Drug-induced DNA hypermethylation: A potential mediator of acquired drug resistance during cancer chemotherapy

https://doi.org/10.1016/S1383-5742(96)00051-8Get rights and content

Introduction

Enhanced genetic diversity within tumor cell populations is well documented [1]. Tumor cells are known to possess higher rates of mutation than their normal counterparts 2, 3, possibly due to inactivation of critical cell cycle check point proteins [4], DNA repair mechanisms [5], and other cellular processes [6]. Such genetic instability in tumor cells is thought to play a major role in tumor progression, metastasis, and the development of resistance to anticancer agents. While less well documented, there is a growing literature which suggests that epigenetic diversity is also enhanced within tumor cell populations 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28. Epigenetic instability may also play a significant role in tumor progression and metastasis.

My laboratory has studied epigenetic changes within tumor cells that are caused by exposure to cancer chemotherapy drugs, and which confer a state of drug resistance upon the cancer cells 29, 30. Specifically, we have found that:

  • Exposure of tumor cells to DNA synthesis-inhibiting concentrations of anticancer agents is associated with extensive DNA hypermethylation;

  • Such drug-induced DNA hypermethylation is very much part of the response to drug-induced toxicity since it is observed primarily in tumor cell populations experiencing >95% lethality;

  • Such drug-induced DNA hypermethylation can silence the expression of genes being synthesized during the window of drug-induced toxicity;

  • Such drug-induced DNA hypermethylation can create drug resistance by randomly inactivating genes whose products are required to activate cancer chemotherapy agents to their cancer-killing forms;

  • Drug-induced DNA hypermethylation can be blocked, in a dose-dependent manner, by inhibitors of DNA methylation such as 5-aza-2′-deoxycytidine;

  • Drug-induced DNA hypermethylation is not a tissue culture phenomenon since it occurs in patients undergoing cancer chemotherapy;

  • Cisplatin is the most potent inducer of DNA hypermethylation, possibly due to conformational changes induced by cisplatin adducts which render the affected DNA a better substrate for DNA cytosine 5-methyltransferase;

  • Topoisomerase inhibitors generally inhibit DNA methylation, possibly by inducing conformational changes which render the affected DNA a poor substrate for DNA cytosine 5-methyltransferase;

  • Drug-induced changes in DNA methylation, both hyper- and hypomethylation, contribute to tumor cell epigenetic and genetic instability.

A discussion of each of these points follows.

A broad selection of anticancer drugs were found to induce DNA hypermethylation when applied at toxic concentrations in vitro (Fig. 1). For example, cytosine arabinoside (araC), 5-fluorouracil and methotrexate each produced dose-dependent increases in the methylation content of DNA. As expected, the methylation inhibitor 5-azadeoxycytidine (5-azadC) decreased methylation in treated tumor cells. Unexpected was the finding that 6-thioguanine inhibited DNA methylation (Fig. 1). One possible explanation for this is suggested by recent work which demonstrated that both O6-methylguanine (a DNA adduct induced by exposure to endogenous and exogenous methylating agents) and 6-thioguanine are repaired by the same short patch repair mechanism [31]. Delayed remethylation of the repair patch could account for the hypomethylation observed [32]. Alternatively, 6-thioguanine could inhibit methylation of 5′ cytosines directly.

Most of the drug-induced DNA hypermethylation appears within newly replicated DNA, and most (but not all), within the canonical methylation target dinucleotide CpG [33]. Methylation occurring in dinucleotide fractions other than CpG (e.g., CpA, CpT, CpC) appears to be gradually lost over several cell cycles, but a significant percentage remains even after five cell cycle times [34]. The simplest explanation for drug-induced DNA hypermethylation is that DNA methylase enzymes, which typically trail replication forks, `catch up' to replication forks stalled as a result of drug-induced DNA synthesis inhibition. Due to their increased contact time with newly replicated cytosines in the stalled replication forks, methylase enzymes (according to this view) modify a much larger percentage of cytosines than they normally would. Both cytosines within CpG dinucleotides and cytosines in other dinucleotide fractions apparently become methylated during this process 29, 30, 34.

