GADD45 Deregulation in Cancer: Frequently Methylated Tumor Suppressors and Potential Therapeutic Targets

Methylation of cytosines is an evolutionary conserved mechanism of gene regulation that plays a critical role during normal development and genetic defects associated with methylation result in multiple developmental diseases. Because cancer cells many times express various features of undifferentiated or dedifferentiated cells, it is not surprising that methylation of genes is frequently observed in cancer, although a loss of global methylation is observed concomitant with methylation of specific CpG islands. A vast amount of information has been gained about aberrant methylation patterns in human cancers. Tumor-specific methylation changes in different genes have been identified and the potential clinical applications of these data had an effect on cancer diagnosis, prognosis, and therapeutics. Consequently, several DNA methylase inhibitors such as 5-azacytidine, decitabine, and MG98 are currently in clinical trials for various types of cancer. For example, a phase III clinical trial of 5-azacitidine in myelodysplastic syndrome patients resulted in significantly improved response rate, quality of life, and enhanced overall survival (5).

What is unexpected and thus far unexplained is the tumor-specific methylation of specific subsets of genes, in particular, tumor suppressor genes such as GADD45γ in the current study. The precise mechanisms are not known, although some increase in DNMT1 (2, 6) expression has been observed in some cancers. Epigenetics can be described as a stable alteration in gene expression potential that takes place during development and cell proliferation, without any change in gene sequence. DNA methylation is one of the most common epigenetic events in the mammalian genome. Although heritable, this process is reversible, making it an interesting therapeutic target. Because cancer is a result of aberrant gene expression, it is no surprise that methylation plays a crucial role.

DNA methylation is a chemical modification, resulting in the addition of a methyl group at the carbon 5 position of the cytosine ring. Although most cytosine methylation occurs in the sequence context 5′CpG3′, it can also engage CpA and CpT dinucleotides (6). Clusters of CpG sites are occasionally present in the genome, designated as CpG islands, ranging from 0.5 to 5 kb and occurring on average every 100 kb, with distinctive properties. These regions are unmethylated, GC rich (60-70%), with no suppression of the frequency of the dinucleotide CpG (7). Approximately half of all the genes in humans have CpG islands (8), and many of them are present in promoter regions critical for gene regulation, although CpG islands are also located at various positions throughout genes (9), such as in exons and introns, or further downstream (10). Cancer-specific gene methylations primarily affect CpG islands in promoter regions of tumor suppressor genes and other, presumably critical genes, leading to silencing of multiple genes in any given cancer cell. These methylations are gene and cancer specific, resulting in cancer-specific methylation patterns that not only correlate with the cancer type but may also correlate with stage, outcome, and response to therapy. Therefore, identification of cancer-specific gene methylation such as for GADD45γ is an important field of cancer research and has resulted in the identification of novel tumor suppressor genes and diagnostic and prognostic methylation patterns.

The procedures and techniques to investigate the epigenetic process are diverse, complex, and often conflicting to some degree. Different methodologies are employed to detect methylation at CpG sites in specific target areas of these genes. Each approach may use one or the other of two principal techniques. Recent advances in the techniques for detection of methylation include powerful tools such as sodium bisulfite conversion, cDNA microarray, restriction genomic scanning, and CpG island microarray (11). The sensitivity and specificity of DNA methylation markers in cancer diagnosis depends on several factors, including the type of cancer and the gene to be studied and the technique involved (12).

Because DNA methylation is closely related to the development of cancer, it would be interesting to know whether its presence or absence affects the prognosis as well. This would help in modifying initial treatment options, monitoring patient response to therapy, and predicting survival. Recently, several studies have shown methylation of certain genes to be closely related to the prognosis of cancer including for colorectal, lung, and prostate cancer (9). Therefore, the findings by Ying et al. (1) about the frequent methylation of the GADD45γ gene in a variety of cancers is an important first step to determine the relevance of this methylation event for cancer prognosis.

The GADD45 gene family encodes three structurally highly related growth arrest– and DNA damage–inducible proteins, GADD45α, GADD45β, and GADD45γ (Fig. 2; ref. 13). They contain only one discernible homology domain to other proteins (i.e., to ribosomal proteins). Although the role of GADD45 in growth arrest and apoptosis is not entirely clear, several, sometimes conflicting aspects of GADD45 function have been revealed. GADD45 proteins are primarily nuclear proteins that interact with various cell cycle–related proteins such as the G₂ cell cycle–specific kinase, Cdc2/Cdk1, proliferating cell nuclear antigen, and the cell cycle kinase inhibitor p21(waf1) (13).

