drug resistance and dna repair

21
Cancer and Metastasis Reviews6:261-281 (1987) © Martinus Nijhoff Publishers. Boston - Printed in the Nctherlands. Drug resistance and DNA repair Margaret Fox ~ and John J. Roberts ~- Paterson Institute for Cancer Research, Christie Hospital and Holt Radium Institute, Manchester :~,I209BX, UK; 2Institute of Cancer Research, Royal Cancer ttospital, Clifton A venue. Sutton, Surrey SM2 5PX, UK Key words: DNA repair, drug resistance, molecular cloning Summary DNA repair confers resistance to anticancer drugs which kill cells by reacting with DNA. A review of our current information on the topic will be presented here. Our understanding of the molecular biology of repair of 06-alkylguanine adducts in DNA has advanced as a result of the molecular cloning of the E. coli ada gene but the precise role of this lesion in the cytotoxic effects of alkylating agents in mammalian cells is not completely understood. Less progress has been made in understanding the enzymology and molecular biology of DNA cross-link repair even though such lesions are important for the cytotoxic effects of the widely used bifunctional alkylating agents and platinum compounds. It is evident that drug sensitive or resistant phenotypes are as highly complex as are the effects of DNA damage on cell metabolism and various aspects of lhese effects are discussed. Few clear correlations have been made between quantitative differen- ces in DNA repair capacity and cellular sensitivity but assays which were developed to measure fidelity and intragenomic heterogeneity in DNA repair are beginning to be applied. Such studies may reveal subtle differences between sensitive and resistant cell lines. The molecular cloning of human DNA repair genes by transfection into drug sensitive rodent cells has been attempted. Some success has been achieved in this area but the functions of the cloned genes have yet to be identified. Introduction In spite of recent advances in treatment of cancer as a result of the introduction of multidrug chemothe- rapy, the development of drug-resistant tumour cell populations is still a major factor in treatment failure. The basic premise on which the develop- ment of multidrug chemotherapy is based is that the acquisition of resistance (by somatic mutation) to a combination of many different drugs with dif- ferent modes of action occurs at an extremely low frequency. Non-genetic factors make a considera- ble contribution to overall non-responsiveness of tumours [1]. Many features of the mode of action of the various drugs and drug combinations must be taken into consideration in order to understand the reasons for development of drug resistance at the cellular level. Evidence now indicates that tumour cells arc inherently unstable genetically [2, 3] and it is im- portant to remember that ionising radiation and alkylating agents commonly used in chemotherapy are in themselves mutagenic and clastogcnic [4]. Exposure of tumour cells to chemotherapy or radi- otherapy, in addition to selecting for intrinsically resistant cells, generates increased genetic diver- sity which may express drug resistance and in- creased metastatic potential. Changes in DNA re- pair capacity are only one of the many metabolic changes which may make a contribution to the drug resistant phenotype. The mechanisms of acquired and intrinsic resistance to all major classes of anti- cancer drugs have been recently reviewed [5]. Many antimetabolitc drugs and other inhibitors of DNA synthesis including alkylating agents cause

Upload: margaret-fox

Post on 06-Jul-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Drug resistance and DNA repair

Cancer and Metastasis Reviews 6:261-281 (1987) © Martinus Nijhoff Publishers. Boston - Printed in the Nctherlands.

Drug resistance and DNA repair

Margaret Fox ~ and John J. Roberts ~- Paterson Institute for Cancer Research, Christie Hospital and Holt Radium Institute, Manchester :~,I20 9BX,

UK; 2 Institute of Cancer Research, Royal Cancer ttospital, Clifton A venue. Sutton, Surrey SM2 5PX, UK

Key words: DNA repair, drug resistance, molecular cloning

Summary

DNA repair confers resistance to anticancer drugs which kill cells by reacting with DNA. A review of our current information on the topic will be presented here. Our understanding of the molecular biology of repair of 06-alkylguanine adducts in DNA has advanced as a result of the molecular cloning of the E. coli ada gene but the precise role of this lesion in the cytotoxic effects of alkylating agents in mammalian cells is not completely understood. Less progress has been made in understanding the enzymology and molecular biology of DNA cross-link repair even though such lesions are important for the cytotoxic effects of the widely used bifunctional alkylating agents and platinum compounds. It is evident that drug sensitive or resistant phenotypes are as highly complex as are the effects of DNA damage on cell metabolism and various aspects of lhese effects are discussed. Few clear correlations have been made between quantitative differen-

ces in DNA repair capacity and cellular sensitivity but assays which were developed to measure fidelity and intragenomic heterogeneity in DNA repair are beginning to be applied. Such studies may reveal subtle differences between sensitive and resistant cell lines. The molecular cloning of human DNA repair genes by transfection into drug sensitive rodent cells has been attempted. Some success has been achieved in this area but the functions of the cloned genes have yet to be identified.

Introduction

In spite of recent advances in treatment of cancer as a result of the introduction of multidrug chemothe- rapy, the development of drug-resistant tumour cell populations is still a major factor in treatment failure. The basic premise on which the develop- ment of multidrug chemotherapy is based is that the acquisition of resistance (by somatic mutation) to a combination of many different drugs with dif- ferent modes of action occurs at an extremely low frequency. Non-genetic factors make a considera- ble contribution to overall non-responsiveness of tumours [1]. Many features of the mode of action of the various drugs and drug combinations must be taken into consideration in order to understand the reasons for development of drug resistance at the cellular level.

Evidence now indicates that tumour cells arc inherently unstable genetically [2, 3] and it is im- portant to remember that ionising radiation and alkylating agents commonly used in chemotherapy are in themselves mutagenic and clastogcnic [4]. Exposure of tumour cells to chemotherapy or radi- otherapy, in addition to selecting for intrinsically resistant cells, generates increased genetic diver- sity which may express drug resistance and in- creased metastatic potential. Changes in DNA re- pair capacity are only one of the many metabolic changes which may make a contribution to the drug resistant phenotype. The mechanisms of acquired and intrinsic resistance to all major classes of anti- cancer drugs have been recently reviewed [5].

Many antimetabolitc drugs and other inhibitors of DNA synthesis including alkylating agents cause

Page 2: Drug resistance and DNA repair

262

significant alterations in gene expression 16-8] which may be related to the known capacity of the former class of drugs to induce differentiation [9]. In some cases altered genc expression has been correlated with alteration in methylation patterns [6]. In others it occurs by as yet undefined mecha- nisms. Such events may be causally related to the development of drug resistance or they may be cpiphenomena.

In the experimental setting, where single agents can be used and the development of non-respon- sive tumours monitored accurately by a variety of experimental techniques [10], resistancc develops significantly more rapidly than can bc accounted for by modcls which propose spontaneous muta- tion to drug resistance and consequent selection of drug resistant cells [11]. Although changes in DNA repair capacity may be demonstrated, they arc often accompanied by other changes which could contribute significantly to the resistant phenotypc.

Interactions between drugs can also contribute to drug resistance. For example, CHO cells ex- posed in vitro to ultraviolet irradiation or drug treatment (EMS or MNNG) prior to selection in methotrexate (MTX) show a transient increase in MTX resistant colonies of which a similar propor- tion are amplified at the dhfr locus irrespective of drug exposure [12].

The mechanism for induction of gene amplifica- tion is still not completely understood but one mechanism is thought to involve inhibition of fork progression and reinitiation of replication at a pre- viously unused origin behind the blocked replica- tion fork [12]. Inhibition of replication could be important in the evolution of drug resistant popula- tions during combination chemotherapy. The effi- ciency of DNA repair may be crucial in this interac- tion as removal of the lesion blocking DNA synthesis prior to exposure to the second drug pre- sumably prevents the increase in frequency of MTX ® colonies. The order of administration of drugs may also have important consequences. A variety of anticancer drugs increase the frequency of mutants at a number of genetic loci, hypox- anthine phosphoribosyl transferase (HPRT) [13- 15[, by induction of point mutations or major ge- netic re-arrangements [15] indicating that the use of

an antimetabolite prior to an alkylating agent is preferable to the other way around.

In an extensive review [16] compiled in 1982 it was concluded that in spite of the vast amount of published data there is little evidence that loss or acquisition of a specific DNA repair pathway is directly related to altered drug sensitivity in tu- mour cells. This conclusion was reached after per- forming experiments based on the biochemical techniques in use for the analysis of DNA repair capacity, which were reviewed in detail in 1981 [17]. The majority of these techniques measure quan- titative differences in rates of DNA strand break rejoining, rates of repair of specific lesions or rate and total amount of unscheduled DNA synthesis or repair replication. However, few quantitative dif- ferences in repair have been reported which correl- ate with a drug sensitive or rcsistant phenotype. Assays have only recently been developed to mea- sure the more qualitative aspects of DNA repair such as fidelity to the original or repair in transcrip- tionally active regions of DNA. An attempt will be made in this review to analyse the contribution of these newer techniques and other recent advances in molecular biology to our current understanding of the complexities of DNA repair and its role in intrinsic and acquired drug resistance in mam- malian cells.

Use of transfected genes to assess repair fidelity

Agents which damage DNA produce a variety of lesions and it is thcrefore difficult to identify the lesion responsible for a particular cytotoxic effect. One approach to this problem is to analyse the functional inactivation of a dominant selectable marker after its transfcction into mammalian cells using appropriate vectors. An important feature of such vectors is that they do not code for their own repair enzymes, so any observed repair reflects that of the host cell. One transfection system uses the plasmid pSV2-gpt in which the bacterial gpt gcne has been engineered for expression in mammalian cells by splicing it between appropriate SV40 se- quences [18]. Mammalian cells cannot normally utilise xanthine as a source of GMP, therefore gpt ÷

Page 3: Drug resistance and DNA repair

transfectants can bc identificd by their growth in media containing mycophenolic acid, aminopterin, thymidine, hypoxanthine and xanthine. Thus if a 7-irradiatcd or drug-treated plasmid is transferred into mammalian cells the ability of the gpt gene to function can be monitored after a suitable cxprcs- sion period by measuring colony formation in ap- propriatc sclective media. The effects of y- and UV-irradiation on pSV2-gpt and pSV2 neo, a plas- mid containing the neo gene which confers resis- tance to the antibiotic G418. have recently been reviewed [191 .

The use of restriction endonucleases to generate DNA damagc in dcfincd sequences has also pro- vided a new analytical method for assessing repair processes in normal or radiation-sensitivc cells. Radiosensitivc human ataxia tclangieclasia (AT) cells [19, 20] can rejoin endonuclease-generated breaks in the gpt gene in immortalised normal and AT-radiosensitive cells. The gene is inactivated by cuts at either the junction of the gene and its con- trolling sequences or within the gene itself. If the cut plasmid is rejoined with fidelity the genc will be expressed and confer resistance to selective drugs (G418 [neo] or mycophenolic acid [gpt]). Mis-re- pair involving some loss of DNA sequences com- monly destroys gpt function. Cuts at sites further away from the coding frame produce less drastic effects. After cuts with one of several restriction enzymes normal cells have relatively high levels of gpt expression. In contrast, after cutting with the same enzymes AT cell expression is much rcduced. Cells which do not express the pgt gene are lost in this system and hence reasons for lack of function cannot be analysed.

In an extension of these studies a plasmid con- taining both neo and pgt and with only one Kpn 1 cut-site in the pgt coding frame is first selected for neo function and subsequently assessed for gpt function [20]. In experimcnts of this type 88% of normal cells express neo and gpt but only 12% of AT cells express both functions. Of 14 normal cell clones whose DNA was analysed after Southern transfer, 10 showed a DNA fragment of the correct size for the intact gpt gcne. One of 11 AT cells analysed had the appropriate fragment. These ana- lyses also revealed that manv cells had integrated

263

multiple copies of the plasmid on which presum- ably there may be only one functional gp~. and several non-functional ones.

