New somatic TERT promoter variants
TERTpmut were detected in 239/301 cases (79.4%), including 14/15 ODG (93%), 12/32 DA/AA (37.5%), and 213/251 GBM (84.8%) (Additional file 4: Table S4). In GBM (= 213) and DA/AA (= 12), TERTpmut were prevalent in IDHwt cases (209/241 GBM IDHwt vs 4/10 GBM IDHmut; 10/12 DA/AA IDHwt vs 2/20 DA/AA IDHmut) (Chi square, P < 0.001) (Additional file 5: Table S5). Thus, in agreement with the diagnostic criteria recommended by the cIMPACT-NOW (Update 3), the 10 DA/AA IDHwt with TERTpmut were referred to as “diffuse astrocytic glioma, IDH-wildtype, with molecular features of glioblastoma, WHO grade IV” [4].
In GBM TERTpmut there was a significant enrichment of cases harbouring EGFR amplification (46% vs 17%) (Chi square, P = 0.001) and/or monosomy 10/PTEN deletions (84% vs 37.5%) (Chi square, P < 0.0001). Likewise, EGFR amplification or gain of whole chromosome 7 in combination with monosomy 10, occurred in 6/10 (60%) of TERTpmut DA/AA IDHwt.
The most common variant, TERTp-124 was detected in 172 cases while the TERTp-146 was found in 65 cases. TERTpmut were mutually exclusive, heterozygous, and equally distributed among the different histological subtypes (Additional file 5: Table S5). Besides the TERTp-124 and TERTp-146, we uncovered two new TERTp variants in two cases of GBM IDHwt (UPN#131 and UPN#171). These novel TERTpmut consisted of a 22 nucleotide tandem duplication, occurring in a genomic region starting at 100 and 110 bp, from the ATG starting site, i.e. c.1-100_1-79dup (TERTp-100-79), in case UPN#131, and c.1-110_1-89dup (TERTp-110-89), in case UPN#171 (Fig. 2a, b) (www.ncbi.nlm.nih.gov/gene, www.ensembl.org/Homo_sapiens, cancer.sanger.ac.uk/cosmic) [5, 7, 20]. They shared a region of duplication of 12 nucleotides, from 1–100 to 1–89 nucleotides from the ATG start site. The absence of TERTp-100-79 in the PB DNA, demonstrated the somatic origin of this variant in case UPN#131.
In silico analysis predicts TERTpmut effects
In silico analysis predicted that both TERTpdup created new binding sites, i.e. 119 for TERTp-100-79 and 108 for TERTp-110-89, which were respectively recognized by 65 and 53 TFs. Instead, TERTp-124 and TERTp-146 were predicted to increase the binding affinity for 40 and 43 sites, and to enhance the probability of binding for 28 and 29 TFs, respectively (Additional file 6: Table S6). Although all TERTpmut affected the binding sites for diverse families of TFs, the ETS group emerged as one of the most frequently involved: 18/65 (28%) in TERTp-100-79, 18/53 (34%) for TERTp-110-89, 23/28 (82%) in TERTp-124, and 25/29 (86%) in TERTp-146, (Fig. 2c, Additional file 7: Table S7). Other recurrently involved TFs in TERTpdup variants were the Specificity Protein/Krüppel-Like Factor (Sp/KLF) family, i.e. 19/65 (29%) in TERTp-100-79 and 16/53 (30%) in TERTp-110-89, and the More than 3 adjacent zinc finger factors family (12/65 in TERTp-100-79 and 7/53 TERTp-110-89) (Additional file 7: Table S7).
The Venn diagram showed a close inter-relationship between all TERTp mutations. Namely, all TERTp mutations shared an increase of the binding affinity, or the number of binding motifs, for 19 common TFs (Fig. 3a), including 18 ETS members (ETS1, ETS2, ERG, ELK1, ETV6, FLI1, ELK4, SPIB, ELF1, ELF3, ETV4, ETV1, FEV, EHF, ETV5, ELF5, SPI1, and GABPA) and TEAD1 (Fig. 3a; Additional file 8: Table S8). The Venn diagram also showed that the new TERTpdup were characterized by the exclusive involvement of 30 common TFs. Specifically, there were 16 Sp/KLF members, i.e. KLF2, KLF3, KLF4, KLF5, KLF10, KLF11, KLF14, KLF15, KLF16, SP1, SP2, SP3, SP4, SP8, SP9, and EGR1, (Fig. 3a, Additional file 8: Table S8) and 14 TFs that belong to 9 different families (Fig. 3a, Additional files 7 and 8: Tables S7 and S8). Matching our TERTpdup with the two cases of TERTpdup previously reported (Additional files 9 and 10: Tables S9 and S10) [21, 23], JASPAR predicted that all variants determined an increase of binding sites for 21 common TFs, and confirmed that the Sp/KLF family was the most frequently involved (14/21) (Fig. 3b, Additional file 11: Table S11).
In vitro analysis confirms the increasing of TERT transcriptional activity induced by its promoter mutations
In vitro luciferase assay was carried out to evaluate whether the new TERTp-110-89 variant induced an increase of TERT transcriptional activity, enhancing its expression, similarly to TERTp-124 and TERTp-146 [12, 21]. In Table S12 (Additional file 12: Table S12) we reported raw data referred to the fluorescence emission values, expressed in Relative Luciferase Activity (RLA), of both Photinus Pyralis and Renilla Reniformis luciferase enzymes, for all samples. Our experiments demonstrated that all three variants caused a significant increase of TERT transcription by 2.3-2.5 fold than wildtype (TERTp-110-89 vs TERTpwt: P < 0,0001; TERTp-124 vs TERTpwt: P < 0,0315; TERTp-146 vs TERTpwt: P < 0,0001; Mann–Whitney U test) (Fig. 4). On the other hand, no differences on the levels of TERT expression were present between the diverse TERTp variants, indicating they may all behave as gain-of-function mutations, likely exerting the same consequences on TERT transcription.