DNA Methylation in Radiation-Induced Carcinogenesis: Experimental Evidence and Clinical Perspectives

Abstract

Ionizing radiation is a valuable tool in many spheres of human life. At the same time, it is a genotoxic agent with a well-established carcinogenic potential. Progress achieved in the last two decades has demonstrated convincingly that ionizing radiation can also target the cellular epigenome. Epigenetics is defined as heritable changes in the expression of genes that are not due to alterations of DNA sequence but consist of specific covalent modifications of chromatin components, such as methylation of DNA, histone modifications, and control performed by non-coding RNAs. Accumulating evidence suggests that DNA methylation, a key epigenetic mechanism involved in the control of expression of genetic information, may serve as one of the driving mechanisms of radiation-induced carcinogenesis. Here, we review the literature on the effects of ionizing radiation on DNA methylation in various biological systems, discuss the role of DNA methylation in radiation carcinogenesis, and provide our opinion on the potential utilization of this knowledge in radiation oncology.

I. INTRODUCTION

Ionizing radiation (IR) remains a pillar of the care and cure of many cancer patients worldwide. At the same time, the carcinogenic potential of IR has been recognized for over a century since the first radiation-induced skin cancers and leukemia in occupationally exposed workers were reported.^1–4 These studies were followed by a plethora of epidemiological studies, including long-term follow-up studies on A-bomb survivors,^5–7 as well as reports on occupational, accidental, and environmental exposures^8–10 and on patients exposed for various diagnostic and treatment purposes.^11–14 Those studies, together with the number of studies that utilized cellular and animal models, highlighted the universal carcinogenic potential of IR, which was shown to cause cancers “in most tissues of most species and at all ages.”¹⁵

The major mechanisms of radiation carcinogenesis are linked to DNA damage associated with it misrepaired DNA lesions and the induction of mutations, our knowledge of which is summarized in excellent reviews elsewhere.^1,15 In addition to these mechanisms, the development of radiation-induced genomic instability, a multifactorial phenomenon exhibited as an increased frequency of mitotically heritable genetic alterations observed in the progeny of irradiated cells multiple generations after exposure,^16–18 suggests that there are other mechanisms that may be implicated in radiation carcinogenesis. The observed frequency of genetic instability induced by IR exposure is substantially higher than that observed for gene mutations at a similar dose; therefore, the latter is considered highly unlikely to be the initiating mechanism.^15,19 In this regard, epigenetic alterations are of particular interest.^18,20

II. EPIGENETIC MECHANISMS THAT REGULATE THE EXPRESSION OF GENETIC INFORMATION

Epigenetics is the study of heritable changes in gene expression that are not associated with alterations in the underlying DNA sequence—in other words, changes in phenotype without actual changes in genotype. The epigenetic mechanisms that regulate the expression of genetic information include DNA methylation, post-translational histone modifications, and nucleosome positioning along DNA. These covalent marks that ensure the proper structure and function of the epigenome are applied by specific enzymes: DNA and histone methyltransferases, the so-called “writers.” The “readers,” proteins that recognize these marks, in turn modulate the gene expression at particular genomic loci. Conversely, the “erasers” are the enzymes that guarantee the reversibility of the previously applied covalent marks, underlining the plasticity of the epigenome.²¹

Epigenetic mechanisms are critical during development, as well as for the maintenance of cellular homeostasis. Regarding epigenetic mechanisms, the expression of genetic information is regulated in a cell-, tissue-, and sex-specific manner.^22,23 Epigenetics also plays a key role in controlling the expression of repetitive elements (REs), which occupy more than 50% of mammalian genomes.²⁴

