Telomeres expand sphere of influence: emerging molecular impact of telomeres in non-telomeric functions

Abstract

Although the impact of telomeres on physiology stands well established, a question remains: how do telomeres impact cellular functions at a molecular level? This is because current understanding limits the influence of telomeres to adjacent subtelomeric regions despite the wide-ranging impact of telomeres. Emerging work in two distinct aspects offers opportunities to bridge this gap. First, telomere-binding factors were found with non-telomeric functions. Second, locally induced DNA secondary structures called G-quadruplexes are notably abundant in telomeres, and gene regulatory regions genome wide. Many telomeric factors bind to G-quadruplexes for non-telomeric functions. Here we discuss a more general model of how telomeres impact the non-telomeric genome – through factors that associate at telomeres and genome wide – and influence cell-intrinsic functions, particularly aging, cancer, and pluripotency.

Introduction

The terminal ends of linear chromosomes called telomeres comprise nucleoprotein complexes, critical for protecting telomere–telomere fusion and DNA-damage-induced genome instability [1]. Human telomeres comprise 5′-TTAGGG-3′ repeats ranging from 3 to 15 kb including ~150–300-bp 3′ single-strand overhang at the end that aids formation of the T-loop protecting telomeres from the cellular DNA-damage-repair machinery [2]. Notably, telomeric repeats were found to also form secondary structural motifs called G-quadruplexes (see Glossary) (also known as G4s) [3] both in solution and in vivo [4]. Further, telomeric G4s were reported to associate with telomeric factors that are critical for telomere homeostasis, implicating G4s in telomere protection or uncapping [5,6].

Glossary.

G4-binding ligands

small molecules that can specifically bind to G-quadruplex structures.

G-quadruplex

a DNA secondary structure formed by G-rich sequences.

Interleukin-1 receptor-1

is a cytokine receptor for IL1A, IL1B, and IL1RN. A mediator of inflammatory and immune responses.

Induced pluripotent stem cells (iPSCs)

mimic ESCs. Somatic cells are induced to form pluripotent cells by expressing the pivotal genes Oct4, Sox2, Klf4, and cMyc (Yamanaka factors).

Myb domain

a helix-turn-helix DNA-binding domain present in several eukaryotic transcription factors; also found in TRF1 and TRF2 proteins.

Potential G4s (PG4s)

computationally identified regions in the genome containing four consecutive guanine tetrads interspersed with 1–15 nucleotides, which can theoretically adopt a G-quadruplex structure.

PRC2

a multiprotein repressor complex. This methyltransferase methylates H3 on lysine 27 (H3K27me3).

Replicative senescence

an event in which, beyond a certain number of divisions, cells undergo irreversible cell cycle arrest.

RGG motif

arginine (R)-glycine (G)-glycine (G) amino acid repeats found in G-quadruplex-binding proteins; crucial to the G4-binding affinity of those proteins.

Telomere looping

occurs when telomeres loop back to the chromatin (~10 Mb away) inducing chromatin compaction.

Telomere position effect (TPE)

repression of genes that are proximal to the telomeres.

Telomeric repeat amplification protocol (TRAP)

assay determining the amount of active telomerase holoenzyme by exploiting the enzyme’s ability to elongate telomeric repeats templates.

TERC

the RNA component of the telomerase holoenzyme. It acts as template for telomere synthesis during DNA replication.

Telomeres widely influence crucial functions that impact cell and organismal biology, including cancer, pluripotency, and aging/senescence. Intriguingly, however, understanding of telomere-dependent molecular mechanisms is largely limited to the subtelomeres (~10 Mb from telomere ends). This is typically through telomere looping, subnuclear compartmentalization, and/or post-translational modifications (PTMs) of telomeric factors (reviewed in [7]). Therefore, how telomeres exert influence beyond subtelomeres at the molecular level – an issue possibly central to many telomere-related functions – remains poorly understood. Understanding from two some-what distinct emerging areas might offer opportunities to bridge this gap.

First, several telomere-associated factors (TAFs) [comprising the multiprotein complexes shelterin (TRF1, TRF2, RAP1, TIN2, TPP1, and POT1) and CST (CTC1, TEN1 and STN1), the reverse transcriptase enzyme TERT along with its RNA template TERC, and telomeric-repeat-containing noncoding RNA (TERRA) (Figure 1)] were reported to interact with DNA and/or other regulatory factors outside telomeres. Second, the presence and genomic distribution of G4s is of interest. G4s were first reported to be present at telomeric ends in the late 1980s [8]. Later, G4s were noted to be enriched in gene regulatory regions throughout the genome [9–11], and a growing body of evidence shows that G4s are involved in gene regulation including epigenetic modifications [12]. Further, the function of TAFs like TRF2 and RAP1 was found to be dependent on association with non-telomeric G4s [13–15]. Together, these observations raise questions with important implications. Why are G4s abundant in distinct regions of the genome: telomeres and gene regulatory regions? With these in mind, we describe here recent work on how telomeres influence function at remote (non-telomeric) sites and how non-telomeric and/or telomeric G4s could be involved.

