Cockayne syndrome: defective repair of transcription?

Alain J. van Gool, Gijsbertus T.J. van der Horst, Elisabetta Citterio, Jan H.J. Hoeijmakers

Author Affiliations

  1. Alain J. van Gool12,
  2. Gijsbertus T.J. van der Horst1,
  3. Elisabetta Citterio1 and
  4. Jan H.J. Hoeijmakers*,1
  1. 1 MGC Department of Cell Biology and Genetics, Erasmus University Rotterdam, PO Box 1738, 3000, DR Rotterdam, The Netherlands
  2. 2 ICRF Clare Hall Laboratories, South Mimms, EN6 3LD, UK
  1. *Corresponding author. E-mail: hoeijmakers{at}


In the past years, it has become increasingly evident that basal metabolic processes within the cell are intimately linked and influenced by one another. One such link that recently has attracted much attention is the close interplay between nucleotide excision DNA repair and transcription. This is illustrated both by the preferential repair of the transcribed strand of active genes (a phenomenon known as transcription‐coupled repair, TCR) as well as by the distinct dual involvement of proteins in both processes. The mechanism of TCR in eukaryotes is still largely unknown. It was first discovered in mammals by the pioneering studies of Hanawalt and colleagues, and subsequently identified in yeast and Escherichia coli. In the latter case, one protein, the transcription repair‐coupling factor, was found to accomplish this function in vitro, and a plausible model for its activity was proposed. While the E.coli model still functions as a paradigm for TCR in eukaryotes, recent observations prompt us to believe that the situation in eukaryotes is much more complex, involving dual functionality of multiple proteins.

Transcription‐coupled repair

Nucleotide excision repair (NER) is a cut and paste mechanism that utilizes the non‐damaged strand as template for an error‐free resynthesis of the excised lesion‐containing part of the damaged strand (for recent reviews on NER, see Hoeijmakers, 1994; Wood, 1996; and, for repair in general, Friedberg et al., 1995). The importance of this repair pathway is illustrated by its remarkable versatility in eliminating a very broad class of structurally diverse lesions, including UV‐induced cyclobutane pyrimidine dimers and 6/4 photoproducts.

One of the most urgent problems originating from damage to the DNA template is that the vital process of transcription becomes hampered. For over a decade it has been known that mammalian cells take special precautions and prioritize elimination of many types of DNA injury from active genes (Bohr et al., 1985; Madhani et al., 1986). The preferential repair of the active genome compartment is accounted for largely by the faster repair of the transcribed strand (Mellon et al., 1987). This specialized strand‐directed form of NER, designated transcription‐coupled repair (TCR), occurs for a number of lesions for which the default NER mechanism, the global genome subpathway, is too slow, such as for the above‐mentioned UV‐induced cyclobutane pyrimidine dimers. 6/4 Photoproducts, on the other hand, exemplify a class of damage for which removal by the global genome system is already very fast; a contribution by TCR only becomes visible when the global genome NER pathway is impaired (van Hoffen et al., 1995).

The universal nature of TCR is suggested by its conservation from man to Saccharomyces cerevisiae (Smerdon and Thoma, 1990; Leadon and Lawrence, 1992; Sweder and Hanawalt, 1992) and Escherichia coli (Mellon and Hanawalt, 1989). Enhanced repair of transcribed strands seems to be restricted to genes transcribed by RNA polymerase II, since no strand bias could be demonstrated for RNA polymerase I‐ and III‐transcribed sequences (Vos and Wauthier, 1991; Christians and Hanawalt, 1993; Damman and Pfeifer, 1997). However, again, inactivation of the global repair system revealed a contribution of TCR to RNA polymerase I‐transcribed genes in yeast (Verhage et al., 1996a) and in human cells (van Hoffen et al., 1995). Finally, the special system for repair of active genes is not only found for in vitro cultured cells but is also observed in UV‐exposed skin in situ (Ruven et al., 1993), underlining its biological relevance.

