Multifaceted roles for thymine DNA glycosylase in embryonic development and human carcinogenesis

Abstract

Thymine DNA glycosylase (TDG) is a multifunctional protein that plays important roles in DNA repair, DNA demethylation, and transcriptional regulation. These diverse functions make TDG a unique enzyme in embryonic development and carcinogenesis. This review discusses the molecular function of TDG in human cancers and the previously unrecognized value of TDG as a potential target for drug therapy.

Introduction

Cancer remains a major public health problem in the world [1], and DNA mutations of tumor suppressor genes or oncogenes [2] are often the underlying molecular causes of this family of diseases. DNA damage in human cells leads to detrimental mutations in the absence of corrective, repair mechanisms [3,4]. For example, the spontaneous or enzymatic deamination of the pyrimidine bases, cytosine, or 5-methylcytosine (5mC) generates uracil or thymine, respectively, and leads to U:G or T:G mismatches (Fig. 1A) [5]. In addition to DNA mutations, epigenetic alterations, which include either DNA hypermethylation or hypomethylation, also contribute to carcinogenesis. DNA methyltransferases and demethylases meticulously regulate DNA methylation patterns in CpG islands during development, gametogenesis, and differentiation of mammalian tissues [6]. DNA mismatches, which occur during DNA demethylation processes (Fig. 1B), undergo the required base excision and repair by the key enzyme, thymine DNA glycosylase (TDG) (Fig. 2) [7]. Recently, we and others found that TDG's involvement in transcriptional regulation makes it a potentially valuable therapeutic target for treating human cancers [8,9].

Figure 1.

DNA mismatches generated from mutation or demethylation (A) Structures of cytosine derivatives. Cytosine is methylated to 5mC by DNMTs. Cytosine can be deaminated to produce uracil, and 5mC can be deaminated to produce thymine. (B) TETs convert 5mC to 5hmC, which is further oxidized to 5fC and 5caC.

Figure 2.

TDG mediates BER The mispaired G:U/G:T/G:5fC/G:5caC, which are either deaminated or demethylated from cytosine or 5mC (Fig. 1), are substrates of TDG. TDG specifically recognizes these mismatches and removed the mispair base, leaving an abasic site. Abasic site is then cleaved by the AP endonuclease and repaired by DNA polymerase β (pol β) and DNA ligase to restore the correct base (cytosine).

TDG Gene and Protein

The human TDG gene locates on chromosome 12q24.1, spans a region of 23 kb (NCBI Gene ID: 6996) and encodes the TDG protein with 410 amino acids (NM_003211). TDG belongs to the mismatch uracil glycosylase subfamily within the monofunctional uracil DNA glycosylase (UDG) superfamily, all of whose members share a common α/β-fold structure [10]. TDG possesses an N-terminal domain (aa 1–111), a C-terminal domain (aa 328–410), and a highly conserved, central, catalytic domain (aa 112–327) (Fig. 3) [11]. The N-terminal and C-terminal domains dictate the specificity of substrate recognition and processing.

Figure 3.

The functional domains of TDG and its binding proteins Dnmt3a, DNA methyltransferase 3a; SIRT1, NAD-dependent protein deacetylase sirtuin-1; TTF-1, thyroid transcription factor-1; PKCα, protein kinase C α; ERα, estrogen receptor α; NCoA-3, nuclear receptor co-activator-3; SRC1, steroid receptor coactivator 1; CBP CH3, CH3 domain fragment of the CREB-binding protein; CBP HAT, HAT domain fragment of the CREB-binding protein; APE1, apyrimidinic endonuclease 1; SUMO-1, small ubiquitin-related modifier 1; TCF4, transcription factor 4; RAR/RXR, retinoic acid receptor and retinoid X receptor; Gadd45a, growth arrest and DNA damage-inducible alpha.

Mutations in the central catalytic domain cause functional disruption in the roles played by TDG. For example, a TDG enzyme with a N140A mutation in the catalytic domain still binds to mismatched DNA but fails to remove the mismatched base (Fig. 2). Interestingly, an M269H mutation did not affect the glycosylase activity but reduced the binding affinity of TDG to the mismatched substrates [12]. Additional features involving the structure of TDG were well described in prior reviews [11,13,14].