The radiosensitivity of AT cells is possibly ex- plained by a disequilibrium between rejoining and exonuclease digestion at the sites of double strand scissions. The disequilibrium is not detectable by conventional analysis of rates and extents of repair of single and double strand breaks. It should be noted however that there are now 5 known comple- mcntation groups in AT cells [21] and only one cell linc has so far been examined in this way. How- ever, few defects in AT cells of other complemen- tation groups have been detected by conventional biochemical assays, so most differences may be quite subtle.

Discquilibrum studies have been done on Walker 256 carcinoma cclls (WS) which are inher- ently sensitive to bifunctional alkylating agents. Resistant Walker cells (WR). on thc other hand, show sensitivity similar to that of conventional cell lines [22]. The comparative study is discussed in detail later.

Using a replica plating technique 4 X-ray sensi- tive mutants of V79 Chinese hamster cells have been isolated after screening 5000 ENU treated clones. Complementation by cell fusion has shown that the three most sensitive mutants are from dif- ferent complemcntation groups and 2 of these are diffcrent from the X-ray sensitive Chinese hamster mutants xrs-1 and EM7-2 isolatcd by Jcggo and Kemp [23] and Thompson and co-workers [24] re- spcctively. The transfected gpt-neo plasmid tcch- nique revealed that the ability to correctly rejoin double strand DNA breaks is significantly reduced in one of the sensitive mutants [25]. No data arc vet available for other mutants, but subtle changes in ability to rejoin strand breaks are correlated with a radiation sensitive phcnotype. Another series of X-ray scnsitive mutants has been isolated after EMS mutagcnesis of CHO cells. All arc defective in double strand breaks rcjoining and are phe- notypically stable but revert with high frequency on treatment with 5-azactydine, suggesting that the xrs genc is under mcthylation control 126].

Page 4: Drug resistance and DNA repair

264

Repair of monofunctional alkylation damage

O~-atkylguanine alkyl transferase

One of the best characterised repair enzymes, par- ticularly in bacterial cells, is 0" alkylguanine alkyl transferase ((~-GAT), which removes methyl, ethyl and possibly higher alkyl groups from guanine by alkyl transfer to a cysteine residue in the protein [27]. The repair process appears essentially similar in both bacterial and mammalian cells. E. coli 06-alkylguanine alkyl transferase preferentially removes methyl groups but the enzyme is known to remove other alkyl groups in the order Me > Et > nPr > nBu > iPr > iBu. The enzyme is also known to dealkylate lYMeT but at a slower rate than 0~-MeG [28]. The gene (ada) coding for the E. coli enzyme has now been cloned by several groups of workers [2%31]. The ada protein has also been purified to homogeneity from E. coli [31], and the complete DNA sequence is known [32]. The gene ada has a molecular weight of 37 Kd with 12 cys- teine residues, one of which (amino acid 321) ac- cepts the methyl group from 06-methyl guanine while another (cysteine residue between amino acids 6%72) acts as an acceptor for the S ster- eoisomer of methyl phosphotriesters. The latter specific methylated protein then acts as trigger for the adaptive response by binding to DNA 35 bases upstream from the transcriptional start of the gene and acting as a positive signal for further transcrip- tion [331.

Much less is known about the molecular biology of the mammalian enzyme. Repeated attempts to purify the protein to homogeneity from rat liver have so far been unsuccessful largely due to in- stability of the enzyme. Partial purification of the enzyme from a variety of sources has been reported [34]. Among the attempts at cloning of the mam- malian genc is one report which suggests the suc- cessful transfection of human DNA from a mer ÷ cell line into a mer- human astrocytoma strain [35]. Protection against cytotoxicity induced by chlo- roethyl nitrosourea (CNU) shows that the human DNA complements the mer- phenotype. Meth- yltransferase levels, determined in transfectcd cells, arc increased relative to those in the mer-

strain. However it is possible that the endogenous human enzyme is re-expressed when cells are sub- jected to sevcral doses of MNNG which itself is a mutagen.

Saffhill, Margison and O'Connor [36] have re- viewed thc varying levels enzyme activity of in different mammalian species and tissues and in normal and tumour derived cell lines in vitro. Be- cause they covered other aspects of repair of al- kylation damage, particularly in vivo [36], we will concentrate on current status in cultured cells.

Some cstablished cell lines, c.g., V79 and CHO (both hamster derived) are completely deficient in 0~-GAT [37, 38] and are dcsignated mer-. The mer- phenotype was originally defincd as lack of capacity to reactivate adenovirus 5 when alkylated virus was grown in the host cell [38] but is now commonly used to describe any cell which lacks enzyme activity. However, in some cases the ability to reactivate viruscs can be demonstrated in the absence of detectable methyl transferase activity indicating that the two phenotypes are not neces- sarily identical. A number of reports indicate that some human tumour-dcrived cell lines have a mer- phenotype, but a number of primary tumours do not exhibit this phenotype [35, 36, 40]. Levels of enzyme activity are either normal or above normal in the majority of primary tumour samples. Such observations suggest that 06-GAT can be lost dur- ing establishment of cell lines in vitro. Transforma- tion of cells by SV 40 virus can lead to loss of enzyme activity and a single sample of lympho- blastoid cells from the same patient can show either mer ÷ or met- phenotype, suggesting that in vitro culture conditions may influence expression. The reasons for these differences in 06-GAT expression are not understood at present but indicate that great caution should be exercised in interpreting data on drug sensitivity of cultured human tumour cells as being indicative of responses of tumours in vivo. The current status of our knowledge of (~'-GAT expression has been rcviewed recently by D.B. Yarosh [40].

A novel approach has been made to understand- ing the roles of 06-methyl guanine methyl trans- ferase activity in protecting mammalian cells from the cytotoxic and mutagenic effects of both mono-

Page 5: Drug resistance and DNA repair

and bifunctional alkylating agents which react at 0h-MeG. The E. coli gene coding for the 37Kd ada gene product and possessing both 06-GAT and methylphosphotriester methyl transferase ac- tivities has been engineered to contain only the protein coding sequences and inserted into a retro- virus-based electable expression vector [41, 42]. The vector was transfected into Chinese hamster V79 cells which lack detectable endogenous 06-GAT activity and clones expressing high levels of the bacterial enzyme selected. Rapid removal of 0~-MeG from alkylated DNA demonstrates the functional activity of the protein in transfected cells.

The survival rate of transfected cells is deter- mined after exposure to increasing doses of N-methyl-N-nitrosourea (MNU), methyl methane sulphonate (MMS), chlorozotocin (CLZ), and ni- trogen mustard (HN2). Those expressing the ada gene are clearly more resistant to MNU and CLZ but not to MMS or HN~. Both MNU and CLZ produce high levels of reaction at the 0G-position of guanine. CLZ gives 06-chloroethylguanine, which has been shown to be a substrate for 06-GAT in vitro. In contrast MMS produces very little 06 methylguanine (06meG/7McG ratio in DNA for MMS is 0.004) and no 06 substituted products have been detected in DNA after HN, exposure. These results in two cell lines which differ, so far as can be ascertained, only in ability to remove alkylation products from 06G and from phosphotriesters, indi- cate an important role for such lesions in the cyto- toxic effects of certain drugs. It is not yet clear from these results whether 0C'-MeG or the methyl phos- photriesters are the major cytotoxic lesions or whether both contribute equally. If 0"MEG is a cytotoxic lesion then the question arises as to how it excrts its cytotoxic effects.

MNU and MNNG are considerably more toxic towards a line of HeLa cells than is MMS but no comparable differences are observed between the cffects of these three agents in hamster cells [43]. As neither HeLa nor V79 ceils are able to excise 0~'-MeG from DNA the vast difference in their sensitivity to the cytotoxic effects and MNU sug- gests that V79 cells possess some mechanism, which HcLa cells lack, whereby they are able to

265

tolerate or bypass unexcised lesions in their DNA. The existence of such tolerance or bypass mecha- nisms also indicates that excision repair may be of secondary importance in some cell lines.

When the effects of MNU on DNA synthesis were compared some remarkable differences were observed between the two cell lines [43]. Exposure of HeLa cells to MNU in early S phase induces minimal initial depression of DNA synthesis in- dicating that the initial alkylation of DNA does not produce lesions which block DNA polymerasc. In the second post-treatment S phase, DNA synthesis is markedly depressed in a dose-dependent man- ner, suggesting the introduction of new damage to DNA. Exposure of synchronised Chinese hamster V79 cells to similarly toxic levels of MNU reveals a marked difference in response. An initial depres-

sion followed by an increase in the rate of DNA synthesis resulted in delay of the peak rate of DNA synthesis but no overall decrease in the amount of DNA synthesised during the first S phase after treatment.

Consideration of the nature of the secondary damage inducd in HeLa cells may give clues as to how 06MEG exerts its cytotoxic effects. The HcLa cell strain used appears to have a mcr- phenotype from 06/N ~ ratios (although mer- strains have been described) [44]. Therefore ceils entering the sec- ond S phase would still carry 06MEG lesions. If such lesions, although not initially constituting a block to the polymerase, arc rccognised by an cndo- nucleasc which incises close to the site of the lesion, then in a subsequent round of DNA synthesis dou- ble strand breaks could be produced.

Pertinent to the above discussion is the observa- tion that in clone 8 cells (transfected with the ada gene and able to remove 06-MEG) MNU fails to induce sister chromatid exchanges at measurable frequencies above background [45]. In contrast, SCE frequencies increase approximately linearly with dose in clone 2 (transfected with the neo gene only). The generation of SCE is known to require some kind of recombination event [46] which also requires nicking at or near the sites of lesions. In the absence of lesions, (i.e.. in clone 8 where the majority are repaired) no nicking would occur and no SCE is generated.

Page 6: Drug resistance and DNA repair

266

From levels of 0"-alkylguanine in V79 cells ex- posed to higher doses of ~4C MNU it can be esti- mated that with the doses used in this study there would be 104 (~'-MeG/cell [47]. Only 40-50 SCE/cell were observed, suggesting that the majority of le- sions are not recognised and nicked to allow recom- bination to occur.

To pursue this line of reasoning, we suggest that V79 Chinese hamster cells which lack the ability to remove 06G have also lost the ability to incise the DNA at sites close to 0 ~ lesions. This hypothesis explains why no secondary damage is observed, and why V79 cells are considerably more resistant to MNU and MNNG. Reduced numbers of double strand breaks form, few chromosome exchanges or aberrations are produced and the chances of cell survival are enhanced. The production of aberrant chromosomes can lead to cell death.

Evidence in favour of an important role of al- kylation at oxygen atom sites in cytotoxicity comes from the observation that the cytotoxicity of al- kylating agents can be increased when endogenous levels of 06-GAT are depleted by pre-treatment of the cells with low (non-toxic) levels of alkylating agents or with the free base @ methylguanine [36, 40, 48]. However , this simple interpretation does not hold in all cases. Pre-treatment with equitoxic doses of MMS, MNU, MNNG, EMS, ENU and ENNG reduced survival in V79 Chinese hamster cells (which lack endogenous 06-GAT activity) to 75%. The V79 cells were then re-exposed to a range of doses of the same agent and their sub-

Table I. Survival ratios for pretreated and non-pretreated V79 cell population.