Figure 1. Structural organization of human telomere.

The 3′ single-strand telomeric overhang comprising TTAGGG repeats folds back into the double-stranded telomeric DNA. This results in the T-loop and D-loop. The multiprotein shelterin complex binds telomeric DNA and protects telomere ends from the DNA-damage-repair machinery preventing telomere–telomere fusions, controls telomere homeostasis, and stabilizes the overall structure. Telomeric-repeat-containing noncoding RNA (TERRA) can form RNA G-quadruplexes and bind to telomeric DNA forming DNA–RNA hybrids and R-loops.

TAFs in non-telomeric roles

Non-telomeric functions of RAP1

Recent findings reveal many non-telomeric functions of TAFs (summarized in Table 1). ChIP-seq in mouse embryonic fibroblasts (MEFs) revealed RAP1 binding to both telomeric and non-telomeric sequences. More than 30 000 non-telomeric sites were identified; ~70% of these mapped in intragenic regions or in the near vicinity of coding regions, suggesting a function of non-telomeric RAP1 in gene regulation. This was further substantiated by deregulated expression of genes harboring RAP1-binding sites in RAP1-deficient MEFs [13]. Non-telomeric RAP1 was also reported to impact the senescence of human mesenchymal stem cells (HMSCs) through epigenetic regulation of RELN, a negative regulator of proliferation. Hypermethylation of the RELN promoter was noted in RAP1-deficient HMSCs, which was rescued on the expression of RAP1 exogenously (Figure 2A) [16]. In addition, RAP1-mediated regulation of PGC1α and PPARα was reported in the cellular metabolism of obesity [17,18]. Independent of DNA binding, RAP1 interacts with IκB kinase (IKK)-activated nuclear factor kappa B (NF-κB) [19] (Figure 2B).

Table 1. Non-telomeric functions of TAFs.

TAF	Function	Cell type	Gene associated	Phenotype	Refs
TRF1	Epigenetic regulation	iPSC	MYC, SOX2, NANOG, BMP	Maintenance of stemness	[35,36,48]
TRF1	Centromeric localization of Aurora B kinase			Maintenance of microtubule–kinetochore attachment	[38]
TRF1	Nek2 directly binds and phosphorylates TRF1 leading to abnormal chromosomal alignment at metaphase	Breast cancer cells – MDA-MB231 and MCF7		Cytokinetic failure and aneuploidization	[39]
TRF2	Transcriptional regulation		PDGFRβ	Regulation of angiogenesis	[21]
TRF2	Transcriptional and epigenetic regulation	Patient-derived glioblastoma	hTERT	Suppresses reactivated human telomerase in glioblastoma	[24]
TRF2	Regulation of interleukin signaling and tumor-associated macrophage (TAM) infiltration in triple-negative breast cancer (TNBC)	Patient-derived TNBC tissue	IL1R1	Antitumor immune response	[25]
TRF2	Transcriptional and epigenetic regulation		PCGF3, OPN4, P21, ANXA2, CHRM2, INHA, KCNH2, OBSL1, SAMD14, SMAD7, THRA	Regulation of gene expression	[20,22,23]
TRF2	DNA damage repair, migration potential maintenance	Cal27 and SCC-131	CD44, OCT4, SOX2, KLF4, c-MYC	Maintenance of oral cancer stemness	[32]
TRF2	Association with lamin proteins	IMR90s	Genome-wide interstitial sites (ITSs)	Chromatin organization	[30,31]
TRF2	Stabilization of REST and hREST protein	SH-SY5Y, H1, and H7		Maintenance of neuronal stemness and formation of NPCs from ESCs	[33,34]
TRF2	Transcriptional regulation	Colon cancer	HS3ST4	Regulates NK cell density and mobilization	[27]
TRF2	Transcriptional regulation for recruitment and activation of MDSCs	B16F10 melanoma and BJcl2 41ANs	GPC6, VCAN, and HS3ST4	Promotes tumor growth via immune suppression	[28]
TIN2	Regulation of oxidative phosphorylation, glucose metabolism	HT-1080 and U2OS		Mitochondrial morphology	[51]
RAP1	Transcriptional regulation	MEF	HS3ST1, NNMT, CAR6, IGF2	Metabolism dysregulation in RAP1 knockout mice	[13]
RAP1	Epigenetic regulation	HMSCs	RELN	Senescence	[16]
RAP1	Transcriptional regulation	Mouse hepatic and adipose tissue	PCG1α, PPARα	Obesity	[17,18]
RAP1	Regulation of IKK-mediated phosphorylation of NF-κB	MDA-231, HeLa, A549		Tumor cell survival and invasion	[19]
TERT	Cofactor of β-catenin transcriptional complex	HeLa, mouse small intestine cells	AXIN2, MYC, HPRT	Cell proliferation, survival, and stemness	[42]
TERT	Interacts with p65 and regulates expression of inflammatory genes	Mouse-derived fibroblasts	NF-κB-regulated genes – IL-6, IL-8, TNF-α	Cell proliferation, immune regulation	[43]
TERT	Interacts with RNA Pol III	hESC-1, A2780, HCT116, P493, BLM, LOX-IMVI	Binding at rRNA- and tRNA-encoding genes in a cell-type-specific manner	Promotes cell proliferation	[44]
TERRA	Transcriptional regulation	PC-3, HBC4, MKN74	STAT1, ISG15, OAS3	Tumor suppression	[47]

