Defective DNA endonuclease activities in Fanconi's anemia cells, complementation groups A and B.

Cells from patients with the inherited disorder, Fanconi's anemia (FA), were analyzed for endonucleases which recognize DNA interstrand cross-links and monoadducts produced by psoralen plus UVA irradiation. Two chromatin-associated DNA endonuclease activities, defective in their ability to incise DNA-containing adducts produced by psoralen plus UVA light, have been identified and isolated in nuclei of FA cells. In FA complementation group A (FA-A) cells, one endonuclease activity, pI 4.6, which recognizes psoralen intercalation and interstrand cross-links, has 25% of the activity of the normal human endonuclease, pI 4.6, on 8-methoxypsoralen (8-MOP) plus UVA-damaged DNA. In FA complementation group B (FA-B) cells, a second endonuclease activity, pI 7.6, which recognizes psoralen monoadducts, has 50% and 55% of the activity, respectively, of the corresponding normal endonuclease on 8-MOP or angelicin plus UVA-damaged DNA. Kinetic analysis reveals that both the FA-A endonuclease activity, pI 4.6, and the FA-B endonuclease activity, pI 7.6, have decreased affinity for psoralen plus UVA-damaged DNA. Both the normal and FA endonucleases showed approximately a 2.5-fold increase in activity on psoralen plus UVA-damaged reconstituted nucleosomal DNA compared to damaged non-nucleosomal DNA, indicating that interaction of these FA endonucleases with nucleosomal DNA is not impaired. These deficiencies in two nuclear DNA endonuclease activities from FA-A and FA-B cells correlate with decreased levels of unscheduled DNA synthesis (UDS), in response to 8-MOP or angelicin plus UVA irradiation, in these cells in culture.     

Two DNA endonuclease activities from normal human and xeroderma pigmentosum chromatin active on psoralen plus ultraviolet light treated DNA

DNA endonuclease activities from the chromatin of normal human and xeroderma pigmentosum, complementation group A (XPA), lymphoblastoid cells were examined on DNA treated with 8-methoxypsoralen (8-MOP) or 4,5',8-trimethylpsoralen (TMP) plus long wavelength ultraviolet (UVA) light, which produce monoadducts and DNA interstrand cross-links, and angelicin plus UVA light, which produces mainly monoadducts. 9 chromatin-associated DNA endonuclease activities were isolated from normal and XPA cells and assayed for activity on PM2 bacteriophage DNA that had been treated with 8-MOP or TMP in the dark and then exposed to UVA light. Unbound psoralen was removed by dialysis and a second dose of UVA light was given. Cross-linking of DNA molecules was confirmed by alkaline gel electrophoresis. In both normal and XPA cells, two DNA endonuclease activities were found which were active on 8-MOP and TMP plus UVA light treated DNA. One of these endonuclease activities, pI 4.6, is also active on intercalated DNA and a second one, pI 7.6, is also active on UVC (254 nm) light irradiated DNA. The major activity against angelicin plus UVA light treated DNA in both normal and XPA cells was found in the fraction, pI 7.6. The levels of activity of both of these fractions on all 3 psoralen-damaged DNAs were similar between normal and XPA cells. These results indicate that in both normal and XPA cells there are at least two different DNA endonucleases which act on both 8-MOP and TMP plus UVA light treated DNA.     

Xeroderma pigmentosum endonuclease complexes show reduced activity on and affinity for psoralen cross-linked nucleosomal DNA.

Two DNA endonuclease complexes have been isolated from the chromatin of normal human and xeroderma pigmentosum, complementation group A (XPA), lymphoblastoid cells which are active on DNA damaged with psoralen plus long wavelength ultraviolet radiation (UVA). In both normal and XPA cells, one endonuclease complex, pI 4.6, recognizes the psoralen cross-link and the other endonuclease complex, pI 7.6, recognizes the psoralen monoadduct. The levels of activity of these complexes from both normal and XPA cells are similar on damaged naked DNA. Kinetic analysis of assays using graduated concentrations of substrate revealed that selective activity of these endonuclease complexes on 8-MOP plus UVA treated DNA correlates with a reduction in Km of these complexes, indicating an increased affinity for, or rate of association with, damaged naked DNA. When the damaged substrates were reconstituted into core nucleosomes (without histone H1), both normal endonuclease complexes showed a 2.5-fold enhancement of activity, which correlated kinetically with a further increase in affinity, or rate of association (decreased Km), for this damaged nucleosomal substrate. This increase in activity and in affinity was reduced but not eliminated when histone H1 was present. By contrast, neither XPA endonuclease complex showed this enhanced activity on, or affinity for, damaged core nucleosomal DNA, and actually showed decreased activity, and affinity, when histone H1 was present. Introduction, via electroporation, of either of the normal complexes into 8-MOP plus UVA treated XPA cells in culture corrected their DNA-repair defect, further confirming the role of these complexes in the repair process.     

