Genetic regulation of TERT splicing contributes to reduced or elevated cancer risk by altering cellular longevity and replicative potential

Abstract

The chromosome 5p15.33 region, which encodes telomerase reverse transcriptase (TERT), harbors multiple germline variants identified by genome-wide association studies (GWAS) as risk for some cancers but protective for others. We characterized a variable number tandem repeat within TERT intron 6 (VNTR6-1, 38-bp repeat unit) and observed a strong association between VNTR6-1 alleles (Short: 24-27 repeats, Long: 40.5-66.5 repeats) and GWAS signals within TERT intron 4. Specifically, VNTR6-1 fully explained the GWAS signals for rs2242652 and partially for rs10069690. VNTR6-1, rs10069690 and their haplotypes were associated with multi-cancer risk and age-related telomere shortening. Both variants reduce TERT expression through alternative splicing and nonsense-mediated decay: rs10069690-T increases intron 4 retention and VNTR6-1-Long expands a polymorphic G quadruplex (G4, 35-113 copies) within intron 6. Treatment with G4-stabilizing ligands decreased the fraction of the functional telomerase-encoding TERT full-length isoform, whereas CRISPR/Cas9 deletion of VNTR6-1 increased this fraction and apoptosis while reducing cell proliferation. Thus, VNTR6-1 and rs10069690 regulate the expression and splicing of TERT transcripts encoding both functional and nonfunctional telomerase. Altered TERT isoform ratios might modulate cellular longevity and replicative potential at homeostasis and in response to environmental factors, thus selectively contributing to the reduced or elevated cancer risk conferred by this locus.

Figure 1.. Analysis of VNTR6–1 and VNTR6–2 within TERT intron 6 in relation to GWAS leads rs10069690 and rs2242652.

a, The chr 5p15.33 genomic region with GWAS leads rs10069690 and rs2242652 within TERT intron 4 and VNTRs within intron 6. b, Distribution of repeat copies of VNTR6–1 (38-bp repeat unit) and c, VNTR6–2 (36-bp repeat unit) in 452 phased long-read genomic assemblies from 226 controls of diverse ancestries. The dots represent repeat copies for each chromosome assembly; box plots show the overall and interquartile range, median (horizontal black line), and mean (black dot, with values above corresponding plots). Half-violin plots show the density distribution of the data. Five VNTR6–1 alleles—24, 25.5, 27, and 40.5 repeat copies—were observed above the 5% frequency threshold and accounted for 90.04% of all alleles in the set; VNTR6–2 alleles were scattered between 8 and 155 repeat copies, all under the 5% threshold. P values were calculated for unpaired two-sample Wilcoxon–Mann–Whitney tests comparing the number of repeat copies between the genotype groups.

Figure 2.. VNTR6–1 affects the splicing ratios of the TERT-FL and TERT-β isoforms.

a, G4-ChIP results within the TERT region in the HEK-293T (VNTR6–1: Long/Long) and NA18507 (YRI, 1000G, VNTR6–1: Short/Short) cell lines display mismatches (%) during DNA synthesis, reflecting polymerase stalling after G stabilization in both the plus (blue) and minus (orange, direction of TERT transcription) genome strands. The genomic region of TERT intron 6 shows VNTR6–1 (24–66.5 copies of the 38-bp repeat unit), VNTR6–2, G4s in the minus strand (polymorphic G4 within VNTR6–1 and constitutive upstream G4), and CRISPR/Cas9 guide RNAs for excising VNTR6–1. The sequence logo shows the consensus of the 38-bp VNTR6–1 repeat unit in UMUC3 cells based on PacBio long-read WGS. b, Agarose gels of RT‒PCR products amplified from cDNA of corresponding samples; gDNA–genomic DNA was used as a negative control; HPRT1 was used as a normalization control. e, Densitometry results of the PCR amplicons in plot b. The differences in the TERT isoform ratios are further explored in Figure S11. Experiments in UMUC3 cells comparing TERT splicing and isoform-specific expression after 72 hrs of treatment with G4 stabilizing ligands, normalized to HPRT1 as an endogenous control in the WT (c, f) and V6.1-KO (d, g) cell lines. c, d, A representative agarose gel of SYBR-Green RT‒qPCR products detecting several isoforms with primers located in exons 6 and 9. The extra PCR band, marked by an arrow in panels c and d, is further explored in Figure S12. f, g, Densitometry analysis of the corresponding agarose gels evaluating the TERT-FL (%) relative to the total PCR products. All analyses are based on three experiments, with one representative gel shown. Comparisons were made against the vehicle control (DMSO). Statistical significance is indicated as follows: **p < 0.01, ***p < 0.001, ****p < 0.0001, Student’s T-test.

Figure 3.. Analysis of TERT expression in 78 Burkitt lymphoma (BL) tumors.