Figure 2. Telomere-associated factors (TAFs) in non-telomeric roles.

(A) RAP1 positively regulates RELN transcription. (B) Extranuclear localization and function of TAFS: Mitochondrial localization of TIN2 causes negative regulation of oxidative phosphorylation. Cytoplasmic RAP1 regulates nuclear factor kappa B (NF-κB) signaling. TRF2 represses CDKN1A (p21) and hTERT and activates PCGF3, PDGFRβ, and IL1R1.

Non-telomeric functions of TRF1 and TRF2

Besides RAP1, TRF1 and TRF2 were reported to bind to ~180 interstitial sites [13–15]. Recently, TRF2 binding was reported at thousands of non-telomeric sites [20], and TRF2-meditated regulation of several genes was observed (described in the following section) [21–25] (Figure 2C). Non-telomeric TRF2 was also shown to regulate vascularization through VEGF-A [26] and PDGFR-β [21], to reduce the recruitment and activation of natural killer (NK) cells [27], and to induce the recruitment of myeloid-derived suppressor cells (MDSCs) that promote tumor growth through immunosuppression [28].

Further, TRF2 induced higher-order DNA looping and chromatin compaction [29] and 3D looping of telomeres with interstitial chromatin was stabilized by TRF2–lamin association [30,31]. TRF2 silencing reduced expression of the cancer stem cell markers Oct4, Sox2, KLF4, and c-Myc [32] and computational modeling suggested direct interaction of KLF4 with TRF2 [32]. Sequestration of the RE-1 silencing factor REST in the cytoplasm by a TRF2 splice isoform results in the activation of REST target genes resulting in differentiation of neural progenitor stem cells (NPCs) [33]. The splice isoform of REST in humans, hREST4 was reported to interact with TRF2 and play a vital role in neural differentiation [34]. Similarly, TRF1 was found to be essential for induced pluripotent stem cell (iPSC) proliferation and directly regulated by Oct3/4 [35]. Accordingly, small-molecule inhibitors of TRF1 reduced the efficiency of reprogramming in mice [36]. Furthermore, the role of telomere length and the functions of TAFs in maintaining the pluripotent or ‘stem-like’ state was reviewed recently [37]. In addition, TRF1 was reported to positively regulate Aurora B’s centromeric function during chromosomal segregation [38] and interaction of TRF1 with Nek2, a cell cycle regulator, induced cytokinetic failure [39].

Non-telomeric functions of telomerase and other TAFs

Although initially only two molecules per cell of hTERT were reported [40], later this was found to be ~60–100 000 molecules per cell supporting the possibility of non-telomeric functions [41]. Importantly, transcriptional regulation of the Wnt/β-catenin genes by hTERT was shown to affect cancer progression [42]. Further, hTERT was found to interact with p65 regulating the expression of NF-κB target genes [43] and with RNA polymerase III (RNA Pol III) to enhance tRNA expression [44,45].

RNA-FISH and CHIRT-seq in mouse embryonic stem cells (mESCs) revealed TERRA-binding sites throughout the genome, implicating another TAF in non-telomeric function. TERRA was reported to regulate the expression of thousands of target genes [46] including the innate immunity genes like STAT1, ISG15, and OAS3 in human cancer cells [47]. TERRA deletion triggered differentiation; overexpression enhanced the self-renewal of mesenchymal stem cells [16]. TERRA maintained the transcriptional landscape of ESCs in a TRF1-dependent manner [48] and aggregation of TERRA was noted in mouse medulloblastoma and proliferating neural progenitor cells [49].

