Human linear chromosomes terminate with TTAGGG repetitive sequences lasting up to 20 kilobases, and these DNA repeats together their binding factors (a six protein-containing shelterin) form special DNA-protein structures so-called telomeres (1–4). Telomeres function as protective caps to maintain genomic stability and integrity by preventing nucleolytic degradation, illegitimate chromosomal recombination or fusion, and DNA damage response/repair (1–4). Telomeric DNA is synthesized by telomerase, an RNA-dependent DNA polymerase (1–4). Telomerase is generally silent in differentiated human somatic cells, and thus these cells undergo progressive telomere attrition with successive divisions due to the end-replication problem (1–4). When such telomere shortening reaches a threshold length to impair its function, dysfunctional telomeres mimic double-stranded DNA breaks to activate the DNA damage response pathway, thereby triggering cellular apoptosis or stable growth arrest named replicative senescence (2–4). The TP53-CDKN1A and/or CDKN2A-pRB checkpoint signalings have been shown as the major players to initiate the senescence program (1–4). Taken together, telomere attrition serves as a mitotic O’clock, recording times of cellular divisions and controlling cellular lifespan (1–4).

Telomere shortening-mediated senescence is believed as an evolutionary trade-off to protect against cancer (4). Indeed, malignant cells proliferate infinitely, which is an essential cancer hallmark (5). Cancer-specific genomic and epigenetic alterations cooperate to evade senescence during oncogenesis, while stabilizing telomere length is the most important mechanism for cancer cells to acquire the capacity of infinite proliferation (1, 3, 4, 6, 7). In the last decades, numerous studies have undoubtedly demonstrated that activation of telomerase is the commonest strategy through which cancer cells maintain their telomere length and are empowered with an immortal phenotype (1, 3, 4, 6, 7). In accordance, telomerase activity is detectable in the vast majority of human cancers (8).

Telomerase enzyme is a ribonucleoprotein complex with a molecular weight of approximately 500 kDa (gel filtration-based estimate), however, its core holyenzyme is only composed of telomerase reverse transcriptase (TERT), the subunit catalyzing telomeric DNA synthesis, and internal template telomerase RNA component (TERC) (9–11). TERC is ubiquitously expressed in human tissues/cells, while the TERT gene is stringently repressed in most normal human cells, which acts as the key mechanism to silence telomerase (4). It has been well-established that the transcriptional de-repression of the TERT gene is an essential step for transformed cells to acquire telomerase activity during the oncogenic process (4, 7). Great efforts have thus been made to elucidate regulatory mechanisms underlying TERT transcription and its role in cancer development and/or progression.

Urothelial carcinomas (UCs) arise from the urothelium in the urinary track including renal pelvis, ureter and bladder, among which the UC of the bladder (UCB) is commonest, accounting for approximately 90% of all UCs, while the rest 10% are UC of renal pelvis and ureter (UCRP and UCU), collectively called upper track UC (UTUC) (12). The accumulated evidence suggests that UCB and UTUC may represent two distinct disease entities, but they share many common characteristics including morphology, histology, and featured genomic alterations (12). Like other malignancies, the majority of UCBs and UTUCs maintain their telomere length via TERT induction and telomerase activation. Moreover, in vitro experiments even showed that TERT alone was sufficient to immortalize and/or transform normal urothelial cells (13–15). The present review article is focused on telomerase activation/TERT transcription in UCB and UTUC tumors. To provide fresh perspectives, we will summarize recent advances with the emphasis on the altered TERT promoter-mediated telomerase activation in UCs and discuss clinical implications in UC managements.

