Arabidopsis UVH3 gene is a homolog of the Saccharomyces cerevisiae RAD2 and human XPG DNA repair genes


For correspondence (fax +1 520 621 3709; e-mail


To identify mechanisms of DNA repair in Arabidopsis thaliana, we have analyzed a mutant (uvh3) which exhibits increased sensitivity to ultraviolet (UV) light, H2O2 and ionizing radiation and displays a premature senescence phenotype. The uvh3 locus was mapped within chromosome III to the GL1 locus. A cosmid contig of the GL1 region was constructed, and individual cosmids were used to transform uvh3 mutant plants. Cosmid N9 was found to confer UV-resistance, H2O2-resistance and a normal senescence phenotype following transformation, indicating that the UVH3 gene is located on this cosmid and that all three phenotypes are due to the same mutation. Analysis of cosmid N9 sequences identified a gene showing strong similarity to two homologous repair genes, RAD2 (Saccharomyces cerevisiae) and XPG (human), which encode an endonuclease required for nucleotide excision repair of UV-damage. The uvh3 mutant was shown to carry a nonsense mutation in the coding region of the AtRAD2/XPG gene, thus revealing that the UVH3 gene encodes the AtRAD2/XPG gene product. In humans, the homologous XPG protein is also involved in removal of oxygen-damaged nucleotides by base excision repair. We discuss the possibility that the increased sensitivity of the uvh3 mutant to H2O2 and the premature senescence phenotype might result from failure to repair oxygen damage in plant tissues. Finally, we show that the AtRAD2/XPG gene is expressed at moderate levels in all plant tissues.


Repair of DNA damage is a significant challenge for plants, since sunlight is both a source of damage and is also required for photosynthesis. The most prevalent lesions produced by the ultraviolet (UV) component of sunlight are cyclobutane pyrimidine dimers (CPDs) and pyrimidine (6–4) pyrimidinone dimers (6–4 products) (Mitchell and Nairn, 1989; Pfeifer, 1997). These photoproducts impair DNA replication and transcription processes and can induce mutations (Lindahl and Wood, 1999; Painter, 1985; Protic-Sabljic and Kraemer, 1985). Other types of detrimental lesions include oxidative damage to the bases in DNA (e.g. thymine glycol, 8-oxoguanine) induced by reactive oxygen molecules during normal cellular metabolism (Demple and Harrison, 1994; Marnett, 2000).

Several pathways for repair of DNA damage have been described in eukaryotes, and at least some of these pathways also exist in plants. These pathways include direct enzymatic photoreversal of UV-induced photoproducts and two excision repair pathways, nucleotide excision repair and base excision repair (Sancar, 1994; Sancar, 1996; Wood, 1996). The photoreversal pathway appears to be the major mechanism for repair of UV-damage in plants. Arabidopsis thaliana contains two photolyases, one specific for CPDs and another for 6–4 products (Ahmad et al., 1997; Jiang et al., 1997a; Landry et al., 1997; Nakajima et al., 1998). These enzymes are activated by visible light, and the majority of photoproducts in plant DNA are only repaired under light conditions, when photoreversal can occur (Britt, 1999).

Nucleotide excision repair acts on UV-induced photoproducts in the absence of visible light and also repairs other lesions that strongly distort the helical structure of DNA. This pathway has been well characterized in yeast (Saccharomyces cerevisiae) and humans (Prakash and Prakash, 2000; Sancar, 1996; Wood, 1996), and the names of several participant genes are derived from yeast repair mutants (RAD) and from humans suffering from the syndrome, xeroderma pigmentosum (XP). These yeast and human mutants are very sensitive to DNA damage, and XP patients exhibit a high incidence of tumors in sun-exposed areas of skin (Prakash and Prakash, 2000; Vessey et al., 2000). Repair by this pathway involves recognition of the DNA damage, incision of the damaged strand on both sides of the lesion, excision of the damaged single-stranded fragment, and subsequent repair of the resultant gap by DNA synthesis and ligation. The incision step requires two nucleases. Incision on the 5′ side of the lesion is performed by a complex of two proteins (called RAD1:RAD10 in S. cerevisiae, XPF:ERCC1 in humans), and incision on the 3′ side is performed by a single protein, RAD2/XPG.