The role of shelterin proteins in noncanonical functions in cancer progression/initiation was reviewed recently [50]. Interestingly, mitochondrial localization of another TAF, TIN2, (Figure 2B) and TIN2-mediated regulation of reactive oxygen species leading to HIF-1 activation were implicated in cancer [51,52]. In addition, non-telomeric function of TAFs was reported in yeast and zebrafish. In yeast, RAP1 binds to promoters and represses histone-coding genes on telomere shortening resulting in the loss of histone-mediated suppression of senescence genes [53,54]. In zebrafish, an ortholog of TRF2 was noted to regulate the expression of neurodevelopment-and aging-related genes [55,56].

Emerging impact of telomeres on non-telomeric functions

Telomere position effect (TPE)

Telomere-dependent suppression of genes, known as the TPE, was first noted in Saccharomyces cerevisiae in URA3, TRP1, HIS3, or ADE2 inserted ~4.9 kb from telomeres. This was due to telomeric heterochromatinization of the adjacent regions [57]. Silencing of genes in the subtelomeric region through the TPE was widely observed in diverse species like Trypanosoma brucei, Plasmodium falciparum, Schizosaccharomyces pombe, Drosophila melanogaster, Pneumocystis carinii, and Candida glabrata [58], including human carcinoma HeLa cells [59].

Later intrachromosomal looping – where telomeres loop to regions ~10 Mb away – was observed in human myoblasts. This induced the repression of genes near telomeres (ISG15, DSP, and C1S), further extending observations on telomere-dependent gene expression, and was called TPE over long distance (TPE-OLD) [60] (Figure 3). Telomere shortening disabled the physical looping of telomeres leading to activation of the genes [60]. The transcription regulation hTERT (located ~1.3 Mb from telomeres) was described by TPE-OLD where the chromosome 5p telomere interacts with the hTERT loci through chromosomal looping in human cancer and immortal cell lines [61]. Notably, TRF2 knockdown disrupted the looping interaction, suggesting the involvement of telomere-bound TRF2 in the interaction of telomeres with hTERT [61].

Figure 3. Telomere position effect over long distance (TPE/TPE-OLD).

Long telomeres loop to subtelomeric regions. This physical association transcriptionally represses genes present in subtelomeres. Looping is lost when telomeres shorten and genes are activated.

Telomere-dependent expression of genes remote from telomeres: the telomere-length-dependent partitioning (TeLPR) model

A mode of telomere-dependent gene expression distinct from TPE and TPE-OLD was recently reported. The expression of genes remote from telomeres (>60 Mb), and spread throughout the genome, altered depending on the length of telomeres [62]. Mechanistically, this was due to increased/decreased non-telomeric TRF2 binding within promoters in cells with relatively short/long telomeres, respectively. Altered TRF2 binding resulted in direct transcriptional regulation of the target gene [62]. Telomere elongation increased TRF2 occupancy at the telomeres; concomitantly, TRF2 occupancy at the non-telomeric sites was reduced. This was in cells with otherwise isogenic backgrounds where TRF2 levels remained relatively unaltered. Moreover, it was shown that promoter TRF2 occupancy induced histone modifications for either permissive (H3K4me3 and H3K4me1) or restrictive (H3K27me3) chromatin marks [62]; for instance, regulation of p21, the cyclin-dependent kinase [also known as cyclin-dependent kinase inhibitor 1 (CDK-1)] or the interleukin-1 receptor-1 IL1R1 described in the following section. Together these suggest a model where TeLPR results in differential TRF2 occupancy between telomeric versus non-telomeric sites (Figure 4). TeLPR interestingly implicates not only telomere-dependent gene expression but also other functions through TRF2, and/or other TAFs, found outside telomeres.

Figure 4. Telomere-length-dependent partitioning (TeLPR).

Relatively long telomeres sequester TRF2 leading to less occupancy at non-telomeric sites. Shorter telomeres, by contrast, result in increased TRF2 binding to promoters. Increase/decrease in promoter-bound TRF2 affects transcription through chromatin modifications.

Independently, the association of non-telomeric RAP1 with IKK was shown to result in the activation of NF-κB [19]. It was further posited that, in cells with short telomeres, the relatively enhanced availability of non-telomeric RAP1 would give enhanced NF-κB signalling [63]. This is consistent with the TeLPR model broadening its scope to include other TAFs.

Notably, non-telomeric TRF2 sites included telomere-like guanine-rich-repeat sequences that could potentially form G4s [20], implicating interstitial G4s in the recruitment of non-telomeric TRF2. Since this would significantly contribute to the TeLPR model, later we discuss studies on telomeric and non-telomeric G4s, particularly, with regard to their role in findings implicating telomeres in functions that are outside telomeres.

