DNA Excision Repair at Telomeres

Abstract

DNA damage is caused by either endogenous cellular metabolic processes such as hydrolysis, oxidation, alkylation, and DNA base mismatches, or exogenous sources including ultraviolet (UV) light, ionizing radiation, and chemical agents. Damaged DNA that is not properly repaired can lead to genomic instability, driving tumorigenesis. To protect genomic stability, mammalian cells have evolved highly conserved DNA repair mechanisms to remove and repair DNA lesions. Telomeres are composed of long tandem TTAGGG repeats located at the ends of chromosomes. Maintenance of functional telomeres is critical for preventing genome instability. The telomeric sequence possesses unique features that predispose telomeres to a variety of DNA damage induced by environmental genotoxins. This review briefly describes the relevance of excision repair pathways in telomere maintenance, with the focus on base excision repair (BER), nucleotide excision repair (NER), and mismatch repair (MMR). By summarizing current knowledge on excision repair of telomere damage and outlining many unanswered questions, it is our hope to stimulate further interest in a better understanding of excision repair processes at telomeres and in how these processes contribute to telomere maintenance.

1. Telomere overview

Telomeres, the physical ends of linear eukaryotic chromosomes, are essential for genomic stability. In most organisms, telomeric DNA contains double-stranded G-rich tandem repeats followed by G-rich single-stranded overhangs at 3′ ends [1]. In mammalian cells, the telomeric DNA sequence is ~3–20 kb of (TTAGGG)_n repeats (see section 4 on description of telomeres in yeast). The six-member telomere binding protein complex termed shelterin (TRF1, TRF2, RAP1, TIN2, POT1, and TPP1) binds to telomeric DNA, forming the unique telomeric chromatin structure called t-loop. One of the important functions of shelterin is blocking the ATM/ATR damage response pathways and suppressing double-strand break repair activities at chromosome ends, thereby preventing linear chromosome ends from erroneously being recognized as damaged DNA by the repair machinery [2, 3]. In addition, shelterin proteins control telomere length via regulating telomerase recruitment and/or modulating telomerase activity at chromosome ends (reviewed in [4]).

Telomeric DNA is vulnerable to damage in a number of ways. In addition to inevitable errors generated in the process of DNA replication during each cell division, telomeric DNA damage can also arise from oxidative stress, ionizing radiation, and carcinogenic chemicals [5–7]. Damaged telomeric DNA causes telomere dysfunction, resulting in aberrant chromosome end-to-end fusions and inappropriate recombinations that lead to genomic instability. Studies on DNA repair processes at non-telomeric chromosomal sequences have demonstrated that multiple molecular mechanisms mend damaged DNA to maintain the integrity of the genome. These mechanisms include NER, BER, MMR, non-homologous end joining (NHEJ), and homologous recombination (HR). Due to the special role of telomeres in protecting chromosome ends, DNA repair at telomeres must be tightly regulated to prevent harmful outcomes caused by unwanted repair activities. Both the NHEJ and HR pathways at telomeres have been discussed at length elsewhere [3, 8, 9]. This review focuses on NER, BER and MMR pathways, which have been loosely and sporadically studied at telomeres.

2. BER at telomeres

2.1 BER overview

BER is a cellular mechanism that corrects discrete small (non-bulky) DNA base lesions caused by oxidation, deamination, and alkylation. Defects in BER result in a higher mutation rate in a variety of organism, leading to predisposition of many types of cancer. BER is initiated by various DNA glycosylases, which recognize and remove damaged bases (see elsewhere in this issue). Eleven DNA glycosylases have been identified in humans. Both OGG1 (8-oxoguanine DNA glycosylase) and SMUG1 (single-strand-selective monofunctional uracil glycosylase 1) can remove oxidized guanine derivatives 8-oxo-7,8-dihydroguanine (8-oxo-G) and 2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyG). UNG (uracil DNA glycosylase) removes uracil, 5-hydroxyuracil, and other uracil derivatives in both single- and double-stranded DNA. AGG/MPG (alkyl guanine glycosylase/methyl purine glycosylase) removes alkylated purines and ethenopurines. TDG (thymine DNA glycosylase) removes uracil, thymine, and 5-hydroxymethyluracil. MUTYH (MutY homolog glycosylase) removes adenine when misincorporated opposite 8-oxo-G or 2,6-diamino-4-hydroxy-5-formamidopyrimidine. NTH1 (Nth Endonuclease III-Like 1), NEIL1 (Nei Endonuclease VIII-Like 1), and NEIL2 can remove oxidized pyrimidines and formamidopyrimidines.