Enhancement of two apurinic/apyrimidinic endonuclease activities from normal but not xeroderma pigmentosum lymphoblastoid cells by nucleosome structure

The influence of nucleosomes on the activity of two chromatin-associated apurinic/apyrimidinic (AP) DNA endonuclease activities, pIs 9.2 and 9.8, from normal and xeroderma pigmentosum, complementation group A (XPA), lymphoblastoid cells was examined. These AP endonuclease activities were studied on non-nucleosomal and nucleosomal plasmid pWT830/pBR322 DNA which had been reconstituted with core (H2A, H2B, H3, H4) or total (core plus H1) histones from normal or XPA cells. Both nucleosomal and non-nucleosomal DNA was rendered partially AP by alkylation with 12.5 mM methyl methanesulfonate, followed by heating it at 70 degrees C, to produce approximately three AP sites per DNA molecule. The activities of both normal lymphoblastoid AP endonuclease activities on nucleosomal AP DNA, reconstituted with core histones, was approximately 2.5 times greater than that on non-nucleosomal AP DNA. When histone H1 was added to the system, this increase was reduced. XPA AP endonuclease activities, on the other hand, did not show any increase in activity on nucleosomal AP DNA reconstituted with core histones. These differences between normal and XPA endonuclease activities on AP nucleosomal DNA were the same regardless of whether histones from normal or XPA cells were used in the reconstituted system.     

Both cross-links and monoadducts induced in DNA by psoralens can lead to sister chromatid exchange formation

The relative importance of DNA-DNA cross-links and bulky monoadducts in sister chromatid exchange (SCE) formation was investigated in three human fibroblast cell lines with different repair capabilities. These cell lines included normal cells, which can repair both classes of lesions; xeroderma pigmentosum (XP) cells, which cannot repair either psoralen-induced cross-links or monoadducts; and an XP revertant that repairs only cross-links and not monoadducts. SCEs were induced by two psoralen derivatives, 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) and 5-methylisopsoralen (5-MIP). After activation with long-wave ultraviolet light, HMT produces cross-links and monoadducts in DNA, whereas 5-MIP produces only monoadducts. In normal human cells both psoralens induced SCEs, but if cells were allowed to repair for 18 h before bromodeoxyuridine (BrdUrd) was added for SCE analysis, the SCE frequency was significantly reduced. XP cells showed an SCE frequency that remained high regardless of whether SCEs were analyzed immediately after psoralen exposure or 18 h later. In the XP revertant that repairs only cross-links, both psoralens induced a high yield of SCEs when BrdUrd was added immediately after psoralen treatment. When XP revertant cells were allowed 18 h to repair before addition of BrdUrd, the SCEs induced by HMT were greatly reduced, whereas those induced by 5-MIP were only slightly reduced. These observations indicate that both cross-links and monoadducts are lesions in DNA that can lead to SCE formation.     

A damage-recognition protein which binds to DNA containing interstrand cross-links is absent or defective in Fanconi anemia, complementation group A, cells.

A DNA binding protein with specificity for DNA containing interstrand cross-links induced by 4,5',8-trimethylpsoralen (TMP) plus long wavelength ultraviolet (UVA) light has been identified in normal human chromatin. Protein binding to DNA was determined using a gel mobility shift assay and an oligonucleotide containing a hot spot for formation of psoralen interstrand cross-links. Specificity of the damage-recognition protein for cross-links was demonstrated both by a positive correlation between level of cross-link formation in DNA and extent of protein binding and by effective competition by treated but not undamaged DNA for the binding protein. Chromatin protein extracts from cells from individuals with the genetic disorder, Fanconi anemia, complementation group A (FA-A), which have decreased ability to repair damage produced by TMP plus UVA light, failed to show any protein binding to TMP plus UVA treated DNA. We have previously shown that these chromatin protein extracts contain a DNA endonuclease complex, pI 4.6, which specifically recognizes and incises DNA containing interstrand cross-links and which in FA-A cells is defective in its ability to incise this damaged DNA (Lambert et al. (1992) Mutation Res., 273, 57-71). Together, these findings suggest that the DNA binding protein identified is involved in recognition and repair of DNA interstrand cross-links.     