G-quadruplex in telomeres

Initial analysis of synthetic guanine-rich sequences derived from the telomeres of Tetrahymena thermophila showed that these telomeric repeat sequences can form intramolecular G4s through non-Watson–Crick G·G base pairing [8] (see Box 1 for a detailed explanation of G4 structure). Human telomere sequences also adopted G4 structure in vitro [64,65]. Later, further studies showed that G4 formation is a general property for telomeric sequences from 15 different species [66]. Eukaryotic telomeres contain 3′ single-strand DNA overhangs. Approximately 150–300 nucleotides long in human telomeres, the overhang may fold back onto telomeres by strand invasion forming a lariat-like structure known as the T-loop [2]. Long single-stranded DNA comprising 5′-TTAGGG-3′ repeats has been used to mimic telomeric overhang in vitro, which condensed into a beaded filament-like structure stabilized by G4-binding ligands, suggesting G4 formation in the telomeric overhang in vivo [67]. Furthermore, TERRA (5′-UUAGGG-3′ repeats) has also been demonstrated to form multiple juxtaposed G4s with three-nucleotide UUA linkers [68]. TERRA hybridizes with the C-rich strand of telomeric DNA forming R-loops (Figure 1). In the ciliated protozoan Stylonychia lemnae, the presence of G4 in vivo was revealed using an antibody (Sty49) that was specific to G4s. Consistent with these, the G4-specific antibodies BG4 and 1H6 localized to the telomeres in human cells, supporting the formation of telomeric G4s in cells [69,70]. The presence of G4s in telomeres, their importance in maintaining genome stability, and targeting for cancer therapeutics have been reviewed [71,72]. TRF2 was reported to bind to telomeric G4s and mediate the interaction of TERRA with telomeres [73,74]. In ciliates, telomere-end-binding proteins regulated the formation of G4 structures in vivo [75]. In vitro observations showed unfolding of the telomeric G4 by multiple TAFs like POT1, TPP1, and the replication protein A [76,77]. In addition, heterogeneous nuclear ribonucleoprotein A1 (hnRNPA1) and unwinding protein-1 (UP1) were also noted to unwind telomeric G4s [78]. Together, these studies support the presence of telomeric G4s and their interaction with intracellular factors.

Box 1. G-quadruplex: intracellular detection.

G-quadruplexes (or G4s) are bookshelf-like structures formed by stacked planar guanine tetrads. A single-stranded DNA or RNA sequence comprising G-repeats can fold into the G4 through the formation of stable Hoogsteen-bonded tetrads (Figure I) [67]. As it involves single-stranded DNA, G4 formation is likely during replication or transcription when the double helix is in a relatively unwound form and/or within the single-strand overhang at the end of eukaryotic telomeres [130]. In addition, the C-rich complementary strand is known to be bound and stabilized by single-strand-binding proteins [131,132]. Based on the orientation of the folding, G4s can be parallel, antiparallel, or mixed (parallel–antiparallel), and depending on the number of strands involved G4s can be intra- or intermolecular. Detection of G4s inside cells started with computationally identified DNA sequences that contributed to PG4s in the genome [92,93]. G4-specific antibodies like sty49 and sty3 were found to bind to telomeric G4 in Stylonychia lemnae macronuclei [133,134]. Later, another G4 antibody, BG4, was reported to bind G4s inside human cells in regions outside telomeres supporting computational predictions [69]. In addition, intracellular fluorescent and radioactive G4-binding ligands [87] were used to detect G4s (extensively reviewed in [135]).

Intracellular G4s are likely to be predominantly protein bound [6,136]. Antibody-based intracellular detection therefore remains challenging as the antibody might be excluded by proteins that are already G4 bound. In addition, it is possible that the high affinity of the antibody promotes the formation of G4s that are otherwise not present. Although BG4-based ChIP-seq has been widely reported, BG4 was used outside cells (cell-free buffer) on sheared chromatin [137]. Therefore, the possibility of G4 formation or deformation during the isolation and purification of DNA, leading to the identification of G4s that are not intracellular, cannot be completely ruled out. With these in mind, an alternative approach used an antibody towards a protein that binds to G4s inside cells: TRF2 ChIP-seq in human cells detected thousands of TRF2-bound G4s throughout the genome [20]. Similar experiments with other G4-binding factors might be useful in revealing a larger repertoire of intracellular G4s.

Figure I. Structure of G-quadruplex (G4).

(A) Sequence pattern of potential G4 sequence with varying loop length. (B) Planar guanine tetrad formed by Hoogsteen hydrogen bonding. (C) Basic topology of the G4 structure.

G-quadruplexes and telomerase

The discovery of the telomerase enzyme raised hopes in oncology and regenerative medicine. Telomerase is not only a tumor marker, but also a potential target, since its activity is essential for the long-term proliferation of cancer cells; most transformed cells in vitro can bypass progressive telomere loss and replicative senescence by activating telomerase. Hahn et al. showed that inactivation of telomerase in cancer cells was sufficient to limit the proliferation and growth of tumor cells [79]. Telomerase activity was therefore initially considered a promising target for the design of anticancer compounds.