Drug Pre- Challcngc Recovery treatment dose mM ratio V79 dose mM

MMS 0.36 0.36 1.9 MNU 0.40 0.40 0.3 MNNG 0.001 0.001 0.12 EMS 8.0 8.0 1.41 ENU 1..0 1.0 1.35 ENNG 0.012 0.012 1.0

Recovery ratios are the ratio of survival at 96 hrs of pretreated and challenged population divided by survival of population given a challenge dose equal to the sum of the pre-treatment and challenge dose.

sequent survival determined. The ratios of survival of the pretreated compared with the non pre- treated population were measured 24-96 hrs after pretreatment (data at 96 hrs arc shown in Table 1). Cells were not significantly sensitised by MMS pre- treatment or pre-treatment with cthylating agents even at early times, but MNU and MNNG pre- treatment did significantly sensitize. These results indicate that sensitization is not due to the depic- tion of 06-GAT and suggest that other persistent DNA lesions or alkylation of repair proteins by MNU, MNNG may play a role.

In another series of experiments V79 cclls on a different pre-treatment schedule with MNU show a transient rise in survival. This does not result from any alteration in the capacity of the cells to remove 0~-MeG as measured by loss of the radi- olabelled product from DNA [49]. Consistent with the suggestion that responses to pretreatment are very cell line dependant are the observations of Karran and Williams [50]. They were unable to sensitise mer + Raji cells or a mer- CNU sensitive derivative to the cytotoxic effects of MNNG or CNU by pre-treatment with {~-MeG at doses which significantly reduced 06-GAT levels. These results clearly imply that in this cell line adducts at the lY' position of G are not lethal lesions and that a defect other than lack of the enzyme is responsible for alkylation sensitivity in this met- strain as in the case of the mer- HeLa cell strain referred to ear- lier. These observations bear on other experiments designed to demonstrate an adaptive response on mammalian cells.

Adaptation

Several groups of workers have attempted to dem- onstrate an adaptive response in mammalian cells [40, 48] and the results of such experiments are conflicting even when nominally the same cell line is used. Data from one laboratory indicates in- creased survival and reduced SCE "after MNNG pre-treatment but there are several other reports of negative results. The mechanisms underlying the increased survival and reduced SCE frequency re- ported in V79 and CHO are unknown, but they seem unlikely to be related to changes in 0~-GAT

Page 7: Drug resistance and DNA repair

activity, as neither cell line expresses thc enzyme constitutively. It is conceivable that MNNG pre- treatment in the particular regime used induces expression of 0"-GAT by inducing hypomethyla- tion [51], however, extensive treatment of V79 cells with azacytidine does not result in expression of 06GAT (Fox and IVlargison, unpublished data). In most other cell lines (with the exception of rat hepatoma cells [52, 53]) attempts to demonstrate adaptation by increase in survival or reduction in mutation have proved negative.

The demonstration of an adaptive response in cultured mammalian cells in vitro seems highly de- pendent on the cell line used, the precise prc-trcat- ment schedule, and the cultural conditions. The factors controlling these variables are far from being clearly understood, and the effects observed are obviously a balance between depletion of 06-GAT activity (resulting in sensitization) and its apparent induction by mechanisms yet to be eluci- dated. Several important factors may be determi- nants.

These are as follows: 1) The constitutive level of expression of ()6-1vleG

transferase 2) The rate of depletion of the cnzvme on exposure

to alkylating agents 3) The rate of resynthesis after treatment 4) Inducibility or otherwise of the enzyme

In human cells the differences in the rate at which various met- human cell lines regenerate ()~GAT activity have been correlated with different sensitivities to MNNG. Those showing a rapid re- generation capacity are classified rein + and those which regenerate levels more slowly rem- [40]. The apparent importance of regeneration rates im- plies that 0~-alkylguanine lesions may, not be toxic at the first round of DNA rcplication but can pro- duce secondary damage during subsequent replica- tion. a conclusion consistent with data discussed earlier.

Overall results available to date raise several important questions: 1) Is 0<methylguanine a major cytotoxic lesion in

some cell lines but not in others'? 2) In cells with constitutive enzyme activity what

are the precise levels of MNU or MNNG to completely saturate (l~'l~teG transferase activity?

267

Other path ways of repair

The possible involvement of excision repair path- ways in removal of 06-AG in cell lines which lack endogenous ff'-GAT activity is only just beginning to be appreciatcd. Human mer- cells and Chinese hamster cell lines V79 and V79/79 which lack 0~-GAT activity are all able to removc 0~'BudG from their DNA as shown by RIA of enzyme di- gests of DNA from cells exposed to BNU [54, 55]. In contrast, fibroblasts from xeroderma pigmen- tosum patients (complementation groups A and G) which show <5% unscheduled DNA synthesis after UV irradiation but are mer' fail to remove 0~'BudG, suggesting that 0~BudG is repaired in cells which lack the alkyl transferase by a process defec- tive in XP cells (presumably nucleotide excision repair). Both Chinese hamster cell lines lost 0~-BudG at approximately equal rates but only V79/79 lost 06-MEG too, suggesting that the 06-BudG is recognised by excision repair endo- nucleases in both Chinese hamster cell lines and that 0<IVleG in some way distorts chromatin struc- ture sufficiently in V79/79 cells to be recognised. Thus it is formally possible that increased loss of 0~-MeG observed in pre-treated cells (see earlier) is due not to enhanced 0O-GAT activity but is in some way due to up-regulation of excision repair bv mechanisms at present unknown.

A knowledge of the complexities of cellular re- sponses to damage induced by monofunctional agents is important for understanding the mecha- nisms of acquired resistance to the bifunctional agents. Bifunctional agents which react at the 0~-position of guanine are generally more cytotoxic and are more commonly used clinically. CLZ BCNU, CCNU mitozolamide and many others are thought to act by a two-stage mechanism. The ini- tial reaction places a chloroethvl group in the 0 ~ position of guanine, which then reacts further to form a DNA interstrand cross link [56]. Resistance to chloroethyl nitrosoureas appears to be consis- tently associated with the mer- phenotype, sug- gesting that removal of the monoadduct before cross link formation (which takes place only slowly) is an important protective mechanism for this class of drugs [57-59]. Evidence in support of

Page 8: Drug resistance and DNA repair

268

this conclusion is provided by the observation that pre-treatment with 0~-McG sensitised the met ÷ cells to the cytotoxic effects of CNU (160] and Day, unpublished). This conclusion is further supported by recent rcsults (referred to earlier) from V79 cells transfected with and expressing bacterial 0"GAT in which the differential survival after exposure to chlorozotocin, mitozolamidc and methazolastone was far greater than aftcr exposure to MNU [42]. Similar conclusions have been reached from cx- periments with BCNU and other chloroethylating agents in mer + and mer- human fibroblasts and human lymphoblastoid ccll lines [48, 57, 58].

Bifunctional alkylating agents

The cytotoxicity of other classcs of bifunctional alkylating agents is generally thought to bc due to their ability to form interstrand and intrastrand cross-links in DNA. DNA-protein cross links are also tormed but these are not lethal events. The many aspects of the cytotoxic, clastogenic and mu- tagenic effects of this class of drugs have been revicwed extensively [5, 43, 61, 62].

Tumour cells can acquire resistancc (or losc sen- sitivity) to this class of alkylating agents by a num- ber of diffcrent mechanisms and in many cases morc than one mechanism may be operating [62]. DNA cxcision repair has bccn studied extensivcly in several pairs of cell lines which exhibit large differences in sensitivity to bifunctional alkylating agents, but no clear-cut correlations bctwcen ca- pacity to repair DNA : I)NA cross links and cellu- lar sensitivity [16] have emerged. It is important to remember that when wild type tumor cells (c.g., Yoshida and Walker rat sarcomas), which show exquisite sensitivity to bifunctional alkylating agents, develop 'resistance' equivalent to that of Chinese hamster V79 cells, it is really a loss of sensitivity. Much evidence indicates that the ma- jority of inter- or intra-strand DNA cross-links are repaired, but the detailed mechanisms have not yet been elucidated.

In a attempt to correlate othcr cellular rcsponses with sensitivity or resistance to nitrogen mustard induced cytotoxicity, the effects of the drug on cell

cycle progression in Raji and TK6 human lympho- blastoid cells are examined by flow cytomctry 163]. TK6 cells arc considerably more sensitive to HN~ than Raji. Delay in S phasc travcrsc is conccntra- tion-dependent in both cell lines and at a given concentration is two fold greater in TK6 than Raji. The number of cross-links increases linearly with dose and is approximatcly two fold higher in TK6 than in Raji. The difference in sensitivity between the two cell lines cannot be adcquatcly cxplained by differences in initial amounts of DNA damage, ratcs of rcpair of cross-links or differential S phase delay, but we are dealing with a complcx pheno- type.

In the same study inhibition and recovery of DNA synthesis is measured by [3H]TdR incorpora- tion and cvidence indicates that intraccllular dNTP pools wcrc pcrturbed over considerable periods after treatment, suggesting that ability to restore normal DNA metabolism is important in cellular recovery. However, no clcar differences were evi- dent in the responses of the two cell lines. TK6 cells lack 06-GAT activity and are considerably more scnsitive to BCNU and MNU (but not to MMS) than arc Raji cells (Dean, Margison and Fox, un- published data). This effect is unlikely to be related to their HN 2 sensitivity, as this drug as mcntiond earlier is not known to react with 06guaninc.

A similar study compares the responses to HN 2 of Fanconi anaemia cells and normal human fibro- blasts. Two Fanconi anaemia cell lines exhibit a 6-10 fold greatcr sensitivity to HN2 than normal fibroblast stains but show no cell cycle delay even when cxposcd to high doses of HN 2 [64]. At equitoxic doses in normal human fibroblasts a marked S phase delay suggests that delay may bc protectivc in allowing timc for repair. No consis- tent association has been reported between cellular sensitivity and the ability of a cell to delay progress through S phase or any other phase of the cell cycle. This supports the suggcstion that thc delay re- sponse is a complex function resulting not only from DNA lesions which inhibit fork progression but also from complex change s in cellular dNTP pools and probably many other as-yet-unidentified changes which occur in response to DNA damage.

Other recent studies in HN~-sensitive and HN~-

Page 9: Drug resistance and DNA repair

resistant Walker rat tumour cells [65] attempt to correlate sensitivity with rates of rcpair of total and intcrstrand DNA cross-links and DNA protein cross links. No differences are dctccted in initial levels or in the ability of the cells to remove HN, or CLZ-induced [651 cross-links. The HN, resistant sub-line is collatcrally sensitive to CNU and shows reduced activity of 06-GA'I -, indicating the com- plexity of drug sensitive and resistant phenotypes in non-isogenic tumour cell lines [65].

Platinum compounds

Cis-diamminedichloroplatinum(lI) (Cisplatin, cis- DDP) is an antitumour drug particularly effective against testicular cancer, both alone and in com- bination with other drugs such as a vinblastine and bleomycin, and active against teratoma, ovarian carcinoma and squamous cell carcinoma of the head and neck [66]. Platinum drugs sccm to exert their cytotoxic and anti tumour effects through an interaction with DNA, although which of the many reactions with DNA is the critical lesion is still a matter for debate. Earlier comparative studies of cis-DDP and its trans isomer (trans-DDP). which is not an anti tumour agcnt and which is far less toxic to mammalian cells in culture, pointed to the likely importance of interstrand crosslinks in inducing cytotoxieity. The higher doses of trans-DDP re- quired to produce toxic effects equal to those of cis- DDP produced higher levels of reaction with DNA but cqual extents of DNA interstrand crosslinking [13, 67, 68]. In addition, studies of both mouse leukemia cells and normal and transformed human cells of differing sensitivity to cis-DDP show that sensitivity correlates wcll with interstrand crosslink formation. It has also been shown that cells can be protcctcd from the toxic effects of cis-DDP (the formation of DNA crosslinks) by incubating cclls in the presence of thiourea immediately after treat- ment [69]. Potentiation of cell killing (accompa- nied by incrcascd crosslinking) can be achieved by post treatment with 1-[-3-D-arabinofuranosyl- cytosine (ara C) [701 and hyperthermia [71]. On the other hand, DNA-protein crosslinks do not simi- larly correlate with eytotoxicity and are thought to

269

reflect the overall level of active drug metabolites in the cell [67, 68].

