-4)-photoproducts ( Sancar 1996).

In plants, photoreactivation efficiently and quickly removes the bulk of UV-induced CPDs and (6-4)-photoproducts ( Britt 1999). This repair depends on the quality, timing and quantity of photoreactivating light, and on damage levels ( Pang & Hays 1991; Stapleton et al. 1997 ; Sutherland et al. 1996 ).

Plants also perform light-independent repair which supports the existence of a pathway equivalent to the well characterized yeast and mammal NER machinery. Dark repair of ultraviolet-induced pyrimidine dimers was measured in carrot protoplasts ( Eastwood & McLennan 1985; Howland 1975) and in Arabidopsis and alfalfa seedlings ( Pang & Hays 1991; Quaite et al. 1994 ). Light-independent repair has also been reported in soybean chloroplasts and leaves ( Cannon et al. 1995 ; Sutherland et al. 1996 ) and in different rice cultivars ( Hidema et al. 1997 ). It was also documented in wheat leaf tissue for both CPDs and (6-4)-photoproducts ( Taylor et al. 1996 ).

Classical genetic techniques led to the identification of light-independent repair Arabidopsis UV-sensitive mutants. Uvr1, uvr5 and uvr7 are deficient in dark repair of (6-4)-photoproducts ( Britt et al. 1993 ; Jenkins et al. 1995 ; Jiang et al. 1997 ), and the uvh1 mutant is defective in repair or tolerance mechanisms that confer protection to different DNA lesions ( Harlow et al. 1994 ).

Despite these biochemical and genetic data in favour of some sort of NER in plants, limited information is available concerning the molecular characterization of plant gene products actively involved in dark repair.

An Arabidopsis XPB ( Ribeiro et al. 1998 ) and a plant homologue of human ERCC1 NER endonuclease ( Xu et al. 1998 ) have been recently cloned. The human heterodimeric complex ERCC1-XPF mediates the endonucleolytic incision at the 5′ side of the DNA lesion ( de Sijbers et al. 1996 ). In humans, inheritable defects in XPF and other components of the NER system are associated with hypersensitivity to sunlight and severe incidence of skin cancer as prominent features of xeroderma pigmentosum (XP) and Cockayne syndrome diseases ( Thompson 1998).

In this study, we describe the isolation and functional characterization of AtRAD1 from Arabidopsis thaliana, a cDNA clone homologue to human XPF and yeast RAD1 NER endonucleases. Transgenic plants depleted in their endogenous AtRAD1 activity were generated and shown to remove CPD lesions less efficiently than wild type. They are more sensitive to UV light and to the cross-linking agent mitomycin C, suggesting an active role of AtRAD1 in a plant NER-like repair pathway and recombination.

Results

AtRAD1 cDNA codes for the plant homologue of yeast RAD1 and human XPF

A RAD1-like DNA fragment derived from Zea mays (EST T25298) was used to screen several Arabidopsis cDNA libraries. Once assembled, the compilation of the sequences from the different positive clones yields an ORF that we have named AtRAD1. The AtRAD1 cDNA sequence is 3151 bases in length with a predicted 2868 base ORF that encodes a 956 amino-acid protein with a molecular mass of 109 kDa and a pI of 8.02 (GenBank accession number AF089003).

The comparison of the deduced amino-acid sequence ( Fig. 1) revealed that the AtRAD1 protein is the plant homologue of animal and yeast endonucleolytic enzymes involved in recombination processes and in the excision repair of UV-induced damages in DNA. AtRAD1 displays an overall 59% sequence similarity to the human protein XPF and Schizosaccharomyces pombe RAD16, and a 53% similarity when aligned to Saccharomyces cerevisiae RAD1 and Drosophila melanogaster MEI-9. No significant overall homology was found with other proteins in the accessible databases.

Figure 1.

Sequence alignment of Arabidopsis AtRAD1 with other DNA-repair enzymes.

Comparison of the deduced amino-acid sequence of AtRAD1 from A. thaliana to human XPF, Saccharomyces cerevisiae RAD1 ( Reynolds et al. 1987 ), Schizosaccharomyces pombe RAD16 ( Carr et al. 1994 ), and Drosophila melanogaster MEI-9 ( Sekelsky et al. 1995 ). Identical amino acids on black background; similar amino acids on gray. Putative nuclear localization signals in AtRAD1 are denoted by horizontal lines below the sequences. The conserved nuclease ERKX2SD signature and the additional conserved 5′ upstream aspartate (D) amino acid are indicated by asterisks. The double horizontal line above the sequences indicates the C-terminal HhH DNA-binding domain.