GADD45 proteins play a role in the G₂-M checkpoint in response to DNA damage. GADD45α activates p53-dependent G₂-M arrest and inhibits cdc2 kinase that can be blocked by overexpression of cyclin B1 providing a link between p53-dependent cell cycle checkpoint and DNA repair (13). Interestingly, unlike GADD45α, neither GADD45β nor GADD45γ seem to inhibit Cdc2 kinase or induce G₂-M arrest. GADD45α-null mice have a phenotype similar to p53-deficient mice, including genomic instability and increased radiation carcinogenesis, suggesting that GADD45α and p53 are involved in similar cellular pathways (13). Indeed GADD45α but not GADD45β and GADD45γ is a direct downstream target of p53.

The role of GADD45 genes in cell cycle arrest has been well established, whereas the role GADD45 genes play in apoptosis remains unclear. Overexpression of GADD45 in normal human fibroblasts causes G₂-M arrest but not apoptosis. Nevertheless, genotoxic stress or Brca1-induced apoptosis seem to involve GADD45α-mediated activation of the stress-responsive c-jun NH₂-terminal kinase and/or p38 mitogen-activated protein kinase (13, 14). NF-κB-induced cell survival has been proposed to be mediated by induction of GADD45β expression and down-regulation of c-jun-NH₂-kinase activity (15, 16). A yeast two-hybrid screen revealed direct interaction of all three GADD45 family members with the upstream kinase MTK1/mitogen-activated protein kinase kinase kinase 4 that activates both p38 and c-jun-NH₂-kinase in response to environmental stresses. All three GADD45 members activate MTK1 kinase activity leading to p38/c-jun-NH₂-kinase activation and apoptosis. Recent findings established the GADD45 family as a critical mediator of apoptosis in cancer cells (3). NF-κB-mediated cell survival in cancer cells is absolutely dependent on two GADD45 family members, GADD45α and GADD45γ (Fig. 3). NF-κB-mediated repression of GADD45α and γ is sufficient for cancer cell survival and repression of the GADD45α and GADD45γ genes is for a large part the result of NF-κB-mediated up-regulation of c-myc, another oncogene frequently overexpressed in a variety of cancers. Whether NF-κB-mediated repression of GADD45α and GADD45γ gene expression in cancer may involve also changes in methylation is not known. Inhibition of NF-κB in cancer cells results in GADD45α- and GADD45γ-dependent induction of apoptosis and inhibition of tumor growth (3) correlating well with the findings of Ying et al. (1) that ectopic expression of GADD45γ in cancer cells inhibits tumor growth.

GADD45α and GADD45γ induction has been shown to be essential for c-jun-NH₂-kinase activation and apoptosis in cancer cells, highlighting the notion that GADD45α and GADD45γ repression via methylation or other mechanisms plays an unambiguous and universal role in the ability of certain types of cancer to escape from programmed cell death (3). Changes in GADD45α and GADD45γ expression are likely to play a role in the proapoptotic functions of various anticancer drugs as well. Whether these drugs act via inhibition of NF-κB or changes in methylation is not clear.

Although members of the GADD45 family seem infrequently mutated in cancer as far as currently known, reduced expression of the three GADD45 family members due to promoter methylation has been frequently observed in several types of human cancer. In resectable invasive pancreatic ductal carcinomas, GADD45α is frequently mutated, and mutation combined with the p53 status correlates with survival of invasive pancreatic ductal carcinoma patients (17). This is the only type of cancer where mutation of a GADD45 gene has been observed thus far. GADD45α expression is also reduced in ovarian cancer, although the precise mechanism has not been elucidated. The GADD45α promoter is methylated in the majority of breast cancers resulting in reduced expression when compared with normal breast epithelium (18). It is thus interesting to notice that in the current publication no differences in GADD45γ methylation were observed in breast cancer cell lines suggesting specific methylation of the GADD45α gene in breast cancer. In pituitary adenomas, silencing of the GADD45γ gene in 67% of patients is primarily associated with methylation of the GADD45γ gene and reversal of this epigenetic change results in re-expression (19). GADD45γ is also down-regulated in anaplastic thyroid cancer and in 65% of hepatocellular carcinomas due to hypermethylation of the GADD45γ promoter. Interestingly, the GADD45β gene is methylated and silenced in hepatocellular carcinoma as well, indicating a strong linkage between at least two GADD45 genes and liver cancer. Our observation that activated NF-κB leads to repression of GADD45α and GADD45γ expression in various types of cancer together with the frequent constitutive activation of NF-κB in cancers, furthermore, suggests that there are at least two mechanisms whereby GADD45 genes become repressed in cancer. Thus, repression of GADD45 gene expression in various types of cancer via methylation or NF-κB activation seems a critical step in oncogenesis and highlights the role of the GADD45 gene family in regulating DNA damage, cell cycle, and cell survival. Because activation of NF-κB is a critical step for many cells to escape programmed cell death and this is dependent on GADD45α and GADD45γ repression, methylation of the GADD45γ gene as reported by Ying et al. may result in a similar phenotype (i.e., helping cancer cells to survive and become resistant to stress and DNA damage).