Dual involvement of proteins in transcription and repair

An additional but distinct link between transcription and repair is represented by proteins with a direct role in both processes. The prototype example of dual involvement is the basal transcription factor TFIIH, consisting of nine subunits. The TFIIH complex is also indispensable for DNA excision repair (reviewed by Hoeijmakers et al., 1996). Mutations in the XPB and XPD subunits of human TFIIH were found previously to cause some forms of the DNA repair disorder xeroderma pigmentosum (XP, see below) (Weber et al., 1990; Weeda et al., 1990; Flejter et al., 1992). Moreover, mutagenesis of several subunits of yeast TFIIH rendered UV‐sensitive alleles, while mammalian and yeast mutants have reduced DNA repair activities in vivo and in vitro, clearly indicating an involvement of TFIIH in nucleotide excision repair (Feaver et al., 1993; Drapkin and Reinberg, 1994; van Vuuren et al., 1994; Sweder et al., 1996a).

It is tempting to assume that the dual involvement of TFIIH in transcription and repair provides the basis for enhanced repair of the transcribed strand of an active gene. However, it should be stressed that in vivo and in vitro repair analysis of yeast and human TFIIH mutants indicated a requirement for TFIIH in both transcription‐coupled and transcription‐independent, global genome DNA repair (van Vuuren et al., 1994; Vermeulen et al., 1994; Wang et al., 1995; Sweder et al., 1996a). Clearly other factors are required for the coupling between transcription and repair, and these will be discussed below.

Transcription‐coupled repair in E.coli

NER in E.coli is performed by the UvrABC endonuclease (Table I) that is able to recognize and repair a surprisingly wide variety of lesions (reviewed by van Houten, 1990). The occurrence of TCR in E.coli was shown by strand‐specific repair analysis of the lactose operon under induced conditions in vivo (Mellon and Hanawalt, 1989). Using an in vitro system, a 130 kDa transcription–repair coupling factor (TRCF) was purified, containing RecG‐like helicase motifs (Selby and Sancar, 1991). TRCF possesses an ATPase activity and was shown to recognize, bind and displace an RNA polymerase stalled on a lesion (Selby and Sancar, 1993). Subsequently, the protein stimulates recruitment of the repair machinery, presumably via an interaction with the damage recognition component UvrA (Selby and Sancar, 1993).

View this table:
Table 1. Proteins involved in each step of the core of nucleotide excision repair in E.coli, yeast and man

Intriguingly, in addition to TRCF, mismatch repair proteins are reported to be involved in vivo in TCR in E.coli and also in man, thus providing a link between mismatch repair and TCR (Mellon and Champe, 1996; Mellon et al., 1996). However, in vitro, purified mismatch repair proteins MutS and MutL could not mediate TCR, nor were mutS and mutL cell‐free extracts found to be deficient in TCR (Selby and Sancar, 1995). Moreover, various yeast mismatch repair mutants failed to exhibit any defect in TCR (Sweder et al., 1996b). It is therefore unclear at present whether the observed TCR defects associated with mismatch repair deficiencies have a direct or an indirect origin.

Transcription‐coupled repair in yeast

The first indications of TCR in S.cerevisiae were provided by repair studies of the two mating‐type loci, MATα and HMLα, which are identical in sequence, but the HMLα locus is transcriptionally silenced by the SIR proteins (Terleth et al., 1989). It was found that MATα was repaired more efficiently than HMLα. However, in this case, the difference in local chromatin structure turned out to be the predominant determinant (Brouwer et al., 1992), indicating that chromatin conformation also influences NER. Bona fide TCR subsequently was demonstrated for URA3 (Smerdon and Thoma, 1990), RPB2 (Sweder and Hanawalt, 1992) and induced GAL7 (Leadon and Lawrence, 1992). The dependency of fast strand‐specific repair on active RNA polymerase II transcription was revealed by the absence of TCR in temperature‐sensitive RNA polymerase II mutants under non‐permissive conditions (Leadon and Lawrence, 1992; Sweder and Hanawalt, 1992).