Although the homology between AtRAD1 and the above-mentioned enzymes is distributed throughout their length, it is especially remarkable at the C-terminal end, being most pronounced between residues 612 and 910, a region which comprises the RAD10 binding domain of S. cerevisiae RAD1 ( Bardwell et al. 1993 ) and the ERCC1-binding domain of XPF ( de Laat et al. 1998 ). This region allows the formation of the heterodimeric protein complexes, RAD1–RAD10 and ERCC1-XPF, with a structure-specific endonuclease activity in charge of performing the 5′ incision in nucleotide excision repair events in yeast and humans.

In the N-terminal half of AtRAD1, several leucine-rich repeats (positions 274–295 and 281–302) that may be involved in tight and specific protein–protein interactions ( Schneider & Schweiger 1991) are conserved. Besides the nuclear targeting consensus signal RR QLDPIWHTLG KRTKQ at position 255, the relatively poorly conserved central region harbors additional putative nuclear localization sequences consistent with the expected AtRAD1 nuclear functionality. The strikingly conserved motif ERKX2SD (positions 767–773) along with an additional conserved aspartate (D) at position 757, both features identified in the RAD1/XPF endonuclease superfamily ( Aravind et al. 1999 ), are present in the predicted AtRAD1 protein. Likewise, the C-terminal HhH DNA-binding domain ( Doherty et al. 1996 ) is also conserved in the AtRAD1 polypeptide. In the case of AtRAD1, a glutamine (Q) replaces the serine residue (S) in the ERKX2SD signature. Nevertheless, the overall domain in which that motif is located significantly preserves the negatively charged residues which are likely to function as nucleophiles in catalysis and in metal ion co-ordination, as well as the nearby prolines (P) and glycines (G) that may be determinants for protein folding. Because of all these sequence properties, we propose the AtRAD1 polypeptide to be functional in a DNA repair context.

AtRAD1 is a single copy gene in A. thaliana

Compilation of data available from the Arabidopsis Sequence Genome Project of the Kazusa DNA Research Institute and our cDNA sequence allowed us to propose the genomic organization of AtRAD1 ( Fig. 2a). The AtRAD1 gene has been mapped in chromosome V between positions 15 000 and 16 000 kb and included in the identifier MEE6.16 (www.kazusa.or.jp). Interestingly, the UV-hypersensitive Arabidopsis mutant uvh1, defective in a repair or tolerance mechanism against different types of DNA damage ( Harlow et al. 1991 ; Jenkins et al. 1995 ) has also been mapped in a close region of chromosome V relative to the SSPL markers nga 106, nga 139 and nga 76 (positions spanned from 5600 to 11 170 kb). AtRAD1 genomic sequence contains eight introns, quite above the average intron number per gene in Arabidopsis ( Delseny & Cooke 1998). This relatively high number of non-coding sequences for AtRAD1 resembles that of the S. pombe RAD16 gene in which seven introns, an unusually high intron number for the S. pombe genome, were identified ( Carr et al. 1994 ). The relative position of these introns between AtRAD1 and RAD16 proteins is in general not conserved, although two of them are located at very similar sequence points (data not shown). The size of the eight AtRAD1 introns ranges between 46 and 113 bp in agreement with the small Arabidopsis intron size reported until now ( Delseny & Cooke 1998).

Figure 2.

Analysis of AtRAD1 gene.

(a) Genomic organization of AtRAD1. Genomic sequence corresponding to mee6.16 identifier was compared to the AtRAD1 cDNA sequence. Empty boxes, exons; black lines, introns. The restriction sites in the cDNA sequence corresponding to the enzymes used in the Southern analysis (see below) and the DNA probe are also shown. All numbers refer to positions in the AtRAD1 cDNA sequence.

(b) Southern analysis of AtRAD1 gene. Molecular sizes of marker DNA are indicated on the left. The 3.5 kb fragment obtained with XhoI has the expected size, including the intron sequences.

Southern blot analysis of the A. thaliana genome digested with several restriction enzymes ( Fig. 2b) revealed a simple hybridization pattern. This indicates that AtRAD1 is the only RAD1 homologue in the Arabidopsis genome.

AtRAD1 is present in all Arabidopsis mature tissues and in shoot meristems

The spatial expression pattern of AtRAD1 in Arabidopsis was addressed in detail by in situ hybridization. As shown in Fig. 3, hybridization with an AtRAD1 antisense probe yielded a homogeneously distributed signal in all mature tissues analyzed: flower (a), shoot (e), root (g) and leaf (i). This indicates that the AtRAD1 transcript accumulates in differentiated tissues where it is required for cell growth. No evidence of transcripts was observed in any of these tissues hybridized with the sense AtRAD1 RNA as control ( Fig. 3b,d,f,h,j). The strong signal obtained in pollen grains and xylem vessels is not gene-specific but is dependent on the biochemical and structural properties of such elements ( Jackson 1991), as it is also revealed and detected in the control preparations elaborated with the AtRAD1 sense probe ( Fig. 3b). AtRAD1 is also expressed in tissues undergoing cell division, as a strong signal was detected in Arabidopsis shoot apical meristems ( Fig. 3c). Interestingly, AtRAD1 is expressed in reproductive tissue: a signal can be seen in the haploid cells inside the embryo sac ( Fig. 3a). Therefore the AtRAD1 polypeptide is present not only in those tissues directly exposed to sunlight and consequently to UV light, such as flower, shoot and leaf, but also in light-protected tissues such as root and shoot apical meristem.