Genetic control of the bypass of mono-adducts and of the repair of cross-links photoinduced by 8-methoxypsoralen in yeast

A large UVA dose by itself induces lethal damage revealed in some repair-deficient strains of Saccharomyces cerevisiae. Following photoaddition of a monofunctional psoralen derivative, 3-carbethoxypsoralen, an extra killing effect is observed by applying a second high UVA dose, in conditions where a fraction of 8-methoxypsoralen (8-MOP) plus UVA-induced monoadducts are transformed into DNA cross-links. In an excision-repair-deficient context, the bypass of 8-MOP plus UVA-induced monoadducts is under the control of the RAD6+ gene product. However, when other steps of the mutagenic pathway are blocked by the rad18-2 or the pso1-1 mutations, bypass occurs. This is also true when in excision-deficient strains the recombinogenic pathway is blocked by the rad52-1 mutation. The recombinogenic pathway may be an alternative to the mutagenic pathway for bypass of monoadducts. The repair of the lesions induced by a second UVA dose applied after a first treatment by 8-MOP plus UVA [i.e. cross-links and other putative lesion(s)] is controlled by at least the RAD2+, RAD6+, RAD52+, PSO2+ and PSO1+ gene products. The role of the pathways involved is discussed according to the nature of the secondarily induced lesions.     

DNA semi-conservative synthesis in normal and Fanconi anemia fibroblasts following treatment with 8-methoxypsoralen and near ultraviolet light or with X-rays

Summary The effect of treatment with 8-methoxypsoralen (8-MOP) plus near-UV radiation (UVA) or with X-rays on the rate of DNA semi-conservative synthesis of fibroblasts from 10 Fanconi anemia (FA), two heterozygous, and three normal cell lines was studied. Following treatments with either X-rays or low doses of 8-MOP plus UVA leading to a majority of monoadducts over cross-links per genome, the FA and hetcrozygous cell lines were indistinguishable from normals: the transient inhibition of semi-conservative DNA synthesis was followed by the recovery of a nomral rate of synthesis. In contrast treatment with higher (but not saturating) doses of 8-MOP plus UVA allowed us to distinguish two classes among the FA cell lines. One class demonstrated a pattern of recovery similar to that of heterozygous and normal cell lines. This indicates that in such cell lines, the predominant lesion in this condition, the cross-links, do not arrest DNA synthesis and are likely to be normally repaired. Another class of FA cell lines did not show a recovery of a normal rate of DNA synthesis even after prolonged post-treatment incubation and although the proportion of cells in S phase was similar to that of the strains of the first category. This indicates that in such cell lines the repair of cross-links is inhibited at some step which is not necessarily the incision one.     

Xeroderma pigmentosum complementation group A protein acts as a processivity factor

We have previously shown that endonucleases present in a protein complex, which has specificity for cyclobutane pyrimidine dimers, locate sites of damage in DNA by a processive mechanism of action in normal human lymphoblastoid cells. In contrast, the endonucleases present in this complex from xeroderma pigmentosum complementation group A (XPA) cells locate damage sites by a distributive or significantly less processive mechanism. Since the XPA protein has been shown to be responsible for the DNA repair defect in XPA cells, this protein was examined for involvement in the mechanism of target site location of these endonucleases. A recombinant XPA protein, produced by expression of the normal XPA cDNA in E. coli, was isolated and purified. The results show that the recombinant XPA protein was able to correct the defect in ability of the XPA endonucleases to act by a processive mechanism of action on UVC irradiated DNA. These studies indicate that the XPA protein, in addition to a role in damage recognition or damage verification, may function as a processivity factor.     