At present, only one factor, Rad26, is known to be involved specifically in TCR in yeast (Table I) (van Gool et al., 1994). The RAD26 gene was isolated based on homology with human CSB whose product, together with CSA, is selectively required for TCR in man (see below). By rad26 disruption, the preferential repair of the transcribed strand of the active RPB2 gene (van Gool et al., 1994) and of the PHO5 PHO3 locus (our unpublished results) is severely diminished but not completely lost. In contrast to its human equivalent (Troelstra et al., 1992), but in analogy to the E.coli trcf mutant (Selby and Sancar, 1993), RAD26 disruption did not lead to enhanced sensitivity to UV irradiation. This can be explained by the higher efficiency of global genome repair in yeast compared with man, which compensates for the loss of TCR. Indeed, evidence was obtained for an overlap between the two pathways in yeast (Verhage et al., 1996b). The rad26 mutant displayed a slower recovery of growth after UV irradiation, which might indicate a functional role for TCR in the biology of yeast under non‐laboratory conditions (van Gool et al., 1994).

Surprisingly, a residual level of TCR in the absence of Rad26 was still found when the global genome repair system was additionally inactivated by disruption of the RAD7 or RAD16 gene (Verhage et al., 1996b). The latter gene products are implicated selectively in repair of inactive DNA, including the non‐transcribed strand of active genes (Verhage et al., 1994). The double mutants displayed a synergistic increase in UV sensitivity, but were not as sensitive as a total repair‐deficient mutant such as rad1 (Verhage et al., 1996b). By analysing repair in triple yeast mutants, it could be excluded that the second TCR activity is encoded by Rad28 (the yeast homologue of human CSA) (Bhatia et al., 1996), by the elongation factor SII (Verhage et al., 1997) or by yeast mismatch repair proteins (Sweder et al., 1996b). Thus, the nature of this second CS‐independent TCR activity remains unknown, stressing the complexity of the transcription–repair interface.

Transcription‐coupled repair in man

At present, three distinct genetically heterogeneous human disorders have been associated with a defect in NER: XP, Cockayne syndrome (CS) and PIBIDS, the photosensitive form of trichothiodystrophy (TTD) (Lehmann, 1987; Nance and Berry, 1992; Stefanini et al., 1993; for a recent review, see Bootsma et al., 1997). Cell complementation studies revealed seven gene products to be involved in XP (XPA–XPG), and two in CS (CSA and CSB) (Table II). In addition, XP groups B, D and G include some patients with CS symptoms in addition to XP features. Furthermore, NER‐deficient TTD patients are assigned to three complementation groups, two of which correspond to XP and combined XP/CS groups (XPB, XPD and TTDA), again revealing overlap of syndromes. Repair analysis of cell lines derived from patients showed that most XP and TTD gene products are required for both NER subpathways, with the notable exception of XPC (Hoeijmakers, 1993). XP‐C cells are deficient in global genome repair in man, while possessing normal TCR activity (Venema et al., 1990b). On the other hand, CS‐A as well as CS‐B cells are deficient in enhanced repair of the transcribed strand of the active genes ADA and DHFR, while having normal global genome repair (Venema et al., 1990a; van Hoffen et al., 1993). Moreover, microinjection of antibodies raised against the CSA and CSB proteins inhibited the residual repair in XP‐C fibroblasts, indicating their requirement for TCR (A.J.van Gool et al., submitted). These findings reveal the existence of NER subpathway‐specific human factors.