Figure 3.

Localization of AtRAD1 in wild-type Arabidopsis tissues.

Longitudinal sections of wild-type flower (a), root (g) and shoot apical meristem (c), and cross-sections of shoot (e) and leaf (i) were hybridized with the antisense AtRAD1 probe. Sections (b,d,f,h,j) were hybridized with the sense AtRAD1 probe as negative controls. Mature tissues were prepared from 8-week-old flowering plants grown in soil, except roots, obtained from 10-day-old liquid cultures. Shoot meristems were prepared from 5-week-old plants grown in soil. ac, Anther chamber; es, embryo sac; ov, ovary; ovu, ovule; pe, petal; ph, phloem; po, pollen grains; sam, shoot apical meristem; lp, leaf primordium; se, sepal; st, stamen; xy, xylem; yl, young leaf.

AtRAD1 partially complements defects of the S. pombe rad16 mutant in repair of UV damage and in mating-type switching

To ascertain whether the structural conservation among AtRAD1 and its human and yeast homologues ( Fig. 1) was conserved at the functional level, we proceeded to overexpress AtRAD1 in S. pombe rad16 mutant cells, as both UV sensitivity and recombination events can be assayed in this organism. Furthermore, the sequence comparison revealed that AtRAD1 shares the highest sequence homology (59%) with the S. pombe RAD16 protein. The rad16 fission yeast mutant is hypersensitive to UV due to defects in the NER pathway, whereas its swi phenotype is consistent with a deficiency in the recombination events involved in mating-type switching ( Carr et al. 1994 ; Schmidt et al. 1989 ).

AtRAD1 corrects S. pombe rad16 UV sensitivity

Arabidopsis AtRAD1 full-length cDNA was introduced in the expression vector pREP1 ( Maundrell 1993). Wild-type heterothallic S. pombe strain (h+ wt) and rad16 mutant cells (h+rad16) were transformed and their survivability upon UV treatment was determined ( Fig. 4a). As controls, the same transformations were conducted using the empty pREP1 vector. The overexpression of AtRAD1 in the wild-type strain did not alter cell survivability. However, a remarkable restoration of resistance to UV was observed for S. pombe rad16 mutant transformed with pREP1 harboring AtRAD1, as a 10-fold increase in cell survivability was observed as compared with the controls ( Fig. 4a). The correction was more pronounced at higher (80 J m−2) than at lower (20 J m−2) UV doses.

Figure 4.

Complementation assays of Schizosaccharomyces pombe rad16 mutant by overexpression of AtRAD1.

(a) Correction of S. pombe rad16 mutant UV sensitivity. S. pombe heterothallic strains were transformed as follows: (♦) h+ wt + pREP1; (●) h+rad16 + pREP1; (▵) h+ wt + pREP1/AtRAD1; (▪) h+rad16 + pREP1/AtRAD1. Error bars represent SD of the means of two or three independent experiments.

(b) Complementation of S. pombe rad16 mutant switching defect. The percentages of ascospores (mean ± SD) are shown for wild-type homothallic strain (h90 wt) and for rad16 mutant strain (h90rad16) transformed with pREP1 vector with or without AtRAD1 cDNA.

AtRAD1 partially corrects S. pombe rad16 mating-type switching defect

The rad16 mutation has pleiotropic effects in S. pombe. As well as conferring UV hypersensitivity to cells, it manifests deficiency in mating-type switching ( Schmidt et al. 1989 ), as rad16 mutant is affected in the proper resolution of the gene conversion step involved in this process. In homothallic strains (h90) that have a reduced efficiency of mating-type switching (swi mutants), the rates of conjugation and meiosis can occur less frequently. Thus in swi mutants the percentage of spores in a colony is an appropriate measure of accurate mating-type switching.