Significant evidence suggests that nucleotide excision repair also occurs in plants. A mutant of Arabidopsis thaliana which carries a mutation in the homolog of RAD1/XPF has been shown to be UV-sensitive (Fidantsef et al., 2000; Liu et al., 2000), and inhibition of expression of this gene by an RNA antisense construct also increases sensitivity (Gallego et al., 2000). In addition, the presumed RAD10/ERCC1 homolog from Lilium longiflorum provides partial complementation of a repair defect in ERCC1-deficient Chinese hamster ovary cells (Xu et al., 1998). A second Arabidopsis gene potentially involved in nucleotide excision repair has been cloned (Ribeiro et al., 1998) and shows homology to the yeast RAD25 and human XPB genes. These gene products are required for helicase-driven excision of damaged DNA. Finally, a potential RAD23/HR23B homolog has been detected in rice and in Arabidopsis EST sequences (Schultz and Quatrano, 1997). This gene encodes a protein which forms a complex with XPC in humans and is required for damage recognition.

Base excision repair, another repair pathway in yeast and humans, is directed towards less bulky DNA lesions (e.g. 8-oxoguanine) which only mildly distort the DNA helical structure (Demple and Harrison, 1994; Wood, 1996). This pathway is initiated by removal of the damaged base using one of several damage-specific DNA-glycosylases. The finding that Arabidopsis contains a DNA-glycosylase specific for 3-methyladenine (Santerre and Britt, 1994) and the prediction of other glycosylase genes in the Arabidopsis genome (The Arabidopsis Genome Initiative, 2000) strongly suggest that base excision repair occurs in plants. Following glycosylase action, incision of the DNA backbone occurs next to the abasic site, using either an AP-lyase activity associated with the glycosylase or an AP-specific endonuclease. Subsequently, the damaged region is excised, and the gap is repaired.

Although most of the enzymes required for initiation of base excision repair differ from those that act in nucleotide excision repair, at least one enzyme, RAD2/XPG has been implicated in both processes. The RAD2/XPG nuclease acts during the incision step of nucleotide excision repair (Prakash and Prakash, 2000; Wood, 1996). In humans, XPG is also involved in repair of oxidized lesions by base excision repair (Dianov et al., 2000; Le Page et al., 2000). Studies in vitro have suggested that human XPG acts, at least in part, by stimulating the activity of the human DNA glycosylase-AP lyase, hNth1 (Bessho, 1999; Klungland et al., 1999). Since the nuclease activity of XPG is not required for this stimulation (Klungland et al., 1999), the function of XPG in base excision repair appears to be distinct from its nuclease function in nucleotide excision repair.

To obtain a better understanding of DNA repair pathways in plants, we have isolated a series of Arabidopsis mutants that exhibit increased sensitivity to both UV-B and UV-C radiation wavelengths (Harlow et al., 1994; Jenkins et al., 1995). In the present study, we have identified the altered gene in one mutant, called uvh3, and have shown that this gene is the Arabidopsis homolog of RAD2/XPG. Since the RAD2/XPG protein acts in both nucleotide excision repair and base excision repair, our findings provide evidence that these pathways play important roles during plant growth.


Mapping of the UVH3 gene

We previously identified a UV-sensitive mutant (uvh3) of Arabidopsis and found that the UVH3 locus is closely linked to the GL1 gene on chromosome III (Jenkins et al., 1995). We have now conducted experiments to clone and identify the UVH3 gene.

As a first step in this analysis, we conducted the mapping experiments outlined in Figure 1. A cross was performed between uvh3/uvh3 homozygous mutant plants (Columbia ecotype) and wild-type plants (Landsberg ecotype). F2 progeny from this cross were screened to identify those that exhibited the UV-sensitivity of the mutant (Columbia) parent. Since the uvh3 mutation is linked to the GL1 locus, we then examined these UV-sensitive F2 progeny for the presence of physical DNA markers in the GL1 region which had been acquired by recombination with the Landsberg parent. As shown in Figures 1, 64 recombinants carrying the Landsberg NIT1 marker and 15 recombinants carrying the Landsberg GAPA marker were observed. Since NIT1 and GAPA flank the GL1 locus, each group of recombinants was further analyzed for the presence of additional Landsberg markers in this region (GL1, 1C10R, and AIG2). Based on the numbers of recombinants found at each position, we concluded that the UVH3 locus is located to the left of marker 1C10R, to the right of GAPA and is tightly linked to marker GL1.

Figure 1.

Mapping and positional cloning of the UVH3 gene.

The UVH3 locus was localized between markers GAPA and 1C10R on chromosome III, based on an analysis of DNA from 315 UV-sensitive F2 progeny (630 chromosomes) obtained from a cross between uvh3 (Columbia ecotype) and wild-type (Landsberg ecotype) plants. Recombinant plants exhibiting the Landsberg NIT1 or GAPA markers were identified, and each group was further examined for the additional markers shown. BAC and cosmid contigs were constructed, as described in the Experimental procedures section. The UVH3 locus was identified as described in the text and is indicated within each contig.