TERT and telomerase activity is undetectable in primary NHUCs, while expressed at low levels in proliferative NHUCs under culture (13, 14, 16, 17). Little is known about in vivo NHUC senescence, but in vitro cultured NHUCs have been well demonstrated to undergo 20 to 30 population doublings before entering senescence (13, 14). Different from other types of normal human cells such as fibroblasts (18, 19), senescent NHUCs exhibit minimal levels of the TP53-CDKN1A pathway activation and lack of substantial telomere erosion, while up-regulation of CDKN2A expression (14) (Figure 1). Although robust telomere shortening does not happen from presenescent to senescent NHUCs, ectopic TERT expression readily immortalizes presenescent cells, and moreover, the telomere-lengthening function of TERT is required for their infinite proliferation (Figure 1) (13, 14). Compared to keratinocytes and ovarian surface epithelial cells in which TP53 or CDKN2A inactivation plus TERT overexpression is required for their immortal phenotype (20–22), NHUCs have fewer barriers to immortalization and the telomere maintenance is likely the only essential demand (13, 14). As further support, Li et al. (15) observed that the induction of endogenous TERT expression in normal urothelial stem cells enabled these cells to undergo immortalization and malignant transformation without any other manipulations (Next section for details).

TERT-NHUCs are genetically stable under optimal culture conditions for a long-observed period (14, 23, 24), as seen in TERT-immortalized fibroblasts (19). However, microenvironments where in vivo oncogenesis occurs are in general unfriendly, which promotes genomic and epigenetic alterations for clone selection of premalignant cells. To mimic an in vivo oncogenic scenario, Chapman et al. cultured TERT-NHUCs at a stressful low-density (23, 24). Indeed, such culture conditions triggered significant karyotypical and epigenetic alterations, and CGH analyses showed 2q loss and 20q gain in those cells (23, 24). 20q gain was frequently observed in HPV-16 E7 oncogene-immortalized NHUCs, too (25). Interestingly, 78% and 61% of UCB-derived cell lines bear + 20q and - 2q, respectively (24, 26). In addition, the LOH of CDKN2A, a critical tumor suppressor gene and senescence effector, occurred in one subline of TERT-NHUCs, while its promoter hypermethylation was more frequent and led to silent CDKN2A expression (24). On the other hand, however, recurrent activating mutations including FGFR3, PIK3CA, HRAS, NRAS or KRAS in UCs are absent in TERT-NHUCs (24). Nevertheless, the cancer-related genomic and epigenetic aberrations identified in TERT-NHUCs at least partially recapitulate in vivo pathogenesis process of UCs (27, 28).

Even under standard culture conditions, TERT-NHUCs undergo phenotypic alterations and one featured change is differentiation blockade (13, 23). Impaired differentiation does not happen immediately after TERT is introduced into NHUCs, while they progressively lose their ability to form an epithelial barrier over time (13). It was observed that TERT-NHUCs overexpressed the polycomb repressor complex (PRC1 and PRC4) components, thereby down-regulating expression of PRC target genes associated with NHUC differentiation (23). Moreover, TERT overexpression compromised NHUCs’ response to differentiation signaling (13). The nuclear receptor peroxisome proliferator-activated receptor γ binds and initiates the cascade of transcriptional and chromatin remodeling events associated with NHUC differentiation (13). Finally, TERT has been shown to directly promote stemness (29, 30), and may contribute to the observed differentiation blockade of NHUCs. Taken together, immortal and immature features of TERT-NHUCs will likely make them more sensitive to oncogenic events or signaling pathways.

The TERT gene is localized at chromosome 5p and contains 16 exons within a 40 kb long region (4). As described above, the TERT gene is transcriptionally repressed in most normal human cells, which leads to telomerase silence, while the induction of TERT expression is required to activate telomerase during oncogenesis (4). Earlier studies have been mainly focused on the molecules controlling TERT transcription in cancer. Indeed, many endogenous and exogenous oncogenic factors are identified to activate the TERT transcription and these findings have greatly contributed to understanding of cancer-specific TERT regulation (4). In 2013, two seminal studies unravelled the hotspot TERT promoter mutation as a novel mechanism for TERT expression and telomerase activation in human cancer (31, 32). Moreover, with the recent development of high-throughput sequencing technology, the massive mapping of cancer genomic and epigenomic landscapes further identified genetic and epigenetic alterations to drive TERT transcription, such as aberrant TERT promoter methylation (33). The activating TERT promoter mutation and hypermethylation are also observed in UC tumors (4, 27, 28), which are our focus for discussion below.