To assay whether AtRAD1 could restore the switching deficiency of S. pombe rad16 mutant, wild-type homothallic S. pombe strain (h90 wt) and rad16 mutant cells (h90rad16) were transformed with the pREP1 vector in which the full-length AtRAD1 had been cloned. As controls, the respective transformations were performed using pREP1 alone. Overexpression of full length AtRAD1 in the wild-type strain did not alter the frequency of sporulation as compared to the control ( Fig. 4b). However, overexpression of AtRAD1 in the S. pombe rad16 mutant could partially compensate for the defect in mating-type switching, as the frequency of sporulation increased by a factor of 2.4, a statistically significant increment (χ2 = 54.31, P < 0.0001).

Plants depleted of their endogenous AtRAD1 activity are more sensitive to UV light

To address the biological functions of AtRAD1 in Arabidopsis, antisense plants were generated. A fragment ≈0.6 kb from AtRAD1 cDNA comprising nucleotides 2104–2679 was used to design a construct encoding antisense RNA to deplete cells of their endogenous AtRAD1 and thus abolish its functionality. Seven different antisense lines were obtained. They were analyzed by Southern blotting to confirm their transgenicity, and by Northern blotting to evaluate their respective expression levels for the transgenic antisense RNA (data not shown).

The antisense lines were examined for their sensitivity to UV irradiation by quantifying the inhibition that a UV pulse inflicted on root growth. In contrast to the effects of UV damage on already formed tissues such as leaves, the inhibition of root elongation on UV treatment is easy to quantify ( Britt et al. 1993 ; Jiang et al. 1997 ; Masson et al. 1997 ). Three-day-old seedlings from the different AtRAD1 antisense lines grown on vertically placed agar plates were subjected to different UV doses (0.2–6 kJ m−2) and allowed to recover in either light or dark conditions. The growth of the seedlings' main root was monitored daily up to day 7. New root growth after irradiation was measured and normalized against the root growth observed for the unirradiated controls. Irradiated wild-type plants and antisense lines kept in the dark upon UV treatment showed a very low recovery that did not result in measurable root growth differences among the different plant populations. This is probably due to a strong UV-induced reduction of the overall fitness of all plantlets, leading to a general growth arrest. Therefore recovery in light and irradiation at 1–2 kJ m−2, which provided the clearest differences in root growth between the transgenic plants and the wild type, were chosen as working conditions to characterize two different lines in more detail. As shown in Fig. 5(a), both antisense lines (A5 and A12) exhibited a general tendency to higher UV sensitivity as compared with the controls at irradiation doses of 1 and 2 kJ m−2. This was visible although, due to the experimental conditions, the population included one-fourth negative segregants. That the numbers are statistically significant indicates greater differences between wild-type and AtRAD1 antisense plants in a homozygote population.

Figure 5.

AtRAD1 antisense Arabidopsis plants display a higher sensitivity to DNA-damaging treatments than wild type and are deficient in dark CPD repair.

(a) Effect of UV-C light on main root growth. A5 and A12 lines express a 0.6 kb fragment from AtRAD1 ORF in antisense orientation. WT, wild type. Error bars, SD of the means from three independent experiments. The difference between WT and transgenic plants was estimated by Student's t-test. Statistical significance, P < 0.05 in each experiment except for A12 at 2 kJ m−2 where the following values were obtained: P= 0.223, P= 0.061, and P= 0.117.

(b) Dark CPD repair. Wild-type (WT) and antisense plants (A2 and A12) were UV-irradiated at 100 J m−2 and allowed to recover in dark for the time periods indicated. Results from six independent experiments. Statistical significance (A12), P= 0.084 and P= 0.017 at 48 and 96 h, respectively.

(c) Sensitivity of antisense line A12 to mitomycin C (MMC) at the doses indicated compared with wild type (WT). Four-day-old seedlings were transferred to liquid medium supplemented with MMC at different concentrations. The picture was taken 10 days after transfer.

Antisense AtRAD1 plants are impaired in dark CPD repair

We investigated the proficiency of the antisense Arabidopsis lines to repair UV-induced DNA damage using the conventional ELISA method with specific antithymidine dimer antibodies. When Arabidopsis seedlings were UV irradiated to generate CPDs with the same UV doses that had been used for the root growth assay (1–2 kJ m−2), even for the wild-type plants no dark repair could be detected (data not shown). We speculated that the proportion of repairable CPDs compared to the total amount of induced thymidine dimers at those high-UV values was too low to be monitored. Therefore we applied lower UV doses to increase the ratio of removed CPDs with respect to the total amount of inflicted CPDs, so that repair could be quantified experimentally. Different UV doses and different amounts of DNA were tested to establish the appropriate UV levels that induce a measurable amount of CPDs without overloading the cellular repair machinery. This resulted in working UV doses of 50–200 J m−2 and 0.5–1 μg of DNA per well as the most convenient values. Detailed experiments were conducted in which CPDs were induced by applying a UV dose of 100 J m−2. Different amounts of DNA (100 ng to 2 μg) were plated to ensure a linear dependence between the ELISA signal and DNA concentration. The relative amount of CPDs per ng DNA was calculated and normalized against non-repaired DNA (taken as 100%).