Next we identified a BAC clone contig, which encompasses the UVH3 site. For this purpose, a BAC library was probed with GL1 and 1C10R markers to identify BAC clones containing these sequences. Contigs encompassing each marker were constructed following identification of overlapping regions within each group of BACs. The contigs were subsequently extended until a contig corresponding to the entire region between the markers was obtained, as shown in Figure 1. Two BACs, T7P17 and T6K18 that carry the GL1 locus were identified. Based on the genetic mapping conducted above, we predicted that the UVH3 gene would be located within one or both of these clones.

Complementation of the uvh3 mutant by cosmid N9

To identify the UVH3 gene within BACs T7P17 and T6K18, we next identified cosmid clones corresponding to these BACs, using the BAC sequences to probe a cosmid library. Several cosmids from this region were identified and assembled into a contig. The portion of the contig corresponding to the T7P17 BAC is shown in Figure 1.

These cosmids were then used to identify sequences that complement the uvh3 defect. For this analysis, cosmid vectors were introduced into Agrobacterium, individual constructs were used to transform uvh3/uvh3 homozygous mutant plants and the transformed plants were tested for UV-sensitivity. We observed that transformation by cosmid N9, but not by any other cosmid, resulted in a large increase in UV-resistance of the mutant plants. As shown in Figure 2(a), the original uvh3 mutant is substantially more UV-sensitive than the wild-type parent (C10), exhibiting severe leaf yellowing and tissue damage after UV irradiation. In contrast, the complemented mutant exhibits resistance to UV-treatment comparable with wild-type. Hence, we conclude that N9 carries the intact UVH3 gene.

Figure 2.

Recovery of UV-resistance, a normal senescence phenotype, and H2O2-resistance following complementation of a uvh3/uvh3 homozygous mutant with cosmid N9.

Mutant plants were transformed with cosmid DNA as described in the Experimental procedures section. The plants shown are wild-type (UVH3/UVH3), mutant (uvh3/uvh3), and T2 generation mutant plants complemented with N9 cosmid clone.

(a) Plants were grown under normal growth conditions and, at two weeks of age, were irradiated with 300 J m−2 of UV-C light, as described in the Experimental procedures section.

(b) Plants were grown under 24 h continuous lighting using 40WT12 Excella bulbs (Industrial) and photographed at 5 wk.

(c) Plants were grown under normal growth conditions and, at 3 wk, were treated with H2O2, as described in the Experimental procedures section.

To identify a likely candidate for the UVH3 gene, we conducted BLAST searches, using sequences from the flanking ends of the T7P17 BAC and N9 clones as queries of the GenBank database (Altschul et al., 1997), and we located these sequences on a PAC clone (MMG15, AB028616), which had been entirely sequenced by the Arabidopsis genome project. The region encompassed by N9 included one gene product that is homologous to the human XPG and yeast (S. cerevisiae) RAD2 proteins. Mutations in the corresponding human and yeast genes result in UV-sensitivity and a defect in the incision step of the nucleotide excision repair pathway (Prakash and Prakash, 2000; Wood, 1996). We further confirmed that the AtRAD2/XPG sequence occurs on the N9 cosmid by PCR analysis (not shown). Based on this analysis, AtRAD2/XPG is a likely candidate gene for UVH3.

Determination of the UVH3 cDNA sequence

To confirm the identification of the UVH3 gene, we amplified the cDNA corresponding to the AtRAD2/XPG gene and then searched for a mutation within this sequence in the uvh3 mutant. AtRAD2/XPG cDNA was amplified from cellular RNA by RT–PCR. In these reactions, the 3′-end of this gene was amplified using a leftward primer which annealed just downstream of the predicted termination codon. The 5′-end of the gene was identified by sequentially testing a nested set of rightward primers, moving progressively upstream until one failed to amplify the cDNA, indicating that the end of the mRNA had been reached. The primer just downstream of this failed primer was then used to amplify the 5′-end of the cDNA. Finally, the amplified cDNA sequence was determined (GenBank accession number AF312711) and compared with the genomic sequence (GenBank accession number AB028616). As illustrated in Figure 3, this comparison reveals a gene containing 17 exons and 16 introns. The predicted protein product consists of 1479 amino acids and should exhibit a molecular weight of 165.7 kilodaltons and a pI of 4.81.

Figure 3.

