Centenarian Exomes as a Tool for Evaluating the Clinical Relevance of Germline Tumor Suppressor Mutations.

OBJECTIVES: The aim of the present study was to evaluate the clinical relevance of mutations in tumor suppressor genes using whole-exome sequencing data from centenarians and young healthy individuals.
METHODS: Two pools, one of centenarians and one of young individuals, were constructed and whole-exome sequencing was performed. We examined the whole-exome sequencing data of Bulgarian individuals for carriership of tumor suppressor gene variants.
RESULTS: Of all variants annotated in both pools, 5080 (0.06%) are variants in tumor suppressor genes but only 46 show significant difference in allele frequencies between the two studied groups. Four variants (0.004%) are pathogenic/risk factors according to single nucleotide polymorphism database: rs1566734 in PTPRJ, rs861539 in XRCC3, rs203462 in AKAP10, and rs486907 in RNASEL. DISCUSSION: Based on their high minor allele frequencies and presence in the centenarian group, we could reclassify them from pathogenic/risk factors to benign. Our study shows that centenarian exomes can be used for re-evaluating the clinically uncertain variants.

Introduction

Tumor suppressor genes (TSGs) are part of the cell cycle regulation mechanism. Mutations in TSGs have long been associated with malignant disorders and often lead to the onset of neoplasms in different organs and systems. Germline genetic disturbances in TSGs could be transmitted and predispose their carriers to inherited forms of cancer. Additional somatic mutations might trigger cell transformation and cancer development.

The introduction of next generation sequencing technology has made it possible many new variants in TSGs to be discovered and their role in different malignant disorders evaluated.

Online publicly available databases have compiled lists of TSGs with germline and somatic mutations associated with oncogenesis. Interpreting the clinical significance of variants in TSGs is however complicated, and databases with information about the functional significance of variants are often incomplete and based on contradictory evidence. For most variants no meta-analyses are performed, which hinders risk estimation and clarification of their role as driver mutations. Yet, such information is very important in medical genetic counseling.

Centenarians are individuals who presumably carry protective variants against risk factors in the environment and are most likely to carry fully functional TSGs. Also, the frequency of risk alleles in TSGs in centenarians should be lower compared to the frequency in the general population. Centenarian exomes/genomes could thus be considered “golden standard” for healthy longevity, making them appropriate for studying the functional significance of variants in TSGs.

The aim of this study is to determine the carriership of risk/pathogenic TSG alleles in healthy Bulgarian individuals and to evaluate their clinical relevance using centenarian exomes. Whole-exome sequencing (WES) data were obtained from pools of individual DNA (pool-seq), an approach which has repeatedly been shown to provide reliable allele frequency estimates.[1]

Methods

Ethics Statement

The study was approved by the Ethics committee of the Medical University No. 1030/08.03.2017. It is conducted in accordance with national and international legislation for conducting research with participation of human subjects. Each participant in the study received information about the aims of the project and signed written informed consent. Interviews conducted with the individuals gathered information about their health status and medical history for major age-related diseases.

Sample Collection, DNA Isolation, and Pool Preparation

Blood or buccal swap samples were collected from 93 healthy Bulgarian individuals. The individuals have been divided into two groups according to their age: 32 centenarians aged 100 to 106 years and 61 healthy individuals aged 18 to 30 years. Whole genome DNA has been extracted using phenol-chloroform extraction protocol, and equimolar amounts of DNA from each individual were used to construct the pools.

Whole-Exome Sequencing

The DNA pools were whole-exome sequenced (BGI Genomic Services, China) at ×250 coverage which is required for pool-seq sequencing to ensure that alleles with low frequency are also detected. The obtained .vcf files were annotated using the web-based service wANNOVAR.[2] Following the “best practice” recommendations for pool-seq data, we performed robust filtering on variant calling: number of individuals per pool >30, genotype quality >99, mapping quality >60, number of reads per minor allele frequency >2, total depth of coverage—above 30.[1,2] The total number of variants annotated in both pools after applying these filters was 89 810 (72 791 in both pools, 8253 in centenarian pool only, and 8766 in control pool only).