However, reactions resulting in DNA inter- strand crosslinks account for only about 1%-2% of the total platinum adducts on cellular DNA and studies during the past five years have established that an intrastrand crosslink between the N-7 atoms of adjacent guanines is the principal adduct formed in DNA by cis-DDP. Sincc trans-DDP is stereoehemically unable to form such an adduct it has been proposed that the ability or inability to form such 1, 2- d(GpG) crosslinks adequately ac- counts for the difference in the antitumour activity of the cis and trans platinum compounds [72]. A difference in the reactions of the other two isomers is that end-labeled fragments of known base se- quences are digested differently at the position of adjacent guanines with cxonuclease Ill whcn DNA is treated with cis-DDP as compared to trans-DDP [72, 73]. Somc additional insight into the possible importance of these various DNA adducts would be gained if it could be shown that cells which were either sensitive or resistant to cisplatin, and which did not differ with respect to the initial damage to D N A , differed in their capacity to repair either total or specific damage to DNA. Earlier studies established that bacterial strains deficient in DNA repair are lcss able to survivc cxposure to cis-DDP than wild type cells [74, 75]. More recently [76] it has been confirmed that proficiency in excision repair is required for survival of cis-DDP- but not apparently for trans-DDP- treated bacterial cells. Purified proteins coded by the uvr genes have been uscd to reconstitute the UVRABC nuclease and to study the incision step of the excision repair of DNA containing various lesions. The UVRABC nuclease cut plasmid DNA containing damage caused by either cis-DDP or trans-DDP. However, adducts caused by the two compounds are recog- nised differently by the nuclease. A specific cutting pattern involving incisions at the 8th phosphodies- ter bond to the 5' and at the 4th phosphodiester bond to the 3' of the adjacent GGs is observed for DNA damage by cis-DDP but not for damage by trans-DDP [77].

In mammalian cells the techniques of alkaline sucrose gradient sedimentation, alkaline elution

Page 10: Drug resistance and DNA repair

270

and DNA renaturation [68, 78, 79] have clearly demonstrated repair of DNA-protein and DNA- DNA crosslinks. Interstrand crosslinking is indi- cated by a shift in alkaline sucrose gradients of DNA molecules towards the high molecular weight end of the gradient, by a decrease in the rate of filter elution of DNA from X-irradiated cells or by an increase in the rapidly renaturing fraction of DNA in cell lysates. All of these drug induced phenomena reach a maximum 6-12 hours after drug treatment and then decline. The half-life of DNA interstrand crosslinks is usually between 12 and 24 hours. Since only minimal degradation of DNA occurs during this time and because cross- links are stable in isolated DNA under physiologi- cal conditions, the reversal of drug-induced phe- nomena may be attributed to DNA repair of un- known mechanism.

The half-life of platinum bound to the DNA of cis-DDP-treated exponentially-growing Chinese hamster cells is approximately 28 hours [79, 80]. Again, since such DNA-bound platinum is stable chemically under physiological conditions its loss is presumed to represent excision repair of DNA ad- ducts. A recent report claims that the different kinetics of accumulation and loss of platinum on the DNA of African green monkey CV-1 cells treated continuously with cis-DDP or trans-DDP is due to a difference in the repair of the respective DNA-bound adducts [81] and this difference pro- vides a rationale for the enhanced biological ac- tivity of cis-DDP over trans-DDP. However, trans- DDP-induced DNA bound platinum is not lost more rapidly than that induced by cis-DDP in ei- ther Chinese hamster or CV-1 cells. In Chinese hamster cells, the different kinetics of accumula- tion and loss of platinum on DNA could be ade- quately explained by the different effects of the two isomers on DNA replication and cell cycle pro- gression [82].

Relationship between DNA excision repair and cytotoxicity

Despite correlations between the extent of drug- DNA interactions and cytotoxicity and the indica-

tions that cis-DDP-induced damage can be re- paired, the relationship between excision repair of DNA damage and toxicity is not always clear. Cells taken from patients suffering from the rare skin condition Xeroderma pigmentosum (XP) are morc sensitive to UV-irradiation than normal cells and are deficient in excision repair of UV-induced damage. These repair deficient XP cells are also more sensitive than normal foetal lung cells to cis- DDP when the lethal effects of thc drug are ex- pressed as a function of reaction with DNA rather than as a function of the dose of the reagent 1831. It has yet to be shown, however, that these cells are in fact defective in their ability to remove DNA- bound platinum. Cclls from paticnts with the ge- netic disease Fanconi's anaemia show unusual sen- sitivity to the cytotoxic and clastogenic effects of difunctional alkylating agcnts and arc also un- usually sensitive to cis-DDP. Such sensitivity is not the result of increased binding of platinum to DNA or, apparently, to a decreased abilitv to remove cis- DDP-induced DNA interstrand crosslinks (Pera and Roberts, unpublished results).

The role of DNA excision repair in protecting cells from the toxic effects of cis-DDP has emerged from studies on stationary phase cell cultures. It has been proposed that DNA synthesis on a dam- aged template is responsible for the toxic effects of cis-DDP [8(I] and that cells allowed to repair DNA prior to S-phase and cell division should show less toxicity than cells entering the proliferative cycle immediately after treatment. Stationary Chinese hamster cells are trcatcd with cis-DDP and their toxicity and platinum-DNA interaction are mea- sured in the non-dividing state after various hold- ing periods. The cells slowly excise platinum and the survival of these cells (when subsequently plated) is increased. The relationship betwcen ccll survival and the amount of platinum bound to DNA with several doses of cis-DDP (determined immediately aftcr treatment) is similar to that de- termined after varying periods of recovery follow- ing treatment with one dose [79]. Human fibro- blasts similarly recover from toxicity when held in a non-dividing state and excised platinum lesions from DNA with a half-life of 2.5 days and DNA- DNA and DNA-protein crosslinks with a half-life

Page 11: Drug resistance and DNA repair

of about 36 hours, and there was no evidence for any degradation of DNA. These results strongly support the hypothesis that damage present on the DNA template at the time of entry into the pro- liferative cycle is responsible for cellular toxicity. Further, the results show that the repair process is effective in achieving biological recovery: they do not however indicate that a particular lesion in DNA was specifically responsible for toxicity.

When, however, attempts are made to relate the inherent drug sensitivity of a given cell line to its repair capacity clear-cut results do not generally emerge. Walker 256 carcinoma cells display an tmusual sensitivity to a variety of bifunctional agents, including cis-DDP, but not UV- or X-irra- diation or to monofunctional agents. A subline of this tumour shows a 30-fold relative resistance to cis-DDP. Nevertheless, following treatment with a given dose the two sublines bind the same amount of platinum to their DNA, they excise platinum lesions from their DNA at a similar rate, and they remove DNA-protein and DNA interstrand crosslinks from DNA at a similar rate [84]. A simi- lar lack of correlation between crosslinking of DNA and cellular sensitivity has been found for lines of murine leukemia cells [85]. Another study of crosslink formation and removal indicates that while there is no significant difference between the crosslink removal rates in the parent and resistant lines, the kinetics are consistent with enhanced mechanism(s) for quenching the cis-DDP-DNA monoadduct in the resistant line [86]. Micetich, Zwelling and Kohn note that strand breaks do not accumulate in DNA during loss of Cisplatin-in- duced DNA adducts or DNA intcrstrand cross- links. It can therefore be presumed that if, as seems likely, the loss of platinum or crosslinks is indica- tive of the operation of a cellular excision repair process, then subsequent steps in this repair pro- cess are equally efficient in both sensitive and resis- tant cells and restore the integrity of cellular DNA.

It is conceivable that the fidelity of DNA repair is defective in sensitive but not in resistant cells. This possibility can be studied by assaying for expres- sion of bacterial genes that have been damaged in vitro prior to transfection into recipient cells. Knox et al. describe the mixed success of a study of DNA

271

repair in sensitive and resistant Walker tumour cells and reveal some possible difficulties with the system [22]. Expression of XGPRT is inhibited in a dose-dependent manner by cis-DDP treatment of pSV2gpt in both sensitive and resistant cell lines. Similar inhibition implies similar levels of plasmid repair in both cell lines, a result consistent with the findings in whole cells. More importantly, similar inhibition could imply comparable fidelity of repair in both cell lines. However. at a level of reaction with plasmid DNA that reduces the expression of XGPRT to less than 10% of the control level, fewer than 100 platinations were seen on the gpt region of the plasmid. Since DNA interstrand crosslinks oc- cur with a frequency of only approximately 1% it follows that they cannot be the lesion in the plas- mid DNA responsible for the inactivation of this genc: rather, the inactivating lesion is likely to bc the major product of reaction with the plasmid, namely an intrastrand crosslink. Therefore this transfection system is unable to compare the re- pairability of DNA interstrand crosslinks present in plasmid DNA by sensitive or resistant Walker cells. On the other hand it seems likely that it is the decreased ability of Walker sensitive cells to toler- ate DNA interstrand crosslinks that is the basis for their unique sensitivity to difunctional agents [87].

The above considerations recall earlier observa- tions in bacteria which showed that bacteriophage T7 and phage can be inactivated bv levels of plati- nation that do not produce, on average, one crosslinking event in their DNAs, and which there- fore indicate that intrastrand erosslinks are likely inactivating lesions for these exogenous DNA mol- ecules. Clearly the relevance of these in vitro sys- tems to the mechanism of the cytotoxic action of platinum compounds in mammalian cells must be questionable, but the transfection system still indi- cates the relative effcctivene~,s (in inactivating mammalian cells) of other (non-intcrstrand crosslinking) reactions with plasmid DNA. The level of reaction between the monofunctional plati- num compound Pt(dien) and the above plasmid which was required to produce the same level of inactivation was approximately ten times that re- quired with cis-DDP, paralleling the relative amounts of platinum bound to cellular DNA by

Page 12: Drug resistance and DNA repair

272

these compounds at equitoxic doses. These dif- ferent levels of reaction are a reflection of a cell's ability to repair or tolerate certain types of damage to its DNA. It would therefore appcar that damage to exogenous DNA, introduced by some but not all agents, is rcpaired or toleratcd in a similar manner to that in genomic DNA. For such damaging agents it would seem possible to use the cxprcssion of transfected genes as a measure of the repair capac- ity of the recipient cell, as discussed carlier. How- ever further studies are required to validate this approach.