TDG in DNA Repair

Mammalian DNA suffers damage by endogenous events and environmental factors, such as free radicals, ultraviolet light, and gamma radiation. As mentioned previously, the deamination of cytosine generates U:G mispairing in double-stranded DNA, and the analogous deamination of 5mC leads to T:G mispairing (Fig. 1A) [15]. TDG was the first, mismatch-specific DNA glycosylase discovered in human cells. It was identified from HeLa cells during a search for DNA repair enzymes capable of excising the T:G mispairing and restoring the canonical C:G base pairing [7,13,16].

TDG recognizes a mismatched thymine in a G:T lesion (Fig. 1) and excises T from this mispairing as the first step in the replacement of T by C in the process called base excision repair (BER) (Fig. 2) [16]. TDG also excises U from a G:U mismatch [12,17]. After thymine or uracil base removal, TDG binds to the apyrimidinic (AP) site across from the unpaired G with an affinity that exceeds its affinity for any other substrates [18]. The N-terminal domain of TDG serves as a flexible ‘clamp’ that binds TDG to the DNA undergoing the BER process. An apurinic/apyrimidinic endonuclease (APE1 or human AP endonuclease-1) subsequently binds to the TDG domain between aa 92 and aa 121 [19], stimulates the displacement of TDG [18], and facilitates the completion of the BER process (Fig. 2).

TDG is one of four mammalian UDGs capable of processing mismatches involving uracil as well as 5-fluorouracil (5-FU) [12]. In fact, TDG possesses the capability to excise a range of mismatched pyrimidine bases, which include oxidized pyrimidines such as thymine glycol Tg [20,21], 5-formyl-U, 5-hydroxy-U, 5-hydroxymethyl-U [22], N⁴-ethenocytosine [21,23,24], and 5-hydroxycytosine (Fig. 1A) [21,25]. N⁴-Ethenocytosine derives from either lipid peroxidation or the exposure to chemical carcinogens such as vinyl chloride [23]. In addition to excising 5-FU, TDG also excises other halogenated uracil analogs such as 5-chlorouracil and 5-bromocytosine [25]. TDG also excises T when paired with a ‘damaged’ adenine [21]. Finally, this ability to recognize a broad range of abnormal pairings, U:G, T:G, or T:A* (where A* represents a chemically altered A), makes TDG an important participant in the active DNA demethylations where the recognition of specific sites plays an equally important role.

TDG in DNA Demethylation

In mammalian cells, 5mC appears principally in the context of CpG dinucleotide patterns, often called CpG islands. Approximately 70%–80% of CpGs undergo methylation [26] as an essential feature in support of embryonic viability [27], aging and various human diseases including cancers that display aberrant DNA methylation patterns in adult tissues [28,29]. TDG regulates both DNA methylation and demethylation. DNA methyltransferases (Dnmt1, Dnmt3a, and Dnmt3b) maintain DNA methylation patterns using S-adenosylmethionine as the primary methyl donor [30]. The N-terminus of TDG interacts with Dnmt3a through its catalytic domain and the proline–tryptophan–tryptophan–proline domain in Dnmt3a that in turn targets Dnmt3a to heterochromatin [31]. Dnmt3a stimulates TDG glycosylase activity while conversely TDG inhibits the methylation activity of Dnmt3a [31,32]. In this capacity, TDG protects CpG-rich promoters from de novo DNA methylation and aberrant hypermethylation, but the role of TDG in regulating DNA demethylation, as described below, overshadows its regulatory role in DNA methylation.

TDG participates in two types of demethylation events: passive demethylation and active demethylation. Passive DNA demethylation occurs during the suppression of normal DNA methylation because Dnmt activity is either absent or repressed. Under these circumstances, DNA methylation occurs across the genome or at specific loci during DNA replication as seen, for example, in the erasure of parental imprints in primordial germ cells [33]. TDG interacts with Dnmt3a and inhibits DNA methylation activity of Dnmt3a [31,32], and consequently, TDG contributes to passive DNA demethylation.

Methylation on carbon 5 of cytosine leads to 5mC (Fig. 1A), the best-studied, epigenetic, DNA hallmark. The so-called ten–eleven translocation (TET) enzymes are α-ketoglutarate-dependent dioxygenases that catalyze the active demethylation of 5mC [26]. Each of the TET family members (TET1, 2, and 3) contains a core structure consisting of an oxygenase domain, which is a double-stranded β-helix (DSβH), and a cysteine-rich domain. TET converts 5mC in DNA to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) (Fig. 1B). TDG subsequently excises the mismatched 5fC:G and 5caC:G and restores the desired cytidine base in the BER process (Fig. 2) [34,35]. The DNA demethylation mechanisms involving TET were well described in recent reviews [36–39].