Following the removal of the damaged base, APE1 (apurinic/apyrimidinic endonuclease 1) cuts the DNA backbone at abasic sites, leaving a single nucleotide gap. The gap is then repaired by short-patch or long-patch repair depending on the gap length. In short-patch repair, the repair gap is only one nucleotide, while in long-patch repair the gap is 2~8 nucleotides in size. In both pathways, the overhanging single-strand flap is removed by flap endonuclease 1 (FEN1). Following this incision of DNA, DNA polymerase β (Polβ) fills the gap and the single-strand break is sealed in a ligation step completed either by the DNA repair enzyme DNA ligase III in association with XRCC1 or by the replicative DNA ligase I (reviewed in [10], and see elsewhere in this issue).

2.2 BER at telomeres

The long arrays of TTAGGG repeats make telomeres particularly susceptible to oxidative lesion formation. Oxidative damages has been implicated as a primary cause of telomere shortening, evidenced by the observations that telomere attrition rate is significantly decreased when cells are grown in hypoxia condition or in the presence of antioxidant [11, 12]. In addition, both uracil residues and oxidized guanine derivatives are commonly present at telomeres [13–15]. Mounting evidence from both in vivo and in vitro studies suggests that BER is actively promoted at telomeres. Here we will review functions of each BER factor in telomere maintenance.

2.2.1 OGG1

8-oxo-G is the most abundant base lesion at telomeres. OGG1 deficiency increases oxidative guanine lesions at telomeres, leading to disruption in telomere length homeostasis [16]. Using OGG1 knockout mice, Wang et al. report telomere shortening in OGG1−/− mouse hematopoietic cells and primary MEFs cultivated in normoxia condition (20% oxygen) or in the presence of an oxidant. Further analysis reveals other telomere abnormalities such as altered telomere sister chromatid exchanges, increased telomere single- and double-strand breaks, and preferential telomere G-strand losses in OGG1−/− mouse cells. These findings indicate that oxidative guanine damage affects telomere maintenance in diverse ways, and the BER pathway is required for maintaining telomere integrity in mammals [16].

The exact BER mechanism for removing telomeric 8-oxo-G residues is still unclear, and it is unknown how BER factors interplay with shelterin proteins. While the binding ability of TRF1 and TRF2 to a defined telomeric substrate is sharply reduced or abolished by 8-oxo-G incorporation [17], OGG1 incision activity is not impaired by TRF1 or TRF2, suggesting that the sequence context of telomere repeats and certain telomere configurations may be involved in telomere vulnerability due to oxidative damage [18]. Further studies are needed to elucidate BER mechanism for removing 8-oxo-G residues at telomeres.

2.2.2 UNG

Deficiency in UNG accumulates telomeric uracils and causes telomere defects including increased telomere fragility and aberrant telomere recombination in mouse hematopoietic cells [13]. Substituting thymine residues in a telomeric DNA substrate with uracils weakens POT1/TPP1 binding to substrates in vitro [13], indicating that telomere defects caused by UNG deficiency may be partially attributed to interference of POT1/TPP1 binding to damaged telomeric DNA.

2.2.3 NTH1

Deficiency in NTH1, another DNA glycosylase that primarily removes oxidative thymine derivative like thymine glycol, results in inefficient repair of oxidative DNA damage at telomeres, leading to preferential accumulation of oxidative base lesions at these DNA sequences [19]. In hematopoietic tissues of NTH1^−/− adult mice, telomeres suffer from multiple defects including telomere loss, increased telomere recombination, elevated telomere fragility, and DNA damage foci under normoxia. These results suggest that telomere DNA damage caused by oxidative thymine derivatives can interfere with telomere maintenance, and that NTH1 plays an important role in repairing such oxidative damage at telomeres [19].

2.2.4 NEIL

Biochemical studies show that telomeric DNA containing thymine glycol (Tg), 8-oxo-G, guanidinohydantoin (Gh), or spiroiminodihydantoin (Sp) can form quadruplex DNA structures in vitro [20]. The NEIL3 glycosylase has excision activity on Tg in quadruplex DNA, whereas the NEIL3 and NEIL1 glycosylases have excision activity on Sp and Gh in quadruplex DNA. These initial findings suggest that the NEIL1 and NEIL3 DNA glycosylases may remove oxidized guanine lesions from telomeres, which is crucial in telomere maintenance by preventing accumulation of oxidized guanine damages in telomeres [20, 21].

2.2.5 APE1

APE1 is the major apurinic/apyrimidinic endonuclease protecting cells from oxidative stress. It associates with telomeres and interacts with shelterin components TRF2 [22–24]. Recent report demonstrates that APE1 plays a unique role in stabilizing telomeric DNA [24]. APE1 depletion results in severe telomere defects including DNA damage response at telomeres, telomere signal loss, chromosome end fragmentation, chromosome end-to-end fusions, and extra telomeric signals regardless of cell type or telomerase status [24]. Additionally, APE1 deficiency induces telomere shortening in telomerase-expressing cells [24], consistent with reports that oxidative stress induces telomere shortening. APE1 depletion reduces TRF2 binding to telomeres [24], likely caused by the accumulation of lesions in telomeric DNA that interfere with the recruitment of factors required for telomere protection [17].