The relative sensitivities of Walker tumour cells and a derived resistant subline to cis-DDP and to a numbcr of other bifunctional agents is of thc order of twenty to thirty. Moreover it is apparent from the response of the resistant Walker cells that they have a sensitivity to cis-DDP and to various other bifunctional agents, such as sulphur mustard, that is essentially the same as that of a number of other normal or tumour cell lines. The responses of both cell lines to a number of monofunctional agents or to uv- or X-irradiation are about the same and similar to those of normal cells. Therefore the so-called resistant cells represent the prevalent phenotype, while the 'sensitive' cells are a mutant variant uniquely sensitive only to bifunctionai agents. Similarly, other experimental animal tu- mours such as the Yoshida sarcoma have the phe- notype of mutant variants sensitive to specific types of DNA damagc [16]. In addition, a number of current studies [88, 89] compare the DNA repair characteristics of selected sensitive rodent cell vari- ants (isolated by mutagcn treatment of normal cells followed by replica plating) with those of the par- ent cells. The sensitivities of these selected cells towards certain agents can even exceed that of the sensitive Walker cell to bifunctional agents. Such variants are potentially vital tools for the elucida- tion of DNA repair pathways.

Levels of complexity in mutagen sensitivity

The complexity of thc phenotype of some mutagen sensitive cell lines is illustrated by comparative studies on Chinese hamster V79 and its mutagen

sensitive derivative V79/79. This latter cell line is sensitive to X-rays, ultraviolet irradiation (UV) and bi and monofunctional alkylating agents. In contrast to the parental V79 line, 0~MeG was re- moved from DNA with a tL/2 of 23.2hrs but no 06-GAT activity was detectable. In the sensitive cell line (V79/79) various defects in DNA synthesis in both control and treated cells were detected using alkaline sucrose gradient centrifugation which overall suggested a defect in ligation of low to high molecular weight DNA [90]. In addition, after MMS and HN 2 exposure thc sensitive cells failed to dclay progress through the cell cycle [91]; delay in DNA synthesis appears to be associated with a tolerance mechanism to bypass DNA lesions prior to replication (see prcvious section).

Several differences exist between the sensitive V79/79 cell line and its parent, all of which presum- ably make varying contributions to its drug sensi- tive phenotype. In this rcspect it is important to remember that the V79/79 cell line was isolated as a slow growing survivor after X-irradiation and may thus carry multiple mutations and/or chromosome rearrangements.

A similar complex phenotypc has been described for UV-I isolatcd on the basis of hypersensitiviy to UV-radiation [88]. UV-1 is hyperscnsitive to some DNA methylating and cross-linking agcnts and moderately hypersensitive to other classes of muta- gens. No differences in uptake and binding of la- bclled L~C MNU have been dctccted. Revertants of UV-1 are resistant to MNNG but still retain sen- sitivity to cross-linking agcnts. Fusion of UV-1 to 2 different UV sensitive CHO mutants results in hybrids resistant to mitomycin C but not to MMS. Methylation and cross-link sensitivity appear to be due to more than one genetic alteration. If this is so then it is difficult to understand how reversion occurs at measurable frequency unless the original variant is sensitive due to altered methylation pat- terns. In this context it is interesting to note that Xrs 'mutants' have been shown to revert to wild- type phenotype on exposure to 5-azacytidine, in- dicating that at least some repair genes are under methylation control [23].

Complex phenotypes are also evident in a series of 10 clones isolated as having greater than 5-fold

Page 13: Drug resistance and DNA repair

sensitivity to mitomycin C than wild type cells [89]. The mutants differ in their cross sensitivity to other DNA damaging agents (e.g., UV, cis-DDP, chlo- rambucil and melphalan). Only one mutant is UV sensitive and is also sensitive to the monofunctional decarbomyl mitomycin C. All clones MMC 1-5 have a recessive phenotypc and preliminary analy- sis indicates hat they fall into at least four comple- mentation groups.

At least 130 UV and EMS sensitive mutants have bcen isolated [24, 92, 93] after mutagencsis of CHO cells with EMS and ICR 17(I. The UV sensi- tive mutants are assigned to 5 genetic complcmen- tation groups on the basis of their response to UV light, The majority (123"130) of mutants are as- signed to groups I and 2 and the remaining 7 clones to another 3 complemcntation groups. The major- ity of the mutants show increased sensitivity to a varicty of othcr drugs, c.g., mitomycin C. An EMS sensitive mutant also shows a 10-fold increase in sensitivity to MMS. A 2-fold increasc in sensitivity was seen for ENU, MNNG and X-rays. When mutants from complementation groups 1-5 (UV sensitive) are fused with human lymphocytes the defects arc complemented by human chromosomes 19, 19, 2, 16 and 13 respectively. The complcmenta- tion of defects in groups 1 and 2 bv chromosomc 19 appears to be due to differcnt genes, as the phe- notypcs of these two groups differed markedly. Group I shows little or no hypcrsensitivity to DNA cross-linking agents, whcreas group 2 mutants are approximately 10-fold more sensitivc.

These studics in Chincse hamstcr cells parallel those in XP cells in which at lcast 8 complementa- tion groups have now been identified indicating a level of complexity approaching that sccn in Sac- charomyces cerevisiae [94]. Indeed an even grcater level of complexity might be expected in mam- malian cells than in lower eukaryotes when the higher order chromatin structure and thc possible preferential repair in some genes are taken into account. Some of the Chinese hamster UV-sensi- tive mutants have been used as recipients for trans- fection of human DNA in attempts to clone human DNA sequences which complcment the repair defi- cient phenotype. A cloned human repair gene ERCC-I has been located on chromosome 19 using

273

a panel of hamster x human hybrids and has been characteried at the molecular levcl [95-97]. It is 15 Kb long with 9 exons encoding two largely iden- tical mRNAs of 1.0 and 1.1 Kb. These have been shown by sequencing the corresponding cDNAs to be largely generated by differential splicing. The function of the 297 amino acid protein encoded by the eDNA is still unknown. Although transfection of the eDNA in an expressing vector conferred UV and mitomycin C resistance to Chinese hamster cells it did not correct the repair defect in SV40 transformed XP cells (complementation groups A and F). The Chinese hamster mutants used are known to be excision defective. Thus although it seems to be possible to clone human DNA which complements repair defective hamster cell lines, the identification of the function of the gene is made more complex by the possibility that the recipient cell line carries multiple repair defects. We are obviously some wav from understanding the molecular biology of DNA repair.

Influence of chromatin structure on repair

Several lines of evidence now indicate that euk- aryotic DNA is arranged in higher order chromatin loops by attachment to the nuclear matrix [98]. Transcribing gene sequences are generally thought to be in close proximity to the nuclear matrix and biochemical and autoradiographic studies on single cells indicate that excision repair does not require the attachment of the DNA to the matrix. However in xeroderma pigmentosum cells belonging to group C(XPC) DNA near to nuclear matrix was prcferentially repaircd 199 i. This non-random dis- tribution was not found in a number of other cell lines possessing low excision repair capacity (e.g.. XP group D. Syrian hamster embryo cells. HeLa cells) [98].

Other studies have demonstrated preferential repair of pyrimidine dimers induced by UV-irra- diation [100, 101]. Both CHO and mouse cells re- move few dimers from their genome overall. How- ever. dimers are removed efficiently from the dihydrofolic acid reductase gene (dhfr), the HMGCoA reductase gene in CHO cells and from

Page 14: Drug resistance and DNA repair

274

the transcribed c-abl, but not from thc silent c-mos genes in mouse cells [102]. In rcpair-proficient human cells thc removal of dimers appears much more rapid in the dhfr gcne than in the genomc as a whole. But in XPC cells repair is as deficicnt in the dhfr gene as in the genome as a whole.

The enhanced repair of essential active genes relative to the genome as a whole may play a sig- nificant role in overall survival of some cell lincs particularly in vitro where probably many genes are not actively transcribed. What happes in tumour cells in vivo is still unknown, as techniques for examining this problem are only just being de- veloped.

q'hc subtleties of control of this apparently selec- tive repair arc far from being understood, as are the possible roles such factors may have in the develop- ment of drug resistant cell lines. In several pairs of drug sensitive and resistant lines, in which no dif- ferences in the total amounts of DNA excision repair have bccn detected, it is possible that there have been redistributions of thc normal pattern of repair prefcrences or changes in chromatin struc- ture which affcct the efficiency of repair. Now that probcs for a number of specific gcnes are available, it is technically feasible to look at pattern changes and repair preferences. The intragenomic hetero- geneity in processing DNA damage is thought to be one of the factors that explain why Chinese hams- tcr cells, which remove only 10% of UV-induccd pyrimidinc dimers, arc as resistant to UV induced cytotoxicity as the significantly more repair-profi- cient human cells [100]. The sequence specificity of DNA damage by alkylating agents is only just be- ginning to be explored. The relative extent of al- kylation at any given guanine residue depends on the sequcnce flanking that site [103-105]. In the case of melphalan (phenyl alanine mustard) the alkylation of guanine is depressed by a 3' flanking cytosine. In a GpG sequence the 3' guanine is alkylated 2-3 times more than the 5'G except when in the sequence GpCpC when thc 5' guanine is preferentially alkylatcd. The GpG, Gp(3) and Gp(4) sequences are alkylated to a greater extent than isolated guanines by all the nitrogen mustards tested. Many of the 'housekeeping" genes (HPRT, dhfr, Tk) and a number of viral transforming genes

(c-Ha-ras 1 proto-oncogene) havc large G-C rich regions in their 5' flanking regions which could act as 'hot spots' for mutation in non-essential genes (HPRT) or cytotoxicity in essential genes (dhfr). Preferential alkylation of 5' flanking sequences in proto-oncogenes could result in mutation or al- tercd expression of such genes, which may be the molccular mechanism underlying increases in ma- lignancy observed in some cases when tumours become refractory to chemotherapy.

Superimposed on this specificity at the DNA level are factors operating at higher levels of DNA organisation which may also be important. Prefer- ential binding of a number of carcinogens to B DNA (aflatoxin BI) or Z DNA (N-hydroxy AAF) has been demonstrated, as has preferential binding to linker DNA as opposed to nucleosome DNA. The functional significance of such specific effects remains to be elucidated. DNA conformation can also modulate the extent of DNA repair as has been shown to be the casc for 0~-GAT. An alternat- ing polynucleotide can undergo conformational transition from thc B to the Z form. The E. coli 06-GAT efficiently removed 0O-guanine residues when the polynucleotide was in the B form but repair activity was totally inhibited on transition to the Z form [106].

Cell metabolism and other factors which may influ- ence sensitivity

The metabolic balance within the cell will also be of prime importance in its overall survival, and the precise mcchanisms involved in the coupling of DNA repair processes to DNA replication are still far from understood.

Intracellular pool sizes

Exposure of CHO-KI cells and other cell lines to chemical mutagens results in a rapid imbalance of cellular dCTP and dqq'P concentrations and recov- cry to normal Icvels takes several hours [107]. Dur- ing the recovery period dNTP pool imbalances prc- sumably make a contribution to the reduction of

Page 15: Drug resistance and DNA repair

semi-conservative replication. Addition of CdR (2 mM) results in a 2-3 fold greater depression of DNA synthesis and the time to rccoverv is corre- spondingly extended. The body of data revicwed in [108-110] indicates that post-treatment of cells with non-toxic concentrations of TdR (also known to produce imbalance of dNTP pools) can potentiate the lethal effects of monofunctional alkylating agents but not of UV or nitrogen mustard, suggest- ing that such effects could interfere with DNA repair.

Drugs other than thymidine, such as hydrox- yurea, which inhibits ribonucleotide reductase, or 5-fluorouracil, which inhibits thymidylate syn- thetase, can also have profound effects on intra- cellular pool sizes and may potentiate or protect from the lethal effects of subsequent exposure to alkylating agents. Such interactions may be impor- tant in the development of non-responsive tu- mours. The importance of differences in the con- trol of dNTP pools in drug resistant and sensitive cells is an area which is at present relatively unex- plored but is of obvious importance in some pairs of cell lines in which different rates of recovery from post-treatment DNA synthesis inhibition may re- flect differences in pool sizes.