G4 formation at the G-rich 3′ overhang of human telomeres (comprising multiple repeats of the GGGTTA hexanucleotide) generally inhibits telomerase extension in vitro, and this inhibition is reinforced if this G4 is stabilized by interaction with specific ligands. The G4 structure allows the design of small molecules that recognize them selectively, which leads to indirect inhibition of telomerase [80]. This observation stimulated the quest for (telomeric) G4 ligands and a number of natural or synthetic compounds have been shown to be active in this way (for a review see [81]). There are now over 3000 different compounds known to interact with G4s [82]. Many, but not all, of these ligands share structural characteristics such as a large, flat aromatic surface, allowing stacking on a terminal quartet, and positive charges favoring electrostatic interactions. The introduction of many positive charges may result in a high-affinity ligand, but with limited specificity, as found for core-extended naphthalene diimide (cex-NDI) derivatives, in which diethylene glycol (DEG) side chains were replaced by positively charged substituents [83]. There is therefore a trade-off to find between affinity and selectity.

Unfortunately, although there are more than 100 clinical trials mentioning telomerase on ClinicalTrials.gov, there is still no validated cancer treatment that directly targets telomerase activity [84]. As feared from the outset, the inhibition of telomere extension by telomerase alone is probably not sufficient to lead to a significant antitumor effect; its association with other treatments is undoubtedly necessary. In addition, the antitelomerase effect of G4s is not as straightforward as initially expected.

First, quadruplexes may help to recruit telomerase rather than inhibit it [85] and only a few molecules interfere with the processivity of telomerase. G4 ligands were noted to be more effective inhibitors of telomeric DNA amplification than extension by telomerase and therefore may not be considered simple telomerase inhibitors. The telomeric repeat amplification protocol (TRAP) assay is inappropriate for the determination of telomerase inhibition by G4 ligands and the inhibitory effect of many G4 ligands has often been overestimated [86].

Second, many G4 sites are present in other regions of the genome, and the antiproliferative effect of some of these compounds could be explained by their action on other sites, such as the promoter of oncogenes. Compounds that exclusively target telomeric G4s are yet to be discovered. For example, a radiolabeled bisquinolinium compound was shown to bind to the terminal parts of chromosomes in metaphase spreads, but binding sites in internal chromosomal regions were also observed [87]. The polymorphic nature of telomeric G4s poses further challenges for the design of telomeric-G4-specific ligands [88]. Molecules that target the hTERT promoter specifically, by contrast, might be interesting, as hTERT repression would affect telomere maintenance. Further, perhaps, strategies that target molecules/factors in a locus-specific way (e.g., as in CRISPR-guided function) could be meaningful.

Finally, G4 ligands are also active in ALT cells, in which telomerase activity is absent; for example, the sensitivity of various osteosarcoma cell lines to a G4 ligand is the same whether these cells are ALT positive or telomerase positive, supporting the idea that G4 ligands are not simple telomerase inhibitors [89].

Non-telomeric G-quadruplexes and G-quadruplex-mediated gene regulation

Outside telomeres, guanine-repeat sequences that formed potential G4s (PG4s) were noted in immunoglobulin switch regions [90], in the promoter of the oncogene c-myc [91], and later throughout the human genome [9]. A key finding in 2006 showed that PG4s were predominant in the promoters of Escherichia coli genes; importantly, further analysis revealed PG4s to be evolutionarily conserved in thousands of microbes in orthologously related genes [10,92,93]. Based on these, authors proposed global G4-mediated gene regulation [10]. Furthermore, PG4 sequences were found to be conserved in rat, mouse, chimpanzee, and human promoters, underscoring the significance of G4s in gene regulation [92], which was supported by genome-wide alteration in gene expression in the presence of intracellular G4-binding ligands [94].