Gene organization of UVH3 and site of the uvh3 mutation.

The cDNA sequence of the UVH3 (AtRAD2/XPG) gene was determined as described in the text and compared with the known genomic sequence. Based on this comparison, the UVH3 gene is diagrammed with exons (thick bars), introns (thin lines), and initiation and termination codons. The uvh3 mutation, identified as described in the text, is located at amino acid 1157 and produces a new termination codon, as shown.

Identification of the uvh3 mutation

To identify the mutated site within uvh3 plants, AtRAD2/XPG cDNA was amplified from mutant RNA, sequenced in its entirety, and compared with the wild-type sequence. As shown in Figure 3, a single G to A mutation was observed in exon 15 of the mutant gene, producing a nonsense codon, which should result in a truncated protein product. The presence of this mutation in the genome was confirmed by amplifying and sequencing the appropriate genomic region from uvh3 mutant plants. The identification of this mutation strongly supports the conclusions that AtRAD2/XPG is the UVH3 gene and that the encoded protein is required for UV-resistance.

Comparison of the UVH3 protein sequence with yeast and human homologs

A multiple sequence alignment comparing the human XPG, yeast (S. cerevisiae) RAD2, and Arabidopsis UVH3 proteins is shown in Figure 4. This analysis reveals high conservation among these sequences in three regions. Two regions (domains N and I) have been implicated in the nuclease function of XPG and are conserved in the FEN-1 family of structure-specific endonucleases (Harrington and Lieber, 1994; Lieber, 1997). Conservation of these domains in the AtRAD2/XPG protein strongly suggests that this plant protein utilizes a nuclease function for repair of DNA damage. The third region (D2) has been described as one of two domains (D1 and D2) which are conserved in eukaryotes but which have no known function (Houle and Friedberg, 1999). D1 was previously found only in higher eukaryotes (Xenopus, Drosophila, mouse, human) but not in yeast (S. cerevisiae, Schizosaccharomyces pombe). Our analysis reveals that it is also not present in Arabidopsis. However, D2 is highly conserved in Arabidopsis, suggesting that this region is important for RAD2/XPG function. The fact that the uvh3 mutation occurs at an invariant residue within this domain gives further support to this possibility.

Figure 4.

Comparison of the UVH3 (AtRAD2/XPG) protein sequence with homologous sequences from human and yeast.

(a) An alignment of UVH3 (GenBank accession number AF312711), human XPG (SwissProt P28715), and S. cerevisiaeRAD2(SwissProt P07276) is represented. Thin lines depict regions of variable length that do not align or align too poorly to identify a domain. Black boxes depict highly conserved domains (N, I, D2). Numbers refer to amino acid residues flanking conserved domains and the C-terminal residue in the UVH3 sequence.

Full alignments are also shown for conserved domains N (b), I (c), and D2 (d). Highly conserved residues are shaded and identities are boxed in black. The position of the uvh3 nonsense mutation within D2 is marked by a star.

An early senescence phenotype exhibited by uvh3 plants is also complemented by UVH3/AtRAD2/XPG

We had previously observed that uvh3 mutant plants exhibit premature senescence (i.e. yellowing of leaf rosettes) and reduced seed production (Jenkins et al., 1995). This early senescence, as exhibited by 5-wk-old uvh3 plants, is illustrated in Figure 2(b) (center panel). To investigate a possible role for UV damage in promoting the early senescence phenotype, we also examined the growth of uvh3 plants under lighting conditions in which UV wavelengths produced by cool white fluorescent lights were removed by filtering the light through three layers of mylar (Jiang et al., 1997a). In this experiment (data not shown), the extent of leaf yellowing was considerably reduced compared with results in Figure 2(b), although early senescence was still present. Consequently, we conclude that UV damage likely plays a role in the early senescence phenotype but that other factors (e.g. oxygen damage) could also contribute.

In the complementation test shown in Figure 2(b) (right panel), we observed that transformation of uvh3 plants by cosmid N9 resulted in recovery of the normal wild-type senescence phenotype and rescue of premature leaf yellowing. As described above, this complementation also resulted in recovery of UV-resistance (Figure 2a). Hence, we conclude that the early senescence phenotype and the increased UV-sensitivity seen in uvh3 plants are due to the same mutation. This result further suggests that the UVH3 gene is required both for repair and for the normal senescence program in plants.