DNA methylation, the most studied and characterized epigenetic modification, is a covalent addition of a methyl group to the fifth position of carbon. This process is enabled by a complex interplay among the DNA methyltransferases, methyl-CpG-binding proteins, and the recently characterized protein ubiquitin-like with PHD and RING finger domains 1 (UHRF1). More than 50% of eukaryotic genes, as well as a mosaic palette of REs, contain CpGrich regions (also known as CpG islands, or CGIs). Depending on the location of CGIs, DNA methylation can have differential regulatory functions. For instance, methylation of DNA at CGIs located within the gene promoter/transcription start site is usually associated with gene silencing.²⁵ However, it remains unknown whether this mechanism is repressive by nature or is simply a lack of activation.²⁶ Conversely, DNA methylation of gene bodies (introns and exons) is not associated with gene silencing and either leads to stimulated elongation and splicing or prevention of initiation of aberrant transcription from alternative transcription start sites.^22,27,28

Alterations in DNA methylation may lead to cellular epigenetic-based reprogramming, resulting in altered gene and RE expressions, genomic instability, and the development of pathological states, including cancer. In the 1970s and early 1980s, epigenetic mechanisms were proposed to be involved in carcinogenesis, with the loss of global DNA methylation being reported as the first epigenetic alteration detected in several human cancers.^29,30 These studies were followed by the identification of DNA hypermethylation at the promoter regions of tumor-suppressor genes in cancerous tissue.^31–33 Further studies have demonstrated that these alterations in global and gene-specific DNA methylation can often be detected at early stages of carcinogenesis, leading to the development of the hypothesis that epigenetic mechanisms may serve as drivers of carcinogenesis.^34–38

III. EPIGENETIC MECHANISMS OF CARCINOGENESIS

Today, several decades since the first report of global DNA hypomethylation in cancerous tissue,^29,30,39 virtually all human cancers have been characterized by this epigenetic alteration. Currently, it is generally accepted that global genomic hypomethylation is a recognized hallmark of cancer.^40,41 Loss of global genomic methylation is usually associated with the hypomethylation of REs, many of which, such as the LINE-1 and Alu elements, are retrotransposons by nature.⁴² Retrotransposons are mobile DNA elements that, during evolution had heavily populated mammalian genomes due to their ability to propagate via the “copy-paste” mechanism.⁴³ DNA hypomethylation in retrotransposons may lead to the loss of epigenetic control over those REs, resulting in their reactivation and subsequent retrotransposition. The latter is usually exhibited as insertional mutagenesis, in which the retrotransposon randomly inserts its copy at a different genomic location, often leading to a shifted open reading frame of the protein-encoding genes.²⁴

It remains unclear whether the retrotransposition is a carcinogenesis-driven event or merely the consequence of the overall genomic instability observed in cancer cells. However, LINE-1 retrotransposition has been reported recently in numerous human cancers.^44–46 Increased retrontransposition may also result in another unwanted effect, genomic amplification, which may further have a negative effect on genome stability.

Even without the subsequent reactivation of retrotransposons, DNA hypomethylation may have substantial influence on carcinogenesis. For instance, it has been shown that the MET oncogene has evolutionary acquired a LINE-1 insertion within its gene body together with the CGI containing an alternative transcription start site. Loss of DNA methylation from this CGI results in aberrant transcription and increased MET copy numbers and is usually associated with a poor prognosis and rapid development of distant metastasis.^47,48

DNA hypermethylation, which is observed in a number of tumor-suppressor genes, is another frequently observed epigenetic alteration in cancer cells. Often, promoter DNA hypermethylation in such genes as cyclin-dependent kinase 2A (CDKN2A, also known as P16^INK4A), O⁶-methylguanine-DNA methyltransferase (MGMT), and phosphatase and tensin homolog (PTEN), to name a few, is associated with their transcriptional silencing in the tumor. The latter effect, however, remains controversial because a number of studies reported on gene silencing preceding DNA hypermethylation, as well as a lack of influence of promoter DNA methylation on gene and RE expression.^22,28,49