Notable in this regard was the description of the intracellular G4-binding protein non-metastatic factor NME2. Interaction of the c-myc promoter G4 with NME2 was observed to be critical for the regulation of c-myc. Destabilization of the G4 gave reduced NME2 binding at the promoter resulting in altered transcription of c-myc in human cancer cell lines [95]. In addition to NME2, c-myc was reported to be regulated by nucleolin, cellular nucleic-acid-binding protein (CNBP), and poly (ADP-ribose)-polymerase-1 (PARP1) through G4 interactions [95–98]. Several other reports support G4-mediated gene regulation: NME2–G4 association in the regulation of hTERT [99]; interaction of PARP-1 with promoter G4s in the regulation of c-Kit, KRAS, and PARP-1 itself [100–102]; myc-associated zinc finger (MAZ) binding to the KRAS promoter G4 [103]; and the zinc-finger protein SP1 binding to the c-kit promoter [104], including computational analysis that revealed enrichment of SP1 binding sites at many promoter PG4s [105,106]. Interestingly, a recent analysis curated 77 G4-binding proteins and reported the glycine/arginine-rich RGG motifs to be important for interaction with G4-forming sequences [107,108]. Besides these, RecQ family helicases BLM, WRN, FANCJ, G4-resolvase-1, and Pif1 helicases were shown to unwind G4s and implicated in the facilitation of replication [109–113]. Together these findings support the genomic presence of G4s and underline the function of G4s in gene regulation in association with regulatory factors. Additionally, PG4s in 5′ untranslated regions (UTRs) in mRNA were shown to regulate alternative splicing, suggesting a role in translation control [114,115].

TRF2 binding to telomeric as well as non-telomeric sites genome wide

TRF2 comprises two DNA-binding domains: a C-terminal Myb domain and a RGG glycine/arginine-rich basic N-terminal domain. DNA binding by TRF2 was noted at sites/motifs that were not limited to the canonical telomeric TTAGGG repeats [116]. TRF2 ChIP-seq using human cancer cells found >20 000 TRF2 sites spread throughout the genome: A majority (38%) of the non-telomeric TRF2 sites were found to harbor G-repeats [20]. Similarly, earlier TRF2 ChIP-seq [14,15] – which recovered ~180 non-telomeric TRF2 sites due to lower read depth and/or antibody efficiency [20] – also found most binding motifs to be G-repeats. Together these show that the majority of TRF2 binding sites include G-repeats, a common feature of telomeric and non-telomeric TRF2 binding. Intriguingly, however, besides the guanine bias, there was no similarity in the compositions of the non-telomeric sequence motifs [20], suggesting that TRF2 binding might be dependent on the G4 DNA structure comprising G-repeats instead of a conserved sequence. Consistent with this, an affinity of recombinant purified TRF2 for both telomeric and non-telomeric G4s has been reported [20,24,73].

The role of TRF2 in T-loop formation at telomeres and telomere protection has been widely studied [2,117]. Of note, recent work, however, showed that T-loops can form independent of TRF2, and TRF2-mediated telomere protection is not essential in mESCs [118,119], indicating the possibility of TRF2 telomeric functions that are dependent on cell type or the developmental stage of cells.

Transcription regulation by non-telomeric TRF2 is G-quadruplex dependent

TRF2 was reported to bind to G4s in the proximal promoter of the human telomerase gene (hTERT) [24]. The TRF2–G4 interaction was necessary for the recruitment of the RE-1-silencing-factor REST and the canonical repressor complex PRC2. This induced histone H3K27 trimethylation at the hTERT promoter leading to epigenetic repression of hTERT in both cancer and normal human cells in a TRF2–G4-dependent way [24]. Furthermore, a large number of human cancers [120], particularly glioblastoma and melanoma [121,122], with two somatic guanine mutations in the hTERT promoter (−124G>A and −146G>A, 124 and 146 nucleotides upstream of the transcription start site of hTERT), were found to be clinically prevalent and causally implicated in hTERT reactivation. It was shown that the two mutations in G-repeats resulted in disruption of promoter G4s compromising TRF2 binding, and resulting in hTERT reactivation. Stabilization of the promoter G4 using G4-binding ligand or by reversing the mutation from A to G using CRISPR in cells reinstated TRF2 binding on the hTERT promoter. This, in turn resulted in re-suppression of hTERT in glioblastoma cell lines including patient-derived primary glioblastoma cells that were treated with intracellular G4-binding ligands [24,123].

In addition, GA-binding protein α/β (GABPA) was shown to be specifically recruited to the mutant hTERT promoter and mediate long-range chromatin interaction between the promoter and a region 300 kb upstream (T-INT) of the promoter [124]. This enhanced hTERT transcription through enriched active histone marks and POL2 recruitment to the promoter. Therefore, loss of TRF2 binding, and increased GABPA recruitment, possibly sequentially, at the mutant hTERT promoter might result in hTERT reactivation. In cancer cells devoid of the hTERT promoter mutation, Jun-D-mediated CTCF recruitment regulated the long-range hTERT interaction with T-INT [125].

TRF2 binding genome wide is telomere dependent

TRF2 binding at the promoter and suppression of p21, crucial for the linking of cycle arrest to DNA damage, was reported to be G4 dependent [23]. Furthermore, interestingly, the regulation of p21 through TRF2–G4 interaction was dependent on telomere length [62]. In cells with relatively long telomeres, TRF2 binding on the p21 promoter was lower than in cells with short telomeres, with otherwise isogenous backgrounds. This resulted in relatively high expression of p21 in cells with long telomeres. Mechanistically, TRF2-induced recruitment of the epigenetic repressor complex REST/LSD1 was attenuated in cells with longer telomeres leading to reduced H3K27me3. Together these contributed to restrictive/permissive chromatin at the p21 promoter in a telomere-dependent fashion.