TERT promoter mutations were first identified in sporadic and familiar melanomas (31, 32), and subsequent investigations showed that they were present widely in many types of human malignancies. Two hotspot mutations occur at the proximal region of the TERT promoter (−124 and −146 bp from the ATG) with a cytidine-to-thymidine (C>T) dipyrimidine transition, which are called as C228T (-124C>T) and C250T (-146C>T), respectively (31, 32). Across UC subtypes, the frequency of TERT promoter mutations is different, varying from less than 20% in UCU, 45% in UCRPs, to ~85% in UCBs (28, 34–49). The C228T mutation is predominant and the presence of C228T and C250T is mutually exclusive, indicating a functional redundancy of these two mutations (Figure 2). Indeed, it has been well established that either C228T or C250T mutation activates TERT transcription (31). C228T- or C250T-bearing TERT promoter reporters exhibit a robustly increased activity compared to their wt counterparts (31). Consistently, mutation-carrying primary UC tumors express higher levels of TERT mRNA and telomerase activity (15, 47, 48, 50). Li et al. (15) further provided the direct evidence by manipulating the endogenous TERT promoter in normal and malignant bladder cells. The C228T mutation in the endogenous TERT promoter was created in normal human bladder stem cells using a CRISPR technique, and the authors then observed overexpression of TERT coupled with increased telomerase activity in these cells. Remarkably, introducing C228T mutation sufficiently drove malignant transformation of normal bladder stem cells. In sharp contrast, switching C228T to a WT sequence in cancer cells resulted in downregulation of TERT expression and loss of tumorigenesis (51). These observations demonstrate a critical role for TERT promoter mutations in telomerase activation and oncogenesis.

Further mechanistic insights have revealed that C228T and C250T mutations in the TERT promoter create de novo binding motifs for ETS transcription factors (Figure 2). In UCB-derived cell lines, these motifs were shown to be bound by the ETS family member GABPA through which the TERT gene was transcriptionally activated (52, 53). Because GABPA only contains the DNA binding domain, it needs to form a complex with its partner GABPB1 or GABPB2 that bears the transactivation domain, and the resultant complex then exerts its effect on target gene transcription (54). In glioblastomas and thyroid carcinomas, the GABPA-GABPB1 complex was shown to activate the mutated TERT promoter (55, 56). Likely, this is also the case in UC tumors. In addition, the heterotetramer of the GABPA-GABPB1 complex exhibits the strongest transcription activity, and it is generated when two or more ETS sites are adjacent or brought into proximity via chromatin looping (54). It was previously identified that the TERT promoter harbors an endogenous ETS site, however, a polymorphism rs2853669 there (-245 from ATG) could disrupt it (45). In C228T-bearing UCB cells, the rs2853669 T/T genotype maintains an intact ETS site and the mutated TERT promoter displays higher activity (Figure 2) (45). Whereas the promoter activity dropped significantly when the rs2853669 C/C genotype was introduced (45). The observed interdependence on the TERT promoter activation between C228T and res2853669T/T indicate that the endogenous ETS motif is bound by the GABPA-GABPB1 complex through which heterotetramers are formed as the strongest version of the transcription activator. In addition, other SNPs in the TERT locus were also observed to affect the occurrence of TERT promoter mutations, but underlying mechanisms remain to be defined (57).