Uvh3 mutants have increased sensitivity to oxygen damage by H2O2

The above results suggested a possible defect in repair of oxidative DNA damage in uvh3 mutant plants. In addition, defects in the human homolog of this gene (XPG) are defective in repair of such lesions (Le Page et al., 2000). Hence, to investigate the possibility that uvh3 plants are defective in repair of oxidative damage, we examined the response of these plants to H2O2, a compound that produces DNA lesions induced by reactive oxygen species. As shown in Figure 2(c), uvh3 mutant plants were substantially more sensitive to treatment (center panel) than wild-type plants (left panel), as indicated by leaf yellowing and desiccation. Complementation of uvh3 plants by cosmid N9 rescued the H2O2-sensitivity (right panel), resulting in the wild-type level of resistance. Thus, the uvh3 gene also appears to be important for repair of oxidative damage.

Expression of the UVH3 gene

The mRNA expression pattern of the UVH3 gene was analyzed as shown in Figure 5. Total RNA was extracted from plant tissues, specific mRNA sequences were amplified by RT–PCR and PCR products were visualized following electrophoresis. To verify that the amount of PCR product reflects the abundance of the corresponding mRNA transcripts, we determined the RNA concentrations that produced a level of product proportional to input RNA. For this analysis, we used the ubiquitin 3 (UBQ3) gene, which is expressed at relatively high levels compared with UVH3. As shown in Figure 5(a), the amount of UBQ3 product is proportional to input RNA up to and beyond 50 ng. Consequently, we used 50 ng of input RNA in subsequent experiments. Finally, we compared the expression of UVH3 in different plant tissues. As shown in Figure 5(b), similar amounts of RT–PCR products were observed using flower bud, root, stem, leaf and apical meristem tissues. Hence, UVH3 appears to be expressed in all tissues at moderate levels.

Figure 5.

Expression of mRNA from Arabidopsis tissues, as measured by RT–PCR.

Total RNA was isolated from the tissues indicated and assayed, as described in the Experimental procedures section. RT–PCR products were detected following agarose gel electrophoresis, and the product amplified from the fully spliced message is shown in each case. Controls without reverse transcriptase were run in each experiment (not shown) and failed to produce a product, thus ruling out any contribution of genomic DNA to amplification.

(a) A calibration of the RT–PCR assay is shown, using the indicated amounts of RNA from 3-wk-old leaves and primers targeted to the UBQ3 gene.

(b) Assays are shown for reactions with 50 ng RNA and primers directed towards the UVH3 or UBQ3 genes. RNA was prepared from unbolted flower buds (F), roots (R), stems (S), or leaves (L) using 4-wk-old plants or from 2-wk-old meristem tissue (M).


Our findings strongly support the identification of the UVH3 DNA repair gene as the Arabidopsis homolog of yeast (S. cerevisiae) RAD2 and human XPG genes. The UVH3 locus was originally identified by the uvh3 mutation, which produced a UV-sensitive phenotype and mapped to the GL1 region of chromosome III. In the present study, we have used genetic crosses to further localize the UVH3 locus and have shown that the N9 cosmid clone complements the UV-sensitivity of the uvh3 mutant. Database similarity searches, using the N9 sequence as a query, were used to identify a gene, AtRAD2/XPG, that is homologous to genes required for DNA repair in yeast and humans. Finally, we have sequenced the AtRAD2/XPG cDNA from the uvh3 mutant and identified a mutation in this gene which should result in a truncated AtRAD2/XPG protein. Taken together, these results demonstrate that AtRAD2/XPG is UVH3.

RAD2/XPG protein is required for incision on the 3′ side of the lesion during nucleotide excision repair in yeast and humans. Hence, our finding that the AtRAD2/XPG gene is required for normal UV-resistance in Arabidopsis strongly suggests that the UV-sensitive phenotype in uvh3 mutants is due to a defect in nucleotide excision repair. In support of this conclusion, uvr1 mutants, which fail to complement uhv3 mutants (unpublished observations), are defective in repair of 6–4 UV-photoproducts (Britt et al., 1993). Further support for a repair defect is provided by the identification of other Arabidopsis genes which appear to be homologs of additional nucleotide excision genes from yeast and humans. UVH1/AtRAD1 protein has been identified as a homolog of the yeast RAD1 and human XPF proteins (Fidantsef et al., 2000; Gallego et al., 2000; Liu et al., 2000), required for the incision on the 5′ side of the lesion during nucleotide excision repair. Homologs of RAD25/XPB (Ribeiro et al., 1998), required for removal of the incised DNA, and RAD23/HR23B (Schultz and Quatrano, 1997), involved in damage recognition, have also been described. A number of additional candidate genes have been described in a recent analysis of the Arabidopsis genome (The Arabidopsis Genome Initiative, 2000).