Although it occurs in a dose-dependent fashion, drug-induced DNA hypermethylation is only readily apparent in tumor cells responding to extremely toxic concentrations of anticancer agents. Typically, greater than 95% cell kill is required, and analytical techniques must be employed that detect methylation only during the window of toxic drug exposure [33]. One such technique is the use of radiolabelled [14C]deoxycytidine which is incorporated into DNA and the subsequent methylation of which can then be determined by HPLC or HPTLC [29]. Such assays, nevertheless, remain difficult because drug-induced DNA hypermethylation is inversely proportional to the rate of DNA synthesis, and the highest levels of hypermethylation are detected at the lowest levels of [14C]deoxycytidine incorporation. Alternatively, the drug-stalled replication forks in which hypermethylation occurs may be isolated from bulk DNA and examined directly for 5-methylcytosine content [33].

We showed recently that 2′,3′-dideoxy-3′-azidothymidine (AZT), although a weak inducer of DNA hypermethylation, was nevertheless capable of leading to a dose-dependent (both with respect to micromolar concentration of AZT and level of DNA hypermethylation) induction of thymidine kinase epimutants (Fig. 2). Such epimutants were observed to have transcriptionally inactivated thymidine kinase genes (TK) that were rapidly reactivatable upon exposure to the demethylating agent 5-azadC [30]. The fact that the AZT-induced TK clones were reactivatable to the TK+ phenotype at extremely high frequency (typically >50%), rules out classical mutation as the mechanism of phenotypic reversion, and provides strong evidence that the initial gene silencing event was epigenetic rather than genetic in origin.

AZT is inactive until it is metabolized to its phosphorylated forms by thymidine kinase and thymidylate kinase. Induction of TK, AZT-resistant epimutants correlated with large decreases in TK-specific mRNA levels in AZT-treated cells. Subsequent exposure to 5-azadC, which resensitizes such epimutants to AZT, also is associated with large scale increases in TK-specific mRNA transcripts [30].

Drug-induced DNA hypermethylation is thus capable of transcriptionally inactivating genes whose products are required for activation of drugs to their toxic form. In the above example, hypermethylation-mediated TK gene inactivation creates a situation of resistance to AZT, as well as to FudR [30], which also requires activation to its phosphorylated form in order to be toxic. Other examples, potential and realized, include the hypermethylation-mediated transcriptional inactivation of the deoxycytidine kinase gene and resistance to araC; of the hypoxanthine phosphoribosyl transferase (HPRT) gene and 6-thioguanine resistance [34]; of topoisomerase II gene inactivation and resistance to etoposide [35]; and of an unidentified gene and resistance to interferon α [36].

In two separate series of experiments we have shown that drug-induced DNA hypermethylation occurs in patients undergoing cancer chemotherapy. In the first study, a pediatric patient undergoing unsuccessful high dose araC chemotherapy for acute myelomonocytic leukemia (Fig. 3) and an adult patient undergoing hydroxyurea chemotherapy for acute myelocytic leukemia (Fig. 4) were observed to have 4- to 5-fold increases in their 5-methylcytosine content which, in the former case, had not returned to normal one month post treatment [29]. In a subsequent study, a series of patients with advanced acute lymphocytic leukemia treated with carboplatin, an isomer of cisplatin, were observed to have significant tumor-associated DNA hypermethylation up to 1 month following carboplatin exposure. This study is described in more detail below.

Cisplatin-induced DNA hypermethylation in HTB-54 human lung adenocarcinoma cells was found to be preventable, in a dose-dependent fashion, by simultaneous exposure to 5-azadC at doses ranging from 0.001 to 10 μM (Fig. 5). AraC-induced DNA hypermethylation was similarly found to be inhibited by simultaneous exposure to 5-azadC within the same dose range [29]. In the clinical study of acute lymphocytic leukemia mentioned above, a series of patients were treated with 5-azacytidine (50 mg/m2/day on days 1 through 5) in addition to carboplatin (250 mg/m2/day on days 3 through 7). Such exposure to 5-azacytidine was observed to transiently prevent carboplatin-induced DNA hypermethylation (Table 1). The transient nature of this effect was probably due to the short half-life of 5-azacytidine as compared to carboplatin, which permitted carboplatin to induce hypermethylation once effective DNA methylation-inhibiting levels of 5-azacytidine had diminished.