W-reactivation and W-mutagenesis in lambda and T7 bacteriophages: a comparative study of the action of ultraviolet radiation (254 nm) and of the photosensitizing agents, 8-methoxypsoralen and angelicin

Monoadducts and interstrand cross-links are formed in DNA after psoralen plus light treatment of bacteriophage lambda . Survival and clear plaque mutation frequency of lambda after photosensitization with 8-methoxypsoralen (8-MOP) are increased when the wild type host is slightly UV-irradiated (W-reactivation and W-mutagenesis). The recA13, lexA1 and uvrA6 mutations block W-reactivation and W-mutagenesis of lambda treated with 8-MOP plus light. Using the technique of "repeated irradiation" we showed that the mutagenic effect of 8-MOP plus light treatment on phage is due mainly to formation of cross-links in DNA. The mutagenic activity of monoadducts had been studied by using angular furocoumarin, angelicin which forms mainly monoadducts in DNA. Upon W-mutagenesis of phage lambda treated with angelicin plus light a high mutagenic effect is observed. The results indicate that the mutagenic activity of monoadducts is 15-20 fold slower as compared to that of cross-links. W-reactivation and W-mutagenesis of UV-irradiated (254 nm) bacteriophage lambda are also observed after 8-MOP plus light treatment of Escherichia coli uvrA and wild type hosts. It is possible that the difference in mutagenic activity of psoralen adducts could depend on the repair mechanism of adducts: cross-links repair in bacterial and lambda DNA is controlled by lexA gene (error-prone SOS-repair mechanism), while monoadducts can be efficiently repaired by error-free excision and recombination.     

Repair in E. coli of transforming plasmid DNA damaged by psoralen plus near-ultraviolet irradiation

Treatment of DNA with psoralen plus near-ultraviolet irradiation gives rise to both monoadducts and cross-links. We have examined the repair of plasmid NTP16 DNA treated in this way in vitro and then used to transform E. coli. Monoadducts are found to be potentially lethal, and can be repaired by uvr-dependent and recA-dependent pathways. The presence of a related resident plasmid in the transformed cells can enhance the survival of the incoming damaged NTP16 DNA. This effect is not recA-dependent, and a similar effect (designated "resident enhanced repair") has been observed previously with UV-irradiated plasmids of this particular incompatibility group. Removal of unbound psoralen from the plasmid DNA and exposure to further NUV is known to increase the ratio of cross-links to monoadducts, and we demonstrate that such cross-linked plasmid DNA is not readily repaired following transformation. However in the presence of homologous DNA (related resident plasmid) there is evidence for the repair, and hence uptake by the cell, of cross-linked DNA.     

Defective replication of psoralen adducts detected at the gene-specific level in xeroderma pigmentosum variant cells.

Replication of damaged DNA is suspected to play an important role in cell cycle, genetic stability, and survival pathways. Using psoralen photoaddition as prototype DNA damage and the renaturing agarose gel electrophoresis technique to measure DNA cross-linking in individual genes, Vos and Hanawalt previously observed efficient bypass replication of psoralen monoadducts in human genes (J.-M. H. Vos and P. C. Hanawalt, Cell 50:789-799, 1987). To understand the mechanism of bypass replication in human cells, mutants affected in such a process would be useful. We now report that cells from individuals suffering from the hereditary recessive syndrome xeroderma pigmentosum variant (XPV) are hypersensitive to killing induced by photoactivated psoralen. In addition, analysis of psoralen-mediated DNA cross-linking in the rRNA genes indicated that although repair of psoralen adducts was similar to that of normal individuals, XPV cells were markedly deficient in the ability to bypass psoralen adducts during replication; in comparison with normal cells, approximately half as many monoadducts were bypassed during replication in XPV cells. Furthermore, in contrast to normal cells, replication of interstrand cross-links was not detected in XPV. This is the first demonstration of a deficiency in bypass replication detected at the gene-specific level in vivo. A model involving a strand-specific defect in recombinational bypass in XPV is proposed.     

Delivery of photoreactive psoralen derivatives to specific biological targets

Psoralens are bifunctional molecules which photoreact with the pyrimidine bases of nucleic acids to form monoadducts and diadducts, or interstrand cross-links. We have prepared psoralen derivatives with additional functional groups which can be specifically directed to chosen biological targets. A sulfhydryl-containing psoralen which can form site-specific cross-links in plasmid DNA has been used to study psoralen repair and mutagenesis. Cloned DNA containing psoralen monoadducts has been cross-linked to specific regions of viral RNA and used to probe virus assembly. A biotinylated psoralen derivative which binds specifically to avidin has been used to detect small amounts of DNA. Finally, a psoralen derivative of insulin has been used to deliver psoralen specifically to activated lymphocytes.     