Total TERT expression analyzed as transcripts per million (TPMs) a, overall and in relation to the b, rs10069690, c, rs2242652 and d, VNTR6–1 genotypes. Group means are shown as red dots with values above corresponding violin plots. e, association of the VNTR6–1 and rs10069690 haplotypes with total TERT expression. The reference Short-C haplotype corresponds to the telomerase-encoding TERT-FL isoform, while the INS1 and TERT-β isoforms encode truncated proteins without telomerase activity. Effect alleles in haplotypes are marked in red; white boxes – exons; gray boxes – introns; black boxes – intron 4 retention; blue boxes – alternative exons 7 and 8; and red lollipops – stop codons. The direction of the TERT exons is from right to left, corresponding to the minus strand, as presented in the UCSC browser. “ATG” indicates translation start codons. P values and β-values are for linear regression models adjusted for sex and age.

Figure 4.. VNTR6–1 affects the proliferation and apoptosis of UMUC3 cells

a, Real-time monitoring of cell population growth dynamics (cell index) for 283 hrs in UMUC3 WT and V6.1-KO cells cultured in media supplemented with full serum or charcoal-stripped (CS) serum revealed significantly greater proliferation rates in WT cells than in V6.1-KO cells under both culture conditions. Statistical significance and β-values for differences in the cell index during the visually determined growth phase (gray highlighting between 50 and 183 hours) were calculated using linear mixed-effects models based on six replicates, **p < 0.01, ***p < 0.001, ****p < 0.0001. b, Quantification of cell doubling events in CFSE-stained cells cultured with CS serum or c, full serum medium for four days in three replicates (****p < 0.0001, Student’s t test). d, Cells were treated with 10 μM cisplatin or e, full serum medium for 48 hrs, followed by Annexin V-FITC staining to determine the percentage of apoptotic cells in three replicates (***p < 0.001, Student’s t test). Differential expression of genes involved in pathways related to the downregulation of f, cell proliferation (positive regulation of cell population proliferation pathway, GO:0008284, Table S11) and g, apoptosis resistance (negative regulation of programmed cell death pathway, GO:0043069, Table S11) according to RNA-seq analysis of V6.1-KO UMUC3 cells compared to WT UMUC3 cells. Genes highlighted in blue are common to both pathways. The data shown in all panels except f and g represent one of three independent experiments.

Figure 5.. The functional differences between the TERT-FL and TERT-β isoforms

a, Structured illumination microscopy images of A549 cells co-transfected with TERT-FL and TERT-β expression constructs at a 50:50% ratio. For individual channels, staining is shown as black/white images for better contrast. On tri-color merged panels, green – FLAG (TERT-β) or HA (TERT-FL), blue – DAPI (nuclei). On the quad-color merged panel, purple - HA (TERT-FL), green - FLAG (TERT-β), red - TOM20 (mitochondria), blue - DAPI (nuclei). The yellow inset in the TERT-β-FLAG panel is shown at a higher magnification to demonstrate colocalization with mitochondria (yellow staining). b, The overview of the VNTR6–1, TERT-FL:TERT-β ratio, proliferation, and apoptosis.

Figure 6.. Association analyses for cancer risk in PLCO and relative leukocyte telomere length (rLTL) in UKB cancer-free individuals.

a, Evaluation of cancer risk associated with the VNTR6–1-Long and rs10069690-T alleles and the composite marker (VNTR6–1/rs10069690) in the PLCO dataset (n=102,708). Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated for comparisons between patients with the indicated cancers and a common group of cancer-free controls using logistic regression analysis with an additive genetic model adjusted for sex and age. b, Evaluation of the relationships between rLTL and VNTR6–1/rs10069690 and rs7705526 in UKB cancer-free individuals (n=339,103, LD, r²=0.33 between VNTR6–1/rs10069690 and rs7705526). P values and β-values were derived from linear regression models adjusted for sex, age, and smoking status. P_int represents the interaction between genotypes and 5-year age groups; P_cond represents the mutual adjustment for rs7705526 or VNTR6–1/rs10069690. The graphs display regression lines with 95% confidence intervals and regression equations. The analysis revealed a decrease in the rLTLs with more copies of the Short-C haplotype. The results of sex-specific analyses of VNTR6–1/rs10069690 are presented in Figure S20b.

Figure 7.. The interaction model of factors affecting cancer risk

a, TERT genetic variants VNTR6–1 and rs10069690 and environmental factors define the relative ratios of the isoforms encoding telomerase-functional TERT-FL and telomerase-nonfunctional TERT-β and INS1b isoforms. These isoforms affect cell proliferation, apoptosis and telomere length, thus modulating cellular longevity and replicative potential, including homeostatic proliferation, which maintains tissue self-renewal, and regenerative proliferation, which responds to environmental factors and tissue damage. b, Cancer risk as a product of G x E x R interactions. The VNTR6.1-Long and rs10069690-T alleles, or their haplotype (Long-T), are associated with reduced cancer risk in tissues with low homeostatic but high regenerative potential (e.g., bladder). The anti-apoptotic effect of the TERT-β isoform reduces the need for regenerative proliferation, thus decreasing the risk of acquiring mutations due to replicative mutagenesis. In tissues with no/low homeostatic and regenerative proliferation (e.g., brain, thyroid, ovary), the same alleles and Long-T haplotype are associated with elevated cancer risk. The anti-apoptotic effect of TERT-β extends cellular longevity, allowing the accumulation of more mutations due to environmental mutagenesis, such as through exposure to reactive oxygen species (ROS), cellular metabolites, etc.