In addition to UV-damage, plants may be especially vulnerable to oxygen damage, such as 8-oxoguanine and thymine glycol residues in DNA. Pathways of both photosynthesis and mitochondrial-based ATP synthesis involve free-radical intermediates which are likely to damage chromosomes during plant growth. In human cells, oxygen damage is repaired by base excision repair and requires XPG protein (Demple and Harrison, 1994; Dianov et al., 2000; Le Page et al., 2000). Hence, the equivalent protein in Arabidopsis, namely UVH3, may also be required for repair of oxygen damage.

In support of this possibility, our uvh3 Arabidopsis mutant is hypersensitive to H2O2, a compound known to cause oxygen-damage. In addition, we previously reported (Jenkins et al., 1995) that uvh3 plants are sensitive to γ-irradiation, a treatment that causes several types of lesions, including DNA strand breaks and oxidative base damage produced indirectly by reactive oxygen species (Riley, 1994; Ward, 1988). While we presently cannot rule out the possibility that the γ-ray-sensitivity of our mutant is due to a second mutation, other than uvh3, we think this possibility is unlikely. Our argument is based on the finding that independently isolated mutants (called uvr1) from the same complementation group also exhibit a γ-ray-sensitive phenotype (unpublished results; Jiang et al., 1997b). Hence, sensitivities to UV light, H2O2 and γ-rays all appear to result from a single mutation in both the uvh3 and uvr1 mutants.

Further evidence supporting the possibility that base excision repair of oxygen damage occurs in Arabidopsis comes from similarity searches of the Arabidopsis database, using as queries yeast and human proteins involved in base excision repair. These searches have predicted the presence of Arabidopsis homologs of glycosylases required for repair of oxygen damage (The Arabidopsis Genome Initiative, 2000).

Our present study also indicates that a deficiency in UVH3 protein produces an early senescence phenotype. We previously reported that uvh3 plants exhibit premature yellowing of rosette leaves and reduction in seed production (Jenkins et al., 1995). This phenotype cosegregated with UV-hypersensitivity in 556 backcrossed lines, suggesting that the two phenotypes are closely linked. We have observed in the present study that transformation of uvh3 plants by cosmid N9 resulted in complementation of both UV-sensitivity and the early senescence phenotype, further confirming that early senescence is due to the uvh3 mutation.

We suggest that the early senescence phenotype is due to long-term accumulation of DNA damage in plant tissues and results from the DNA repair deficiency in uvh3 mutants. On the one hand, we favor the possibility that UV damage contributes to this phenomenon, since the early senescence phenotype is partially alleviated when uvh3 plants are grown under conditions that filter out UV wavelengths. However, premature senescence is not totally suppressed under these conditions, suggesting that other types of DNA damage are also involved. Since uvh3 mutants are also hypersensitive to oxidative damage, one plausible explanation for the continued senescence in the absence of UV light is that uvh3 plants accumulate oxidative damage in their DNA with age, due to a defect in repair of these lesions. A possible model to explain the premature senescence is that the accumulated DNA damage inhibits transcription, leading to reduced levels of enzymes required for leaf viability. This model is supported by the observations that both UV-induced pyrimidine dimers and 8-oxoguanine lesions severely block transcription in mammalian cells (Donahue et al., 1994; Le Page et al., 2000).

Further insight into a possible cause of premature senescence in the uvh3 mutant comes from studies of human XPG mutations, which cause the combined diseases of xeroderma pigmentosum and Cockayne syndrome. Patients carrying these mutations exhibit both a defect in nucleotide excision repair and the symptoms of Cockayne syndrome, which include impaired development, short stature and premature death (Nance and Berry, 1992). Cells from such patients are defective in a specialized type of oxidative repair, called transcription-coupled base excision repair (Le Page et al., 2000). This pathway targets repair of oxidative damage to the transcribed strands of active genes and might therefore be essential to maintain transcription levels during growth. Hence, one possible scenario to explain the premature senescence in our Arabidopsis uvh3 mutant is a defect in transcription-coupled repair of oxygen damage.

This possibility is made more likely by our prediction that the uvh3 mutation affects the AtRAD2/XPG protein structure in a similar manner to that expected for the dual-syndrome human XPG mutations. Like the human mutations, the uvh3 mutation results in protein truncation due to introduction of a nonsense codon. A comparison of the locations of three known XPG/Cockayne syndrome mutations with that of uvh3 reveals that two of the human mutations truncate upstream of the highly conserved D2 region in XPG (at amino acid residues 262 or 659 in the human sequence; see Figure 4), while both uvh3 and the third human mutation (at amino acid residue 980 in the human sequence) occur within D2. The effects of these truncating mutations indicate that the C-terminus of XPG in both humans and Arabidopsis is important for normal function of this enzyme.