We investigated two independent antisense lines (A2 and A12) for their proficiency in repairing CPDs in a light-independent reaction. Firstly, CPD levels were evaluated from wild-type and both antisense plants that had been irradiated and allowed to recover in the presence of light (data not shown). No remaining CPDs were detected for any of the three lines at the first time point measured 24 h after UV treatment, indicating that the fast and efficient light-dependent DNA repair mechanism was functional in the AtRAD1 antisense plants as in the wild type. This fast repair under photoreactivating conditions is well documented for many organisms, including higher plants ( Taylor et al. 1996 ). Secondly, we investigated the kinetics of CPD dark removal in wild-type and antisense plants. The results are presented in Fig. 5(b). In contrast to photoreactivation, CPD dark removal was remarkably poor, even in the wild type, repairing only 45% of the UV-induced damage after 96 h recovery. However, the antisense lines A2 and A12 showed a pronounced decrease in efficiency of CPD dark repair as only about 25% of the induced damages could be removed in the same recovery period, suggesting that AtRAD1 antisense plants are deficient in CPD dark repair. We therefore conclude that the phenotypical impairment observed in AtRAD1 antisense plants in the root growth test is correlated with a biochemical deficiency in dark removal of UV-induced lesions.

Recombinational role of AtRAD1

AtRAD1 antisense lines A2, A5 and A12 were analyzed for their growth behaviour in presence of mitomycin C (MMC), a potent chemical agent that cross-links the complementary strands of DNA. Since the repair of such interstrand cross-links is thought to require recombinational repair ( de Sijbers et al. 1996 ), this assay aims to provide some indications of the eventual role of AtRAD1 in recombination processes. Doses between 0 and 20 mg l−1 were used. The results for the wild type and the A12 antisense line are presented in Fig. 5(c). Wild-type plants do not exhibit symptoms of slight sensitivity to MMC, in terms of the root system and leaf size, up to a concentration of 5 mg l−1. In contrast, 2.5 mg l−1 MMC strongly inhibited the growth of AtRAD1 antisense plants, which showed an important reduction in the size of the leaves. These antisense plantlets were already severely arrested in growth at MMC doses of 5 mg l−1 and became chlorotic at lower MMC concentrations than wild-type seedlings, which indicates that antisense plantlets are impaired in recombinational repair processes.

Discussion

In this study, the AtRAD1 cDNA from A. thaliana was isolated. AtRAD1 codes for a protein with a remarkable amino-acid sequence similarity to the DNA excision repair enzymes human XPF, S. pombe Rad16, S. cerevisiae RAD1 and the D. melanogaster MEI-9. In S. pombe rad16 mutant cells, AtRAD1 is able to partially complement sensitivity to UV ( Fig. 4a) as well as the recombinational defect in mating-type switching ( Fig. 4b). This implies that AtRAD1 has the potential to repair UV-induced damage by excision and to mediate in recombinational events, as shown for yeast RAD1 ( Davies et al. 1995 ). It is noteworthy that excision repair is a key process in S. pombe as, in contrast to S. cerevisiae, fission yeast has no photoreactivation pathway ( Griffiths & Carr 1998). The highly complex and co-ordinated machinery of NER, in which quite a large number of proteins are involved, may impose severe constraints when full correction of a defective phenotype is pursued in a heterologous system. In this sense the partial correction of S. pombe rad16 mutant defects by overexpressing AtRAD1 is an invaluable indication of the AtRAD1 function in vivo.

The biological role of AtRAD1 was addressed more precisely by generating transgenic Arabidopsis plants. The AtRAD1 antisense plants showed altered DNA repair activities, since they were more sensitive than the wild type to UV treatment ( Fig. 5a), indicating that they are down-regulated in their NER pathway. This was confirmed biochemically by monitoring their inability to remove CPDs efficiently in dark conditions ( Fig. 5b). Our functional analyses demonstrate for the first time that the dark repair already reported in plants, but so far not characterized at the molecular level, can (at least in part) be assigned to nucleotide excision repair.

Our results indicate that transgenic plants down-regulated in their endogenous AtRAD1 activity are more sensitive than the wild type to the interstrand cross-linking agent MMC ( Fig. 5c). This observation, along with the complementation of the mating-type switching defect in S. pombe rad16, provides evidence for an active role of AtRAD1 in recombination revents. In plant development the germlines are defined at a later stage than for animals, therefore any modification in somatic plant cells has the potential to exert an effect if transmitted to the progeny. It is essential to assess in detail the involvement of plant NER enzymes in recombination, as it is known that intragenic recombination contributes to promoting sequence diversification in plant resistance gene evolution, a strategy that enriches the host's responses to counteract pathogenic pressure ( McDowell et al. 1998 ). The recombinational activity of AtRAD1 has interesting biological implications, as recombination events are thought to be tightly regulated and mechanistically important for applied purposes such as those related to the generation of transgenic plants by T-DNA transformation ( Puchta & Hohn 1996; Tinland 1996).