Comparison of histones in normal and xeroderma pigmentosum lymphoblastoid cells

Histones from normal human and xeroderma pigmentosum, complementation group A (XPA), lymphoblastoid cells were compared both quantitatively, qualitatively and for binding affinity for DNA. Electrophoretic examination of the histones showed that all five major histone species were present in both cell groups and that there were no quantitative differences between normal and XPA histones. Binding affinity to [3H] mammalian DNA of the histones was determined. No significant differences were observed in binding of either normal or XPA histones to DNA.     

Repair of 8-methoxypsoralen monoadducts in mouse lymphoma cells

Studies of the repair of DNA lesions at biologically important doses is extremely difficult for most mutagens. With 8-methoxypsoralen (8-MOP) plus longwave ultraviolet light (UVA) as the lesion-inducing agent, however, it is easy to manipulate the relative frequency of different DNA adducts by means of a special experimental protocol (the tap-and-test protocol) and this can be used to measure repair of DNA adducts. Three classes of photoadducts are produced by 8-MOP plus UVA treatment: 3,4-cyclobutane monoadducts, 4',5'-cyclobutane monoadducts, and 8-MOP-DNA interstrand crosslinks. A monoadduct is formed when a photoactivated 8-MOP molecule reacts with a pyrimidine base. An 8-MOP-DNA interstrand crosslink is formed when an existing monoadduct is photoactivated to react with another pyrimidine base on the opposite DNA strand. Thus monoadducts are formed by absorption of one photon of light and crosslinks by absorption of two. In the tap-and-test experiments, cells were exposed to UVA in the presence of 8-MOP and then re-exposed to UVA in the absence of free 8-MOP so that only crosslinks can be produced by the second UVA treatment. By means of this technique we have previously shown that DNA crosslinks are much more effective than monoadducts at producing chromosomal damage (sister-chromatid exchanges and micronuclei) but not mutations (Liu-Lee et al., 1984). If L5178Y mouse lymphoma cells were able to remove monoadducts, incubation prior to the second UVA treatment should lead to decreases in the effect of re-irradiation, because fewer monoadducts would be available for crosslink formation. In this way, we have found that psoralen monoadducts are repaired in these cells and that about 70% of those capable of crosslink formation are removed or otherwise made unavailable for crosslink formation in 6 h.     

gamma-H2AX formation in response to interstrand crosslinks requires XPF in human cells

To further define the molecular mechanisms involved in processing interstrand crosslinks, we monitored the formation of phosphorylated histone H2AX (gamma-H2AX), which is generated in chromatin near double strand break sites, following DNA damage in normal and repair-deficient human cells. Following treatment with a psoralen derivative and ultraviolet A radiation doses that produce significant numbers of crosslinks, gamma-H2AX levels in nucleotide excision repair-deficient XP-A fibroblasts (XP12RO-SV) increased to levels that were twice those observed in normal control GM637 fibroblasts. A partial XPA revertant cell line (XP129) that is proficient in crosslink removal, exhibited reduced gamma-H2AX levels that were intermediate between those of GM637 and XP-A cells. XP-F fibroblasts (XP2YO-SV and XP3YO) that are also repair-deficient exhibited gamma-H2AX levels below even control fibroblasts following treatment with psoralen and ultraviolet A radiation. Similarly, another crosslinking agent, mitomycin C, did not induce gamma-H2AX in XP-F cells, although it did induce equivalent levels of gamma-H2AX in XPA and control GM637 cells. Ectopic expression of XPF in XP-F fibroblasts restored gamma-H2AX induction following treatment with crosslinking agents. Angelicin, a furocoumarin which forms only monoadducts and not crosslinks following ultraviolet A radiation, as well as ultraviolet C radiation, resulted only in weak induction of gamma-H2AX in all cells, suggesting that the double strand breaks observed with psoralen and ultraviolet A treatment result preferentially following crosslink formation. These results indicate that XPF is required to form gamma-H2AX and likely double strand breaks in response to interstrand crosslinks in human cells. Furthermore, XPA may be important to allow psoralen interstrand crosslinks to be processed without forming a double strand break intermediate.     