View this table:
Table 2. Features of NER‐deficient XP, CS and TTD complementation groups

The consequences of TCR deficiency in man are severe, in contrast to the same mutations in E.coli and yeast (Selby and Sancar, 1993; van Gool et al., 1994). CS patients typically suffer from developmental and neurological abnormalities, including neuro‐dysmyelination, immature sexual development, mental retardation and impaired physical development manifested by cachectic dwarfism, microcephaly, skeletal and retinal abnormalities and a characteristic ‘bird‐like’ face. Death results from progressive neurological degeneration, in most cases before the age of 20 (Lehmann, 1987; Nance and Berry, 1992). In accordance with their defect in DNA repair, most CS patients display an increased photosensitivity of the skin. However, in contrast to XP, they have not been reported to develop skin tumours (Lehmann, 1987; Nance and Berry, 1992). When compared with totally NER‐deficient XP patients, it is remarkable that CS individuals exhibit many additional symptoms whereas their NER defect is only restricted to the mechanism for faster repair of the transcribed strand. Most XP individuals have a normal development but display predominantly cutaneous features, including photosensitivity, pigmentation abnormalities and predisposition to skin cancer. The non‐XP characteristics in CS point to an additional function of the CS proteins.

The Cockayne syndrome A and Cockayne syndrome B proteins

The CSA gene recently was isolated and predicted to encode a protein of 44 kDa containing five WD repeats (Henning et al., 1995). Such domains are found in a large family of proteins implicated in a diverse range of cellular activities, and may function by stimulating formation of multi‐protein complexes (Neer et al., 1994). The CSB gene, previously isolated as ERCC6 (Troelstra et al., 1990), encodes a protein of 168 kDa with a strongly conserved middle part (Troelstra et al., 1992; van Gool et al., 1994). This region contains motifs shared with a large family of helicases and, more specifically, with the entire ATPase domain of the still expanding Swi2/Snf2 subfamily (Troelstra et al., 1992; Gorbalenya and Koonin, 1993). Interestingly, this family includes participants in all major multi‐enzyme DNA repair systems (including both NER subpathways), transcription activation and repression as well as preservation of chromosome stability (reviewed in Carlson and Laurent, 1994; Eisen et al., 1995). No overt DNA unwinding function has been shown for any of these proteins but, recently, compelling evidence was obtained that the common domain may disrupt protein–DNA interactions in an ATP‐dependent fashion (Hirschhorn et al., 1992; Peterson and Tamkun, 1995). The yeast Swi2/Snf2 protein is part of a large complex of at least 10 proteins (Cairns et al., 1994) which is able to disrupt a nucleosome in vitro, thus allowing binding of the GAL4 transcription activator (Côté et al., 1994; Kwon et al., 1994) or the TATA‐binding protein (Imbalzano et al., 1994). Nucleosome disruption might be induced by changes in helical twist that are generated by Swi/Snf binding to promoter DNA (Quinn et al., 1996). A similar, yet distinct activity has been reported for the human nucleosome remodelling factor as well as for the yeast RSC complex, containing the Swi2/Snf2‐like ATPases ISWI and STH1 respectively (Tsukiyama et al., 1995; Cairns et al., 1996). Finally, the yeast Mot1 protein is able to disrupt binding of the TATA‐binding protein to a promoter sequence, and thus inhibits transcription initiation (Auble et al., 1994). In all cases, disruption required the ATPase activity of the Swi2/Snf2 family members, which is stimulated by naked DNA (Swi2/Snf2), or a fully assembled nucleosome (ISWI), indicating a target specificity. Recently, it was shown that purified recombinant Rad26 (Guzder et al., 1996) as well as purified CSB protein (Selby and Sancar, 1997; our unpublished results) possesses a strong ATPase activity which is stimulated by double‐ (and to a lesser extent single‐) stranded DNA. However, in none of the above cases has a standard helicase activity been found. The recent report on the crystal structure of a DEXX helicase (Subramanya et al., 1996) reveals a structural similarity of the helicase domain with the ATP‐binding core of the strand exchange protein RecA. All together, this opens up the possibility that the CSB protein also performs some type of local DNA strand separation, influencing DNA topology and thus remodelling its target protein–DNA complex.