In summary, we have identified an Arabidopsis homolog of yeast RAD2 and human XPG, designated UVH3 (also UVR1). A mutation in this gene (uvh3) results in hypersensitivity to UV-irradiation and H2O2, a premature senescence phenotype, and a presumed sensitivity to γ-rays. Our results support a role for the Arabidopsis RAD2/XPG protein in nucleotide excision repair and suggest that this protein also acts to repair oxidative damage by base excision repair. Our results further suggest that repair of both UV-photoproducts and oxidative damage is essential for normal timing of tissue senescence. Our observation that expression of the UVH3 gene occurs in all plant tissues is consistent with a role for this gene in repair of spontaneous DNA damage during plant metabolism.

Experimental procedures

Strains and growth conditions

The wild-type line (C10) was generated as a single plant isolate of Arabidopsis (Columbia ecotype) and served as the parent of the uvh3 mutant, which was isolated as described (Harlow et al., 1994; Jenkins et al., 1995). Plants were normally grown at 20–23°C in Sunshine 3 soil (Sun Gro Horticulture, Inc., Bellevue WA, USA) under continuous lighting (a combination of 40WT12 Excella and F40 agro bulbs) at an approximate distance of 40 cm from the light source.

DNA-damaging treatments

To conduct tests of UV-sensitivity, plants were irradiated with UV-C light, incubated for 3 days under F40 GO gold fluorescent lights, which lack photoreactivation wavelengths, and then transferred to standard growth conditions for 10 days (Harlow et al., 1994). UV-C radiation was supplied by a UV Stratalinker (Stratagene model 2400) by selecting the desired UV dose. The dose rate (30 J m−2 sec−1) was determined using a recently calibrated UVX Digital Radiometer equipped with a UVX-25 UV-C sensor (UV-Products, San Gabriel, CA, USA), and the time required for the proper dose was programmed into the Stratalinker. H2O2 treatment was conducted by dipping plants in a 1-m H2O2 solution for 10 min, washing them thoroughly with tap water, and growing them under F40 agro lights for 1 wk. UV-or H2O2-sensitivities were detected by measuring leaf yellowing and tissue death.

Genetic and physical mapping of the UVH3 gene

To generate recombinant plants for mapping of the UVH3 locus, a uvh3 mutant plant (Columbia ecotype) was crossed with a wild-type Landsberg erecta plant, and F2 progeny were screened for UV-sensitivity, as described in Harlow et al. (1994). Briefly, trays of 2–3 wk-old plants were irradiated with 200 J m−2 of UV light following protection of the plant meristems with a temporary foam suspension of para-amino benzoic acid (PABA). The plants were then washed with water to remove the PABA and grown for several days (as described under ‘Strains, growth conditions, and UV-irradiation treatment’) to detect UV-sensitive plants. DNA from UV-sensitive (uvh3/uvh3) F2 plants was isolated, as described in Harlow et al. (1994), and examined for the presence of physical polymorphisms, which differ between the Columbia and Landsberg ecotypes. Polymorphisms were detected by PCR assays and agarose gel electrophoresis or by Southern blotting of restriction digests (Harlow et al., 1994). PCR-based markers (GAPA, GL1, NIT1) and the RFLP-based marker (AIG2) are listed in the TAIR database (, and the PCR-based marker 1C10R was derived from the right end of YAC CIC1C10 (Arabidopsis Biological Research Center, Ohio State University, Columbus, OH, USA).

To identify BAC clones encompassing the UVH3 locus, a filter array of the TAMU BAC library (Arabidopsis Biological Research Center) was hybridized by Southern blotting (Harlow et al., 1994) with GL1 or 1C10R probes that had been amplified by PCR from C10 genomic DNA and end-labeled, using a Primer II Kit (Stratagene, La Jolla CA, USA). The extent of overlap between pairs of related BACs was determined, using one BAC as a probe to detect overlapping fragments in a restriction digest of the second. The contig was extended by amplifying the ends of identified BACs (Woo et al., 1994), sequencing these regions and using the sequences in BLAST database searches to identify overlapping BACs.

