Coding and noncoding somatic mutations in candidate genes in basal cell carcinoma.

Basal cell carcinoma (BCC) represents the most commonly diagnosed human cancer among persons of European ancestry with etiology mainly attributed to sun-exposure. In this study we investigated mutations in coding and flanking regions of PTCH1 and TP53 and noncoding alterations in the TERT and DPH3 promoters in 191 BCC tumors. In addition, we measured CpG methylation within the TERT hypermethylated oncological region (THOR) and transcription levels of the reverse transcriptase subunit. We observed mutations in PTCH1 in 58.6% and TP53 in 31.4% of the tumors. Noncoding mutations in TERT and DPH3 promoters were detected in 59.2% and 38.2% of the tumors, respectively. We observed a statistically significant co-occurrence of mutations at the four investigated loci. While PTCH1 mutations tended to associate with decreased patient age at diagnosis; TP53 mutations were associated with light skin color and increased number of nevi; TERT and DPH3 promoter with history of cutaneous neoplasms in BCC patients. Increased reverse transcriptase subunit expression was observed in tumors with TERT promoter mutations and not with THOR methylation. Our study signifies, in addition to the protein altering mutations in the PTCH1 and TP53 genes, the importance of noncoding mutations in BCC, particularly functional alterations in the TERT promoter.

Basal cell carcinoma (BCC) accounts for about 80 percent of all skin cancers and it is the most commonly diagnosed neoplasm among the Caucasian population [1][2][3] . The tumor originates from stem cells within hair follicles or the interfollicular epidermis and infundibulum undergoing ultraviolet (UV) radiation induced mutagenesis [4][5][6] . Basal cell carcinoma rarely metastasizes; however, due to sheer number of people affected, the disease poses a considerable health hazard as it causes extensive morbidity through local invasion and tissue destruction 7 . The most common genetic alterations in BCC involve the Hedgehog receptor patched 1 (PTCH1) and the cell cycle regulator TP53 and the mutations in both genes predominantly reflect UV etiology 8 . Exome sequencing based studies, besides denoting BCC as the cancer with highest mutation burden, have also identified less frequent alterations in genes other than PTCH1 and TP53 8,9 . Germline mutations PTCH1 also occur in patients with Gorlin syndrome, a Mendelian disease with a high prevalence of BCC 10 . Aberrant activation of the Hedgehog pathway is common in sporadic BCC either through mutations in the PTCH1 gene, activating mutations in smoothened (SMO) or loss-of-function mutations in suppressor of fused homolog (SUFU) 9,11,12 . Ubiquitous hyper-activation of the Hedgehog has led to development of inhibitors targeting G-protein coupled receptor Smoothened for treatment of advanced BCC 11,12 . While, the studies based on exome sequencing have confirmed the centrality of PTCH1 mutations in BCC, the variants within smoothened are reportedly causal in resistance to the inhibitors 8 .
TP53 encodes a transcription factor, which is a tumor suppressor involved in cellular stress responses including DNA damage, oxidative stress, oncogenic hyperproliferation, pathogenic stimuli, UV-induced pigmentation 13,14 .
Due to the criticality of p53 mediated transcription in tumor suppression, mutations in TP53 mainly occur within the DNA binding domain that cluster at several hotspot amino-acid residues 15,16 . A large proportion of TP53 mutations in skin cancers including BCC are typically C > T and CC > TT transitions at dipyrimidines localized within several specific mutational hotspots due to hyper proneness to UV-induced damage 15,[17][18][19] . BCC tumors have been shown to arise in skin containing clones of cells with stabilized and mostly mutant p53 protein 20,21 .
Besides the frequent canonical coding region alterations in PTCH1 and TP53, noncoding mutations within the core promoter of the gene encoding telomerase reverse transcriptase (TERT) occur at elevated frequencies in BCC [22][23][24][25] . Reported initially in melanoma, the TERT promoter mutations are frequent in multiple cancers, with some exceptions, arising from tissues with low rates of self-renewal 22,[26][27][28] . Those promoter mutations create de novo sites where binding of ETS transcription factors leads to increased transcription through epigenetic histone modification 26,29 . Methylation within the TERT hypermethylated oncological region (THOR) of the same promoter also de-represses TERT transcription 30 . Noncoding mutations in a bidirectional DPH3 promoter region with unknown functional consequences are also frequent in different skin cancers including BCC 31 .