It is becoming increasingly recognized that both genetic and epigenetic alterations in concert contribute to the process of carcinogenesis. For instance, a number of recent studies have convincingly demonstrated that the vast majority of human cancers harbor mutations in the genes that belong to the epigenetic machinery (i.e., DNA methyltransferases).^50,51 Alterations in the expression of those genes may compromise the cellular epigenome, subsequently leading to altered gene expression and genomic instability.^21,25,52 Conversely, epigenetic alterations may further predispose to mutations because DNA hypermethylation-mediated silencing of the critical DNA repair gene MLH1 has led to the development of new mutations due to inefficient DNA repair.⁵³

IV. DNA methylation in radiationinduced cancers

In 2004, Belinsky et al. reported DNA hypermethylation of the p16^INK4A gene in lung adenocarcinomas of plutonium-exposed workers at the Russian nuclear enterprise MAYAK.⁵⁴ Interestingly, the levels of p16^INK4A DNA hypermethylation were 3.5-fold higher in the adenocarcinomas of exposed workers compared with non-IR worker controls (confidence interval = 1.5–8.5; p = 0.001). The investigators also reported that the increased probability for gene-specific methylation approximated a 4-fold increase in relative risk for adenocarcinoma in workers exposed to plutonium.⁵⁴ In another study, Su et al. detected DNA hypermethylation of the P16^INK4A (z = 2.844, p = 0.005) and MGMT (z = 3.034, p = 0.002) genes in the sputum of uranium miners.⁵⁵ The degree of DNA hypermethylation in the promoter of p16 and MGMT significantly correlated with the cumulative doses of radon exposure, with the cumulative exposure dose range of 12 ± 6 to 294 ± 132 (z = 3.859, p = 0.0001). Interestingly, a recent study using a mouse model also identified DNA hypermethylation of p16^INK4A and its transcriptional silencing in radiation-induced thymic lymphoma.⁵⁶

Other studies performed with the cohort of MAYAK workers diagnosed with lung adenocarcinoma demonstrated hypermethylation of GATA5, a gene that plays a critical role in cellular differentiation.⁵⁷ The investigators have also acknowledged the higher incidence of DNA hypermethylation in adenocarcinomas from MAYAK workers, in whom at least one of five investigated genes was hypermethylated in 93% of cases, whereas in non-IR workers, this effect was observed in only 66% of cases. However, it must be emphasized that, to date, no radiation-specific DNA hypermethylation signatures were reported.

Results from epidemiological studies suggested that epigenetic alterations, aberrant DNA methylation in particular, may be involved in radiation carcinogenesis, which inspired the investigation of effects of IR on DNA methylation in experimental systems.

A number of studies showed that exposure to IR may substantially affect the cellular epigenome and result in the loss of global DNA methylation, especially in organs and systems known to be sensitive to radiation-induced carcinogenesis. For instance, Giotopoulos et al. reported loss of DNA methylation in bone marrow of mice exposed to 3 Gy of IR.⁵⁸ Similar losses of DNA methylation were observed in other cancer-prone tissues such as the thymus, mammary gland, and spleen,^59–61 but not in the lung or muscle tissues.⁶² Importantly, those effects could be detected at extended time points after irradiation, even when the radiation-induced DNA damage was long since repaired.⁵⁹ These effects were also detectable in the organ and systems not directly exposed to IR.⁶³ A decrease in global DNA methylation was reported in a rat model shortly after irradiation,⁶¹ as well as in rats with radiation-induced mammary tumors,⁶⁴ further establishing a link between radiation-induced global genomic hypomethylation and carcinogenesis. Another important finding from those studies was that the IR-induced changes in DNA methylation primarily stem from REs rather than from individual genes.^42,63,65,66

Given the abundance of REs in mammalian genomes and their extensive DNA methylation, the effects of IR on DNA methylation on a global scale can be detected with a high degree of reproducibility. Conversely, studies investigating IR-induced gene-specific DNA methylation often present controversial results. The introduction of next-generation approaches into DNA methylation analysis promises to shed more light on the effects of IR at the level of single-gene resolution.