The current publication by Ying et al. (1) analyzes the methylation status of two regions in the GADD45γ promoter in a total of 75 cell lines as well as primary tissues and tumors. They show that promoter hypermethylation is frequently detected in tumors cell lines, including 85% of non-Hodgkin, 50% of Hodgkin lymphoma, 73% of nasopharyngeal, 50% of cervical, 29% of esophageal, and 40% of lung carcinoma but not in immortalized normal epithelial cell lines, normal tissues, or peripheral blood mononuclear cells. To gain more insight into the GADD45γ methylation, they also did high-resolution bisulfite genomic sequencing. They found that densely methylated CpG sites were detected in all silenced cell lines, indicating that epigenetic silencing of GADD45γ could be involved in the pathogenesis of tumors.

A hypermethylation of CpG island in cancer is strongly associated with transcriptional silencing of distinct subsets of genes (2). Increased methylation in the promoter region of a gene leads to reduced expression, whereas methylation in the transcribed region has variable effects on gene expression (20). Several mechanisms have been proposed to explain the transcriptional repression by DNA methylation. The first mechanism involves direct interference with the binding of specific transcription factors to their recognition sites in their respective promoters (21). The second mode of repression involves a direct binding of specific transcriptional repressors to methylated DNA (22). The reduction in affinity of certain factors, however, is often insufficient to account for the inactivity of promoters in vivo (2).

Methylation of non-core regions within a promoter CpG island does not block gene transcription and is frequently observed in cancer and even in normal cells (23). When a core region is methylated in a cell, it is almost always associated with that of non-core regions (23, 24). Genome-wide screening of methylated regions in tumor DNA has led to the identification of several tumor suppressor genes (25).

Numerous steps are necessary to identify methylation associated with a tumor suppressor gene. Initially, methylated core regions associated with specific gene silencing must be identified. Subsequently, expression of a gene in normal cells must be confirmed. Analysis of gene expression levels in a range of tissues is sometimes helpful to evaluate its consequential expression level (24) Thereafter, the connecting role of methylation in gene silencing is analyzed by treating cells with a demethylating agent and observing the resulting up-regulation of gene expression. Many genes have been shown to be silenced in various cancers through MS-RDA32 (24) or MCA-RDA (24) combined with 5-azadeoxycytidine treatment (25, 26). The ultimate central step is functional analysis of the candidate tumor suppressor genes. The presence of mutations in these genes in some tumor samples, if even only a few, is strong evidence for a tumor suppressor gene.

In conclusion, the identification of CpG hypermethylation events as tumor biomarkers can provide useful information for early diagnosis, cancer risk assessment, and prognosis (2). Methylated CpG islands are associated with virtually every type of tumor (2), may be grouped into tumor-specific marker panels (2), and can be detected with a high degree of sensitivity (2). The mapping of hundreds to thousands of CpG island–associated tumor suppressors and the discovery of the repressive properties of methylation have had a great effect on disease prevention and treatment (2). Differences in methylation patterns among tumors seem associated with patient outcome or other clinical responses and can be used as markers to classify tumors in association with more effective treatments for patients. Most importantly, the results provided by Ying et al. (1) establish the GADD45γ gene as a major target for methylation in various types of cancer and further support the notion that modulation of GADD45 gene family expression and function may be critical steps in cancer development and progression. These results also imply that therapeutic strategies targeting the epigenetic repression of the tumor suppressor GADD45γ have the potential to result in strong anticancer effects.