TDG also plays a central role in active DNA demethylation. TDG forms a ternary complex with activation-induced deaminase and growth arrest and DNA damage-inducible alpha (Gadd45a), according to co-IP experiments [6]. The N-terminal domain of Gadd45a (i.e. aa 1–132) interacts with the N-terminal and C-terminal domains of TDG, as determined by yeast two-hybrid assays. Gadd45a promotes DNA demethylation in a TDG-dependent manner [35].

TDG in Transcriptional Regulation

Growing evidence suggests that TDG interacts with an increasing number of transcription factors and thereby regulates gene expression (Fig. 4) [11,13,14]. TDG was first reported to interact with the transcription factor, c-Jun, according to a yeast two-hybrid screen [40]. In addition, TDG also interacts with retinoic acid receptor (RARα) and the retinoid-X-receptor (RXRα) in a ligand-independent manner. The central catalytic region of TDG is necessary and sufficient to promote binding to either RARα or RXRα. The attenuation of retinoic acid-dependent RAR/RXR transcriptional activities in TDG-null mouse embryo fibroblasts (MEFs) [6] indicates that TDG enhanced both RXR and combined RXR/RAR activities [41]. Activated RARα forms a ternary complex with TDG and the CREB-binding protein (CBP). CBP and its paralog, p300, are histone acetyltransferases and act as co-activators for RAR/RXR. Acetylation of TDG affects the stability of the ternary complex and retinoic acid-dependent gene expression, a finding which suggests that TDG may couple the transcription co-activator, CBP/p300, with nuclear receptors [42]. In GST pull-down assays, TDG interacts with other nuclear receptors, including receptors for estrogens, androgens, glucocorticoids, progesterone, all-trans-retinoic acid α, peroxisome proliferator-activated receptor gamma, thyroid hormones, and vitamin D₃ [41–45]. For example, TDG binds to ERα and stimulates ERα activity in a ligand-dependent manner, which suggests that TDG is a co-activator of ERα [43].

Figure 4.

TDG in transcriptional regulation (A) TDG interacts with nuclear receptors and acts as a transcriptional co-activator. (B) TDG interacts with TCF4 and enhances Wnt/β-catenin signaling. (C) TDG represses the transcriptional activity of TTF-1.

Nuclear receptors contain functional, ligand-binding domains, DNA-binding domain, and two activation domains (AF1 and AF2) [46]. During the activation process, nuclear receptors recruit p160 proteins and/or nuclear receptor co-activators (NCoAs). TDG interacts directly with SRC1/NCoA-1, a member of the p160 family, and cooperatively stimulates the ERα activity in the presence of estrogen [44]. In addition, TDG also interacts with other members of p160 family, such as NCoA-3 [45]. The interaction between TDG and NCoA-3 is particularly important for the transcriptional activation of steroid hormone receptors including the estrogen, androgen, and progesterone receptors (ER, AR, and PR, respectively) [45].

TDG also interacts with transcription factor TCF4 and acts as a positive regulator in the Wnt pathway [8]. As in other interactions between TDG and its participating partners, it is the N-terminus of TDG, which binds TCF4 and the HAT domain of CBP/p300. Expression of TDG enhances Wnt signaling. Knocking down TDG by shRNAs inhibits Wnt signaling and represses the proliferation of colorectal cancer (CRC) cells in vitro and in vivo [8]. Independently, another group also reports that TDG promotes transactivation of β-catenin/TCFs by cooperating with CBP in Wnt signaling [9].

TDG also acts as a transcriptional repressor. Thyroid transcription factor-1 (TTF-1) plays an important role in early embryonic development [47,48], and the C-terminal domain of TDG binds to the C-terminal activation domain of TTF-1. This binding represses the transcriptional activity of TTF-1 in a dose-dependent manner [47]. Because TDG is highly expressed during early mouse embryogenesis at 14.5 gestational days [49], it plays an important role in preventing TTF-1-mediated transcription [47] during early stages of development.