In this study, we sequenced 191 BCC lesions and 115 corresponding normal skin surrounding tumors for mutations in the PTCH1 and TP53 genes, and the TERT and DPH3 promoters. We also investigated CpG methylation at the TERT promoter between −560 to −774 bp region from the ATG start site that falls within the THOR. The effect of the TERT promoter mutations and THOR methylation on the transcription of the catalytic subunit of the telomerase was also measured. In addition, we also investigated the correlation between TERT promoter mutations and telomere length in BCC tumors.

PTCH1 mutations.
We detected 137 PTCH1 mutations in 105 tumors; of those 44 tumors also showed loss of heterozygosity including focal deletions in 4 tumors (Table 1; Supplementary Figure 1; Supplementary Table 1 and 2). In addition 7 tumors showed only loss of heterozygosity without a mutation on the remaining allele. One mutation each was detected in 79 (41%) tumors, 21 (11%) tumors carried two mutations each, four tumors had three mutations each and one tumor had four mutations.
The C > T base change was the most frequent mutation with 64 transitions in 56 (29%) tumors, followed by CC > TT tandem mutations, and C > A mutations in 8 tumors each. Thirty-six mutations were missense, 33 were nonsense and 17 mutations were intronic (Fig. 1). We also detected 37 insertion-deletions (indels) that included 25 frame-shift alterations, 6 in-frame deletions, two nonsense, and three intronic. Out of three duplications, two were frame-shift and one in-frame. We also detected 14 mutations at 5′ splice sites and 7 mutations at 3′ splice sites (Supplementary Table 3 Table 4).
Analysis of bisulfite converted DNA after cloning also showed statistically significantly higher methylation in tumors without mutations than in those with mutations (Fisher's exact P: < 0.00001, Mann-Whitney U-test P: < 0.00001; Supplementary Fig. 2). Pairwise sequence alignment of cloned bisulfite sequence and the target genomic sequence using QUMA showed more than 95% homology for each clone. Individually at the each CpG sites, tumors without the TERT promoter mutations had statistically significantly higher methylation than the tumors with mutations. The methylation across all 14 CpG sites ranged between 85% − 98% in tumors without mutations compared to 44% − 91% in tumors with mutations. tumors. In addition to the frequent C > T transitions observed in 35 (18.3%) tumors at −8 bp and in 14 (7.3%) tumors at −9 bp, CC > TT tandem mutations at −8_9 bp were present in 17 tumors (8.9%). Additional mutations included two C > T transitions at −12 bp, and two C > A transversions at −9 bp. The C > T mutation at −12 bp in two BCCs co-occurred with −8C > T and −8/−9CC > TT mutations, respectively, and the C > T mutation at −13 bp in one BCC co-occurred with −8/−9 CC > TT ( Mutations a 60 31.4 Tandem mutations (CC > TT) 16 0.001, t-test; Fig. 3). Further stratification showed that tumors with TERT promoter mutations and complete methylation (n = 23) www.nature.com/scientificreports www.nature.com/scientificreports/ had statistically significantly (P = 0.003) higher TERT expression than the tumors with complete methylation and without TERT promoter mutations (n = 14) (Supplementary Fig. 3). Similarly, the tumors with partial methylation and TERT promoter mutations (n = 25) had statistically significantly (P = 0.004) higher TERT expression than corresponding tumors with partial methylation and without the TERT promoter mutations (n = 2). The difference in TERT expression in tumors based on only methylation status was not statistically significant (P = 0.82).
Telomere length. Results from the measurement of relative telomere length were available for 174 BCCs and 104 tumor-surrounding skin tissues. Of 174 BCCs, 100 were with and 74 were without TERT promoter mutations. Relative telomere length ranged between 0.19 and 3.62 with median values of 1.02 in tumor-surrounding skin, 0.81 in tumors without TERT promoter mutations and 0.72 in tumors with the promoter mutations. Tumors had shorter telomeres than the surrounding corresponding skin tissues. Telomere with TERT promoter mutations had shorter telomeres than the tumors without those mutations; however, the difference was not statistically significant ( Supplementary Fig. 4).