Cisplatin is one of the most commonly used anticancer drugs. However, the development of resistance to cisplatin represents a major obstacle in its effective use. The extreme level of DNA hypermethylation induced by cisplatin exposure is noteworthy [29]. It is in contrast to alkylating agents, which also directly modify DNA, but which inhibit DNA methylation [29], possibly by inactivating the sulfhydryl-rich DNA methylase enzyme [37]. Cisplatin exposure increases the 5-methylcytosine content of DNA to a greater degree than any other compound tested. This suggests that some process other than DNA synthesis inhibition might be contributing.

Recent studies have shown that the major adduct induced by cisplatin in DNA is an intrastrand crosslink between two adjacent guanines on the same DNA strand [38]. The presence of such d(G*pG*) cisplatin adducts in DNA significantly alters the DNase I cleavage pattern 39, 40. It is thought that the enhanced Dnase I cleavage of DNA containing d(G*pG*) cisplatin adducts is due to distortions induced in DNA by the adducts. The d(G*pG*) adduct bends the double helix 32–34° towards the major groove, thus widening the minor groove, and unwinds the DNA by approximately 13° [41]. The crystallographic structure of the Dnase I-DNA complex indicates that a loop of the enzyme penetrates into the minor groove of the DNA and interacts asymmetrically with the backbone of both strands, contacting two phosphate groups on each side of the cleaved bond and two phosphate groups on the adjacent strand across the minor groove [42]. We suggest that the cisplatin adduct-induced distortions in DNA, in a manner similar to that for DNase I, may render DNA a better substrate for DNA methyltransferase. This could occur by virtue of the modified DNA offering better access to the five carbon atom of cytosines within the major groove, or greater affinity of the enzyme for cisplatin-kinked DNA. An alternative perspective is that methylation marks `suspect' DNA for scrutiny by other enzymes, such as those involved in DNA repair processes [43]. Precedent for this perspective might be obtained from experiments demonstrating that transfected DNA is rapidly methylated and inactivated [44], and that viral DNA is a target for DNA methylase and gene inactivation [45], [46]. However, the recent finding that 5-methylcytosine inhibits repair of 3′ nearest neighbor O6-methylguanines by O6-methylguanine DNA methyltransferase [47]suggests that drug-induced DNA hypermethylation is unlikely to play a general role in marking DNA containing adducts in need of repair.

We have shown that topoisomerase II inhibitors, including nalidixic acid, novobiocin, etoposide and teniposide, induce DNA hypomethylation 33, 48. This is interesting in view of data showing that exposure to topoisomerase II inhibitors alters gene expression and induces differentiation in a number of experimental systems 49, 50and the strong association of DNA hypomethylation with gene activation [44]. One explanation of our findings that could relate them to the effects of topoisomerase inhibitors upon gene expression is that topoisomerase II releases DNA from that topological conformation that constitutes an efficient substrate for DNA methylase. Thus, even in the presence of increased amounts of DNA methylase in some malignant cells [25], large stretches of DNA might be rendered methylation-resistant by topoisomerase activity. (It should be pointed out that topoisomerase II inhibitors generally permit topoisomerase activity (strand breaking, relief of torsional stress) but prevent the strand rejoining activity of the enzyme.)

The ability of topoisomerase inhibitors to induce DNA hypomethylation could offer a potential explanation for their demonstrated effects upon gene expression and differentiation. Since demethylating agents have been demonstrated to be capable of activating transcriptionally repressed genes, it is reasonable to postulate that the ability of topoisomerase inhibitors to activate gene expression might be related to their ability to inhibit DNA methylation [48]. We have suggested that the enhanced topoisomerase activity observed in tumor cells 51, 52, by virtue of its potential to inhibit DNA methylation, might account for the ectopic gene expression long considered to be a hallmark of the malignant state. We and others have shown that exposure of mammalian cells to topoisomerase II inhibitors leads to the generation of polyploid DNA content, another hallmark of the malignant cell. Increased levels of topoisomerase enzymes in malignant cells could thus provide a theoretical link between the induction of polyploidy, DNA hypomethylation and ectopic gene expression, three markers for genetic instability in malignant cells.