Given the requirement of GABPA and GABPB1 to activate the mutated TERT promoter, they have been suggested as targets for telomerase-based cancer therapy (55). Indeed, following stable GABPB1 knockdown, glioblastoma cells carrying C228T or C250T TERT promoters undergo diminished TERT/telomerase expression, progressive telomere erosion and eventual loss of tumorigenic potential (55). However, GABPA and GABPB1 stimulate TERT transcription, but several lines of evidence suggest that they may serve as tumor suppressors in other cancer types regardless the presence or absence of TERT promoter mutations (56, 58–63). In UCBs, TCs and renal cell carcinomas (RCCs), GABPA or GABPB1 inhibition promotes their stemness and invasiveness, despite downregulation of TERT expression (59, 60, 63). Moreover, GABPA-depleted UCB cells resisted cisplatin-induced apoptosis (59). Mechanistically, GABPA activates the transcription of its target genes Fox-A1 and GATA3, two important differentiation-promoters of urothelial cells, while insufficiency or deficiency of GABPA expression results in blockade of cellular differentiation, thereby inducing an immature UCB status characterized by enriched stemness and EMT phenotypes (59). In primary tumors derived from patients with UCB, TC and RCC, GABPA expression is inversely correlated with advanced stage, aggressive or metastatic diseases and survival (59, 63). Intriguingly, GABPA and TERT expression even anti-correlated with each other in these tumors, suggesting a more complicated in vivo relationship between them (59). Like other tumor suppressors, the aberrant promoter hypermethylation and/or copy number loss contribute to the downregulation of GABPA or GABPB1 in the cancer types above (56, 59–61). Collectively, GABPA and GABPB1 functions are context- or cell type-dependent, and caution should be taken to target them for telomerase-based cancer therapy.

In a recent study, GABPA and GABPB1 were further shown to recruit tripartite motif containing 28 (TRIM28) to the mutant TERT promoter for TERT transcription in UCB cells (Figure 2) (64). TRIM28 is a nuclear factor that serves as a scaffold protein complexes regulating gene transcription. Under a unphosphorylated status, TRIM28 interacts with TRIM24 through which TRIM28 activity is inhibited, while TRIM28 phosphorylation at Ser824 by the mTORC1 disrupts its association with TRIM24 and in turn induces TERT transcription. Mechanistically, phosphorylated TRIM28 enhances TERT transcription by releasing Poly II RNA polymerase pause at the transcription site (Figure 2) (64). By doing so, TRIM28 also stimulates proliferation of UCB cells harboring the TERT promoter mutation. Cell growth was inhibited upon TRIM28 depletion, which could be partially rescued by TERT over-expression (64).

TERT induction and telomerase activation take place in general at the late stage of a stepwise oncogenesis process (4). However, many studies have demonstrated that TERT promoter mutations and detectable telomerase activity are present in premalignant lesions or even benign tumors (41, 65–67). Weyerer et al. performed TERT promoter mutation analyses on different sites in bladder organs from UCB patients, and they identified the mutation in not only tumor tissues, but also non-invasive urothelial lesion as well as adjacent non-tumor (normal) tissues (68). A recent analysis of UCB patients showed that the mutated TERT promoter could be detected in urine even 10 years prior to clinical diagnosis of the disease (39). In addition, urothelial papilloma (UP) of the urinary bladder, a benign entity, and papillary urothelial neoplasm of low malignant potential (PUNLMP) were observed to carry the C228T mutation with high frequencies, 46% and 43%, respectively (69). Similar findings were reported by others (70). These results suggest that the TERT promoter is targeted for mutation at an early stage of carcinogenesis, and even in benign tumors lacking malignant characteristics at pathological and morphological levels (63). The early onset of TERT promoter mutations is highly consistent with recent observations that oncogenic genomic alterations may take place in childhood or adolescent periods long before cancer formation (71). The biological and clinical significance underlying TERT promoter mutations in precursor lesions and even in normal cells are worthy of further investigations.

