The identification of two regulatory ESCC susceptibility genetic variants in the TERT-CLPTM1L loci

The chromosome 5p15.33 TERT-CLPTM1L region has been identified by genome-wide association studies as a susceptibility locus of multiple malignancies. However, the involvement of this locus in esophageal squamous cell carcinoma (ESCC) development is still largely unclear. We fine-mapped the TERT-CLPTM1L region through genotyping 15 haplotype-tagging single nucleotide polymorphisms (htSNPs) using a two stage case-control strategy. After analyzing 2098 ESCC patients and frequency-matched 2150 unaffected controls, we found that rs2853691, rs2736100 and rs451360 genetic polymorphisms are significantly associated with ESCC risk in Chinese (all P<0.05). Reporter gene assays indicated that the ESCC susceptibility SNP rs2736100 locating in a potential TERT intronic promoter has a genotype-specific effect on TERT expression. Similarly, the CLPTM1L rs451360 SNP also showed allelic impacts on gene expression. After measuring TERT and CLPTM1L expression in sixty-six pairs of esophageal cancer and normal tissues, we observed that the rs2736100 G risk allele carriers showed elevated oncogene TERT expression. Also, subjects with the rs451360 protective T allele had much lower oncogene CLPTM1L expression than those with G allele in tissue specimens. Results of these analyses underline the complexity of genetic regulation of telomere biology and further support the important role of telomerase in carcinogenesis. Our data also support the involvement of CLPTM1L in ESCC susceptibility.

As one of the most common and fatal malignant tumors in the world, esophageal squamous cell carcinoma (ESCC) was diagnosed at a relatively high frequency in China [24]. It has been shown that heavy alcohol drinking, tobacco smoking, micronutrient deficiency and dietary carcinogen exposure are risk factors of this lethal disease [25,26]. However, only a portion of exposed individuals develop ESCC, indicating that genetic factors may also impact esophageal malignant transformation. Considering the involvement of the 5p15. 33 TERT-CLPTM1L locus in ESCC is still largely unclear, we examined the associations between 15 haplotype-tagging SNPs (htSNP) across the TERT-CLPTM1L locus and ESCC risk in three large independent hospital-based casecontrol studies. To investigate the biological function of three ESCC susceptibility SNPs, we examined impacts of these genotypes on TERT or CLPTM1L expression using luciferase reporter gene assays and inspected the association between these polymorphisms and gene expression in esophageal tissues.

Stratified analyses of associations between rs2853691, rs2736100 or rs451360 SNP and ESCC risk
The risk of ESCC associated with the rs2853691, rs2736100 or rs451360 SNP was further investigated by stratifying for age, sex, smoking and alcohol drinking status using the combined data of three case-control sets ( Table 3). For rs2853691, a significantly increased risk of ESCC associated with the rs2853691 GG genotype compared with the AA genotype was observed for both groups stratified by sex, smoking and drinking status (all P < 0.05) or the group aged 58 years or younger www.impactjournals.com/oncotarget (P = 5.4 × 10 −5 ). Additionally, the rs2853691 AG genotype was only associated with ESCC risk in the male group (P = 0.026), the smoking group (P = 0.008) or the drinking group (P = 0.013). For rs2736100, significant associations between TG or GG genotype and ESCC risk were observed in all stratified groups (all P < 0.05), but not in the drinking group (P = 0.974). There was a significantly multiplicative gene-drinking interaction (P interaction = 0.012). For rs451360, the TT genotype was only associated with ESCC risk in the male group (P = 0.006), the group aged 58 years or younger (P = 0.016), the non-smoking group (P = 0.025) or the drinking group (P = 0.005). However, significant associations between the rs451360 GT genotype and ESCC risk were observed in all stratified groups (all P < 0.05), but not in the female group (P = 0.326).