2.3 Impacts of shelterin proteins on BER activities

Nucleosomes and chromatin remodeling influence BER activity (reviewed in [25], and see elsewhere in this issue). Due to special configuration of telomere chromatin and abundant telomere binding proteins, it is likely that BER activities at telomeres are impacted by shelterin proteins. Indeed, elegant biochemical assays using purified proteins and DNA substrates demonstrate that shelterin proteins POT1, TRF1 and TRF2 physically interact with BER components Polβ, FEN1, and APE1. The shelterin proteins facilitate Polβ primer extension on actively synthesized telomeric substrates, stimulate Polα strand displacement, increase the percent of long-patch BER intermediates on telomeric substrates, as well as improve APE1 cleavage and binding [26, 27]. In addition, POT1, TRF1 and TRF2 increase FEN1 endonuclease activity on flap substrates and also increase Ligase I binding to DNA to help ligation. The stimulating effect of shelterin on BER activities may potentiate the role of telomere binding proteins in promoting efficient BER at telomeres [27].

3. NER at telomeres

3.1 NER overview

NER is an important DNA repair pathway involved in the removal of a wide spectrum of DNA lesions including UV-induced pyrimidine dimers, bulky chemical adducts, and certain forms of oxidative damage (see elsewhere in this issue). Deficiencies in various proteins participating in the NER pathway lead to three rare recessive genetic disorders: xeroderma pigmentosum (XP), Cockayne syndrome (CS), and the photosensitive form of the brittle hair disorder trichothiodystrophy (TTD). XP patients are extremely sensitive to ultraviolet rays from sunlight and are predisposed to skin cancer. CS is characterized by neurodevelopmental abnormalities, premature aging with a short lifespan (~12 years), and hypersensitivity to UV light. TTD is associated with ichthyosis and brittle hair, intellectual disability and delayed development, and recurrent infections (reviewed in [28]).

NER proteins recognize damaged sites and excise a 24–32 nt DNA fragment containing the damaged residue with extreme accuracy. Then, DNA polymerases use the undamaged strand as template to synthesize a new complementary strand to fill the gap, which is followed by ligation mediated by DNA ligase I or III [29–31]. Depending on the initial damage recognition, NER can be divided into two distinct pathways — global genome NER (GG-NER) and transcription coupled NER (TC-NER). GG-NER detects and removes lesions from the non-transcribed regions and silent chromatin over the entire genome, while TC-NER ensures faster repair of many lesions located on the sense strand of actively transcribed genes [29, 32]. GG-NER and TC-NER differ in initial damage recognition but proceed in a similar manner after recognition.

In GG-NER, DNA lesions are detected directly by the XPC/hHR23B-CETN2 complex through the structural distortion in DNA [33, 34]. This complex is the only NER factor known to be dispensable for TC-NER [35, 36]. The DDB complex, a damaged DNA-binding heterodimer consisting of DDB1 and DDB2/XPE, then recognizes DNA double-helix distortion caused by the lesion [34, 37]. Subsequently, DNA around the lesion is opened up by the concerted action of RPA, XPA and bi-directional XPB/XPG. The oligonucleotide at the lesion site is then removed, and the resulting gap is filled by DNA polymerase Polδ and Polε associated with the “sliding clamp” PCNA [34, 36].

In TC-NER, RNA polymerase II (RNAP II) recognizes a lesion while it transcribes a gene. Typically, the lesion is located in the transcribed strand, causing the RNAP II to stall or arrest [38]. When RNAP II is stalled by DNA damage, it recruits Cockayne Syndrome group B (CSB, also known as ERCC6, a member of the SNF2/SWI2 family of DNA-dependent ATPases) protein, which removes RNAP II from DNA so that it no longer acts as an obstacle to DNA repair [39, 40]. CSB, acting as a coupling factor, then recruits the CSA complex, core NER factors (XPA, TFIIH, XPG, XPF-ERCC1, and RPA), and histone acetyltransferase p300 to damaged sites. Subsequently, the promoter region is opened by the XPB and XPD helicases of TFIIH, and transcription resumes after the removal of the lesion [41].

