Mutation Research, 250 (1991) 223-228

223

© 1991 Elsevier Science Publishers B.V. All rights reserved 0027-5107/91/$03.50 ADONIS 002751079100181R MUT 02546

Hierarchies of DNA repair in mammalian cells: biological consequences Leon H.F. Mullenders l,z, Harry Vrieling 1,2, Jaap Venema l and Albert A. van Zeeland 1,2 t MGC-Department of Radiation Genetics and Chemical Mutagenesis, University of Leiden, Leiden and 2 J.A. Cohen Institute, Interunicersity Research Institute for Radiopathology and Radiation Protection (The Netherlands)

(Accepted 24 May 1991)

Keywords: DNA repair; Hierarchy; Mammalian cells; Biologicalconsequences

Summary Mammalian cells exposed to genotoxic agents exhibit heterogeneous levels of repair of certain types of DNA damage in various genomic regions. For UV-induced cyclobutane pyrimidine dimers we propose that at least three levels of repair exist: (1) slow repair of inactive (X-chromosomal) genes, (2) fast repair of active housekeeping genes, and (3) accelerated repair of the transcribed strand of active genes. These hierarchies of repair may be related to chromosomal banding patterns as obtained by Giemsa staining. The possible consequences of defective DNA repair in one or more of these levels may be manifested in different clinical features associated with UV-sensitive human syndromes. Moreover, molecular analysis of hprt mutations reveals that mutations are primarily generated by DNA damage in the poorly repaired non-transcribed strand of the gene.

The complexity of enzymatic mechanisms that have evolved for recognition and repair of DNA damage is undoubtedly related to the complexity of the substrate itself, namely the condensed structure of DNA in chromosomes and chromatin. The complex structure of chromatin may influence both the induction and processing of DNA damage within various parts of the genome that exhibit diverse molecular structures and activities. Different levels of organization of the genome can be observed both in mitotic chromosomes and in interphase chromatin. Mammalian Correspondence: Dr. L.H.F. Mullenders, MGC-Department of Radiation Genetics and Chemical Mutagenesis, University of Leiden, Wassenaarseweg 72, 2333 AL Leiden (The Netherlands).

chromosomes can be stained with Giemsa to reveal a longitudinal pattern of alternating dark (G) and light (R) bands, due to an alternating pattern of differences in DNA and protein composition along the chromosomes (Yunis, 1981; Holmquist et al., 1982). Early replicating constitutively expressed genes reside in R-bands whereas the late replicating tissue-spi~cific genes are localized in G-bands. Based on replication banding patterns obtained by adding bromodeoxyuridine during various stages of the cell cycle (Latt, 1973), bands have been equated to replicon clusters, units of chromosomal replication (Hand, 1978). At the chromatin level, replicon clusters are composed of DNA loops which are attached to the nuclear matrix and there is a large body of evidence to suggest that the DNA loop is the unit of replica-

224 tion, i.e., the replicon (Dijkwel et al., 1979; Aelen et al., 1983). Various experimental approaches have revealed that the attachment sites of DNA loops at the nuclear matrix encompass origins of replication (Dijkwel et al., 1989; BuongiorniNardelli et al., 1982). Moreover, the matrix-associated DNA is enriched in transcriptionally active genes with their attachment sites usually near the 5' ends. Since transcriptional activity and early gene replication appear to be coupled (Goldman et al., 1984), it has been proposed by Zehnbauer and Vogelstein (1985)that transcriptional activity of a gene is associated with the activation and matrix association of a neighboring replication origin. In mammalian cells the activity of DNA-repair processes is not homogeneously distributed over the gcnome. For ultraviolet light as well as for certain chemical carcinogens repair of transcriptionally active housekeeping genes has been demonstrated to be more rapid than that of tissue-specific genes, regions of non-coding DNA and the genome overall (Bohr et al., 1987). The preferential repair of UV-induced photolesions in active genes correlates well with the preferential occurrence of repair sites at the attachment sites of DNA loops at the nuclear matrix (Mullenders et al., 1988). The interesting possibility that different levels of repair in genomic regions are related to the chromosomal banding pattern has not been addressed experimentally so far. From a biological point of view, processes of DNA-excision repair in mammalian cells serve to alleviate the lethal, mutagenic and carcinogenic consequences of DNA damage. Therefore, intragenomic heterogeneity of repair could have major implications for DNA-damage frequencies in various genomic regions. As a consequence poorly repaired non-expressed DNA sequences might accumulate DNA damage or mutations. However, effects of heterogeneous repair on mutation induction can currently only be deduced for some expressed genes. In this study we briefly review the current knowledge on heterogeneity of DNA repair, propose a model describing the existence of hierarchies of DNA repair and discuss the ultimate consequences of heterogeneous repair for mutagenesis in UV-irradiated mammalian cells.