Screening cosmid libraries and mutant complementation

To identify cosmid clones carrying the UVH3 gene, inserts of approximately 100 Kb were released from potential BAC clones by digestion with NotI, labeled with a Primer II Kit, and used to hybridize to a cosmid library (Harlow et al., 1994; Olszewski et al., 1988). Cosmids recovered from positive clones were assembled into a contig, using one cosmid as a probe to detect overlapping fragments in a restriction digest of a second.

To conduct complementation tests, cosmid DNA was introduced into Agrobacterium strain AGL1 by electroporation (Mozo and Hooykaas, 1991). Transformed bacteria were selected on LB medium (Sambrook et al., 1989) containing 30 mg l−1 carbenicillin and 60 mg l−1 kanamycin. Cosmid integrity was verified by restriction analysis of cosmid DNA isolated from transformed Agrobacterium cells. DNA from the bacteria was transferred to uvh3 mutant plants by dipping flowers in the bacterial culture, and transformed kanamycin-resistant plants were selected from the next generation seed as described by Bechtold et al. (1993), except that the vacuum step was omitted. Primary transformants were irradiated with 300 J m−2 of UV-C light and then grown as described under ‘Strains, growth conditions, and UV-irradiation treatment’ to identify UV-resistant, complemented transgenic lines. The UV-resistant phenotypes were further confirmed in T2 and T3 generation plants.

Amplification and sequencing of UVH3 cDNA and genomic fragments

UVH3 cDNA was amplified from C10 wild-type and uvh3 mutant plants by RT–PCR, using total RNA prepared from leaf tissue. The RT–PCR procedure was identical to that described below under ‘RT–PCR Analysis’ and utilized several primer pairs, designed from the known genomic DNA sequence. Upper primers were: H3U906 (5′-GCTTTTAATTACACGGCGAAG), H3U1024 (5′-AGAAA CCCTAGCCAACAAACGA), and H3U3770 (5′-ATTCAGCGCCTGGG AGAACTT). Lower primers were: H3L4338 (5′-TATCTTCGCTTG GCTTGTCAAC), H3L7204 (5′-TTTGCCTCCAATTCCACTACCT), and H3L7858 (5′-TTACTCCCCTAATTACTTTTACCA). PCR products were separated on 1% agarose gels, extracted using a Prep-A-Gene DNA purification kit (Bio-Rad, Hercules, CA, USA), and subjected to DNA sequencing of both strands. Sequencing primers were designed to anneal at approximately 500 base intervals along the sequence To verify the uvh3 mutational change in genomic DNA, the region surrounding the mutation was amplified by PCR from both wild-type and mutant plant DNAs, using primers H3U3770 and H3L7858, and the reaction products were purified and sequenced as above.

RT–PCR analysis

Total RNA was isolated from plant tissues using an Rneasy Plant Mini Kit (Qiagen, Valencia, CA, USA). RNA quality was examined by gel electrophoresis, and RNA concentration was determined by spectrophotometry. Analysis of gene expression was performed using a OneStep RT–PCR Kit (Qiagen). Reverse transcription was conducted at 50°C for 30 min. The reactions were then incubated at 95°C for 15 min and subjected to 25–30 cycles of 95°C for 1 min and 60–65°C for 1–3 min, followed by a final extension at 72°C for 10 min. Ten microliters of each 25 µl reaction mixture was subjected to electrophoresis on 3% agarose, and the identity of each RT–PCR product was confirmed by Southern blot analysis and sequencing. Primers UBQ3U1297 (5′-CAATCTCTCCCAAAG CCTAAAG) and UBQ3L2132 (5′-TCGACTCCTTTTGAATGTTGTAG) were used to amplify the UBQ3 gene, and primers H3U3370 and H3L4849 (5′-CGCGATCTGACGACATTACGAG) were used to amplify UVH3.

Computer analysis

Multiple sequence alignments were obtained using the Clustal W program with default parameters (Higgins et al., 1996) and the BOXSHADE program ( The BLAST server ( was used for database similarity searches (Altschul et al., 1997). cDNA and genomic sequences were aligned by the local alignment program, LALIGN (Pearson and Miller, 1992), with manual adjustments to allow for coincidence of the splice junctions with the known consensus for Arabidopsis. Molecular weight and charge of the AtRAD2/XPG protein was calculated using the expasy web site (


The authors wish to thank Dr Michael Jenkins and Gazi Showkat Hossain for assisting in the mapping experiments, the Arabidopsis Biological Research Center for the TAMU BAC filter and BAC and YAC clones, and Dr N.E. Olszewski for the cosmid library. This study was supported by National Science Foundation grant (MCB-9728125).

Genbank accession numbers AF312711and AB028616.