An important point is the ubiquitous expression of AtRAD1 in all the tissues analyzed ( Fig. 3), seen even in those that are protected from UV light. The wide distribution of AtRAD1 expression in Arabidopsis, both in UV-exposed and unexposed tissues, in already differentiated and in meristematic ones ( Fig. 3), could account for functions other than the elimination of UV-induced pyrimidine dimers. This is further supported by the finding that CPDs were poorly removed after a long recovery period in wild-type plants, less than 50% in 96 h, most of it being repaired during the first 48 h ( Fig. 5b). Plant NER may play an important role in removing damage that is (i) not accessible to photolyases; or (ii) not pyrimidine dimers. (i) It is known that in other organisms NER is tightly connected to the transcription machinery and is the predominant pathway to remove UV lesions from regions that are being transcribed, allowing the excision of damage hidden by stalled transcription factors ( Aboussekhra & Thoma 1999; Livingstone-Zatchej et al. 1997 ). (ii) Due to special plant metabolic and lifestyle properties, substrates different from pyrimidine dimers might be the target for NER. Oxidative lesions could be such substrates. Activated oxygen species are harmful to DNA and can be generated intracellularly during respiration and photosynthesis, and as a result of air pollutants such as ozone. Also, oxidative stress is an important genotoxic agent against which natural defences can be overwhelmed in the course of the hypersensitive response triggered during plant–pathogen interactions ( Dangl et al. 1996 ). These circumstances can inflict oxidative damage in plant cells if their scavenging defences become overloaded ( Britt 1996). It will be the next challenge to evaluate in detail different types of DNA damages as potential targets for NER in plant cells.

Concerted interactions between plant nucleotide excision repair components and other pathways can be envisaged, so that NER would not only be an alternative to photoreactivation, but also a component of a broader network. In this sense, recent reports on S. cerevisiae have proved the physical interaction between DNA mismatch repair protein MSH2 and several NER enzymes ( Bertrand et al. 1998 ), as well as a synergy between base and nucleotide excision repair to confer protection against DNA methylating damage ( Xiao & Chow 1998). Overlapping functions of some components of the base-excision and nucleotide-excision repair pathways have recently been reported in human cells. Concretely, XPG protein promotes binding to damaged DNA of the DNA glycosylase–AP lyase hNth1, which initiates the removal of oxidized pyrimidines in DNA ( Klungland et al. 1999 ). Importantly, a function of NER in mismatch correction has unambiguously been demonstrated in S. pombe ( Fleck et al. 1999 ).

Plant mismatch repair proteins such as AtMSH2 ( Culligan & Hays 1997), homologous recombination enzymes such as AtDMC1 ( Klimyuk & Jones 1997), and BER glycosylases ( Shi et al. 1997 ) have recently been described and characterized. As plant components from NER such as lily ERCC1 ( Xu et al. 1998 ), AtXPB ( Ribeiro et al. 1998 ), DCR23 ( Sturm & Lienhard 1998) and AtRAD1 (in this study) become identified, it will be feasible to determine the existence and relevance of these cross-interactions between NER and the other repair pathways in the plant cell context.

Plants differ greatly from animals in terms of cell differentiation. Therefore DNA repair strategies, although conserved among higher eukaryotes, might have a different impact on cell fate. The understanding of the balanced contribution of plant NER machinery in relation to photoreactivation, combined with the unique plant physiological features, will certainly provide insights about the role of the different strategies that plants set in action to cope with the wide range of environmental challenges to which they are inevitably exposed.

Experimental procedures

Isolation of AtRAD1 full-length cDNA

An A. thaliana cDNA library (ecotype Columbia) from 5-week-old whole seedlings and the Arabidopsis library CD4-16 (ABRC, Ohio State University, Columbia, USA) were screened following standard procedures ( Sambrook et al. 1989 ). A 380 bp maize RAD1-like DNA fragment (clone 5D10D10, Zea mays EST T25298; Maize RFLP Laboratory, University of Missouri–Columbia, USA) was initially used as probe. The screenings yielded eight cDNAs. The inserts (0.7–3.3 kb) were excised through the ExAssist/SOLR system (Stratagene). Nucleotide sequences were derived using an ABI 373 automated sequencer (Applied Biosystems/Perkin Elmer). DNA sequence analyses were performed using the gcg package (Genetics Computer Group, University of Wisconsin, Madison, USA) and compared to sequences in GenBank using the blast program (www.ncbi.nlm.nih.gov) for alignment outputs and arranged with the boxshade program (www.isrec.isb-sib.ch/software/BOX_form.html). Protein sequences were analyzed at http sites genome.ad.ja and psort.nibb.ac.pj.