In one of the pioneering studies, Antwih et al., using the 450 K methylation array approach, observed substantial changes in gene-specific DNA methylation in human breast cancer cell lines.⁶⁷ Interestingly, gene ontology analysis revealed that a large fraction of affected genes belonged to radiation response pathways. Similarly, radiation-induced changes in DNA methylation were reported by Bae et al., who investigated the response in HCT116 human colorectal cells.⁶⁸ Conversely, two studies performed on normal human fibroblasts reported a lack of radiation-induced changes in DNA methylation.^69,70 The results of these studies clearly demonstrated that the IR-induced effects on DNA methylation critically depend on the cell/ tissue type. For example, even in one of the most sensitive organs to IR exposure, the bone marrow, clear patterns of cell specificity regarding the magnitude of response were observed.^71,72 The observed changes were primarily detected in genetically unstable cancer cells and cell lines with compromised DNA repair, which may explain the considerably higher degree of changes in DNA methylation after irradiation. Other contributing factors may also be associated with different doses, quality of radiation, time points after irradiation at which DNA methylation was evaluated, and differences in the approaches for data analysis.^66,73,74 Furthermore, in vitro systems do not represent the whole organismal response and lack the influence of other important factors that may predetermine the tissue/cell epigenetic response. For instance, it has been shown that male C57BL/6J mice exhibiting a robust IR-induced epigenetic response have a clear lack of this response when a gonadectomy is performed.⁷⁵

V. ARE THE OBSERVED EPIGENETIC EFFECTS DRIVERS OR PASSENGERS?: HIGH-LINEAR ENERGY TRANSFER (LET) RADIATION AS A MODEL TO STUDY THE EPIGENETIC MECHANISMS OF RADIATION CARCINOGENESIS

It is becoming increasingly recognized that alterations in DNA methylation are not just the passive bystanders in the process of carcinogenesis or consequences of neoplastic transformation and, very possibly, they are the active players that shape the tumor landscape. Indeed, as discussed above, altered DNA methylation can be detected very early during the process of carcinogenesis and may influence numerous biological processes. It is also becoming increasingly recognized that both genetic and epigenetic alterations in concert contribute to carcinogenesis. However, one of the most challenging aspects of investigating the role of epigenetic alterations in genotoxic carcinogenesis is determining the truly epigenetically driven mechanisms. For instance, the loss of DNA methylation that is observed in the vast majority of cases after exposure to doses of 1 Gy and above may be also mediated by the substantial damage to DNA or preoccupation of DNA methylatransferases (DNMT1) and DNA methylation accessory protein (UHRF1) in recruiting the repair complexes to the sites of damaged DNA instead of a direct response in the maintenance of DNA methylation. One could thus consider that the genotoxic effects of IR simply predetermine the epigenetic alterations.

In this regard, of particular interest are the model systems that utilize exposure to low mean absorbed doses of high-LET radiation, such as protons and heavy ions, types of IR that are dominant in the space environment. The necessity of understanding the effects of IR exposure during the space missions and the introduction of high-LET radiation into clinical practice have triggered the investigation of biological and molecular mechanisms of response to high-LET radiation.^76–78

Studies in in vitro and animal experimental systems clearly indicate that exposure to heavy ions results in clustered DNA damage compared with low-LET terrestrial radiation.^76,79–81 This more complex DNA damage is frequently irreparable and usually promotes apoptosis via p53 at the S/G₂ checkpoint, leading to greater relative biological effectiveness.⁸²

Cytogenetic studies report a much higher complexity of chromosomal rearrangements caused by exposure to heavy ions compared with sparsely ionizing IR.^83,84 The complexity of these rearrangements, however, determines the lethality of the vast majority of them. Both in vitro and in vivo studies have reported a very low number of complex rearrangements within a short time after irradiation.^79,85 Furthermore, it has been shown that the levels of chromosomal aberrations in astronauts with a total time of 2 years spent in space were not substantially higher than the background measurements taken before the first flight.⁸⁶