3.2 NER at telomeres

Telomeric DNA sequence is predicted to form pyrimidine dimers after UV irradiation, and indeed, UV-induced pyrimidine dimers are detectable in the telomeric region [42, 43]. Due to the significant role of telomeres in the maintenance of chromosomal stability, NER is expected to be enhanced at telomeres. However, studies on NER at telomeres are surprisingly limited [42]. Inconsistent results have been reported on whether UV-induced pyrimidine dimers are repaired at telomeres. An early study finds that normal human somatic cells from different donor ages and disease states are capable of efficiently repairing UV-damaged telomeres, and the rate and extent of telomeric repair decline with donor’s age [42]. This study also shows that in normal human fibroblasts, UV-induced pyrimidine dimers are repaired more efficiently at telomeres than those at inactive, noncoding genomic locus such as X chromosome-linked locus 754, but less efficiently than those at the actively transcribed genomic region such as the dihydrofolate reductase gene [42]. This difference may be due to the intragenomic heterogeneity of DNA repair processes that relate to the level of genomic transcription [44–48]. Supporting the conclusion that NER is proficient at telomeres, a recent finding shows that UV-induced telomeric CPDs (cyclobutane pyrimidine dimers) are repaired at a faster rate than CPDs at the bulk genome region in human skin fibroblasts expressing exogenous telomerase [49]. In striking contrast, Rochette et al [43] report that telomeres do not repair CPDs [43]. The same study shows that in two primary human fibroblast cell lines, the frequency of CPD formation at human telomeres is ~7 fold higher than that at non-telomeric region [43]. Surprisingly, cells tolerate persistent high levels of telomeric CPDs and continue proliferating without apparent acceleration in shortening of mean telomere lengths [43]. Clearly, more investigations are needed to explain the disparate results, and more importantly, to determine the NER process at telomeres.

It remains a possibility that some unrepaired lesions may be tolerated at telomeres. Tolerance of unrepaired lesions likely requires mechanisms bypassing CPDs to fully replicate telomeric DNA as well as to avoid DNA breakage and accumulation of single-stranded DNA. One such mechanism relies on specialized DNA polymerases such as Polη, which is coded by the XPV gene (this gene is defective in a variant type of xeroderma pigmentosum) and is capable of bypassing CPDs by incorporating A opposite a T or C in a CPD (reviewed in [50]). Consistently, a recent study reports that exposure to UV or hexavalent chromium (Cr(VI)) that generates bulky DNA lesions induces accumulation of Polη at telomere regions, which suppresses DNA damage foci formation at telomeres [51]. Deficiency in Polη elevates replication-associated telomere aberrations, suggesting that Polη is required for proper replication of telomeres containing bulky DNA adducts [51].

While it remains to be determined whether NER is promoted at telomeres, various NER factors have been found to play pivotal roles in telomere maintenance, implying the connection between telomere regulation and NER (see below).

3.2.1 ERCC1/XPF

ERCC1/XPF is a structure-specific endonuclease formed by ERCC1 and XPF that localizes at telomeres and interacts with TRF2 [52]. ERCC1−/− MEFs retain the telomeric overhang after TRF2 inhibition, suggesting that this nuclease is involved in the removal of the G-rich 3′-overhang, likely because unprotected t-loop structure resembles the substrates for ERCC1/XPF [52]. ERCC1/XPF has also been implicated in telomere loss in mice overexpressing TRF2 [53]. Such mice show increased skin cancer after UV exposure as well as severe skin pathologies resembling human xeroderma pigmentosum with NER deficiency [53]. Keratinocytes derived from these mice are hypersensitive to UV irradiation and mitomycin C but not to ionizing radiation [53]. The TRF2 overexpressing mice have short telomeres and it appears that such telomere shortening is mediated by ERCC1/XPF [53]. Interestingly, increased TRF2 expression has been observed in human skin cancer [53]. These findings potentiate a role of TRF2 in regulating NER activity in mammalian cells.

3.2.2 XPB, XPC, XPD

Loss of XPB or XPD increases telomere attrition and enhances DNA damage at telomeres under oxidative stress, suggesting that XPB and XPD may have an important role in maintaining telomere integrity [54, 55]. Another key NER component, XPC, has been implicated in telomere maintenance as well, particularly upon UV exposure. Shorter telomeres have been observed in XPC−/− mice skin after chronic UV exposure, which is thought to be associated with an increased tumor incidence [56]. Interestingly, the telomere shortening effect can be reversed by an additional deficiency in telomerase. Double knockout mice deficient in both XPC and telomerase show aberrantly long telomeres that presumably results from activation of the alternative lengthening of telomeres (ALT) pathway, suggesting that XPC may have a role in suppressing aberrant telomere recombination [56]. Overall, the functions of XPB, XPC, XPD at telomeres are largely unclear.