Methods and materials

Diploid normal human, xerodcrma pigmentosum and Cockayne's syndrome fibroblasts were cultured as described previously (Venema ct al., 1990a,b). UV-irradiation was performed with a Philips TUV lamp (predominantly 254 nm) and cells were incubated for various periods of time to allow repair. The removal of cyclobutane pyrimidine dimers in specific sequences was analyzed with the dimer-specific enzyme T4 endonuclease V as described previously (Vcnema et al.. 1990a,b). For mutation studies Chinese hamster ceils were cultured under standard conditions, UVirradiated and independent H P R T mutants were isolated. Sequence analysis of H P R T c-DNA was performed as described previously (Vrieling ct al.. 1988, 1991). Results and discussion

Hierarchies of DNA repair in mammalian cells Intragenomic heterogeneity in UV-induced repair has been observed both in rodent and in human cells (Bohr et al., 1987). The methodology for measuring repair of cyclobutanc pyrimidine dimers (CPD) in defined sequences was initially developed and described for C H O cells containing amplified dihydrofolatc reductase ( D H F R ) sequences. Within the amplified domain CPD were removed faster and more efficiently from the active D H F R gene than from inactive flanking sequences or the genome overall. Further insight into the role of transcription in DNA repair was provided by the discovery that preferential repair of CPD in the D H F R gene in CHO cells was confined to the transcribed strand only. Preferential and strand-specific repair of CPD in transcriptionally active genes has also been reported for immortalized and primary human cells (Mellon et al., 1986, 1987; Venema et a[., 1990; Kantor et al., 199(I). In this case preferential repair concerns differences in the rate of repair, since human cells are able to perform complete removal of CPD from their genome. The kinetics and extent of repair of the non-transcribed strand of active genes was very similar to that of the genome overall suggesting the absence

225 TABLE 1 PERCENT REPAIR OF CYCLOBUTANE PYRIMIDINE DIMERS IN RESTRICTION FRAGMENTS OF (IN)ACTIVE GENES IN CONFLUENT PRIMARY HUMAN FIBROBLASTS Cell line ~ VH25, VHI6 c GM38 d ADA - SCID c

Gene b ADA 90 90

DHFR 90 90

754 50 30 40

Factor IX 40

a Cells were UV-irradiated with 10 J/m 2 and allowed to repair for 24 h. ADA (EcoRI): 18.5 kb; DHFR (Hindlll): 22 kb; 754 (EcoRl): 14 kb; Factor 1X (Kpn): 16 kb. c From Venema et al., 1990. d From Kantor et al., 1990. e Absence of transcription of the ADA gene due to a deletion in promotor region (Berkvens et al., 1987).

of gross variations in repair among the different genomic regions (Mellon et al., 1987). However, direct comparison of genomic sequences reveals distinct differences in rate and extent of repair between expressed housekeeping genes such as adenosine deaminase ( A D A ) and D H F R and two non-expressed X-chromosomal loci, i.e., 754 and factor IX (Table 1). In three different normal human fibroblast cell lines derived from male donors, only 3 0 - 5 0 % of C P D were removed from the 754 locus, compared to an almost complete repair of the A D A and D H F R genes. The low efficiency of repair was also observed for the inactive factor IX gene. The question whether the different levels of repair efficiencies in active and inactive genes are mediated by the transcription process, was addressed by investigating repair in the absence of transcription of the A D A gene. As shown in Table 1, a profound difference in repair efficiency of the A D A and 754 genes is still observed in the absence of A D A transcription, implying that other factors than transcription underlie the more efficient repair of housekeeping genes. The data currently available suggest the existence of several hierarchies of D N A repair of CPD: (1) a slow repair of transcriptionally inactive (X-chromosomal) chromatin; (2) fast repair of transcriptionally poised or active chromatin;