Complementation assays in S. pombe

Schizosaccharomyces pombe strains, media and classical genetic techniques for S. pombe are described elsewhere ( Gutz & Schmidt 1985). The S. pombe strains used in this study were h+leu1-32 ura4-D18 (h+ wt); h90leu1-32 ura4-D18 (h90 wt); OL454: h+rad16::ura4+ura4-D18 leu1-32 and OL458: h90rad16::ura4+ura4-D18 leu1-32. For simplification, strains OL454 and OL458 were named h+rad16 and h90rad16, respectively. Media were supplemented with uracil if necessary (100 mg l−1).

Constructions of plasmids for complementation analyses

An NdeI site was introduced by PCR at the 5′ end of the AtRAD1 ORF using the start ATG codon. The full-length AtRAD1 coding region between unique NdeI–DraI sites was inserted into an NdeI/SmaI-digested derivative of the expression vector pREP1 ( Maundrell 1993) in frame with the inducible fission yeast nmt1 promoter.

UV inactivation curves

Quantitative UV tests were performed as described previously ( Schmidt et al. 1989 ) with slight modifications using the heterothallic strains h+ wt and h+rad16. Cells were plated on EMM medium (between 103 and 107 cells ml−1) and irradiated with UV-C light (λmax = 254 nm) selecting doses from 20 to 80 J m−2. Colonies, corresponding to surviving cells, were counted after 5 days' incubation at 30°C. Experiments were done twice for the S. pombe wild-type transformants, and three times for the rad16 transformants containing pREP1, either with or without AtRAD1. Each set of experiments yielded eproducible results. Therefore the mean values were used to draw the inactivation curves.

Determination of sporulation frequencies

Homothallic S. pombe strains h90 wt and h90rad16 were transformed with pREP1 vector harboring the full-length cDNA AtRAD1 or the pREP1 vector alone. Approximately 500 cells from each sporulating colony were counted in a haemocytometer. The percentage of spores was calculated as the number of spores per total number of cells. Frequencies of sporulation for the different populations of transformants were compared using the ranking method ( Wierdl et al. 1996 ). Statistical significance of the data was tested by χ2 analysis.

Southern blot analyses

Plant genomic DNA was extracted with the Nucleon PhytoPure® kit (Amersham, Little Chalfont, UK) and digested (5 μg) with appropriate restriction enzymes. DNA fragments were separated in 0.8% agarose gels, transferred to a nylon membrane (Hybond N, Amersham) and cross-linked with UV light. PCR-amplified, DIG-labeled (Roche, Mannheim, Germany) fragments (560 or 850 bp long) of the coding region of the AtRAD1 cDNA were used as probes. Hybridization, stringency washes and detection were performed according to the manufacturer's instructions (Roche).

In situ hybridizations

Eight-μm sections of inflorescence, shoot, leaf and root tissues were prepared as described ( Jackson 1991). In situ hybridization was performed as previously described ( Coen et al. 1990 ) with slight modifications. The 211 bp DNA fragment of AtRAD1 between BamHI and EcoRV unique sites was cloned in pBluescript SK(–) vector (Stratagene, Heidelberg, Germany) and used to prepared digoxigenin-labeled antisense and sense probes (Roche). Specific activity of probes was determined by dot-blot analyses and used at working dilutions of 1 : 1500–1 : 2000 (for mature tissues) and 1 : 100 (for shoot apical meristems) in 120 μl hybridization buffer. Dilutions with the same specific activity were used for each probe to hybridize a given tissue. The hybridization signal was detected by NBT/BCIP (Roche). The color reaction was stopped with 10 m m Tris–HCl, 1 m m EDTA pH 8 after 3–4 days for the preparations corresponding to mature tissues, and after an overnight incubation for the meristems. Sections were mounted in Euparal (Schmid GmbH + Co, Koengen/N., Germany).