Using immunoprecipitations of in vitro translated proteins or GST pull‐down assays, interactions between CSA and p44 (a subunit of TFIIH), CSA and CSB (Henning et al., 1995), CSB and XPG (Iyer et al., 1996), CSB and XPA, CSB and the p34 subunit of TFIIE, and CSB and XPB (Selby and Sancar, 1997) were claimed. On the other hand, no physical interactions of significant quantities of these proteins with CSA and CSB could be demonstrated in cell‐free extracts that are competent in performing NER and transcription in vitro (A.J.van Gool et al., submitted). Thus, associations between the CS proteins and some of the other transcription and repair proteins may have a transient character, which occur, for example, when the cell is challenged with genotoxic agents.

A tentative model for TCR in humans, which is analogous to the proposed mechanism of TCR in E.coli, has been suggested previously (Hanawalt et al., 1994). The initial step is formed by an elongating RNA polymerase II complex blocked by a lesion in the transcribed strand. The stalled RNA polymerase has to retract or dissociate to permit access of repair proteins to the injury (Selby and Sancar, 1990; Donahue et al., 1994). In vitro, the elongation factor SII is able to stimulate retraction, preceding transcriptional read‐through, by promoting the 3′→5′ exonuclease activity of the RNA polymerase II molecule (Reines et al., 1989; Izban and Luse, 1992). However, by itself, this appeared insufficient for repair proteins to reach the lesion (Donahue et al., 1994). In analogy with other Swi2/Snf2‐like proteins, CSB may mediate the partial disruption of RNA polymerase binding to DNA, thereby facilitating the backtracking. The recent report that recombinant CSB, on its own or with CSA, was unable to remove a stalled RNA polymerase in vitro (Selby and Sancar, 1997) suggests that a combination of proteins is required to mediate efficient upstream translocation. The WD repeats in CSA could serve to stabilize transient interactions between CSB and the stalled transcription complex. In addition, these domains, perhaps together with unidentified regions in CSB, may be involved in the transient interaction with the repair machinery, and thus stimulate repair of the lesion.

Dual functionality of the CSA and CSB proteins?

The question remains whether the requirement for CSA and CSB in TCR reflects their function as factors that couple repair to blocked transcription, as described above, or whether they are (also) required for the transcription process itself. In the latter scenario, mutations in CSA and CSB would abolish transcription, and consequently also TCR. However, the fact that both CS genes can be disrupted completely in yeast, mice and man (Troelstra et al., 1992; van Gool et al., 1994; Henning et al., 1995; Bhatia et al., 1996; van der Horst et al., 1997, and unpublished results) directly indicates that neither CSA nor CSB is essential for transcription. Moreover, microinjection of CSA and CSB antibodies had no effect on transcription in vivo, whereas they did inhibit recovery of RNA synthesis after UV exposure. Furthermore, complete depletion of CSB from active cell‐free extracts did not significantly affect transcription in vitro (A.J.van Gool et al., submitted), indicating that CSA and CSB do not largely contribute to basal transcription. However, several observations support the possibility that these proteins may have a non‐essential, auxiliary function in transcription.

(i) The phenotype of CS patients cannot be rationalized easily on the sole basis of a repair defect (Bootsma and Hoeijmakers, 1993). As outlined above, the characteristic clinical features of CS merely reflect neurological and developmental abnormalities, which are not apparent in XP‐A patients that are totally deficient in NER (Bootsma et al., 1997). Comparable features have also been found in XP‐B, XP‐D, TTD‐A and XP‐G patients, of which the first three have been shown to carry mutations in subunits of the transcription/repair factor TFIIH (Schaeffer et al., 1993; van Vuuren et al., 1994; Vermeulen et al., 1994). Moreover, there are patients that display many of the CS hallmarks except for sun sensitivity and underlying TCR‐related defect in RNA synthesis recovery after UV exposure (Nance and Berry, 1992; our unpublished results). This is consistent with the idea that the CS proteins have an additional function beyond their involvement in NER. The striking phenotypic parallels with the CS symptoms resulting from mutations in the repair/transcription factor TFIIH suggests that the non‐XP CS features are derived from influencing basal transcription.