(3) accelerated repair of the transcribed strand of transcriptionally active genes. The localization of the active housekeeping genes and the inactive factor IX gene in chromosomal R- and G-bands, respectively, challenges us to propose that repair efficiencies of various genomic regions might be related to the cytogenetically determined chromosomal banding pattern. In general terms D N A sequences that reside in early replicating R-bands may be repaired fast and efficiently, whereas repair of late replicating G-band D N A is slow and less efficient. In support of this hypothesis are recent data on repair of (in)active sequences in embryonic cell lines of Drosophila melanogaster (de Cock et al., 1991). No differences in kinetics or extent of CPD repair were found between the active GART and Notch genes, the inactive white gene and the genome overall. In contrast to mammalian cells S-phase in Drosophila appears to be unimodal and G-bands or replication bands are lacking in diploid chromosomes (Steinemann et al., 1981). Thus it seems that chromosomal D N A in Drosophila is organized differently from that of mammals, and does not have the properties associated with chromosomal banding pattern. Although this may account for the observed lack of preferential repair, other factors such as the absence of cytidine methylation in Drosophila could play a key role as well.

Biological consequences of preferential repair The biological relevance of preferential repair of transcriptionally active genes has been illustrated in studies with UV-sensitive Cockayne's syndrome (CS), and xeroderma pigmentosum group C (XP-C) cells (Venema et al., 1990a,b). Cells from patients with CS have an apparently normal capacity to remove CPD from the genome overall, but are hypersensitive to UV. However neither CS strain from two independent complementation groups was able to repair transcriptionally active D N A with the same rate and to the same extent as normal cells. XP-C cells are able to perform a normal repair level of the transcribed strand of active genes despite being deficient in repair of inactive genes or the genome overall (Venema et al., 1991). CS and XP-C patients differ with regard to the incidence of neu-

22¢~ TABLE 2 REPAIR

('HARA('TERISTI(/S

AND CLINICAl. FEATURES

OF SOME UV-SENSITIVE

Normal/human disorder

G c n o m e overall repair

Preferential rcpair

Neurodcgeneration

Normal XP-l) XP-("

10()of 15-20r,; 15-20r ~

yes rio " yes

--

('S

100r';

no

+

ltUMAN

DISORI)ERS Tumor formation

'

" V c n e m a , u n p u b l i s h e d results.

rodcgcneration and tumor formation (Tablc 2). The clinical manifcstations of CS and XP-C may bc a conscquence of the.differential ability to perform preferential repair of activc gcnes and efficient repair of thc gcnome ovcrall. Thc ability to perform prcfcrcntially repair may bc a key factor in neurodcgeneration since central nervous system neurons do not divide and transcribc a larger fraction of their gcnome than other cell typcs. XP-C cells are able to perform prcferential repair and patients virtually never suffer from neurological abnormalities, whcreas the opposite is true for CS patients. On the other hand tumor formation may primarily depend on the overall rcpair capacity. Poorly repaircd inactivc scqucnees might accumulatc DNA damagc and non-expressed mutations, which may intcrfere with normal DNA-protcin interactions at regulatory sequcnccs anti may promote I)NA rearrangcmcnts. Such processes may be rclcvant for the activation of certain oncogencs involved in carcinogenesis. Heterogeneity of DNA-repair processes could have major implications for induction of mutations as the frequency of mutations primarily depends on the extent of repair of mutagenic lesions which can occur before fixation during DNA replication. It is conceivable that the observed exclusive rcmoval of CPD from the transcribed strand of active genes in Chinese hamster cells will have its reflection on ultraviolet light-induced mutations. The molecular nature of mutations induced by UV light was investigated in hprt mutants from various Chinese hamster cell lines with different repair capacities. Among the mutants analyzed from repair-proficient cells (two cell lines, i.e., V79 and CHO) all possible classes of base-pair changes were present, the majority

being transversions. Since almost all mutations occurred at dipyrimidine sites, the assumption was made that they were caused by UV-induccd photoproducts (CPD a n d / o r (6-4) photoproducts) at these sites. In repair-proficient cells, after UV irradiation with 2 J / m 2, over 85% of the mutatk)ns could be attributcd to lcsions in the non-transcribed strand of the hprt gcne (Vricling et al., 1991; Mcnichini ct al., 1991). This was 65c/~ at 12 J / m 2 (Vrieling et al., 1989). Analysis of DNA repair in thc hprt gene of V79 cells rcvealed that more than 90% of CPD were removed from the transcribed strand within 8 h aftcr irradiation, whcrcas virtually no dimcr removal could be detected in the non-transcribed strand even up to 24 h after UV (Vrieling et al., 1991). These data presented the first proof that strand-specific repair of DNA lesions in an expressed mammalian genc is associated with a strand specificity for mutation induction. Also in human cells the effect of preferential removal of photoproducts from the transcribed strand of thc hprt gcne on mutation induction has bccn observed (McGregor et al., 1991). Synchronized primary human fibroblasts were irradiated either in early GI- (6 h prior to thc S-phase) or in the S-phase of the cell cyclc. 80% of the hprt mutations after irradiation in G~ were duc to photolesions in the non-transcribed strand, suggesting that photoproducts were preferentially removed from the transcribed strand in the time period between G) and S. Of the mutants from cells irradiated in S, 71% were caused by lesions in thc transcribed strand, indicating that fixation of mutations happened very soon after UV leaving too little time for rcpair effects on mutagenesis. Molecular analysis of hprt mutations in three repair-deficient hamster cell lines, i.e., V-HI,