TDG and Posttranslational Modifications

The regulation of the various activities of TDG involves either posttranslational modifications or associations with other co-activators. For example, the acetylation [19] or sumoylation [50] of TDG regulates its function in DNA repair. TDG also recruits histone acetyltransferases in the CBP/p300 family, which act as transcriptional co-activators. Although CBP is involved in both DNA repair and transcription, the individual roles for TDG in these two processes are separable. The N151A mutation in TDG abrogates its glycosylase activity, which leads to lethality in mouse embryos, and abrogates its DNA demethylase activity, which regulates the Tat enhancer [6]. However, the N151A mutation permits the CBP-dependent transcription. Still other mutations in TDG, such as the N140A mutant, have no effect on TDG function in activation of Wnt signaling [8].

Phosphorylation and acetylation are mutually exclusive covalent modifications of the N-terminus of TDG that produced different effects on TDG activity. Acetylation by CBP/p300 at the N-terminal lysine-rich region [19] of TDG reduces DNA-binding affinity and abrogates G:T processing. Protein kinase C α (PKCα) interacts with TDG and phosphorylates the N-terminal serine residues adjacent to the lysines that undergo acetylation by CBP/p300. Phosphorylation preserves the G:T processing activity presumably by preventing the CBP/p300 acetylation [51]. TDG also interacts with histone/protein deacetylases, such as SIRT1. SIRT1 interacts with aa 67–110 of human TDG, deacetylates TDG, and enhances its glycosylase activity in vitro and in vivo [22].

TDG also undergoes covalent conjugation with small ubiquitin-related modifiers (SUMOs) and with ubiquitin as part of the regulatory process controlling DNA repair. TDG sumoylation by SUMO-1 or SUMO-2/3 [52] takes place in the nucleus and/or at the nuclear membrane [50]. During BER, the N-terminus of TDG binds tightly to the AP sites and protects these base-deficient sites in DNA from enzymatic degradation. SUMO-1 conjugation at the C-terminus of TDG causes the dissociation of the N-terminus of TDG from the AP sites and coordinates the recruitment of APE1 to complete the repair process [53,54]. This C-terminal sumoylation also inhibits the interaction of TDG with CBP [50], which alters transcription, but enhances the interaction of TDG with APE1. These effects suggest that sumoylation of TDG may have other, as yet undiscovered, differential effects on DNA repair and transcription [52].

TDG in Development and Cancer

TDG is essential for early embryonic development [6], and TDG-null mouse embryos die at approximately day E11.5 [55]. TDG expression appears in specific tissues of the developing fetus at 14.5 gestational days during development of the nervous system, thymus, lung, liver, kidney, and intestine. TDG also appears during late stages of development, where high levels of TDG expression emerge in the thymus, brain, and nasal epithelium [49]. The DNA repair function of TDG is only a minor function in mouse embryogenesis. In embryogenesis, TDG interacts with activated histone modifiers, such as CBP/p300, to maintain states of active and bivalent chromatin during cell differentiation. TDG associates with the promoters of essential developmental genes to protect them from aberrant CpG methylation [55]. Finally, TDG also plays an important role in stem cells. Reprogramming of fibroblasts to pluripotency requires Tet/TDG-mediated DNA demethylation, and Tet/TDG-deficient MEFs fail to experience reprogramming [56].

The functions of TDG in human cancer are an active, even controversial topic with respect to the role that TDG plays in tumorigenesis [57]. TDG's role in BER and TDG's interactions with the p53 ‘watchman’ proteins originally suggests that TDG acts solely as a tumor suppressor. The p53 proteins recruit over-expressed TDG to the p21Waf1 promoter and enhance its transcriptional activity [58,59]. Moreover, p53 transcriptionally regulates TDG expression by binding to the TDG promoter and regulating TDG nuclear translocation. These findings certainly support the hypothesis that TDG is a candidate for inclusion in the family of tumor suppressors [60].

Several research groups studying the associations between TDG and human cancers suggest that sequence variants of TDG and other BER pathway genes may act as susceptibility alleles in CRC. Screening the coding sequence and intron–exon boundaries of TDG genes in 94 familial CRC case revealed a TDG variant, namely A196G, in a 66-year-old male with rectal cancer [61]. A more recent study reported a heterozygous mutation in TDG in a rectal cancer patient [62].