Recent work further shows the regulation of IL1R1 expression to be telomere dependent through TRF2 binding at the IL1R1 promoter [25]. Interaction of TRF2 with the promoter G4 was necessary for recruitment of the histone acetyl-transferase p300 leading to increased acetylation at histone K27 and the induction of IL1R1. In the case of cells with relatively short telomeres, TRF2 binding at the IL1R1 promoter was greater than with long telomeres, consistent with the TeLPR model. Consequently, IL1R1 expression was enhanced in cancer cells, with short vis-à-vis long telomeres resulting, interestingly, in telomere-dependent activation of interleukin-1 signaling [25].

It is of interest to note that at the hTERT or the p21 promoters, TRF2 promotes repressive chromatin through interaction with REST and canonical repressor complexes [23,24]. By contrast, at the IL1R1 promoter TRF2–p300 interaction induces active chromatin [25]. This is likely to be due to distinct PTMs of TRF2 at the respective promoters: we noted acetylation at the 293-lysine residue in the case of the IL1R1-promoter-bound TRF2 [25]. In the case of the hTERT promoter, by contrast, methylation at the TRF2 arginine-17 residue was evident (Sengupta et al., unpublished). An earlier paper described non-telomeric TRF2 to be predominantly methylated vis-à-vis telomeric TRF2 [126]. Overall, these findings support the notion that interstitial binding of TAFs would require extensive interactions with histone modifier/other regulatory cofactors. While further work will be necessary for TRF2, and other TAFs like RAP1/TRF1, it is possible that PTMs distinguish telomeric and non-telomeric binding/function.

Importantly, these emerging findings suggest a broader interplay where TRF2 binding to telomeric and non-telomeric G4s is in line mechanistically with the TeLPR model. Furthermore, other TAFs that associate with interstitial regions might function in a similar mode.

Concluding remarks and future perspectives

In conclusion, emerging evidence increasingly shows that telomeres or TAFs can influence gene expression and other functions throughout the genome. For instance, TRF2 binds to promoters throughout the genome (e.g., genes ~60 Mb away from the nearest telomere) and regulation of the target genes was telomere dependent [62]. Telomere-dependent binding of non-telomeric TRF2 (the TeLPR model described earlier) posits signaling from the wings to the entire chromosome [127]. G4s in telomeres and the enriched presence of G4s in gene regulatory regions genome wide is notable contextually. The binding of TAFs like TRF2 to G4s, both telomeric and non-telomeric, therefore supports intrinsic genetic links between telomeres and the non-telomeric genome through G4s. Together with further work, this conceptual advance might strengthen the understanding that incremental telomeric shortening impacts biology across the genome in a continuous and gradual fashion. This therefore modifies the prevailing broad understanding that cellular triggers – for instance, replicative senescence – are activated when telomeres reach a critically short limit.

Further, as novel non-telomeric functions of TAFs are discovered, an understanding of the mechanistic details of how (and to what extent) telomeres influence non-telomeric biology would be of interest (see Outstanding questions). Here the newly released Telomere-to-Telomere (T2T) genome sequence is expected to be of significance [128]. Re-analysis of ChIP-seq with the current genomic data may reveal novel binding sites, especially comprising repetitive DNA, which was relatively poorly covered until recently and may reveal novel G4s [129]. Together these will further deepen our understanding of how telomeres impact non-telomeric functions.

Outstanding questions.

What are the consequences of the TeLPR model – which shows how non-telomeric function is telomere dependent – in cellular events like differentiation and neoplastic progression that are closely associated with change in telomere length? Could insights from TeLPR aid in better understanding of aging-related disorders?

Do the telomeric shelterin proteins undergo PTMs for non-telomeric binding?

Can shelterin complex proteins undergo different PTMs across cell types? More importantly, do they vary in stem cells vis-à-vis the differentiated state?

Can DNA topology, the G-quadruplex structure in particular, be a key element in the regulation of cellular events associated with altered telomere length?

Highlights.

Shelterin proteins are known for telomeric functions only. Emerging evidence shows non-telomeric presence and function of shelterin factors.

DNA secondary structure G-quadruplex motifs are enriched in gene promoters. Recent work shows how G-quadruplex-dependent non-telomeric binding of telomere-associated proteins regulates transcription.

Interstitial G-quadruplex motifs associate with telomeric protein in a telomere-length-dependent fashion.