The dinucleotide CpG is strongly underrepresented in human and animal DNA, being present at only about 20–25% of its expected frequency. There is good evidence that this `CpG suppression' is due to the fact that CpG is the target for DNA methylation, and that the product, 5-methylcytosine, undergoes spontaneous deamination to thymidine. Over time, then, CpG tends to be replaced by TpG and its replication product CpA. Evidence has arisen that such spontaneous deamination of 5-methylcytosine in DNA may represent one of the most important types of mutation in human disease 53, 54, 55, 56, 7, 57.

To the extent that cells containing hypermethylated replication forks recover from toxic drug exposure, they would be expected to have an increased mutation rate due to the subsequent spontaneous deamination of the newly methylated cytosines. Thus, epigenetic changes occurring during the carcinogenic process or as a result of exposure to drugs during cancer chemotherapy, should increase the level of genetic instability within the tumor cell population. Changes in DNA methylation may, in fact, play a major role in promoting genetic instability within tumor cell populations. Methylation-mediated changes in genomic stability could result from gene activation in regions of hypomethylation (e.g., of genes associated with metastasis, angiogenesis, recombination, etc.), gene inactivation within regions of hypermethylation (e.g., of tumor suppresser genes or genes whose products are required to activate anticancer drugs, etc.), reduced ability to identify the parental template strand during DNA mismatch repair, or through the generation of mutation `hotspots' in hypermethylated regions of the tumor cell genome. As mentioned above, it is also possible that changes in DNA methylation occurring in tumor cells might set in motion a cascade of ectopic gene expression events that might release tumor cells from normal homeostatic controls, as well as transducing epigenetic events into genetic ones by the induction of polyploidy or via mutagenic deamination of 5-methylcytosine.

Drug-induced DNA hypermethylation represents one potential mechanism of drug resistance for those agents requiring enzymatic activation to achieve cytotoxicity. Since not all drugs induce DNA hypermethylation, but in fact induce DNA hypomethylation (e.g., alkylating agents, 6-thioguanine, 5-azadeoxycytidine, etc.), it may be possible to reduce that portion of drug resistance that results from drug-induced DNA hypermethylation by appropriately combining DNA methylation inhibitors with hypermethylating drugs. In our limited experience in vivo, however, it has not been possible to achieve long lasting blockade of DNA hypermethylation in patients undergoing sequential treatment with 5-azacytidine and carboplatin. This was probably due to the differences in half lives of the two drugs. Alternative regimens employing longer lasting methylation inhibitors, or methylation inhibitors applied by continuous infusion, might achieve durable inhibition of drug-induced DNA hypermethylation. However, precise titration of the hypermethylation/hypomethylation phenomenon to achieve steady state levels of DNA methylation during cytotoxic chemotherapy may be difficult or impossible to achieve. An additional problem is that hypomethylation itself might contribute to the problem of acquired drug resistance. Perry et al. have shown, for example, that 5-aza-2′-deoxycytidine exposure significantly enhances both the frequency and the rate of gene amplification in vitro [58]. Gene amplification is thought to underly the development of resistance to a wide array of anticancer agents 59, 60. Attempts to moderate drug-induced DNA hypermethylation-mediated drug resistance by addition of hypomethylating agents to chemotherapy protocols could result in the amplification of genes associated with drug resistance, e.g., the MDR1 gene associated with multi drug resistance.

The principal paradigm for anticancer drug design for the past 50 years has been cytotoxicity – the search for ever more potent toxins – and an increasing clinical focus on applying anticancer drugs at their maximum tolerable dosage. Our results suggest that such a treatment strategy may be problematic in the absence of 100% tumor cell kill since the surviving cancer cell population, because of its enhanced epigenetic/genetic diversity, may be substantially superior to normal cell populations with respect to their ability to withstand further selective pressure, i.e., drug exposure.