Discussion
In this study, other than confirming the high frequency of PTCH1 and TP53 mutations in BCC tumors, we showed recurrent noncoding mutations within the TERT and DPH3 promoters. TERT promoter mutations were the most frequent alterations followed by PTCH1 mutations; the simultaneous occurrence of mutations in the four investigated loci was more frequent than per chance. Tumors with TERT promoter mutations had higher transcription levels of telomerase reverse transcriptase subunit than the lesions without those alterations. We observed no statistically significantly increase in TERT transcription levels in tumors due to an increased methylation within the THOR.  Inactivation of PTCH1 through mutations confers Sonic Hedgehog independent growth, genomic instability and tumor development potential. The full-length human PCTH1 consisting of 1447 amino-acids and 23 exons is predicted to contain 12 transmembrane segments, two extracellular domains and two intracellular domains 32,33 . The prevalent mutations in PTCH1 block protein maturation and consequent abolition of its function but the degree of impairments is dependent on the mutation-type 34,35 . In this study, we observed inactivating alterations in PTCH1 in 58.6% of tumors. In accordance with a central role of UV radiation in the pathogenesis of BCC, the majority of point mutations were represented by C > T and CC > TT transitions as also reported in previous studies 9, 36,37 . The observed frequency of mutations and loss of heterozygosity at the PTCH1 locus in this study was similar to earlier studies 38 . However, the mutation frequency was lower than that reported in an exome sequencing based genomic analysis 8 .
Approximately 31% of tumors carried mutations in the TP53 gene and similar to PTCH1, a majority of those alterations reflected UV etiology. Most of the base changes in TP53 were predominantly protein altering missense mutations and to lesser extent nonsense alterations. The mutations mainly affected the DNA binding domain with typical hotspot mutations as reported earlier 8 . The selection for missense mutations within the DNA binding domain of the gene is driven through dominant negative effect where mutant forms hinder functioning of the protein from intact alleles 16 . The observed frequency of the mutations in our study was closer to that reported in non-aggressive than in aggressive BCC tumors 38,39 . However, in an exome sequencing study on a large series of BCC tumors, the frequency of TP53 mutations was about 61% 9 . The frequencies of mutations detected in BCC tumors at known hotspots like codons 177, 196, 245, 248, 258 and 278 in TP53 were in conformity with earlier reports 18,19 . Impaired p53 in mice has been shown to result in lack of tanning response and addiction to sunlight; in our data, we observed a strong association between TP53 mutations and fair skin 13,40 .
TERT promoter mutations were detected in 59.2% of tumors and in accordance with earlier reports it was the most frequently altered locus in BCC [23][24][25] . Of the two main TERT promoter mutations, −124 C > T and −146 C > T, the former, with the exception of skin neoplasms, is overwhelmingly predominant in most cancers 22 . However, in melanoma and keratinocyte cancers, the −146 C > T is the most frequent TERT promoter mutation 22 . Skin cancers are also characterized by the presence of CC > TT tandem mutations at −124/−125 and −138/−139 bp positions that lead to the similar CCGGAA consensus binding site for ETS transcription factors 41 . The −138/−139 CC > TT tandem mutation in melanoma was shown to be associated with the worst melanoma-specific patient survival 42 . The altered site due to the −146 C > T mutation specifically involves non-canonical NF-kB signaling with cooperative binding between p52/RelB and ETS1 rather than general binding of the multimeric GABPA/GABPB1 complex as reported in gliobastoma, liver and bladder cancer cell lines 29,[43][44][45] . In accordance with previous studies, we detected increased transcription of the reverse transcriptase subunit in BCC tumors with than without the TERT promoter mutations 46,47 .
We also investigated the THOR region where methylation of CpG sites is generally associated with increased TERT expression 30,48 . However, our data show that increased TERT expression in BCC tumors was primarily attributed to the promoter mutations similar to adult gliomas 49 . Telomeres in tumors with the TERT promoter mutations are usually shorter than in tumors without those mutations as observed in melanoma and gliomas 47,50 . Though, we did observe that telomere were shorter in BCC tumors with the promoter mutations than in tumors without those mutations, the difference was not statistically significant. The decreased telomere length in tumors with somatic TERT promoter mutations signifies delayed effect in stabilization of telomeres until after generation of genomic instability 22,51 .