Databases of TSG and Variants

The Tumor Suppressor Gene Database (TSGene 2.0) is an online comprehensive resource for data based on pan-cancer analysis of TSGs and contains a list of 1217 TSGs (1018 protein-coding and 199 non-coding genes).[3]

We examined the WES data of Bulgarian individuals for the presence and frequency of variants in TSGs from this database.

Additionally, we compiled a list of 30 161 variants in 248 genes from the following publicly available TSG databases:

UniProt, containing 181 human TSGs.[4] The 29 557 variants in these TSGs were found in single nucleotide polymorphism database (dbSNP)[5]

DisGeNet database (604 additional variants in 67 additional genes)[6]

Results

TSG Database

The TSG database contains a list of 1217 TSGs. The obtained WES data from Bulgarian samples was inspected for the presence and frequency of variants from the genes in this database.

Altogether 5042 variants in 851 TSGs—4092 variants in both pools, 424 variants in centenarians only, and 526 in controls only—were found in the Bulgarian WES data. For the 4092 variants found in both pools, we used Fisher exact test to determine the significance of allele frequency differences between the two pools (Figure 1).The frequencies of variants with significant difference (false discovery rate > 1.0 × 10−5) in allele frequency, altogether 46 variants, are plotted on Figure 2.

Figure 1.

Manhattan plot of varants in TSGs called in both centenarian and control pools. The horizontal lines correspond to P = 1.0 × 10−5 and 5.0 × 10−7. TSGs indicate tumor suppressor genes.

Figure 2.

Forty-six variants in TSGs showing significant difference (FDR > 1.0 × 10−5) in allele frequency between centenarians and controls. FDR indicates false discovery rate; TSG, tumor suppressor gene.

The variants below the diagonal identity line (n = 24) have significantly higher allele frequency in the control pool. Three of these variants are benign (rs558114 in MUS81, rs741810 in FUS, and rs1057090 in MCPH1), but for the remaining 21 variants no information is reported in dbSNP database. It can be speculated that these variants can potentially have negative impact on longevity, and further analyses on them are needed.

Analysis of all variants called in both pools of the Bulgarian WES data, regardless of their allele frequency differences, revealed only 2 pathogenic/risk alleles, rs1566734 in PTPRJ gene and rs861539 in XRCC3 (Table 1). There were no variants designated as pathogenic or risk alleles among those called in the centenarian pool only (containing 424 variants) or in the control pool only (containing 526 variants).

Table 1.

Risk Variants in TSGs in Bulgarian Healthy Individuals.

Pool	Ref	Alt	Func	Gene	dbSNP	dbSNP Significance	MAFCent	MAFCont	P Value	FDR
Risk variants in TSGs from TSG database
Both	A	C	Exonic	PTPRJ	rs1566734	Pathogenic	0.198	0.246	.246	0.246
Both	C	T	Exonic	XRCC3	rs861539	Risk factor	0.474	0.375	.089	0.201
Risk variants from UniProt and DisGeNet
Both	T	C	Exonic	AKAP10	rs203462	Risk factor	0.375	0.350	.491	0.466
Both	C	T	Exonic	RNASEL	rs486907	Risk factor	0.328	0.349	.420	0.442

Abbreviations: AKAP10, A-kinase anchoring protein 10; Alt, alternative allele; cent, centenarians; cont, controls; dbSNP, single nucleotide polymorphism database; FDR, false discovery rate adjusted P value; Func, functional consequence; MAF, minor allele frequency; PTPRJ, protein tyrosine phosphatase receptor type J; Ref, referent allele; RNASEL, ribonuclease L; TSGs, tumor suppressor genes; XRCC3, X-ray repair cross complementing 3.

UniProt and DisGeNet Databases

Of the 30 161 variants in TSGs (assembled from UniProt and DisGeNet), 38 were found in the pooled exome sequencing data of Bulgarian individuals.

Two of 38 variants were found as risk factors in the dbSNP (rs203462 in AKAP10 and rs486907 in RNASEL; Table 1).

Among TSG variants found in the centenarian or young individual pools only, none were pathogenic or risk factors.