Activation of poly (ADP ribose) polymerase

Other metabolic perturbations which occur as a result of DNA damage (eithcr direct strand breaks or brcaks generatcd during excision repair [111, 112]) includc the activation of the chromatin bound enzyme (poly ADP ribose) polymerase. The cn- zyme uses NAD as substrate and its continued activation can lead to drastic depletion of NAI) and ATP pools. The activation of the polymerasc and utilisation of NAD arc dependent on the amount of DNA damage induced. The cffccts are reversible, however, if nucleotide pools arc sufficient to per- mit regencration of ATP. Cells of different lineage (normal and tumour cells) may show large diffcrcn- ces in their intrinsic ability to survive such meta- bolic alterations. Tumour cells with their greater genetic diversity and often polyploid karyotype may be better fitted to survive such metabolic dis-

275

turbances and this factor may contibute signifi- cantly to the rapid development of drug resistance in tumours.

Hypomethylation

The two carcinogens MNU and MNNG which methylate DNA at the (16 position of guanine have been shown to causc extensive hypomethylation of DNA of Raji human lymphoblastoid cells two days after addition of the drug. The lower level of meth- ylation pcrsists for several cycles after treatment [113]. Treatment of K562 human erythroleukemia cells with CCRG 81045 an antitumour imidazo- tetrazine causes an increase in number of cells pro- ducing hemoglobin after 3 days and this incrcase is directly proportional to the decrease in 5Me cytosine content in [14] DNA. All ethyl analogue (CCRG 82019) failcd to cause any change in 5MedC content. These observations suggest a di- rect correlation between hypomethylation and the induction of a differentiated phenotype in K562 cells. One possible mechanism for hypomethyla- tion of DNA is the insertion of thymidine in place of cytosine opposite I)6 alkylated guanines. In sup- port of this hypothesis, evidence has been pre- sented indicating that the ratio of incorporation of 5-(methyl) thymidine to cytidine rises significantly in CCRG 8103,5 treated cultures after 3 days. The critical result of such a transition is loss of potential sites of enzymatic methylation and hence disor- dered gene expression [112].

Alkylation of DNA may also impair the action of proteins which recognisc specific sequences in DNA. Reaction of DNA with MNU has been shown to impair the ability of restriction enzymes to cleave this substrate. Since enzymatic methyla- tion is catalysed by sequence specific enzymes, al- kylation in the parental strand may result in an aberrant methylation pattern in the progeny strand. It has been suggested that N-7 guanine alkylation, a far more frequent event than 0<al - kylation, may be an important lesion in this re- spect. This area has been reviewed recently [7.8]. Changes in DNA methvlation patterns have also been reported after exposure of cells to cytosine

Page 16: Drug resistance and DNA repair

276

arabinoside, hydroxyurea (hypermethylation) and UV irradiation (hypomethylation). UV irradiation has also been shown directly to activate the tran- scription of thc metaliothionine gene which, on the basis of experiments with azacytidine, is thought to be under methylation control [114]. Hypermcthyla- tion after cytosine arabinoside and hydroxyurea exposure appears to require direct inhibition of DNA synthesis. Indirect inhibition via cyclohcx- amide-mediated inhibition of protein synthesis or by depletion of growth factors by serum depriva- tion did not result in hypermethylation.

In cells exposed to alkylating agents and other drugs used in chemotherapy extensive changes in methylation patterns may lead to changes in gene expression, either silencing of transcriptionally ac- tive genes or activation of those previously re- pressed. Whethcr any of these changes contribute to drug induced cytotoxicity or to the development of drug resistant phenotypes and increased malig- nancy remains to be determined.

Conclusions

Drug resistance (intrinsic or acquired) is commonly considered to be of genetic origin. Available evi- dence indicates that defined identifiable genetic changes are involved in many types of drug resis- tance, particularly in the case of resistance to anti- metabolitcs where it can be due either to deletion or mutation of an enzyme necessary for incorpora- tion of the drug into DNA (HPRT for resistance to 6-thioguaninc or 6-mercaptopurine) or amplifica- tion of a target enzyme (dhfr in the case of meth- otrexatc). In cell lines sensitive or resistant to al- kylating agents or other drugs which covalently interact with DNA, available evidencc also sug- gests their mutational origin, that is, they show long term genetic stability in the absence of selec- tive pressure. However, the exact mutations in- volved have yet to bc idcntified.

Exposure to DNA-damaging agents also triggers a highly complex chain of epigenetic events involv- ing multiple changes in cell metabolism, including inhibition of DNA synthesis and DNA mcthylation with resultant changes in gene expression. Which

of these changes contribute to cell dcath is as yet unclcar; indeed it is probably the synergistic effect of many events which is ultimately lethal. This level of complexity has so far precluded any precise iden- tification of a causal relationship between altered DNA rcpair capacity and altered cellular sen- sitivity. The correlations exist but most drug resis- tant (or sensitive) ccll lines exhibit other changes intrinsic or induced, transient or permanent which may contribute equally to their altered phenotypc.

It is important to know whether cells of a human tumour can possess a sensitivity to drugs which is comparable to that of the highly sensitivcr odent cell variants and whether tumours exhibit a com- parable decrease in sensitivity following chemothe- rapy, due either to selection of pre-existing resis- tant cells or to mutational change. To our knowledge it has yet to be demonstrated that cclls of a primary human tumour can have a sensitivity to cytotoxic agents that cquates with that of the sensitive rodent variants, other, that is, than those occurring in individuals suffering from a genetically determined repair defcct. Likewise it has also yet to be demonstrated whether a so-called resistant human cell derived from an originally sensitive tumour has a sensitivity which equates with 'nor- real cells' or whether it is more resistant than nor- mal cells due to acquisition of repair processes.

One of the main difficulties in assigning a defini- tive role to specific DNA lesions in the cytotoxic and mutagenic effects of DNA damaging agents is that comparisons of differentially sensitive cells are not comparisons of otherwise isogcnic cell lines. Similarly many of the UV and drug sensitive vari- ants that are available have been isolated after mutagencsis and could carry many mutations other than those conferring the DNA repair defect se- lected for.

Thc response to and recovery from DNA dam- age is governed by a large number of genes in- volved not only in excision repair, but in DNA replication, control of poly ADP ribose activity and lcvcls of intraccllular DNA pools. Small differen- ces in expression of one or more of these genes may make a profound difference to cellular sensitivity. The transfection of cloned DNA repair genes into freshly cloned cell lines of rodent or human origin

Page 17: Drug resistance and DNA repair

which lack specific repair funct ions may allow

more detai led analysis of such complex interac-

t ions.

It is to be hoped that the appl icat ion of the

powerful tools of r e combinan t D N A technology

will in the nea r future help to elucidate some of the

complexit ies of the drug resistant phcnotypes dis-

cussed, and that such knowledge when available

can be applied in the clinic.

Key unans~vered questions

I. How closely are excision repair and replicat ion

coupled and what is the mechanism of this coupling

at the molecular level?

2. What are the precise funct ions of the proteins

encoded by the m a m m a l i a n repair genes c loned to

date?

3. How many genes are involved in D N A repair in

m a m m a l i a n cells?

4. Is the cytotoxicity of D N A damaging agents dttc

directly to lesions in D N A and as a corollary, is

resistance duc to more efficient repair or do other

events , e .g. , dcmethv la t ion of D N A , and per turba-

t ions of cellular pools play an impor tan t role?

5. Are D N A repair genes, in part icular that for

06-(}A'I" inducible in m a m m a l i a n cells?

6. Do sensit ive and resistant cells differ in their

abil i ty to tolerate damage to D N A and what is the

precise role of D N A repair?

Acknowledgements

The authors are suppor ted by grants from the Med-

ical Research Counci l and the Canccr Rcscarch

Campaign .

References

1. Henderson JF: The experimental setting. In: Fox BW and Fox M (cds) Antitumour drug resistance. Springer Vcrlag: Berlin Hcidelburg. New York, Tokyo, 1984. pp 23-35

2. Sager R, Gadi 1K, Stephens L. Grabowy CT: Gcnc ampli- fication: an example of accelerated evolution in tu- mourigenic cells. Proc Nat Acad Sci (USA) 82: 71H5-7019, 1985

277

3. Cifone M/,,.. Fidler 1J: Increasing metastatic potential is associated x~.lth increasing genetic mstabilitv of clones iso- lated from routine neoplasms. Proc Nat Acad Sci (USA) 78: 6949-6952, 1981

4. Kerbal RS. Davies AJS: Facilitation of turnout pro- grcssion by cancer therapy. Lancet 3(1: 977-978. 1982

5. Fox BW, Fox M (cds) Antitumour drug resistance. Springer Vcrlag: Berlin Hcidclburg, Nc~. York, Tokyo, 1984

6. Jones AP: Altering gene expression with 5-azacvtidine. Cell 4t): 48.%-486, 1985

7. Jones PA: DNA mcthylation and cancer. Cancer Rcs 46: 461-466, 1986

S. Wilson VL Jones PA: Inhibition of DNA methylation by chemical carcinogens in vitro. Cell 32:23%246.19,~4

9. Sartorclli AC: Malignant cell differentiation as a potential therapeutic approach. Brit J Cancer 52: 293-302, 19S5

It). Stcphans TC, Adams K. Peacock JH: Emcrgence of nitro- sourca resistant sublines of Lewis lung tumor following McCCNU treatment in vivo. Brit J Cancer 53: 237-2-1-5. 1986

11. Goldic J[ !, Goldman AJ, Gudauskas GA: Rationale for the ttsc of alternating non-cross-resistant chemotherapy. Cancer Treatment Repts 66: 439-449, 1982

12. Shimkc RT, Sherwood SW, llill AB, John_¢<m RN: ()vet- replication and recombination of DNA in higher eukarvotcs: Potential consequences and biological im- plications. Proc Natl Acad Sci (USA) 83: 2157-21hl, 19,.'56

13. Johnson NP. Hocschcle JD, Rahn RO. O'Neill .IP. Hsie AW: Mutagenieity cytotoxicity and DNA binding of Plati- num I ll)-chloroummines in Chinese hamster ovary cells. Cancer Rcs 40: 1463-.14(,,8. 1980

14. Singh B, Gupta RS: Mutagcnic responses of thirteen anti- cancer drugs on mutation induction at multiple genetic loci and on sister chromatid exchanges in Chinese hamster ovary cells. Cancer Res 43: 577-584, 19~3

15. Thackcr J: The molecular nalure of mutations in cultured mammalian cells: A review. Mut Rcs 150: 431-442, 1985

16. Fox M: Drug resistance and DNA repair. In: Fox BW. Fox M (cds) Antitumour drug Resistance. Springer Verlag: Berlin, Heidclburg, New York, Tokyo. 1984. pp 335-362

17. Fricdbcrg EC, Hanawalt PC (eds): DNA repair: A Labo- ratory Manual of Research Procedures, \.'ol 1 Pts A & B. Marcel Decker Inc., New York. Basel, 1981

18. Mulligan RC. Berg P: Expression of a bacterial gene in mammalian cells. Scicnce 209: 1422-1427, 1980

19. Thacker J: The use of recombinant I)NA tcchmques to stud.',' radiation induced damage repair and genetic change in mammalian cells. Int J Rad Biol 50: 1-31, 1986

20. Cox R. Is'Iasson WK, Dcbenham PG. Wcbb MBT: The use of recombinant DNA plasmids for the determination of DNA-rcpair and recombination in cultured mammalian cells. Brit J Cancer 9: Supp VI 67-72, 1984