The mechanism behind TERT promoter mutations remains elusive. Several lines of evidence indicate that shortened or dysfunctional telomeres drive the mutation occurrence. First, in several cancer types including UCBs, telomeres are significantly shorter in mutation-bearing tumors than those with a wt promoter (72, 73). Second, the TERT promoter mutation is much more frequent in old patients than in young ones (72, 73). It is known that old individuals have shorter telomeres in their tissues/cells, and telomere dysfunction occurs earlier during the oncogenic process, which thus experiences strong pressure for telomere stabilization. Third, the presence of TERT promoter mutations are strongly correlated with activating mutations in mitogen-activated protein kinase (MAPK) genes, such as FGFR3, BRAF, and RAS in UCBs, melanoma and TCs, respectively (43, 63, 65, 67). The MAPK hyperactivity accelerates cell proliferation, inducing excessive telomere attrition and dysfunction. Forth, HIV infection leads to accelerated telomere shortening, and HIV-related premalignant lesions have a high frequency of TERT promoter mutations (74). Finally, the C228T mutation was found in hematopoietic cells and lymphocytes from patients with telomere disease due to germline defects in TERT or telomerase accessory factors (75). The acquisition of C228T mutation restored telomerase activity and proliferation capacity of patient’s lymphocytes by lengthening telomeres. Intriguingly, in patients bearing a heterozygous mutation in the TERT coding region, the C228T mutation facilitated the transcription of the WT TERT allele, which showed a clear bias toward positive selection (75). Because these patients were free of cancer, the occurrence of C228T mutation in their lymphocytes dose not result from oncogenic events per se, indicating a causal relationship between shorter dysfunctional telomeres and TERT promoter mutations. The identification of frequent C228T mutation in benign nevi with shorter telomeres also supports for this view (67). Mechanistically, telomere position effect-over long distances (TPE-OLD) may play a role. TPE-OLD was shown to control chromatin structure at the TERT locus in a telomere length-dependent manner (76). When cells have long telomeres, a telomere-loop structure is generated through the shelterin protein TRF2 in the TERT locus, resulting in a repressive chromatin surrounding the TERT promoter region (76). However, very short or dysfunctional telomeres at chromosome 5p impair the loop formation, thereby inducing a loose chromatin that exposes the TERT promoter to oncogenic or mutagenic factors (76). Thus, very short or dysfunction telomeres confer cells strong selection pressure on one hand, while create an epigenetic environment favorable for induction of TERT promoter mutations on the other hand.

The TERT promoter is embedded in a CpG island (CGI) spanning approximately 4 kbs from -1 800 to +2 200 bp relative to the ATG site (33, 53, 63). CpGs in the TERT promoter CGI are in general unmethylated in normal human cells lacking TERT expression (33, 53, 63). The unmethylated status of the TERT promoter in those cells has been proposed as a critical mechanism to repress telomerase activation, because it is accessible to physiological repressors through which TERT transcription is blocked (33, 53). In TERT-expressing cancerous tissues and cells including UCs, the TERT promoter hypermethylation occurs widely, however, the methylation is distributed unevenly throughout the promoter region. Rowland et al. (33) analyzed 823 cancer cell lines derived from 23 different tissue types for the TERT promoter methylation, and they observed a highly consistent TERT promoter methylation pattern across cell lines: hypermethylated upstream of the transcription start site (UTSS) region while hypomethylated (transcription start site) TSS or proximal promoter region (Figure 3A). This was also the case in 23 analyzed UCB cell lines (33). The same methylation profile was further observed in primary UCB tumors. In normal urothelial tissues in which the TERT transcription is suppressed, both UTSS and TSS or proximal TERT promoter regions are hypomethylated (33, 53). Thus, the UTSS hypermethylation is cancer-specific and likely an epigenetic mechanism to induce TERT expression by unlashing repressors from the promoter. However, activated normal human lymphocytes and embryonic stem cells express high levels of TERT and telomerase activity, and their TERT promoter methylation pattern is very similar to that in TERT-negative normal cells. These results unravel different epigenetic mechanisms underlying TERT regulation between cancerous and normal cells. Elucidating this issue will be important both physiologically and carcinogenically.