Noncoding mutations within the DPH3 promoter were detected in 38.2% of the lesions. The effect of the mutations, located within the proximity to an ETS binding motif, has been rather ambiguous 31 . The presence of mutations exclusively at dipyrimidinic sites coupled with typical CC > TT tandem alterations in the DPH3 promoter again signified the role UV radiation through sun-exposure. In the absence of a definite functionality, the DPH3 promoter mutations being mere passenger events cannot be ruled out 52 . However, we did observe that the DPH3 mutations associated with an increased risk of cutaneous neoplasms. DPH3 encodes a short peptide involved in electron transfer during the synthesis of eukaryotic diphthamide and forms a complex with Kti13, which functions in both tRNA and translational elongation factor 2 (EF2) modifications 53,54 . Overexpression of DPH3 has been shown to promote migratory ability of murine melanoma and downregulation of its expression was shown to inhibit cellular invasion and metastasis in vivo 55,56 .
In conclusion, we described noncoding mutations within the TERT and DPH3 promoter at high frequency in BCC tumors in addition to frequent alterations in PTCH1 and TP53 genes that impair protein functions. Although, the mutations in PTCH1, TP53, and TERT promoter have been reported earlier, this is the first communication to report alterations in those loci and the DPH3 promoter simultaneously in BCC tumors and to describe methylation levels within the THOR and transcription of the telomerase reverse transcriptase subunit [23][24][25] . Interestingly, the alterations in PTCH1, TP53 and DPH3 promoter occurred more frequently in tumors with TERT promoter mutations. The simultaneous presence of the mutations in four loci associated with fair skin and history of cutaneous neoplasms. It is likely that the increased cellular proliferation following the activation of Hedgehog signaling or elimination of checkpoints due to p53 loss would require telomere buttressing due to increased cellular proliferation, which is probably attained through telomerase rejuvenation via TERT promoter mutations. The TERT promoter mutations impart a distinctive tumor environment characterized by epithelial-to-mesenchymal transition (EMT) gene expression signature and MAPK signaling 57 . Although, BCC tumors rarely metastasize; a small but significant proportion of tumors progress to an invasive disease with metastasis resulting in patient mortality. Inoperable patients with metastatic disease are treated with vismodegib and sonidegib that impede Hedgehog pathway through inhibition of SMO 58 . Patients with basal cell nevus syndrome respond to the targeted treatment with a low rate of acquired resistance 12 . It is likely the TP53 and TERT promoter mutations could impact the treatment outcome.

Material and Methods
Patients and tissues. Fresh-frozen BCC tumor tissues from 191 patients were included in this study. Of those patients 191 patients, corresponding normal appearing skin from 115 patients was available and included in the study (Supplementary information; Supplementary Table 1). Seventy-one (37.2%) nodular, 42 (22.0%) superficial, and 4 (2.1%) adenoid tumors were grouped as low risk. Twenty-four (12.6%) infiltrative, 7 (3.7%) were morphoeic, 6 (3.1%) micronodular, and one metatypical type were categorized as high risk. For 36 (18.9%) tumors histological data were not available. BCC lesions were retrieved from the Biobank of the Instituto Valenciano de Oncología in Valencia, Spain and collected at the Department of Dermatology of the University of L' Aquila, Italy. The study was approved by the Instituto Valenciano de Oncología Ethics Committee and all methods were performed in accordance with the relevant guidelines and regulations. A written informed consent was signed by all study participants.
Mutational Analysis. DNA and total RNA were extracted from fresh-frozen tissues using the QIAGEN AllPrep DNA/RNA/miRNA Universal Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instruction.