21..laspers NG, Painter RB, Paterson MC. Kidson C, lnouc T: Complcmcntation analysis of ataxia telangicctasia. KROC Found. Ser 9: 147-102. 19~5

Page 18: Drug resistance and DNA repair

278

22. Knox R J, Lydall DA. Fricdlos F, Basham C, Roberts JJ: Thc effect of monofunctional or difunctional platinum adducts and of various other DNA damage on the expres- sion of transfected DNA in mammalian cell lines inher- ently sensitive or resistant to difunctional agents. Biochem Biophys Acta 908: 214-223, 1987

23. Jcggo PA, Kemp LM: X ray sensitive mutants of a Chincsc hamster ovary cell line: isolation and cross sensitivity to DNA damaging agcnts. Mut Res 112: 313-319, 1983

24. Thompson LH, Rubin J8, Clcavcr JE, Whitmorc GF, Brokman K: A screening method for isolating DNA rcpair deficient mutants of CHO cells. Somatic Cell Genet 6: 391-=406, 1981)

25. Jones N J, Debenham PG, Thackcr J : New X ray sensitive mutants of cultured hamster cells. A Brit J Cancer 54: 349. 1986

26. Jeggo PA. Holliday R: Azacytidine-induced reactivation of a DNA repair gcnc in Chinese hamster ovary cells. Mol and Cell Biol 6: 2944-2949, 1986

27. Olsson M, l.indahl T: Repair of alykylated DNA in E. coli: methyl group transfer from 0~methylguanine to a protein cystcinc residue. J Biol Chem 255: 10369-10571, 1980

28. Morimoto K, Dolan ME, Scichitano D. Pegg AE: Repair of 0~'propylguanine and (F, butylguanine in DNA by 0~-al - kylguaninc DNA alkyltranslerases Irom rat liver and E. coli. Carcinogcncsis 6: 1027-1(131, 1985

29. Sedgwick B; Molecular cloning of a gene which regulates the adaptive response to alkylating agents in E. coli. Mol and Gen Genet 191: 460-472, 1983

30. Margison GP, Cooper I)P, Brennand J: Cloning of the E. coli 0°mcthylguaninc and methylphosphotriester meth- yltransferase gene using a functional DNA repair assay. Nucleic Acids Res 13: 1939-1952, 1984

31. Nakabeppu Y. Kondo 1t, Kawabata S, Iwanaga S, Sckiguiehi M: Purification and structure of the intact ada regulatory protein E. coli KI2 (~mcthylguaninc DNA mcthyltransferase. J Biol ( 'hem 260: 7281-7288, 1985

32. Demple B, Sedgwick B. Robins P, Totty N, Waterfield MP, Lindahl T: Active site and complete sequence of the suicidal methyltransferase that counters alkylation muta- genesis. Proc Nat Acad Sci USA 82: 2688-2692, 1985

33. Teo I, Sedgwick G, Demple B, Li B, Liudahl T: Induction of rcsistance of alkylating agents in E. coli: the ada' gene product serves both as a regulatory protein and as an enzyme for repair of mutagenic damage. EMBO J 3: 2151- 2157, 1983

34. Brcnt TP: Inactivation of purified human O"alkylguanine alkyltransferase by alkylating agents or alkylatcd DNA. Cancer Res 46: 2320--2323, 1986

35. Yarosh DB, Barnes D, Erickson LC: Transfection of DNA from a chlorocthylnitrosourea-resistant tumour cell line (mer ' ) to a sensitive cell line ( m e r ) results in a tumour cell line resistant to M N N G and CNU has in- creased (~'-methylguanine DNA methyltransfcrase Icvcls and reduced lcvcls of DNA cross-linking. Carcinogcncsis 7: 160.3-1606, 1986

36. Saffhill R, Margison GP. O'Connor P.I: Mechanisms of carcinogenesis induced by alkylating agents. Biochem Bi- ophys Acta 823: 111-145, 1985

37. Warren W, Crathorn AR, Shooter KV: The stability of methylated purincs and of methylphosphotricstcrs in DNA of V79 cells after treatmcnt with N-methyi-N-nitro- sourca. Biochem Biophys Acta 563: 82-88, 1979

38. Goth-Goldstcin R: Inability of Chinese hamstcr ovary cells to cxcisc 0~alkylguaninc. Cancer Res 41): 2633-2624,

1980 39. Day RS. Ziolkowski CHJ: l luman brain turnout cell

strains with deficient host cell reactivation of N-methyl N-nitro-N-nitrosoguanidinc damaged adenovirus 5. Na- ture (London) 279: 797-799, 1977

41.1. Yarosh DB: The role of (~'methylguanine DNA meth- yltransfcrasc in cell survival mutagenesis and car- cinogencsis. Mut Res 145:1-16. 1985

41. Brcnnand J, Margison GP: Exprcssion of the l'.'. Coli [Y'methylguanine methylphosphotricstcr gene in mam- malian cells. Carcinogenesis 7: 185-188, 1986

42. Margison GP. Brennand J: Reduction of the toxicity and mutagcnicity of alkylating agents in mammalian cells har- bouring the E. coli alkyltransferase gent. Proc Nat Acad Sci 83: 6292-6296, 1986

43. Roberts JJ: The repair of DNA modified by cytotoxic mutagcnic and carcinogenic chemicals. Advances in Rad Biot 7: 211-381, 1978

44. Baker RM, van Voorhis WC, Spencer I.A: HcLa cell varmnts that differ in sensitivity to monofunctional al- kylating agents with indcpcndcncc of cytotoxic anti muta- genic responses. Proc Nat Acad Sci (USA) 76: 524(2-5253, 1979

45. Ockcy CH, Brennand J, Margison GP: The frequency of N-methyl-N-nitrosourea induced sister chromatid ex- changes is reduced in mammalian cells cxpressing the E. coil 0%uaninc alkyltransferase gene. Brit J Cancer 54: 366, 1986

46. Shaffer DA: Replication bypass model of sister chromatid exchanges and implications for Bloom's syndrome and Fanconi's anaemia, tluman Genetics 37: 177-F~, 1977

47. Suttcr W, Brennand J, McMillan S, Fox M: Relative mu- tagcnicity of antineoplastic drugs and other alkylating agents in V79 Chinese hamster cells; independence of cylotoxic and mutagenic responses. Mut Res 73: 171-181, 1980

48. Karran P, l.indahl T: Cellular defence mcchanisms against alkylating agents. Cancer Surveys 4: 58.¢-595, 1985

49. Durrant LG, Margison GI', Boyle .IM: Prctreatmcnt of Chinese hamster V79 cells with MNU increases survival without affecting DNA repair or mutagenicity. Car- cinugenesis 2: 55-61), 1981

50. Karran P. Williams SA: Thc cytotoxic and mutagcnic effects of alkylating agents on human lymphoid cells arc caused by different DNA lesions. Carcinogenesis 6: 789- 792, 1985

51. llolliday R. Jeggo PA: Mechanisms for changing gcnc

Page 19: Drug resistance and DNA repair

expression and their possible relationship to car- cinogencsis. Cancer Survcys 4: 557-581. 1985

52. Laxal F: Inducible repair of alkylation damage in mam- malian cclls. Proc Nail Acad Sci [USA) 81: 1962-1966. 1984

53. I_aval F. Laval J: Adaptive response in mammalian cells: Cross reactivity of different pre-treatments on cytotoxicity as contrasted to mutagcnicity. Proc Natl Acad Sci (USA) 81: 1062-1066, 1984

54. Boyle JM, Saflhill R, Margison GP, Fox M: A comparison of cell survival mutation and persistancc of putati,,c pro- mutagenic lesions in Chinese hamster cells exposed to BN U or MNU. Carcinogenesis 1981-1985. 1986

55. Boyle JM, GP Margison, Saffhill R: Evidence for excision repair of lY'nButyldeoxyguanosine in human cells. Car- cinogencsis 1987-199(I, 1986

56. Erickson L('. Bradley MO, Kohn KW: Measurements ot DNA damage in Chinese hamster cells trcatcd with cquitoxic and equimutagenic doses of nitrosoureas. Can- cer Res 38: 3379-3384, 1978

57. Erickson LC. Laurent G, Sharkey NA, Kohn KW: DNA cross-linking and monoadduct repair in nitrosourea- treated human turnout cells. Nature 288: 727-729. 1980

58. Bodell WJ. Aida T, Berger MS. Rosenblum Mi.: In- creased repair of 0%lkylguanmc adducts in glioma derived human cells resistant to the cytotoxic and eytogenetic effects of 1.3-bis['2-chlorocthyl)-l-nitrosourea. Car- cinogenesis 7: 879-&',13, 1986

59. Bodell WJ. Aida T, Berger MS. Rosenblum ML: Repair of 06-(2-chlorocthyllguaninc mcdmtcs the biological effects of chloroethylnitrosoureas. Environ llealth Persp 62: 119-126. 1985

60. Day RS, Ziolkowski CJH, Scudicro DA, Mvcr SA, Lubiniecki AS, Giradi A J. Galloway SM, Bynum CD: Deleetive repair of alkylated DNA by human tumour and SV40 transformed human ecll strains. Nature 288: 724-- 727, 1980

61. Fox M, Scott D: "l he genetic toxicology of nitrogen and sulphur mustard. Mut Rcs 75: 131-168, 198(I

62. Fox BW: Alkylating agents: Mechanisms of Resistance. A Review. 13th Int. Canccr Congress PtC Biology ol Cancer (2). Alan R Liss. Inc, 1983 p 247-255

63. Dean SW. Fox M: DNA repair. DNA synthesis and cell cycle delay in human lymphoblastoid cells differentially sensitive to nitrogen mustard. Mut Rcs 132: 63-72. 1984

64. Dean SW, Fox M: Investigation of the cell cycle response of normal and Fanconi anaemia fibroblasts to nitrogen

. ~6.",-_79. mustard using flow cytometrv. J Cell Scicncc 64: ~ - "~ 1983

65. Dean SW. Johnson AB, Tew KD: A comparative analysis of drug induced DNA effects in a nitrogen nmstard resis- tam cell line expressing sensitivity to nitrosoureas. Bin- chemical Pharmacol 35: 1171-1176, 1986

66. Prestayko AW. Crookc ST, Carter SK (cds) Cisplatin current status and new devctopments. Academic Press. New York, 1980

279

67. Pascoe JM, Roberts .l J: Interactions between mammalian cell DNA and inorganic platinunl compounds. I. DNA interstrand crosslinkmg and cytotoxic properties of plati- num 11 compounds. Biochemical Pharmaco123: 1345, 1974

68. Plooy ACM. wm Dijk M. Lohman PHM: Induction and repair of DNA crossIinks in Chinese hamster ovary cells treated with various platinum compounds in relation to platinum binding to DNA, cytotoxicity, mutagcnicity and antitumour activity. Cancer Res 44: 2043-2051, 1984

69. Zwelling LA. Anderson T, Kohn KW: DNA-protein and DNA intcrstrand crosslinking by cis- and trans-platinum (II) diammincdichloridc in LI21tl mouse leukemia cells and in relation to cytotoxicity. Canccr Res 39: 365-369. 1979

70. Bcrgcrat J-P, Drcv, inko B. Corry P, Barlogic B, Ho DH: Synergistic lethal effects ofcis-dichlorodiammineplatinum and l-D-arabinofuranosylcytosinc. Cancer Re~ 41: 25-30, 1981