It has recently been identified that cancer cells display an allele-specific activation of the TERT transcription, leading to biallelic or monoallelic expression (BAE or MAE) of TERT (33, 53, 77). In tumor cells expressing TERT with BAE, both alleles exhibit hypermethylated UTSS while the hypomethylated proximal TERT promoter (33, 53). In MAE cells, however, this featured methylation pattern occurs only in the expression-allele (Figure 3B) (33, 53). Comparison between cancer cells with and without TERT promoter mutations further showed a bias in allele-specific TERT expression. Most wt cells express TERT biallelically, although a small fraction of them shows MAE (33, 53, 77). Like most wt cells, homozygous mt cells exhibit BAE (33, 53, 77). Both alleles are characterized by hypermethylated UTSS and hypomethylated TSS in these BAE cells. In cancer cells harboring a heterogenous TERT promoter mutation, the mutant-allele displays the expression-allele methylation profile, and the non-expressing WT allele may have increased TSS methylation and/or reduced UTSS methylation (33, 53). Moreover, the mutant expression allele is associated with histone H3K4 trimethylation (H3K4me3) and H3K9/K14 acetylation that marks active transcription, while the WT non-expressing-allele is enriched with the repressive histone H3K27me3 and H3K9me3 (Figure 3B) (33, 53). Further studies reveal that there are subtle differences in methylation between MAE and BAE cells, and hypomethylation at positions from -271 to -290 while hypermethylation from +109 to +145 are observed in MAE cells bearing WT or mutated TERT promoters (33). These observations indicate that the differential methylation at two positions above may play a part in regulating allele-specific transcription of the TERT gene.

The most striking dissimilarity is the hypermethylated UTSS in cancer cells while hypomethylated UTSS in normal cells, however, Stern et al. observed that decreased levels of the UTSS region methylation were associated with high TERT expression and invasive phenotype in cancer cells with a mutated TERT promoter (53). T24 cells, harboring a heterogenous C228T mutation, are non-metastatic UCB cells and express low levels of TERT mRNA coupled with hypermethylated UTSS, while compared with those in T24 cells, their metastatic variants, including T24T, FL3 and SLT4, have robustly increased TERT expression and reduced UTSS methylation (53).

More recently, the TERT promoter UTSS hypermethylation was further observed to epigenetically inhibit the expression of TERT antisense promoter-associated RNA (TAPAS) (78, 79). TAPAS is a 1.6 kb long non-coding RNA localized 167 nts upstream of the TERT TSS with the antisense direction to the TERT promoter (79). TAPAS depletion and over-expression in cancer cells led to significantly enhanced and diminished TERT expression, respectively (78, 79). In addition, the nuclear accumulation of TERT mRNA occurs frequently in cancer cells, and intron retention is identified as a underlying mechanism (80, 81). TAPAS was similarly observed to induce accumulation of TERT mRNA in nucleus, and such effect prevented access of TERT transcripts to the translational machinery, thereby inhibiting TERT protein translation (79). Ott et al. showed that levels of TAPAS RNA were inversely correlated with TERT mRNA abundances in UCB cell lines and primary tumors (78). The authors further demonstrated that the hypermethylated UTSS contributed to the repression of TAPAS transcription, but TERT promoter mutations did not affect its expression (78, 79). It is thus concluded that the UTSS hypermethylation-mediated TAPAS silencing serves as a key mechanism for TERT upregulation in the UCB pathogenesis (78). Detailed dissections are required to understand their relationship and exact roles in telomerase activation during oncogenesis.

TERT induction and telomerase activation play essential roles in immortalization and malignant transformation of NHUCs, while the TERT promoter mutation is a predominant mechanism to activate telomerase in UCs, especially UCBs. In addition, the aberrant TERT promoter hypermethylation occurs widely in UCs and interacts with the mutated promoter to robustly augment TERT transcription. The mutated or methylated TERT promoter and TERT transcripts are useful biomarkers for non-invasive urinary diagnosis and surveillance of UCs, however, it is important to standardize assay and evaluation system for clinical practice. The impact of the TERT promoter mutation on UC outcomes may be affected by TERT SNPs, promoter methylation status and other variables, which are required for further investigations to achieve TERT-based precision UC management. Finally, TERT promoter mutations and gene expression as predictors for response to UC immunotherapy, the association between TERT-mediated HERV activation and immunotherapy efficacy, and TERT vaccine for UC treatment are all important and worthy of detailed investigations in future. For all these purposes, the standardization of a TERT assay and evaluation system is required.

TL, CX and DX conceived and designed the study. TL, SL, CX and DX participated in the data process, analysis and interpretation. TL, CX and DX wrote and revised the manuscript. All authors contributed to the article and approved the submitted version.