3.2.3 CSB

CSB belongs to the SNF2/SWI2 ATPase family and is a multifunctional protein playing a role in chromatin remodeling, TC-NER, and transcription regulation (reviewed in [57]). The majority of CS patients are defective in the CSB gene [58, 59]. Batenburg et al. found that the CSB protein locates at a subset of telomeres and physically interacts with one shelterin component TRF2 [60], thus it is possible that CSB may be recruited to telomeres by TRF2 [60]. Deficiency in CSB affects multiple aspects of telomere homeostasis [60]. First, deficiency in CSB induces accumulation of fragile telomeres and dysfunctional telomeres [60], suggesting that CSB may be required for promoting efficient replication of telomeres and contributing to telomere protection. Consistent with this notion, treatment with aphidicolin further induces the frequency of fragile telomeres in CS cells [60], suggesting that telomere replication is compromised in CS cells. In addition, CS controls telomere length in a telomerase-dependent manner. While cells derived from CS patients undergo telomere shortening, overexpression of wild-type CSB into CS cells results in telomerase-dependent telomere lengthening [60]. The latter is associated with a reduction of the amount of telomere-bound TRF1 [60], a negative mediator of telomere length maintenance [61–63]. Furthermore, Batenburg et al. also observe that CS cells or CSB knockdown cells reduces the expression of the long non-coding telomere repeat-containing RNA known as TERRA [60]. Given that CSB regulates transcription and can function as a chromatin remodeler [57], the effect of CSB deficiency on TERRA expression is not entirely surprising. Again, current studies on CSB at telomeres are mainly descriptive, and it remains to be determined which function(s) of CSB plays a direct role in telomere maintenance.

4. BER and NER in subtelomeric and telomeric chromatin of yeast

Unlike mammalian cells, budding yeast telomeric DNA is composed of ~350 bp long (TG_1–3)_n tandem repeats that are bound by the Rap1 protein and its interacting proteins Rif1p and Rif2p [1]. These proteins negatively regulate telomere length via a protein-counting mechanism, in which the number of Rap1p-Rif1p-Rif2p bound to telomeres limits the action of telomerase in cis. The yeast telomerase, as its mammalian counterpart, is composed of two core components, a reverse transcriptase protein subunit (encoded by the EST2 gene) and a template RNA (encoded by the TLC1 gene) [64]. Unlike mammalian cells, yeast telomeres lack nucleosomes, and only the subtelomeric repeats contain nucleosomes [64]. Despite the lack of nucleosomes, telomere chromatin is repressive and forms heterochromatin. This is due to the recruitment of Sir proteins, including Sir2p, Sir3p, and Sir4p, by Rap1p and Ku70/Ku80 dimer to telomeres. Sir2p is a histone deacetylase that removes the acetyl groups from H3 and H4 histone tails, generating histone binding sites for Sir3p and Sir4p. This leads to the spreading of the Sir2-3-4 complex from sites of nucleation over a 2–3 kb subtelomeric domain. Consequently, genes near telomeres are transcriptionally silenced — a phenomenon referred to as telomere position effect (TPE) [65, 66]. The extent of silencing decreases with increasing distance from the telomere [66]. In addition, yeast telomeres gather into distinct subcompartments near the nuclear periphery [67, 68]. It is thought that clustering of telomeres helps sequestration of silencing factors to chromosome ends, therefore avoiding inappropriate repression of genes located elsewhere in the genome. However, silencing is not uniformly present at all native telomeres, with some chromosome ends exhibiting very little silencing while others having substantial silencing [69].

Like mammalian telomeres, oxidative damage occurs in telomeric guanines in S. cerevisiae [70]. The Liu group developed an elegant quantitative assay for measuring oxidative guanine damage at telomeres. Using this assay, they found that telomeric oxidative guanine damage can be enhanced by exogenous oxidative stress, and ogg1Δ cells accumulate more telomeric lesions than wild-type cells, suggesting that Ogg1 may be a primary DNA glycosylase in excising oxidized guanines in yeast telomeres [70]. Reminiscent to the effect of 8-oxo-G on TRF1 and TRF2 binding to mammalian telomeric sequences [15, 17, 70], yeast telomeric sequences containing oxidized guanine lesions reduce Rap1p and Rif2p binding to telomeres [70]. In addition, OGG1 deletion leads to telomere lengthening that depends on telomerase and/or homologous recombination pathways, suggesting that both telomerase and recombination pathways become more active upon OGG1 deletion [70]. Since Rap1p and Rif1/2p negatively regulates telomerase, the decreased telomeric binding of Rap1p and Rif2p explains the telomerase-dependent telomere lengthening in ogg1Δ cells. In addition, increased recombination rate has been observed in yeast cells with BER deficiency or harboring high density of 8-oxo-G [71, 72], and therefore, it is conceivable that Ogg1 deficiency-induced accumulation of telomeric 8-oxo-G also contributes to increased telomere recombination.