71. Mcyn RE. Corry PIVI, Fletcher SE. l)cmctriades/vI: Ther- mal enhancement of DNA damage in mammalian cells treated with c/~-diamminedichloroplalinum (Ill. Cancer Rcs 40: 1136--1139, 1980

72. Pinto AL. Lippard S.h Binding of the antitumor drug cis- diamminedichloroplatinum (II) (Cisplatin) to DNA Bio- chem Biophys Acla 780: 167-180. 1985

73. Tullius TD, Lippard S J: Cis-Diamminedichloroplatinum (II) binds in a unique manner to oligo(dG)-oligo(dC) se- quence in D N A - a new assay using Exonuclease II1. J Ant Chem Soc 103: 4620. 1981

74. Rover-Pokora B, Gordon LK, Haseltine WA: Use of cxonuclease III to determine the site of stable lesions in defined sequences of DNA: the cvclobutane pyrimidinc dimer and cis and trans-dichlorodiammineplatinum II ex- amples. Nucleic Acid Res 9: 4595~-1(~19, 1981

75. Drobnik J, Urbankova MM. Krekulova A: The effect ol' cis dichlorodiammincplatinum (ll) on t:5cherichia cob B. Mutation Res 17: 1.3--20, 1973

76. Beck D J, Brubakcr RR: Effect of cis-platinum (II)diam- mine dichloride on wild type and deoxyribonucleic acid repair-deficient mutants of Escherichia coil. J Bact 116: 1247, 1973

77. Beck D J, Popoff S, Sancar A, Rupp D: Reactions of the UVRABC excision ,nucleasc with DNA damaged by di- amminedichloroplalinum (ll). Nucleic Acids Rcs 13: 7395-7412, 1985

78. Van den Berg HW: Roberts J J: Investigatians into the mechanism of action of anti-tumour platinum compounds: time- and dose-dependent changes in the alkaline sucrose gradient sedimentation profiles of I)NA from hamster cells treated with cis-platinum (ll)diamminedichloridc. Chem-Biol Interactions 11: 493-499. 1975

79. Pera Jr MF. Rawlings CJ. Shacklcton J. Roberts .IJ : Quan- titative aspects of the fl~rmation and loss of DNA inter- strand crosslinks in Chinese hamster cells follo'~,ing treat- ment with e/.~--dmmmincdichloroplatinmn (IlL 1I. Comparison ol results from alkalinc elution. DNA re-

Page 20: Drug resistance and DNA repair

280

naturation and DNA sedimentation. Biochcm Biophys Acta 655: 152, 1981

80. Fraval HNA, Roberts J.l: Excision repair of cts-diam- minedichloroplatinum (ll)-induccd damage to DNA in Chinese hamster cells. Cancer Rcs 39: 1793-1797, 1979

81. Ciccarelli RB, Solomon KJ, Varshavsky A, l.ippard S J: hi vivo effects of c/s- and tratt~-diamminedichloroplatinum (ll) on SV40 Chromosomes: Differential repair, DNA- protein crossiinking, and inhibition of replication. Bio- chemistry 24: 7533-7540, 1985

82. Roberts J J, Friedk~s F: The differential toxicity of c/s- and trans-diamminedichloroplatinum (If) towards mam- malian cells lack of influence of any difference in the rates of loss of their DNA-bound adducts. Cancer Res 47: 31-36, 1987

83. Fraval HNA, Rawlings CJ, Roberts Jj: Incrcased sen- sitivity of UV-repair deficient human cells to DNA bound platinum products which unlike thymine dimers arc not recognized by an endonuclease extracted from M. Luteus. Mutation Res 51: 121-132, 1978

84. Rawlings CJ, Roberts JJ: Walker rat carcinoma cells are exceptionally sensitive to cis-diamminedichloroplatimlm (II) (Cisplatin) and other difunctional agents but not dc- fectiVe in the removal of platinum DNA adduets. Muta- tion Res 166: 157-168, 1986

85. Strandberg MC, Bresnick E, Eastman A: The significance of DNA cross-linking to cis-diamminedichloroplatinum (ll)-induced cytotoxicity in sensitive and resistant lines of murine leukemia L1210 cells. Chem-Biol Interactions 39: 16%180, 1982

86. Micetich K, Zwclling LA, Kohn KW: Quenching of DNA: Platinum (II) monoadducts as a possible mecha- nism of resistance to ct~-diamminedichloroplatinum (II) in L1210 cells. Cancer Res 43: 36419-3613, 1983

87. Roberts 11, Fricdlos F, Scott D, Ormerod MG, Rawlings C J: The unique sensitivity of Walker rat tumour cells to difunctional agents is a~ociated with a failure to recover from inhibition of DNA synthesis, and increased chromo- some damage. Mutation Res 166: 16',t-181, 1986

88. Hoy CA. Thompson LH, Salazar EP, Stcwart SA: Dif- ferent genetic alterations underlie dual hypersensitivity of CHO mutant UV-I to DNA mcthylating and cross linking agcnts. Somatic Cell Mol Genctics 11: 523-532, 1985

89. Robson CN, Harris AL, Hickson LD: Isolation and charactcrisation of Chinese hamster ovary cell lines sensi- tive to mitomycin C and bleomycin. Cancer Res 45: 5304- 5309, 1985

90. Fox M, Bloomfield ME, Hopkins J, Boyle JM: Differen- tial responses of nascent DNA synthesis and chain elonga- tion in V79 and V79/79 cells exposed to UV light and chcmical mutagens. Carcinogenesis 4: 261-268, 1983

91. Graham A, Fox M: The role of suppression of DNA synthesis and inhibition of cell cycle progression in cellular sensitivity to alkylation damage. Carcinogenesis 4: 26q- 274, 1983

92. Bush DB, Cleaver JE, Glaser DA: I.argc scale isolation of

UV sensitive clones of CHO cells. Somatic Cell Genetics 6: 407-418, 19811

93. Thompson LIt, Bush DB, Brookman K, Mooney CL, Glaser DA: Genetic diversity of ultraviolet sensitive DNA repair mutants of Chinese hamster ovary cells. Proc Natl Acad Sci (USA) 78: 3734-3737, 1981

94. Friedbcrg EC: Nucleotide excision repair of DNA in eukaryotes: comparisons between human cells and yeast. Cancer Surveys 4: 52%555, 1985

95. Westcrvcld A, Hoeijmakcrs A J, van I)uin M, Odijk I:I, Wood RD, Bootsma D: MolecuIar ckming of a human DNA repair gene. Nature (London) 3111: 425-429,198-t

96. Maclnnes MA, Bingham JM, Thompson LH, Strniste GF: DNA mediated cotransfer of excision repair capacity and drug resistance into Chinese hamster ovary mutant cell line UV 135. Mol and Cell Biol 4:1152-1158

97. Ruhin JS, Prideaux VR, lluntington HF, Dulhanty AM, Whitmorc GF, Bernstein A: Molecular cloning and chro- mosomal location of DNA sequcnccs associated with a human DNA repair gcne. Mol and Cell Biol 5: 398-405, 1985

98. Mullenders LIIF, van Kesteren AC, Bu.~smann CUM, van Zceland AA, Natarajan AT: Distribution of UV in- duced repair events in higher order chromatin loops in human and hamster flbroblasts. Carcinogencsis 7: 995- 1002, 1986

99. Mullendcrs I.HF, van Kestcren AC, Bussmann (]JM, van Zeeland AA, Natarajan AT: Prcfcrcntial rcpair of nu- clear matrix associated DNA in xeroderma pigmentosum group C. Mut Res 141: 75-82, 1984

100. Bohr VA, Okumoto DS, Ifanawalt PC: Survival of UV-ir- radiated mammalian cells correlates with efficient DNA repair in an essential gene. Proc Natl Acad Sci (USA) 83: 3830-3833, 1986

101. Bohr VA, Smith CA, Okumoto DS, Hanawalt PC: DNA repair in active genes: removal of pyrimidine dimcrs from the DHFR genc in ClIO cells is much more efficient than in the genome overall. Cell 44): 359-369, 198.5

102. Madhani til3, Bohr VA, Hanawal! PC: Differential re- pair in transcriptionally active and inactive proto-on- cogenes c abl and c mos. Cell 45: 417-423, 1986

1113. Mattler WB, Hartley .IA, Rahn RO, Gibson N, Kohn KW: Sequence specificity of DNA alkylating agents and their preferential binding to Ha-ras sequences. Abstract Int. Conf. on mechanisms of DNA damage and repair. Implications for carcinogenesis and risk assessment. Gaithersburg, Maryland U.S.A., June, 1985

104. Furois-Corbin S, Pullman B: Specificity of carcinogen DNA interaction: A theoretical exploration of the factors involved in thc effect of ncighbouring bases on N-methyl N-nitrosourca alkylation of DNA. Chem Biol lnt 54: 9-13, 1985

105. Briscoc WT, Cotter I.E: DNA sequence has an effect on the extent and kinds of alkylation of DNA by a potent carcinogen. Chem Biol Int 56: 321-331, 1985

106. Boiteaux S, Costa de Oliveira R, Laval J: The Escherichia

Page 21: Drug resistance and DNA repair

coli (r', methyltransferase does not repair promutagenic 05 mcthylguaninc residues whcn presenl in Z DNA. J Biol Chem 260: 8711-8715, 1985

107. Newman CN, Miller Jll: Mechanism o[ UV induced deoxynucleoside triphosphatc pc~ol imbalance in CI IO-KI cclls. Mutation Res 1-15: 95-101. 1985

108. Kunz BA: Gcnctic eflects of deoxyribonuclcotide pool imb,"danccs. En',ironmcntal Mut 4: 695-727, 1982

109. Fox M: The effects of pyrimidinc t~ucleotides on alkylating agcnt induced cytotoxicity and' spontaneous and indttced mtttation to purine analogue resistance hi V79 cells. [tl: FJ de Scrres (ed) Gcnetic Conscqucnces of nucleotidc pool imbalance. Plenum 1985. pp 435-4-10

110. Weinburg G L Ullman B. Wright CM. IVlartin DW: The effects of exogenous thymidinc on endogenous dcoxy- nuclcotidcs and muta~cncsis in mammalian cells. Somatic Cell and Mol Genetics 11: 413-419. 1985

Ill. Kameshita I, Matsuda Z, lani~,uchi T. Shizuta Y: Poly (ADP ribosc) synthetase. J Biol Chcm 259: 4770-4776. 1984

281

112. Krcimeyer A. Wieckens K. Adamctz P, Hilz II: DNA repair-associated ADP ribosylation in vivo. J Biol Chcm 259: 890-896, 1984

113. Boechan TI.J. Drahovsky D: llypomcthylation of DNA in Raji cells after treatment v,ith N-metbyl-N-nJtrosourea. Carcinogenesis 2: 39.-42, 1981

114. Ti.sdalc M J: Antitumour mlidazotetrazines X effect of 8-catbomyl-3-mcthytimidozo [5-1.-d] 1235 tctrazin-4 (3H) one (CCRG 81045 M & B 39831: NSC 362856) on DNA methylation during indliction of haemoglobin synthesis in human leukemia ccll line K562. Biochem Pharmacol 35: 311-316, 1986

115. Nvcc J~ Lit, 1.. Jones PA: Variable effects of DNA syn- thesis inhibitors upon DNA methylation in mammalian cells. Nucl Acid Rcs 14: 4352--4367, 1986

Addre.~,~ .for oJfprinLs: M. Fox. Paterson lnstilule for Cancer Research, Christie Hospital amt Holt Radium Institute. Man- chester M20 9BX. United Kingdom