(ii) CSB−/− knock‐out mice have been generated to study the function of CSB in a multicellular organism (van der Horst et al., 1997). Fibroblasts derived from these mice display a repair deficiency that is very similar to yeast rad26 and human CS‐B cells, primarily reflected in a severe defect in TCR of UV‐induced CPD lesions and concomitantly in recovery of RNA synthesis. Phenotypically, the CSB−/− mice do not display the characteristic hallmarks of CS as dramatically as the human patients, but minor growth disturbance and neurological deficits have been noted (van der Horst et al., 1997). Unexpectedly, when the CSB−/− mice were crossed with knock‐out mice of either the XPC or the XPA gene, which do not show any obvious developmental abnormalities (De Vries et al., 1995; Nakane et al., 1995; Sands et al., 1995), a strong synergistic effect was observed. The CSB−/−XPC−/− as well as the CSB−/−XPA−/− mice display a very severe growth impairment, suffer from neurological problems and die before weaning (our unpublished observations). Since crossings between XPA–/−XPC−/− mice yielded normally developing offspring, the CSB defect in combination with a total NER deficiency apparently leads to dramatically pronounced CS features. This suggests that CSB has an additional cellular function besides its involvement in NER. As mentioned above, XP complementation group G also includes patients with very severe CS features, which are clearly more pronounced than those associated with classical CS‐A and CS‐B (Vermeulen et al., 1993; our unpublished observations). This suggests that XPG, one of the endonucleases of NER (Table II), also has a CS‐like additional function. Since the XPG NER defect involves TCR as well as a global genome repair, the same synergistic effect may underlie the severe clinical outcome of human XPG mutations as that found in the CSB–XPA and CSB–XPC double knock‐out mice. A prediction of these considerations is that XPG‐deficient mice show essentially the same symptoms as the CSB–XPA or –XPC double mutant mice.

(iii) Ionizing radiation induces a variety of lesions that for a large part are repaired by means other than NER, since totally NER‐deficient XP‐A cells are proficient in repair of such lesions (Leadon and Cooper, 1993). Among these are oxidative damages, such as thymine glycols, that have been shown to block ongoing transcription (Htun and Johnston, 1992) and that are repaired in a transcription‐coupled way (Leadon and Lawrence, 1992). Surprisingly, CS cells display a slightly increased sensitivity to ionizing radiation, and were shown to be impaired in strand‐specific removal of ionizing radiation‐induced lesions (Leadon and Cooper, 1993). Since oxidative damage can also arise from intracellular metabolic processes, it was suggested that failure of transcriptional bypass of these lesions leads to the observed clinical features in CS patients (Leadon and Cooper, 1993). Recently, it was shown that cells from those XP‐G patients that also display severe CS features are defective in TCR of thymine glycols as well (Cooper et al., 1997). Thus, in analogy with CS, this form of XP‐G may also be caused by impaired transcriptional bypass of damaged DNA.

(iv) The genotoxic agent N‐acetoxy‐2‐acetylaminofluorene (NA‐AAF) is converted mainly into a dG‐C8‐AF adduct in human cells (van Oosterwijk et al., 1996). It is known that TCR does not contribute significantly to repair of dG‐C8‐AF lesions in normal cells, and consequently repair rates of active genes in CS cells are very similar to those in normal cells. However, CS‐A and CS‐B cells show clearly hypersensitivity and an inability to recover their RNA synthesis after NA‐AAF treatment (van Oosterwijk et al., 1996). This suggests that the NA‐AAF sensitivity of CS cells is not caused by impaired TCR per se, but instead may reflect trapping of essential transcription factors preventing re‐initiation of transcription after NA‐AAF treatment.