Mutations at different loci were screened using PCR and Sanger sequencing (Supplementary information; Supplementary Table 6). For PTCH1 a set of 21 pairs of primers flanking from exon 3 to exon 24 of the PTCH1 gene was used and TP53 was sequenced for mutations in exons 5, 6, 7, 8 and 9. For TP53 and PTCH1 each PCR was carried out in 10 µl volume containing 10 ng DNA, 0.11 mM dNTP and 0.15 µM of each primer and, 5-µl HotStar Taq Plus Master mix kit (Qiagen). These regions presented a high frequency of mutations as reported in COSMIC database for the transcript ENST00000269305 that encodes the longest isoform. For TERT promoter, PCR was carried out in a 10-µl volume containing 10 ng DNA, 50 mM KCl, 0.11 mM dNTP, 2 mM MgCl 2 and 0.11 mM of each primer and Taq polymerase (GENAXXON biosciences GmbH). PCR for DPH3 was carried out in a 10-µl volume containing 10 ng DNA, 25 mM MgCl 2 , 0.11 mM dNTP, 5% DMSO and 0.11 mM of each primer and Taq polymerase (GENAXXON biosciences GmbH). Denaturation temperatures for PCR were set at 95 °C for 45 sec, annealing temperature were adjusted according to primer sequences, and polymerization at 72 °C for 30 sec each for 35 cycles. Amplified products were purified with ExoSAP (Illustra ExoProStar, GE Healthcare Life Sciences) and were subjected to 35 cycles of sequencing reaction with a di-deoxy terminator kit (BigDye Terminator v3.1 Cycle Sequencing Kit, life technologies, Thermo Fisher Scientific Inc.) and analyzed in a capillary sequencer (AbiPrism 3130xl Genetic Analyzer).
Sequencing data were analyzed using Geneious Pro 5.6.5 software and sequences from the NCBI gene database were used as references, PTCH1  Fig. 5). The point mutations in the PTCH1 and TP53 genes were annotated by using web-based tool Mutalyzer and variant effector predictor (VEP) algorithm from Ensembl database 59 . The sequence topology for the PTCH1 gene was generated with PROTTER 60 . Mutations with intron/exon boundaries were analyzed for the effect on splicing using maximum entropy model 61 .
For splice site analysis on the 5′ end, three nucleotides from exon and 6 nucleotides from the following intron were included in the model; for 3′ splice site, 20 nucleotides from the intron and three nucleotides from the preceding exon were included in the model.

Measurement of TERT promoter methylation.
A 213 bp genomic region within the TERT promoter, from −560 to −774 (chr5:1,295,665-1,295,878; hg19) from ATG start site was screened for CpG sites in-silico using MethPrimer 62 . The selected region included 14 CpG sites and primers were designed to amplify both methylated and unmethylated sequences. PCR product was either sequenced directly or after cloning into T-overhang vector. The sequence data files were further analyzed for CpG methylation status using a web-based bisulfite sequencing analysis tool called QUMA (QUantification tool for Methylation Analysis) under default settings 63 .

Measurement of TERT mRNA expression by quantitative real-time PCR (qRT-PCR).
For measurement of gene expression, each reverse transcription reaction was performed using approximately 1.0 µg RNA and random hexamer primers using a cDNA synthesis kit (Thermo Scientific, Waltham, USA). The real-time PCR was carried out in triplicates on a 384-well layout using primers specific for TERT (Supplementary Table 6) and primers for the GUSB gene (Qiagen). Difference in gene expression levels were calculated following the ΔΔC T method; GUSB expression was used as an internal reference (ΔC T ) and difference in expression levels was calculated between BCCs and matched tumor-surrounding skin (ΔΔC T ) followed by performing a log 2 transformation.
Measurement of telomere length. Relative telomere length in tumor DNA was measured using the monochrome multiplex PCR assay as described previously including minor modifications [64][65][66] . The standard curve was used to quantify the telomere (T) and albumin genes (S) based on the respective average Ct values obtained in triplicate. The relative telomere length was expressed as the ratio between T/S. Inter-assay and intra-assay variation were determined by duplicating the reference DNA for all the dilutions in all the assays performed.
Statistics. The associations between mutations in PTCH1, TP53, TERT promoter, DPH3 promoter and patient age at diagnosis, sex, phototype, nevus count, solar lentigos, history of cutaneous neoplasms, exposure Scientific RepoRtS | (2020) 10:8005 | https://doi.org/10.1038/s41598-020-65057-2 www.nature.com/scientificreports www.nature.com/scientificreports/ to sun, immunosuppressive treatment and histology were determined by χ 2 -tests and size of the effect determined by odds ratio (OR) and 95% confidence intervals (CIs) in logistic regression model. Multivariate logistic regression was also carried and included statistically significant variables from univariate analysis. Box plots were drawn to for TERT expression and telomere length in tumors with and without TERT promoter mutations and differences analyzed by two-tailed t-test. Similar box-plots were also drawn after stratification of tumors based on THOR methylation status.