NER occurs at yeast telomeres to repair UV-induced DNA damage [73, 74], yet very little is known about how NER functions at telomeres in yeast. Histone modifications, especially acetylations, have a profound effect on NER [75, 76]. Nucleosomes with a higher level of histone H4 acetylation stimulate repair synthesis [77]. Several reports investigated the effect of chromatin structure and histone deacetylation on NER efficiency at the subtelomeric region in budding yeast. Irizar et al. report that UV-induced CPDs are repaired significantly more efficiently at non-repressive subtelomeric ends than at repressive subtelomeric ends [74]. Such suppression of NER is modulated by Sir2p, as deletion of SIR2 increases NER efficiency at repressed subtelomeres to a level similar to that at non-repressed subtelomeres. Consistently, Sir2p at repressive subtelomeres inhibits H3 and H4 acetylation [74]. Similarly, when telomere silencing is enhanced by either overexpression of Sir3p or deletion of RPD3 (another histone deacetylase), NER is inhibited at subtelomeres [73]. Therefore, the NER efficiency at yeast subtelomeres appears to be controlled by the silencing state of subtelomeric chromatin structure.

5. MMR at telomeres

5.1 MMR overview

MMR is an evolutionarily conserved process involved in the recognition and correction of erroneous nucleotides including mismatched nucleotides arising during DNA replication, heteroduplexes formed during recombination, as well as DNA lesions induced by various chemical and physical factors [78–80]. Besides repairing mismatches, MMR proteins also play important roles in meiotic and mitotic recombination, DNA damage signaling, triplet-repeat expansion, as well as somatic hypermutation and class-switching recombination (reviewed in [79, 80]). Although the basic functions of MMR pathways are largely conserved from bacteria to humans, the eukaryotic MMR systems are much more complex and composed of multiple MutS and MutL homologs, which form corresponding functional heterodimeric complexes [81]. In humans, MutSα (hMSH2-MSH6) recognizes base-base mismatches and small insertion/deletion loops, whereas MutSβ (hMSH2-hMSH3) can recognize relatively larger insertion/deletion loops. The recognition of unpaired nucleotides by MutSα or MutSβ will recruit MutLα (hMLH1-hPMS2) – the major MutL heterodimer in MMR. The MutSα (or MutSβ) complex is able to search for the strand breaks in an ATP-dependent fashion and then recruits exonucleases, such as EXO1, to eliminate the DNA strand containing the incorrect nucleotide. The single-strand DNA-binding protein RPA binds to and stabilizes the single-stranded DNA, and the single-stranded gap is filled in by Polδ and ligated by Ligase I [80]. Given its biological importance, MMR deficiency frequently leads to an increased mutation rate, microsatellite instability and the development of Lynch syndrome (a hereditary colorectal cancer syndrome). The most frequently mutated MMR genes in Lynch syndrome are hMSH2 and hMLH1 [82, 83].

5.2 MMR at telomeres

5.2.1 MMR and telomere length control

A number of studies suggest that MMR deficiency can be associated with telomere shortening. Analysis of Lynch syndrome families shows that telomere lengths in leukocytes of cancer patients carrying MMR gene mutations are significantly shorter than symptom-free mutation carriers and healthy controls [84]. In addition, cancer patients with MMR gene mutations show the most pronounced telomere attrition with age [84]. However, it is entirely unknown whether the shorter telomere length in cancer patients reflects the effects of MMR deficiency or merely represents a consequence of the cancer syndrome. Another evidence suggesting that defective MMR may impact telomere length comes from a survey of colon carcinomas [85]. The results of this study reveal that tumors with high microsatellite instability (a surrogate marker for MMR deficiency) tend to have shorter telomeres in comparison to those with stable microsatellites [85]. Moreover, down-regulation of hMSH2 in normal primary human lung fibroblasts increases the rate of telomere shortening [86]. It has been proposed that MMR deficiency may result in an accumulation of telomere mutations, thereby leading to telomeric repeat instability and accelerated telomere shortening [86].

Interestingly, results from animal model studies differ significantly from what observed in human cells. Tissues or primary MEFs derived from MSH2−/− mice have normal telomere lengths and telomerase activity [87]. It is highly plausible that the potential effect of MMR deficiency on telomere length differs by species and it can also be a consequence of the disease induced by MMR deficiency.

5.2.2 Role of MMR proteins in telomere DNA recombination

Results from analyzing yeast, mammalian, and human cell systems suggest that defective MMR proteins exert diverse effects on telomere DNA recombination. In budding yeast, loss of MMR proteins such as MLH1, MSH2, or PMS1 (PMS2 in mammals) promotes ALT in telomerase-deficient cells and enhances telomerase-independent cellular proliferation and survival [88]. In humans, hMSH6 deficiency in telomerase-expressing colon cancer cells induces ALT-like telomere elongation, but neither MSH2 nor MLH1 deficiency induces ALT [89]. In mammalian cells, PMS2 deficiency increases sister telomere chromatid exchange — a hallmark of ALT [90]. These initial observations suggest that a subset of but not all MMR proteins may possess MMR-independent functions in ALT, especially in the suppression of aberrant telomere recombination. It is thus very tempting to speculate that the anti-recombination function of MMR proteins may be related to such an action.