Immunochemical measurement of histone H3 in non-nucleosomal compartments of cultured mammalian cells

We measured histone H3 in the non-nucleosomal compartment of cultured mammalian cells by enzyme-linked immunoelectrotransfer blot assay of cytosolic proteins using affinity-purified rabbit anti-H3 IgG, and peroxidase-linked second antibodies. The cytosolic H3 level was estimated to be 0.5-1.0% of the nucleosomal H3 content in MH-134SC cells (mean generation time 11 h) and 3-4% in HeLa cells (mean generation time 22 h). It showed characteristic changes under the inhibitions of DNA and/or protein synthesis and during the cell cycle of HeLa cells. These indicate an inverse relationship between the cytosolic H3 level and the replicating activity of nuclear DNA. The possible implication of the non-nucleosomal histones in the regulation of histone gene expression is discussed.     

Formation by bifunctional furocoumarins of DNA crosslinks and their repair in cultured mouse embryo fibroblasts

Formation of crosslinks in DNA by three bifunctional psoralen derivatives plus UVA light in mouse embryo fibroblasts was evaluated by a NaI density gradient centrifugation method. Psoralen was shown to be a more active cross-linking agent than 8-methoxypsoralen. As for 4,5',8-trimethylpsoralen, it needed much lower concentrations and much less 365 nm light fluence to yield high percentages of crosslinked DNA. Repair of adducts formed by these psoralen derivatives was studied by splitting the irradiation dose into two equal parts separated by variously long dark repair periods. It was shown that essentially only monoadducts formed during the first irradiation period were repaired. These mouse embryo fibroblasts seem unable to repair interstrand DNA crosslinks.     

Induction and repair of psoralen cross-links in DNA of normal human and xeroderma pigmentosum fibroblasts

Skin fibroblasts from normal human subjects were exposed in vitro to long-wave ultraviolet radiation (UVA, 320–400 nm) alone, or in combination with 8-methoxypsoralen (8-MOP). DNA damage was analysed with the alkaline elution technique before and after post-treatment incubation of the cells at 37°C for various times.Cells treated with UVA at 1.1 J/cm² showed an increased DNA elution rate, which returned to the normal level within 30 min of post-treatment incubation. In cells treated with PUVA (8-MOP at 20 μg/ml plus UVA at 0.04 J/cm²), the alkaline elution rate was not different from untreated control cells, either before or after post-treatment incubation for times up to 7 days.When the PUVA treatment was followed first by a washing, to remove any unbound 8-MOP, and then by UVA (PUVA + UVA) at 1.1 J/cm², the alkaline elution rate decreased below the control level. During the post-treatment incubation of the PUVA + UVA-treated cells there was a gradual increase of the alkaline elution rate to a level significantly above that in control cells. This increase was observed after 30 min. It reached a miaximum after 24 h and remained after 7 days of post-treatment incubation. Cells from a patient with xeroderma pigmentosum of complementation group A, which were given the same PUVA + UVA treatment, did not show any change in the alkaline elution rate during the post-treatment incubation.If, as seems likely, an increased alkaline elution rate indicates an increase of DNA breaks, and a decreased alkaline elution rate indicates the sealing of breaks and/or the formation of cross-links, the results would suggest the following: (1) UVA irradiation in itself is capable of inducing DNA breaks, which are rapidly sealed during post-treatment incubation; (2) PUVA treatment induces mono-adducts, some of which appear to remain in the DNA for at least 7 days of post-treatment incubation and can be activated to form DNA cross-links by a second dose of UVA; (3) DNA cross-links induced by PUVA + UVA can be recognized by a repair process that involves the formation of DNA breaks. This process is not observed in xeroderma pigmentosum cells of group A.     

Normal rate of DNA breakage in xeroderma pigmentosum complementation group E cells treated with 8-methoxypsoralen plus near-ultraviolet radiation

Treatment of normal and xeroderma pigmentosum complementation group E skin fibroblasts with 8-methoxypsoralen plus repeated doses of near-ultraviolet radiation elicited a marked increase in DNA strand breakage during a subsequent incubation. No such induction of breaks was noted with cells from xeroderma pigmentosum groups A and D. The results suggest that the gene product which is deficient in xeroderma pigmentosum group E cells is involved in a critical step of DNA repair of far-ultraviolet photoproducts but not so in the repair of psoralen cross-links.