227

UV-5 and UV-24, showed a completely changed mutation spectrum compared to repair-proficient cells, consisting almost solely of GC ~ AT transitions (Vrieling et al., 1989; Menichini et ai., 1991; Vrieling, unpublished results). 90% of the basepair changes in repair-deficient ceils were caused by photoproducts in the transcribed strand of the hprt gene, which we propose to be caused by differences in error rate between synthesis of the leading and the lagging DNA strand on damaged templates. To allow replication forks to proceed incorporation of a nucleotide opposite photoproducts in the template of the leading DNA strand (translesion synthesis) has to occur. However, synthesis of the lagging strand may be allowed to terminate opposite a photoproduct, since DNA synthesis can start from a new initiation site downstream from that photoproduct, thereby generating a gap which can be filled by post-replication processes at a later time. This latter process is likely to cause fewer errors than translesion synthesis. Analysis of hprt mutations in human XP-A cells also showed that in the absence of DNA repair UV-induced mutations are predominantly caused by photoproducts in the transcribed strand of the hprt gene (McGregor et al., 1991; Dorado et al., 1991). It is obvious from the H P R T mutation spectra that in repair-proficient mammalian cells heterogeneity in repair leads to the expected heterogeneity in distribution of mutations within the gene. At the chromosomal level it suggests that biases in rates of chromatin repair could be responsible for the alternating pattern of DNA-sequence differences along the chromosomes reflecting the chromosome's history of fixed mutations as suggested by Holmquist (1991).

Acknowledgements This work was supported by the Association of the University of Leiden with the Netherlands Organization for Scientific Research and Euratom.

References Aelen, J.M.A., R.J.G. Opstelten and F. Wanka (1983) Organization of DNA replication in Physarum polycephalum.

Attachment of the origins of replicons and replication forks to the nuclear matrix, Nucleic Acids Res., I 1, 11811185. Berkvens, Th.M., E.J.A. Gerritsen, M. Oldenburg, C. Breukel, J.Th. Wijnen, !|. van Ormondt, J.M. Vossen, A.J. van der Eb and P. Meera Khan (1987) Severe combined immune deficiency due to a homozygous 3.2 kb deletion spanning the promotor and the first exon of the adenosine deaminase gene, Nucleic Acids Res., 15, 9365-9378. Bohr, V.A., D.H. Philips and P.C. Hanawalt (1987) Heterogeneous DNA damage and repair in the mammalian genome, Cancer Res., 47, 6426-6436. Buongiorni-Nardelli, M., G. Micheli, M.T. Carri and M. Marilly (1982) A relation between replicon size and supercoiled loop domains in the eukaryotic genome, Nature (London), 298, 100-102. de Cock, J.G.R., E.C. Klink, W. Ferro, P.H.M. Lohman and J.C.J. Eeken (1991) Repair of UV-induced pyrimidine dimers in the individual genes Gart, Notch and white from Drosophila melanogaster cell lines, Nucleic Acids Res., in press. Dijkwel, P.A., and J.L. Hamlin (1988) Matrix attachment sites are positioned near replication initiation sites, genes and an interamplicon junction in the amplified DHFR domain of Chinese hamster cells, Mol. Cell. Biol.. 8, 5398-54{)9. Dijkwel, P.A., L.H.F. Mullenders and F. Wanka (1979) Analysis of the attachment of replicating DNA to a nuclear matrix in mammalian interphase nuclei, Nucleic Acids Res., 6, 219-230. Dorado, G., H. Steingrimsdonir, C.F. Arlett and A.R. Lehmann (1991) Molecular analysis of ultraviolet-induced mutations in a xeroderma pigmentosum cell line, J. Mol. Biol., 217, 217-222. Goldman, M.A., G.P. Holmquist, M.C. Gray, L.A. Caston and A. Nag (1984) Replication timing of mammalian genes and middle repetitive sequences, Science, 224, 686-692. Hand, R. (1978) Eukaryotic DNA: organization of the genome for replication, Cell, 15, 317-324. Holmquist, G.P. (1991) Mutational bias, molecular ecology and the chromosome evolution, in: G. Obe (Ed.), Advances in Mutagenesis Research, Springer, pp. 95-123. Holmquist, G., M. Gray, T. Porter and J. Jordan (1982) Characterization of Giemsa dark- and light-band DNA, Cell, 31, 121-129. Kantor, G.J., L.S. Barsalou and P.C. Hanawalt (1990) Selective repair of specific chromatin domains in UV-irradiated cells from xeroderma pigmentosum complementation group C, Mutation Res., 235, 171-180. Latt, S.A. (1973) Microfluorometric detection of deoxyribonucleic acid replication in human metaphase chromosomes, Proc. Natl. Acad. Sci. (U.S.A.), 70, 3395-3399. McGregor, W.G., R.-H. Chen, L. Lukash, V.M. Maher and J.J. McCormick (1991)Cell cycle-dependent strand bias for UV-induced mutations in the transcribed strand of excision repair-proficient human fibroblasts but not in repair-deficient cells, Mol. Cell. Biol., 11, 1927-1934. Mellon, I., V.A. Bohr, C.A. Smith and P.C. llanawalt (1986) Preferential DNA repair of an active gene in human cells, Proc. Natl. Acad. Sci. (U.S.A.), 83, 8878-8882.