Identification and characterization of several SNPs of TDG suggest an association with various cancers. Rs4135054, for example, links to esophageal squamous cell carcinoma (ESCC) [63]. The nonsynonymous coding SNP (rs2888805) and the intronic SNP (rs4135150) of TDG are associated with increased risk of developing nonmelanoma skin cancer and other cancers [64]. Rs2888805 is heterozygous in more than 10 cases in 94 patients with a family history of CRC [61]. However, the V367M mutation in TDG, for which rs2888805 codes, fails to show a statistically significant association with lung cancer risk in a case–control analysis [65]. Other mutations in TDG such as G199S and G994A, which are encoded by rs4135113, show no association with CRC risk [66].

TDG polymorphism is also associated with genomic instability and cellular transformation [67,68]. The D239Y variant occurs in 6.2% of global population. The D239Y-expressing cells are more sensitive to DNA damage and are more active in DNA damage response [67]. The G199S variant occurs in 10% of global population [68]. This variant binds to its substrate and abasic product more tightly. Accumulation of G199S variant leads to double-strand breaks and cell transformation. These TDG variants may be associated with the risk of human cancers.

Tumors in mouse models display differential expression of TDG. Tumors in the lactating mammary epithelium over-expressing MMTV-v-Ha-ras or MMTV-c-myc transgene [49] show increased TDG levels. The elevated TDG expression is also present in osteosarcomas and lymphomas in p53 mutant mice [49]. In contrast with the results from mouse models, TDG in human tissues displays a high expression level only in thymus [69], in which DNA contains the highest percentage of 5mC [70]. TDG appears at lower levels in other normal human tissues [69]. TDG shows no association with human gastric cancer using loss of heterozygosity analysis in 40 gastric tumor samples [69]. Gastric cancers show low mRNA levels of TDG in comparison with the levels of TDG in adjacent nontumor tissues, as detected by real-time RT–PCR. There is no significant correlation between TDG mRNA levels and clinicopathologic factors of gastric cancer [71]. However, recent studies using IHC demonstrated that TDG levels are significantly higher in tumor tissues than in the adjacent normal tissues in CRC patients [8]. In our recent unpublished study, the expression of TDG is higher in gastric cancer tissues compared with the normal tissues (Fig. 5). Other groups, who analyze the associations between DNA repair pathway genes and the risk of ESCC and gastric cancer, using data from a genome-wide association study, found that TDG is significantly associated with ESCC and gastric cancer risks [63]. In addition to gastrointestinal cancers, the mRNA levels of TDG increase in p53-mutated adrenocortical tumors associated with Li-Faumeni syndrome [72], but in multiple myeloma cell lines [73] and pancreatic adenocarcinoma [74], the TDG mRNA levels decrease. However, a larger number of samples than the number investigated to date should be evaluated in order to validate these findings.

Figure 5.

TDG expression in colorectal and gastric cancers TDG levels in tumor, adjacent, and normal tissues from the same patients were analyzed by IHC. TDG expression was significantly increased in human colorectal (A) and gastric cancer tissues (B).

Summary

TDG is a well-known enzyme with a spectrum of interesting and important biological functions. TDG regulates DNA repair, DNA methylation, and DNA transcription. TDG itself is tightly regulated by both genetic and epigenetic factors, and consequently, it is not surprising that TDG plays multifaceted roles in carcinogenesis. The interesting expression pattern of TDG in human cancers further suggests that it may be a valuable biomarker for certain cancers. Our finding that TDG shRNAs inhibit CRC growth in vivo indicates that TDG is a potential therapeutic target for cancer treatment. The potential side-effects of TDG inhibition need to be carefully assessed because of the numerous and essential roles that TDG performs in early development and stem cell regulation. TDG also participates in the reversal of the DNA-dependent toxicity of 5-FU, a common chemotherapeutic drug for the treatment of cancer. TDG removal of 5-FU from modified DNA results in an increase in DNA single-strand breaks [75]. Conversely, resistance to 5-FU treatment may arise in part because of a TDG deficiency. Further studies are needed to evaluate TDG as a target for the new antineoplastic agents.

Funding

This work was supported by the grants from the National Cancer Institute (Nos. R21 CA139359 and R01 CA172379), the Office of the Dean of the College of Medicine, the Kentucky Lung Cancer Research Program, and the National Institutes of Health (No. P30 GM110787).