5.2.3 Role of MMR proteins in telomere damage signaling

It has been well recognized that MMR proteins also mediate DNA damage signaling. Dysfunctional telomeres are recognized as DNA damage and can activate the ATM/ATR DNA damage signaling pathways [9, 91]. Two studies investigate the role of MMR proteins in signaling dysfunctional telomeres. Using MSH2−/−TERC−/− double knockout mice, Martinez et al. show that abrogation of MSH2 rescues aging pathologies and tumor development associated with short telomeres [92]. Such rescue is not due to a rescue of short telomeres per se in telomerase null mice, as loss of MSH2 does not prevent telomere shortening in TERC−/− mice. Instead, it appears that such rescue results from MSH2 deficiency-induced attenuation of p53 and p21 induction [92], suggesting that MSH2 modulates the p53-mediated signaling pathway and cell cycle arrest in response to short and dysfunctional telomeres. This observation is consistent with the direct role of MMR proteins in signaling DNA damage through the p53 pathway [93–95].

A similar role has also been suggested for mammalian PMS2. Investigation of mice doubly deficient for telomerase and PMS2 reveals that PMS2 deficiency prolongs the lifespan and survival of telomerase-deficient mice, rescues degenerative pathologies, and reduces tumor formation [90]. Similar to the loss of MSH2, PMS2 deficiency attenuates p21 induction in response to short telomeres [90]. Together, these results suggest that certain MMR proteins like MSH2 and PMS2 may be in the same pathway with p21 and p53, and regulate signaling of cell cycle arrest associated with telomere dysfunction. These findings highlight the important role of MMR proteins in mediating cellular response to dysfunctional telomeres in vivo, which is fundamental to organismal aging and to tumor development.

5.2.4 Does MMR function at telomeres?

Our current knowledge on MMR at telomeres is exceedingly limited. Many fundamental questions remain to be answered. First, are telomeres hot spots for incorporating DNA mismatches or are they more resistant to mismatch incorporation? While it is expected that the abundant Cs in telomere repeats make telomeres highly susceptible to potential U:G mismatches, telomeres form special chromatin structures and shelterin proteins may have an inhibitory effect on mismatch incorporation. In order to fully understand the telomere maintenance mechanism, it is important to apprehend whether mismatches are present at telomeres and how these mismatches are generated and recognized. Second, if mismatch incorporation is present at telomeres, to what extent do DNA mismatches affect shelterin protein binding to telomeres? To date there has been no report on the binding efficiency of shelterin proteins to DNA substrates containing mismatches. Third, are telomeres functional substrates for MMR? Does normal MMR process occur at telomeres to remove mismatches? Does the presence of shelterin proteins have an inhibitory or stimulatory effect on the MMR process? Fourth, recent research progress suggests that chromatin organization, nucleosome assembly factors, and histone modifications regulate MMR activities [96]. How does the special telomere chromatin organization, nucleosome composition, TERRA, and unique histone modifications impact the MMR process at telomeres? Lastly, it is known that MMR deficiency frequently display microsatellite instability [97, 98]. The tandem hexanucleotide telomeric repeats may conceptually be envisioned as super microsatellites. Thus, could telomere instability caused by MMR deficiency be another potential mechanism driving tumorigenesis? Answers to these questions will not only elucidate the mechanisms by which mismatches are repaired at telomeres, but also offer needed insights into the molecular basis of tumorigenesis resulting from MMR deficiency, ultimately providing guidance to design more effective therapeutic strategies.

6. Concluding remarks and future directions

The cellular and environmental processes that influence the accumulation of DNA damages at telomeres are directly relevant to cancer and aging. Although NER, BER, and MMR pathways play indispensable roles in DNA damage response and repair, excision repair at telomeres is a largely unexplored area. While it is clear that BER is active at telomeres, the roles of NER and MMR at chromosome ends are elusive. This may be partially due to the complications from either the essential functions of the core repair proteins in cell/organismal survival, or from the difficulties in separating telomere-specific roles of repair factors from their genome-wide functions. To this end, we have summarized the known functions of major BER, NER, MMR proteins at telomeres and their interacting shelterin partners in Table 1.

Table 1.

Excision repair protein functions and their interacting partners at telomeres.