Despite the lack of detectable DNA damage and chromosomal aberrations, studies in experimental animal models have shown that heavy ions are not only potent carcinogens, but can induce cancers at much lower doses and even in organs that are not known to be the classical organs for IR-induced carcinogenesis.^87–94 For instance, leukemogenesis studies in mice demonstrate that exposures to as low as a 0.4 Gy mean absorbed dose of heavy iron ions (⁵⁶Fe) were enough to increase the levels of leukemia, whereas doses above 1 Gy of low-LET irradiation were needed to cause the same effect. Other studies also reported lung tumors in mice exposed to ⁵⁶Fe or protons,⁸⁹ as well as enhanced intestinal tumor multiplicity in APC^min mice.^90,91 In addition, high-LET radiation was shown to be a very potent inducer of liver tumors, a site that is not common for radiation carcinogenesis.

Studies using the exposures to low-mean absorbed doses of high-LET IR may aid in better understanding the driving potential of epigenetic mechanisms in radiation-induced carcinogenesis. For instance, in a study assessing the dose-dependent effects of total body irradiation to low-mean absorbed doses of ⁵⁶Fe, Miousse et al. demonstrated an absence of detectable DNA damage, as well as no increases in reactive oxygen species, senescent cells, or apoptotic events in the hematopoietic stem and progenitor cells 1 and 5 months after exposure.⁷¹ At the same time, changes in DNA methylation of transposable elements LINE-1 and SINE B1 (corresponding to Alu elements in humans) and DNA methylation machinery were detected in the pool of hematopoietic stem and progenitor cells after exposure to leukemogenic (0.4 Gy, 1 A GeV), but not lower doses of ⁵⁶Fe. Importantly, those changes were still evident 5 months after exposure and also resulted in reactivation of LINE-1 elements that may further lead to LINE-1 insertional mutagenesis, genome amplification, and the development of genomic instability. The persistence of epigenetic alterations considered as a hallmark of cancer (loss of global and RE-associated DNA methylation paralleled by reactivation of the LINE-1 retrotransposon) in the absence of detectable DNA damage and other cellular and molecular alterations suggests that epigenetic reprogramming may serve as one of the driving forces in IR-induced carcinogenesis. Further studies are clearly needed to confirm this hypothesis and to prove the causative role of epigenetic alterations and DNA methylation related to radiation-induced carcinogenesis.

VI. TARGETING THE CANCER EPIGENOME FOR RADIOSENSITIZATION: CONSIDERATIONS FOR ONE-CARBON METABOLISM RELATED EFFECTS

Given that IR is capable of inducing stable alterations to DNA methylation and that there is a higher degree of IR-induced epigenetic responses in the cancerous cell, it seems reasonable to expect that tumor radiosensitization could be achieved by mod-ulation of the tumor cell epigenome. This notion is strengthened by the finding that DNA methylation levels can regulate the cancer cell response to radiotherapy. Kim et al., using the radiosensitive (H460) and radioresistant (H1299) human non-small-cell lung cancer cell lines, demonstrated the differential DNA methylation patterns in 747 genes.⁹⁵ The investigators have further shown that silencing of SERPINB5 and S100A6 can mediate radioresistance in H460 cells.

The DNA methylation status of LINE-1, the most abundant and usually heavily methylated RE, is becoming a valuable tool in the prognosis of tumor response to therapy. It is generally recognized that a lower degree of LINE-1 DNA methylation is associated with a poor prognosis, advanced metastasis, and weak tumor response to treatment.⁴² Some studies indicate that the DNA methylation status of LINE-1 may also serve as a predictor of tumor response to radiotherapy⁹⁶; however, more basic and clinical research is needed to confirm these findings.