Functional relevance of TERT rs2736100 and CLPTM1L rs451360 genetic variants on gene expression
Considering the chromosome location of the three ESCC susceptibility SNPs, we only investigated the impacts of TERT rs2736100 and CLPTM1L rs451360 SNPs on gene expression. The rs2736100 variant locates in the intron 2 region of TERT. As shown in Figure  1A, reporter gene assays demonstrated that the intron 2 segment containing the rs2736100 flanking sequence showed promoter activities in KYSE30 and KYSE150 ESCC cells. Moreover, the TERT rs2736100G allelic reporter construct (pTERT-G) showed significantly higher luciferase activities compared to the rs920778T allelic reporter construct (pTERT-T) (both P<0.01) ( Figure 1A). We next examined whether the ESCC susceptibility SNP rs451360 has an allele-specific effect on the intronic enhancer activity on CLPTM1L expression in ESCC. Either KYSE30 cells or KYSE150 cells transfected with the CLPTM1L pCL-T allelic plasmid showed significantly lower luciferase activities compared to cells expressing pCL-G allelic reporter construct (both P<0.05) ( Figure 1B). We next examined whether these two ESCC susceptibility SNPs has an allele-specific effect on gene

DISCUSSION
In the current study, we systematically examined the impacts of SNPs in the TERT-CLPTM1L loci on ESCC susceptibility via a case-control design as well as gene expression of TERT or CLPTM1L in vitro and in vivo. After genotyping 15 htSNPs in the discovery stage, we identified three ESCC susceptibility genetic polymorphisms (rs2853691, rs2736100 and rs451360) which were validated in two validation case-control sets. Reporter gene assays indicated that the ESCC susceptibility SNP rs2736100 locating in a potential TERT intronic promoter has a genotype-specific effect on TERT expression. Similarly, the CLPTM1L rs451360 polymorphism also showed allelic effects on gene expression. Genotype-phenotype correlation data supported the regulatory role of these two genetic variants in TERT or CLPTM1L gene expression in vivo. Our observations support the hypothesis that genetic polymorphisms in oncogene regulatory elements might explain a part of ESCC genetic basis besides those genetic variants identified by GWAS [27][28][29][30][31].
Interestingly, TERT rs2736100 has been found to be associated with risk of lung cancer [1,2,24], glioma [9], testicular cancer [12],colorectal cancer [33], acute myeloid leukemia [34], pancreatic cancer [35] and bladder cancer [36]. However, its involvement in ESCC etiology is still largely unclear. To the best of our knowledge, this is the first case-control study to investigate the association between the TERT rs2736100 polymorphism and ESCC risk. We observed a significantly increased ESCC risk among individuals with TERT rs2736100 G allele compared to carriers of rs2736100 T allele. It has been reported that increased TERT expression in ESCC tissues were observed compared to normal tissues [18,19], which indicated the oncogene nature of TERT in ESCC. Since rs2736100 G allele is associated with elevated TERT expression, the associations between the polymorphism and increased cancer risk are biologically plausible.
CLPTM1L appears to act as an oncogene with significantly increased expression in malignant tissues [20][21][22][23]. In line with this, CLPTM1L silencing by miR-494 can inhibit cell growth and invasion and induce ESCC cell apoptosis [23]. The CLPTM1L rs451360 polymorphism has been associated with decreased risk of lung cancer among different ethnic populations, with T allele as a protective allele [5,[37][38][39][40][41]. Here, we provided first evidences that rs451360 SNP also play a part in ESCC susceptibility, which are unlikely to be attributable to unknown confounding factors due to having relatively large sample sizes, significantly increased odd ratios with small P values. Additionally, our genotype-phenotype correlation data between the rs451360 genetic variant and gene expression supports the case-control study since the protective T allele carriers showing less oncogene CLPTM1L expression. Since the TT genotype of the functional rs451360 SNP is relatively rare (about 1-2% among common populations), the potential clinical translation of this genetic variant might be compromised.
In conclusion, we demonstrated that there are three genetic polymorphisms (rs2853691, rs2736100 and rs451360) in the TERT-CLPTM1L loci are significantly associated with ESCC risk in Chinese populations. Our results underline the complexity of genetic regulation of telomere biology and further support the important role of telomerase in carcinogenesis. Our data also support the involvement of CLPTM1L in ESCC susceptibility. These results may lead to better understanding of ESCC etiology in different populations.

Study subjects
This study consisted of three case-control sets: (a) Jiangsu set: 588 ESCC cases from Huaian No. 2 Hospital (Huaian, Jiangsu Province, China) and sexand age-matched 600 controls. (b) Shandong set: 1000 cases with ESCC from Shandong Cancer Hospital, Shandong Academy of Medical Sciences (Jinan, Shandong Province, China) and sex-and age-matched ( ± 5 years) 1000 healthy controls. (c) Hebei study: 510 ESCC patients from Bethune International Peace Hospital (Shijiazhuang, Hebei Province, China) and 550 sex-and age-matched healthy controls. Sixty-six pairs of ESCC specimens and esophagus normal tissues adjacent to the tumors were obtained from surgically removed specimens of patients in Bethune International Peace Hospital and Huaian No. 2 Hospital. All individuals were ethnic Han Chinese. At recruitment, the informed consent was obtained from each subject. The detailed information on subject recruitments can be found in Supplementary  Table S1 and our previous studies [42][43][44]. This study was approved by the institutional Review Boards.