Protein	Repair pathway	Effects on telomere maintenance upon deficiency	Binding partners at telomeres	References
APE1	BER	↑ telomere loss, ↑ end-to-end fusions, radial chromosomes, ↑ extra-telomeric signal.	TRF2	Madlener et al. (2013)
OGG1	BER	telomere length deregulation, ↑ T-SCE	unknown	Lu et al. (2010), Rhee et al. (2011)
UNG	BER	↑ telomere fragility, aberrant telomere recombination	unknown	Vallabhaneni et al. (2015)
NTH1	BER	↑ telomere loss, ↑ telomere recombination, ↑ telomere fragility	unknown	Vallabhaneni et al. (2013)
XPB	NER	↑ telomere attrition rate, ↑ telomere damage	unknown	Gopalakrishnan et al, (2010), Ting et al. (2010)
XPC	NER	Telomere shortening after UV exposure, activation of ALT when telomerase is deficient	unknown	Stout et al. (2013)
XPD	NER	↑ telomere attrition rate, ↑ telomere damage	unknown	Gopalakrishnan et al, (2010)
ERCC1/XPF	NER	↑ telomeric overhang afterTRF2 inhibition, ↓ telomere loss after TRF2 overexpression	TRF2	Munoz et al. (2005), Shell et al. (2008), Zhu et al. (2003)
CSB	NER	↑ fragile telomeres, ↑ telomere damage, ↓ TERRA	TRF2	Batenburg et al. (2010)
MSH2	MMR	↑ telomere attrition rate (only in human lung fibroblasts), ↑ end-to-end fusions, telomere loss, attenuates telomere damage signaling	unknown	Martinez et al. (2009), Campbell et al. (2006)
MSH6	MMR	↑ T-SCE	unknown	Bechter et al. (2004)
PMS2	MMR	↑ T-SCE, attenuates telomere damage signaling	unknown	Siegl-Cachedenier et al. (2007)

T-SCE: Telomere sister chromatid exchange.

Several unique features of telomeres in fact warrant special considerations of excision repair. First, telomeres form the t-loop structure by inserting the single-stranded G-rich overhang back to duplex telomeric DNA [99, 100]. Formation of the t-loop structure is thought to shield chromosome ends from being recognized as double-strand breaks and help protect telomeres from being erroneously repaired. If DNA lesions are produced inside t-loop, how would these lesions be recognized and repaired by excision repair processes in the context of t-loop? Would it require temporary disassembly of t-loop? How would shelterin proteins affect BER, NER, or MMR activities at telomeres? Second, mammalian telomere chromatin contains nucleosomes [101, 102]. It is well established that efficiency of excision repair is profoundly influenced by chromatin remodeling complexes and histone modifications [96, 103–105], and DNA repair process and chromatin remodeling act cooperatively to initiate damage recognition and promote chromatin rearrangement in order to attain efficient removal of lesions [106–108]. Telomeres are specialized heterochromatin characterized by hypermethylation of DNA and repressive histone marks such as hypermethylation of histone H3 at different lysine residues, most notably at lysine 9 (H3K9), hypoacetylation of H3 and H4, and binding of heterochromatin protein 1 (HP1) to telomeres [109–111]. Additionally, both TRF1 and TRF2 are able to remodel nucleosomal spacing at telomeres [112–114]. Changes in epigenetic pattern and nucleosomal organization also occur upon deprotection of telomeres [115–118]. Is the detection of DNA damage affected by the unique telomere chromatin structure? How does the telomere chromatin structure influence the efficiency of excision repair in terms of both the access to damaged sites and the repair synthesis step? Does efficient repair require remodeling of telomere chromatin, and if so, how does such remodeling impact telomere stability? On the other hand, at least NER induces chromatin remodeling. Upon completion of NER, new H3.1 histones are incorporated at sites of newly synthesized DNA [119–121]. How does new histone incorporation impact telomere maintenance? Moreover, one recent study demonstrates that in human cells, telomeres spatially interact with specific chromosome regions and form high-order chromatin structures via chromosome looping [122]. Such chromosome looping brings the telomere close to genes up to 10 Mb away from the telomere when telomeres are long, and looping disappears when telomeres become short [122]. Therefore, telomere shortening likely alters global chromatin organization including regions distal from chromosome ends [122]. Then, does this high-order chromatin structure impact the recognition and repair of DNA lesions residing within the structure? Answers to all these questions will uncover the intricate relationship between telomere maintenance and DNA repair activities.

Lastly, given the unique role of telomeres in protecting genome stability, excision repair activities need to be tightly regulated to avoid unwanted repair activities. Although NER, BER, MMR are presumed to take place at telomeres, how these repair activities are regulated at telomeres remain entirely unknown. Emerging evidence suggests that TRF1 and/or TRF2 serve as a platform for recruiting various repair proteins to telomeres, and meanwhile, these and other shelterin proteins may play a regulatory role in restricting repair to a certain level. Further research is needed to understand the exact mechanism of excision repairs at telomeres, the consequences of excision repair deficiency on telomere stability, and the relation of excision repair to telomere function in cancer, premature aging, and other diseases.

Highlights.

• Telomeres, the unique DNA-protein structures located at the ends of chromosomes, play an important role in protecting genome integrity

• Deficiency in various BER proteins accumulates oxidative lesions at telomeric sequences, disrupting telomere length homeostasis and causing telomere dysfunction

• NER factors play pivotal roles in maintaining telomere integrity, although it is unknown how the NER pathway functions at telomeres

• MMR proteins are involved in suppressing telomeric DNA recombination and modulating telomeric DNA damage signaling