At present, significant effort is being spent by many laboratories on designing new paradigms for anticancer therapy, including therapies targeting tumor cell differentiation and apoptosis. These approaches are based upon the premises that differentiated cells do not divide, and that the normal program of cell death, apoptosis, is aberrantly inactivated in malignant tumor cells. Some of these therapeutic strategies are based on setting in motion epigenetic cascades in order to reprogram aberrant patterns of gene expression in the tumor cell. In exploring these alternative therapeutic strategies, one should consider the dual possibilities that epigenetic instability may contribute to tumor progression in a manner similar to that of genetic instability; and that these two processes intersect in their cause and effect relationships.

Section snippets

Acknowledgements

The technical efforts of Ms. Sherry Leonard, Ms. So Wong, and Ms. Dawn Mylott are gratefully acknowledged. Patient specimens were kindly provided by Drs. Spencer Raab (deceased), C. Tate Holbrook, and Alan D. Kritz. This work was performed with support from grant CA RO1 47217 from the National Cancer Institute.

First page preview

First page preview
Click to open first page preview

References (60)

  • Usmani, B.A. (1993) Genomic instability and metastatic progression, Pathobiology 61(2),...
  • Sweezy, M.A. and Fishel, R. (1994) Multiple pathways leading to genomic instability and tumorigenesis, Ann. NY Acad....
  • Cheng, K.C. and Loeb, L. (1993) Genomic instability and tumor progression: mechanistic considerations, Adv. Cancer...
  • Hartwell, L. (1992) Defects in a cell cycle checkpoint may be responsible for the genomic instability of cancer cells,...
  • Honchel, R., Halling, K.C. and Thibodeau, S.N. (1995) Genomic instability in neoplasia, Sem. Cell Biol., 6(1),...
  • Russo, C.A., Weber, T.K., Volpe, C.M., Stoler, D.L., Petrelli, N.J., Rodriguez-Bigas, M., Burhans, W.C. and Anderson,...
  • Nyce, J., Weinhouse, S. and Magee, P.N. (1983) 5-Methylcytosine depletion during tumor development: an extension of the...
  • Yamasaki, H., Mesnil, M. and Nakazawa, H. (1992) Interaction and distinction of genotoxic and non-genotoxic events in...
  • Nyce, J., Leonard, S., Canupp, D., Schulz, S. and Wong, S. (1993) Epigenetic mechanisms of drug resistance:...
  • Ribieras, S., Song-Wang, X.G., Martin, V., Lointier, P., Frappart, L. and Dante, R. (1994) Human breast and colon...
  • Rideout, W.M. 3rd, Eversole-Cire, P., Spruck, C.H. 3rd, Hustad, C.M., Coetzee, G.A., Gonzales, F.A. and Jones, P.A....
  • Herman, J.G., Latif, F., Weng, Y., Lerman, M.I., Zbar, B., Liu, S., Samid, D., Duan, D.S., Gnarra, J.R. and Linehan,...
  • Lee, Y.W., Klein, C.B., Kargacin, B., Salnikow, K., Kitahara, J., Dowjat, K., Zhitkovich, A., Christie, N.T. and Costa,...
  • Greger, V., Debus, N., Lohmann, D., Hopping, W., Passarge, E. and Horsthemke, B. (1994) Frequency and parental origin...
  • Costa, M. (1995) Model for the epigenetic mechanism of action of nongenotoxic carcinogens, Am. J. Clin. Nutrition 61(3...
  • Ferguson, A.T., Lapidus, R.G., Baylin, S.B. and Davidson, N.E. (1995) Demethylation of the estrogen receptor gene in...
  • Yoshiura, K., Kanai, Y., Ochiai, A., Shimoyama, Y., Sugimura, T. and Hirohashi, S. (1995) Silencing of the E-cadherin...
  • Pieretti, M., Powell, D.E., Gallion, H.H., Conway, P.S., Case, E.A. and Turker, M.S. (1995) Hypermethylation at a...
  • Moulton, T., Crenshaw, T., Hao, Y., Moosikasuwan, J., Lin, N., Dembitzer, F., Hensle, T., Weiss, L., McMorrow, L. and...
  • Laird, P.W. and Jaenisch, R. (1994) DNA methylation and cancer, Hum. Mol. Genet. 3 Spec. No....
  • Matsuo, K., Tang, S.H., Zeki, K., Gutman, R.A. and Fagin, J.A. (1993) Aberrant deoxyribonucleic acid methylation in...
  • Mills, K.I., Sproul, A.M. and Burnett, A.K. (1993) Methylation of the major breakpoint cluster region (M-bcr) in...
  • Issa, J.P., Vertino, P.M., Wu, J., Sazawal, S., Celano, P., Nelkin, B.D., Hamilton, S.R. and Baylin, S.B. (1993)...
  • Ohtani-Fujita, N., Fujita, T., Aoike, A., Osifchin, N.E., Robbins, P.D. and Sakai, T. (1993) CpG methylation...
  • Baylin, S. (1992), Abnormal regional hypermethylation in cancer cells, AIDS Res. Hum. Retroviruses, 8(5),...
  • Makos, M., Nelkin, B.D., Chazin, V.R., Cavenee, W.K., Brodeur, G.M. and Baylin, S.B. (1993) DNA hypermethylation is...
  • Sakai, T., Toguchida, J., Ohtani, N., Yandell, D.W., Rapaport, J.M. and Dryja, T.P. (1991) Allele-specific...
  • Peltomaki, P. (1991) DNA methylation changes in human testicular cancer, Biochim. Biophys. Acta, 1096(3),...
  • Nyce, J., Mylott, D., Leonard, S., Willis, L. and Kataria, A. (1989) Detection of drug-induced DNA hypermethylation in...
  • Nyce, J.W. (1989) Drug-induced DNA hypermethylation and drug resistance in human tumors, Cancer Res., 49(21),...
  • Cited by (49)