SNP selection and genotyping
The TERT-CLPTM1L gene loci cover a 91716bp region of chromosome 5p15.33 and contain a great number of SNPs. An htSNP approach was utilized to analyze the TERT-CLPTM1L genetic polymorphisms globally [45]. Genotyped HapMap SNPs among Han Chinese and Japanese populations (HapMap Rel 21, NCBI B36) with a minor allele frequency >5% were included in the selection. The htSNPs were chosen in a 95716bp region (91716bp TERT-CLPTM1L loci and 2kb up-stream as well as 2kb down-stream regions of the TERT-CLPTM1L gene loci). Using a method described previously with the sample size inflation factor, Rh 2 , of ≥ 0.8, fifteen htSNPs were selected with Haploview 4.2 software on a block-by-block basis (Supplementary Table S2). www.impactjournals.com/oncotarget TERT-CLPTM1L htSNPs were genotyped through the MassArray system (Sequenom Inc., San Diego, California, USA). A 5% blind, random DNA samples was analyzed in duplicates and the reproducibility was 99%. To reduce the costs of the study, we genotyped the TERT-CLPTM1L rs2853691 A>G, rs2736100 T>G and rs451360 G>T SNPs in two validation sets using the PCR-based restriction fragment length polymorphism (RFLP) as described in Supplementary Table S3. A 5% samples were genotyped by two investigators and the reproducibility was 98.0%.

Luciferase reporter gene constructs
Specific primer pairs (Supplementary Table S4) with the KpnI and XhoI restriction sites were used to amplify the intron 2 segment of TERT (chr.5: 1319429~1319865 bp [GRCh38.p2] including the rs2736100 flanking region) from human genomic DNA carrying TERT rs2736100 TT genotype or GG genotype. Similarly, the intron 16 segment of CLPTM1L (chr.5: 1285954~1286844 bp [GRCh38.p2] including the rs451360 flanking region) was amplified with human genomic DNA carrying CLPTM1L rs451360 GG genotype or TT genotype. The PCR products were then digested with KpnI and XhoI (New England Biolabs) and ligated into an appropriately digested pGL3-Basic vector (TERT) or pGL3-Promoter vector (CLPTM1L). The resultant TERT reporter gene plasmids were designated pTERT-T or pTERT-G, which were only different at rs2736100 polymorphic site. The resultant CLPTM1L reporter gene plasmids were named as pCL-T or pCL-G, which were identical except for the different allele at rs451360 polymorphic site. Restriction analysis and complete DNA sequencing confirmed the orientation and integrity of these constructs.

Real-time analysis of TERT and CLPTM1L mRNA
Total cellular RNA was isolated from sixty-six pairs of ESCC specimens and esophagus normal tissues adjacent to the tumors with TRIzol Reagent (Invitrogen) and converted to cDNA using the PrimeScript RT Master Mix (TAKARA). TERT and CLPTM1L mRNA expression in cancerous and normal esophagus tissues was examined using the TaqMan real-time quantity PCR method.
Relative gene expression quantization for TERT (ABI, Assay ID Hs00972656_m1) and CLPTM1L (ABI, Assay ID Hs00363947_m1) was calculated using β-actin (ABI, Assay ID 4333762T) as an internal reference gene was carried out using the ABI 7500 real-time PCR system in triplicates.

Statistics
Pearson's χ 2 test was used to examine the differences in demographic variables, smoking status, drinking status, and genotype distributions of rs2853691, rs2736100 or rs451360 SNP between ESCC cases and healthy controls. The associations between rs2853691, rs2736100 or rs451360 genotypes and ESCC risk were estimated by odds ratios and their 95% confidence intervals computed by logistic regression models. All ORs were adjusted for age, sex, smoking or drinking status, where it was appropriate. A P value of less than 0.05 was used as the criterion of statistical significance, and all statistical tests were two-sided. All analyses were performed using SPSS 16.0 (SPSS Inc.).