The potential of DNMT inhibitors such as nucleoside analogs (5-azacytidine, decitabine, and zebularine) in tumor radiosensitization is becoming increasingly recognized. Those drugs were first introduced into clinical practice several decades ago, but were only shown to improve the blood cell count and survival in patients with myelodysplastic syndrome and acute myeloid leukemia after dose optimization was achieved in recent years.^97,98 At the same time, the results of the clinical trials in patients with solid tumors were less promising.⁹⁹

A series of in vitro studies demonstrated increased sensitivity to radiotherapy in gastric, glioblastoma, head and neck, colorectal, and nasopharyngeal cancer cell lines when treated with these nucleoside analogs.^100–104 Emerging evidence also exists on the success of combined nucleoside analog/ radiotherapy treatment in glioblastoma U251 and nasopharyngeal carcinoma xenograft models.^100,104

Another promising and still unexplored avenue in tumor radiosensitization is via targeting one-carbon metabolism, one of the major biochemical pathways in living organisms that affects nearly all cellular functions and more than 100 specific biomethylation reactions. One-carbon metabolism involves the reactions that surround the transfer of the methyl group from S-adenosylmethionine (SAM) to acceptor molecules and the regeneration of SAM. The latter ties together gene regulation, amino acid synthesis, purine and pyrimidine synthesis, antioxidants, and four vitamins. One of the central molecules involved in one-carbon metabolism is the essential amino acid methionine. Methionine is critical for a number of vital processes, including the synthesis of SAM, a universal donor of methyl groups for DNA, RNA, protein, and lipid methylation. Methionine is also needed for the synthesis of glutathione and is indispensable for protein synthesis.

Interestingly, there is a remarkable difference in the needs for methionine between normal tumor cells; rapidly proliferating cancer cells require much higher levels of methionine to maintain function. Therefore, as would be expected, tumor cells are extremely sensitive to methionine restriction. Although the normal cell has a capacity for re-methylation and further utilization of methionine from homocysteine, the cancer cell is incapable of proper synthesis and utilization of endogenous methionine.¹⁰⁵ For instance, it has been shown that plating normal fibroblasts and tumor cells together in methionine-deficient homocysteine-supplemented medium results in cell cycle arrest and apoptosis of tumor cells, whereas normal fibroblasts grow abundantly.¹⁰⁶

Although the potentiation of the chemotherapy effect of methionine deprivation has been investigated both in vitro¹⁰⁷ and in vivo^108,109 and even in clinical trials,^110,111 the potential combination of methionine dietary deprivation with radiotherapy has yet to be addressed. Accumulating evidence indicates that cancer cells are radiosensitized by the deprivation of other methyl group donors,^112,113 suggesting that methionine deprivation combined with radiotherapy may have beneficial effects for cancer treatment.

Whereas the potential of this approach seems clear, several critical issues need to be addressed due to the toxicity associated with long-term methionine restriction. Classical cancer therapy regimens, chemotherapy or radiotherapy, are usually a lengthy process over several months. Long-term methionine deficiency is associated with substantial weight loss both in rodent models and in clinical trials, as well as thrombocytopenia, neutropenia, and the development of hepatosteatosis.^{110,111,114,115} Recent advances both in tumor imaging and radiation techniques, as well as the development and widespread implementation of stereotactic body radiation therapy regimens into clinical practice^116–118 may, along with improved local tumor control, significantly decrease the duration of treatment and thus allow for effective and safe methionine dietary interventions.

ABBREVIATIONS:

CDKN2A

cyclin-dependent kinase 2A

CGI

CpG island

DNMT

DNA methyltransferase

IR

ionizing radiation

LET

linear energy transfer

LINE-1

Long Interspersed Nucleotide Element 1

MGMT

O⁶-methylguanine-DNA methyltransferase

PTEN

phosphatase and tensin homolog

RE

repetitive element

SAM

S-adenosylmethionine

SINE B1

Short Interspersed Nucleotide Element B1

UHRF1

ubiquitin-like with PHD and RING finger domains 1