    • Epigenetic reprogramming and potential application of epigenetic-modifying drugs in acquired chemotherapeutic resistance

      2020, Advances in Clinical Chemistry
      Citation Excerpt :

      DNA synthesis inhibitors such as doxorubicin stall the replication fork where DNMTs can come in and have persistent contact allowing for increased methylation. Chemotherapy-induced DNA hyper-methylation modulate tumor response to cytotoxicity and also observed to induce resistance in in vitro conditions and clinical patients undergoing chemotherapy [42]. Transcriptional repression (indicated by alteration in epigenetic metabolic donor molecules, elevated DNMTs and absolute percentage of 5-methyl cytosine contents) and transcriptional activation (indicated by loss of global DNA methylation) were reported in chemoresistant phenotypes [38,43–45].

    • Reversal of epigenetic aberrations associated with the acquisition of doxorubicin resistance restores drug sensitivity in breast cancer cells

      2018, European Journal of Pharmaceutical Sciences
      Citation Excerpt :

      Both genetic and epigenetic alterations could be involved not only in breast cancer development but also in evolution of therapeutic resistance (Longacre et al., 2016). Chemotherapy-induced changes in DNA methylation and chromatin environment have previously been reported in alteration of genes in doxorubicin resistance development (Nyce, 1997; Chekhun et al., 2006; Boettcher et al., 2010). Increased DNA methylation may reportedly increase survival pathways, evasion of apoptosis and resulted in chemo-resistance (Montenegro et al., 2016).

    • Epigenetic regulation of PAX5 expression in acute T-cell lymphoblastic leukemia

      2011, Leukemia Research
      Citation Excerpt :

      In many cancers promoter hypermethylation was detected in genes involved in differentiation, drug resistance and cell cycle control. A correlation between these epigenetic changes and clinical course of disease has been suggested [4–6]. In several reports, the presence of CpG methylation in T-ALL has been described previously (Table 1) but its role as prognostic value or therapeutic target still has to be determined.

    • 5-Aza-2'-deoxycytidine sensitizes busulfan-resistant myeloid leukemia cells by regulating expression of genes involved in cell cycle checkpoint and apoptosis

      2010, Leukemia Research
      Citation Excerpt :

      Acquired resistance to anticancer agents is mediated by both genetic and epigenetic mechanisms [1].

    View all citing articles on Scopus
    1

    Tel.: +1 (919) 752-7036; Fax: +1 (919) 752-9148.

    View full text