Generation of Arabidopsis plants down-regulated in AtRAD1

Two oligonucleotides containing an NcoI site were used as primers to amplify the AtRAD1 576 bp region between positions 2104 and 2679. The amplified fragment was introduced in front of the CaMV 35S promoter in antisense orientation into the NcoI site of pTd3300 (obtained by exchanging the ORF of the β-glucuronidase gene from plasmid pTd33; Tinland et al. 1995 ) , giving rise to pTd3304. The construct was introduced into Agrobacterium tumefaciens strain GV3101(pPM6000) ( Bonnard et al. 1989 ). Wild-type A. thaliana plants (ecotype Columbia, RI) were infiltrated with the strain GV3101 (pPM6000, pTd3304) in the presence of 0.03% Silwet L-77 (Osi Specialities, Danbury, CT, USA) as described previously ( Bechtold et al. 1993 ). Plants were selected on kanamycin (50 mg l−1), transformation was verified by Southern blotting, and expression levels of the antisense constructs were checked by Northern analysis. Transgenic lines shown contain one or two copies of the T-DNA inserted fragment per genome with some T-DNA rearrangements. Expression levels of the transgene were between 50 and 100 times higher than the endogenous levels of the full-length AtRAD1 mRNA.

Determination of sensitivity to DNA-damaging agents

UV sensitivity tests

Ten seeds from the segregating population T2 were sown in a row on 1% agar square plates containing 70 ml 0.5 × MS (Murashige and Skoog with vitamins) medium supplemented with 1% sucrose (pH 5.8). Plates were incubated in the growth chamber at 23°C (16 h light/8 h dark; Osram L 58 W/21, Lumilux Cool White). Plates were placed vertically to allow the roots of the seedlings to grow downward on the surface of the agar. After 3 days they were exposed to different doses of UV-C light (0.2–6 kJ m−2) and allowed to recover in the growth chamber either in light or dark. The elongation of the main root of each seedling was measured daily, up to day 7.

Assays of mitomycin C sensitivity

Four-day-old T2 seedlings, germinated as described above, were transferred into 24-well plastic plates. Each well contained one seedling in 1 ml liquid 0.5 × MS medium plus 1% sucrose and 500 mg l−1 MES pH 5.8. The medium was supplemented with mitomycin C (Kyowa, Japan) at concentrations of 2.5, 5, 10 and 20 mg l−1 ( Masson et al. 1997 ). Plates were incubated at 23°C (16 h light/8 h dark) for up to 17 days.

DNA damage analysis by ELISA

Four-week-old wild-type and AtRAD1 antisense Arabidopsis seedlings from the T2 generation, were grown in Petri dishes, in the presence of kanamycin (50 mg l−1) in the case of transgenic lines, and UV irradiated at 254 nm with different doses using a short pulse. After UV treatment they were incubated in light or complete darkness. Two different plates containing 100 seeds were used per line and per experimental condition (UV dose and recovery time in dark). After recovery, plants were harvested and the total genomic DNA was isolated. For samples allowed to recover in the dark, harvesting and DNA extraction manipulations were conducted under red dark-room light to avoid photoreactivation. The quality and quantity of DNA were tested by agarose gel electrophoresis and by absorbance. Increasing amounts of DNA were plated on 96-well flat-bottomed microtiter plates. Detection was performed by ELISA using monoclonal antithymidine dimer antibodies MC-062 (Kamiya Biochemical Co., Seattle, WA, USA) as described ( Mori et al. 1991 ) using 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulphonic acid)-ABTS (Sigma, St. Louis, MO, USA) as a peroxidase substrate. Absorbance was measured at 410 nm. Four microplate wells containing the same amount of DNA were read as four measurements to be averaged.

Acknowledgements

We gratefully acknowledge Esther Vogt and Stéphane Pien from ETH-Zürich for kindly providing the A. thaliana cDNA library and the in situ hybridization preparations of Arabidopsis shoot apical meristems, respectively. We are indebted to Xavier Fernàndez-Busquets for critical comments on the manuscript and to Esther Betran for helping with the statistical analyses. We are also grateful to Johannes Fütterer, Nania Schärer-Hernández, Peter Macheroux, Andrew Fleming and Siegbert Melzer, ETH-Zürich, for providing helpful comments on the manuscript and for encouraging discussions. We particularly acknowledge Professor Ingo Potrykus for his constant support. We thank Dr H. Schmidt for supplying the pREP1 vector and Dr U. Feister for comments on UV effects. F.G. was supported by a post-doctoral fellowship from the Spanish Ministry of Education and Culture and by the Ciba-Geigy–Jubiläums-Stiftung (Basel). This project was supported by Swiss National Foundation for Scientific Research Grant no. 31-51′077.97.

Footnotes

  1. GenBank accession number AF089003.

Note added in proof

The paper Repair of UV damage in plants by nucleotide excision repair: Arabidopsis UVH1 DNA repair gene is a homolog of Saccharomyces cerevisae Rad 1 by Zongrang Liu, Gazi S. Hossain, Maria A. Islas-Osuna, David L. Mitchell and David W. Mount, which is based on independent research, also reports the identification of Arabidopsis thaliana AtRAD1 and is published in pages 519–528 in this issue of The Plant Journal.

Ancillary