(v) A recent study revealed a role for CSA and CSB in UV‐induced modification of RNA polymerase II. After exposure of cells to UV irradiation or cisplatin, the large subunit of RNA polymerase II becomes ubiquitinated (Bregman et al., 1996). Ubiquitin conjugation of proteins has pleiotropic consequences for many cellular pathways, including DNA repair, sporulation, cell cycle progression and transcription (reviewed in Jentsch, 1992). Besides proteolytic degradation, ubiquitination can also result in modification and alternative processing of protein complexes (Chen et al., 1996; Hicke and Riezman, 1996; Hochstrasser, 1996; Wang et al., 1996). Surprisingly, the UV‐induced ubiquitination of RNA polymerase II was specifically reduced in CS‐A and CS‐B cells, but could be restored by transfection of the corresponding cDNAs, strongly implying a direct or indirect role for CSA and CSB (Bregman et al., 1996). The RNA polymerase II ubiquitination defect seems to be correlated with absence of CSA or CSB rather than with defective TCR, since wild‐type ubiquitination is observed in totally repair‐deficient XP‐A, XP‐B and XP‐D fibroblasts (D.B. Bregman, personal communication). Interestingly, a low level of ubiquitination of RNA polymerase II is also observed without genotoxic treatment. Assuming that this type of modification has a functional role, mutations in CSA or CSB might affect the regulation of the enzyme and thus influence transcription. In agreement with these observations, we recently found evidence that in whole cell extracts a significant portion of RNA polymerase II is associated specifically with CSB (A.J.Van Gool et al., submitted).

The above considerations strongly suggest that the CS and XPG proteins have additional functions besides their role in DNA repair. The fact that the non‐XP CS symptoms become more severe when a CSB defect is placed in a total NER‐deficient background indicates that unremoved damage synergizes with the additional CSB function. One non‐NER function therefore might be to detrap blocked transcription machinery either by dissociation of a stalled RNA polymerase complex or by promoting bypass via translesion RNA synthesis. This step could involve the CSA/CSB‐dependent ubiquitination of RNA polymerase. The main objective of this procedure might be the release of ‘trapped’ transcription to make it again available for RNA synthesis. The second (NER) function would be to recruit the NER machinery to accomplish TCR.

The non‐NER CS features may arise primarily from transcription insufficiency that develops in the course of time in specific tissues as the consequence of the accumulation of persisting lesions in the genome. Without TCR, such lesions trap essential transcription factors, e.g. elongating RNA polymerase complex, thereby interfering with transcription of the same and possibly other genes. Endogenous cellular metabolism is known to be a cause of DNA injury (e.g. oxidative damage) that may comprise, in part, substrates for the NER system and, in part, for other repair processes such as base excision repair. When global genome repair is still intact (like in the case of classical CSA and CSB mutations), this NER subpathway can function as a backup that takes care of removal of accessible lesions from transcribed strands, albeit more slowly. The rate of removal then depends on the activity of this NER subpathway in the specific cell type for the specific lesion. However, not all types of DNA damage are recognized by global genome repair or other repair mechanisms. Consequently, in CS, inevitably such types of lesions will still accumulate, particularly in post‐mitotic cells with a long lifespan and no cell renewal system, such as neurones. Tissues like skin with continual proliferation, will ‘dilute’ their damage because of DNA replication and thus not suffer from this problem.

In conclusion, we hypothesize that CS patients suffer from two defective processes (Figure 1): impaired TCR and an inability to release trapped transcription. In an XP‐A patient, lacking both TCR and global genome repair, the latter ‘backup’ system is not available but the transcription trapping problem does not occur because the CSA/CSB‐dependent turnover of transcription components is still active. In XP‐C, the active TCR mechanism compensates for the detrapping as well as for efficient removal of the lesions from the transcribed strand of active genes. Finally, as argued above, the severe clinical CS features resulting from XPG mutations might be explainable by the synergistic combination of a defect in both NER subpathways as well as in the release of trapped transcription. Future research should reveal whether this scenario is a reality.

Figure 1.

Speculative model for the functions and clinical impact of factors involved in transcription‐coupled repair, global genome repair and release of the transcription machinery blocked by a DNA lesion in the transcribed strand. See text for further explanation.


We thank all members of our laboratory for helpful discussions. The work reported here is supported by the Dutch Scientific Organization (Section Medical Sciences, project 900‐501‐093), Human Frontier Science Program (RG373) and an EC‐TMR network (contract No. CHRX‐CT94‐0043).