228 Mellon, 1., G. Spivak and P.C. Hanawalt (1987) Selective removal of transcription blocking DNA damage from the transcribed strand of the mammalian DHFR gene, Cell, 51, 241-249. Menichini, P., H. Vrieling and A.A. van Zeeland (1991) Strand specific mutation spectra in repair proficient and repair deficient hamster cells, Mutation Res., 251,143-155. Mullenders, L.H.F., A.C. van Kesteren-van Leeuwen, A.A. van Zeeland and A.T. Natarajan (1988) Nuclear matrix associated DNA is preferentially repaired in normal human fibroblasts, exposed to a low dose of UV light, but not in Cockayne's syndrome fibroblasts, Nucleic Acids Res.. 16, 10607-10622. Steinemann, M. (1981) Chromosomal replication in Drosophila virilis. III. Organization of active origins in the highly polytene salivary gland cells, Chromosoma, 82, 289-301. Venema, J., A. van Hoffen, A.T. Natarajan, A.A. van Zeeland and L.H.F. Mullenders (1990a) The residual repair capacity of xeroderma pigmentosum group C fibroblasts is highly specific for transcriptionally active DNA, Nucleic Acids Res., 18, 443-448. Venema, J., L.H.F. Mullenders, A.T. Natarajan, A.A. van Zeeland and L.V. Mayne (1990b) The genetic defect in

Cockayne syndrome is associated with a defect in repair of UV-induced damage in transcriptionally active DNA, Proc. Natl. Acad. Sci. (U.S.A.), 87, 4707-4711. Venema, J., A. van Hoffen, V. Karcagi, A.T. Natarajan. A.A. van Zeeland and L.H.F. Mullenders (1991) Xeroderma pigmentosum complementation group C cells remove pyrimidine dimers selectively from the transcribed strand of active genes, Mol. Cell. Biol., in press. Vrieling, H., M.J. Niericker, J.W.I.M. Simons and A.A. van Zeeland (1988) Molecular analysis of mutations induced by N-ethyI-N-nitrosourea at the hprt locus in mouse lymphoma cells, Mutation Res., 198, 99-1(15. Vrieling, H., J. Venema, M.L. van Rooijen, A. van Hoffen, P. Menichini, M.Z. Zdzienicka, J.W.I.M. Simons, L.H.F. Mullenders and A.A. van Zeeland (1991) Strand specificity for UV-induced DNA repair and mutations in the Chinese hamster hprt gene, Nucleic Acids Res., in press. Yunis, J.J. (1981) Mid-prophase human chromosomes. The attainment of 2000 bands, Hum. Genet., 56, 293-298. Zehnbauer, B.A., and B. Vogelstein (1985) Supercoiled loops and the organization of replication and transcription in eukaryotes, Bioassays, 2, 52.