Discussion

Pathogenic variants in TSGs can be inherited and entail risk for oncogenesis. Different pathogenic alleles predispose to tumor development to a different degree. Assessment of the risk in the healthy carriers of pathogenic alleles is important for medical genetic counseling and in setting up preventive measures for reducing cancer risk.

For this study, we used exome data from Bulgarian centenarians as a tool for evaluation of the clinical effect of germline pathogenic tumor suppressor mutations in healthy carriers.

Data for 89 810 variants in 17 133 genes was obtained by performing WES on two DNA pools, one composed of Bulgarian centenarians and one of coethnic young and healthy individuals. The 1217 genes listed in TSGene 2.0 database were screened for variants in our WES data and 5042 were found (see Figure 3, left side of the workflow chart). Two of these are nominated as having clinical relevance in dbSNP database—pathogenic/risk factor (cf Table 1).

Figure 3.

Flowchart for pathogenic (TSVs) in Bulgarian WES data. TSVs indicate tumor suppressor variants; WES, whole-exome sequencing.

From the 30 161 tumor suppressor variants (TSVs) from UniProt/DisGeNet databases, 38 were found in Bulgarian WES data, and 2 of these are proposed by dbSNP as risk factors (Figure 3, right side of the workflow chart).

All in all, from 89 810 variants in Bulgarian healthy individuals—young individuals and centenarians, 5080 are in TSGs (0.06%), and only 4 of these are pathogenic/risk TS variants (0.004%).

None of these 4 variants show significant difference in allele frequencies between the two pools, raising the question about the risk entailed by these variants.

The rs1566734 (Gln276Pro) polymorphism in PTPRJ (protein tyrosine phosphatase receptor type J) is a missense single nucleotide polymorphism, which is classified as pathogenic TSV by DisGeNet and as associated with colon cancer (variant disease-associated ([VDA] score = 0.700), colorectal cancer (VDA score = 0.020), papillary thyroid carcinoma (VDA score = 0.010). Our data show no significant difference in allele frequency of this variant between Bulgarian centenarians and young individuals, and the frequency is relatively high (0.198/0.246, respectively). As this variant is not associated with tumor development in centenarians and could thus be reclassified from pathogenic to benign in Bulgarian population.

This polymorphism in PTPRJ has previously been genotyped in Japanese population and was found that in combination with another variant in the same gene (rs1503185, Arg326Gln) increases the risk for developing lung, head and neck, colorectal, and esophageal cancers.[7-10] On the other hand, Toland et al [11] had found that this variant was not associated with increased colorectal cancer risk. Iuliano et al [7] also observed a nonsignificant increased frequency of homozygotes for Gln276Pro polymorphism in papillary thyroid carcinoma cases in 2 distinct Caucasian populations. A meta-analysis has also found this variant not to be associated with increased risk of colorectal cancer.[12]

The findings for nonassociation of rs1566734 polymorphism are in line with our proposal this variant to be reclassified from pathogenic to benign.

The rs861539 (Thr241Met) polymorphism in the DNA repair XRCC3 (X-ray repair cross-complementing 3) gene is classified as risk factor in DisGeNet database based on repeated findings of association with increased risk of melanoma (VDA score = 0.800), breast carcinoma (VDA score = 0.100), lung cancer (VDA score = 0.100). Our results show that Bulgarian healthy individuals have higher, albeit nonsignificantly different, frequency in the centenarian group compared to the young individuals. This variant thus seems to be benign in Bulgarian population.

A study on Taiwanese population has previously found that carriers of this variant, in combination with certain other variants in XRCC group genes, had up to ×10 increased risk of developing oral cancer,[13] and that TT genotype was more prevalent in patients with breast cancer.[14] Also, a Belgian study found that the T allele of this variant is not a risk allele for breast cancer.[15] A meta-analysis[16] concludes that this variant is the risk factor for melanoma, but the association is not significant in Caucasians. There was no significant association found between this polymorphism and lung cancer risk[17] in a population from northern Spain and prostate cancer in north Indian population.[18]

The high frequency of this variant in centenarians, who do not have cancer, indicates that this variant is not a risk factor in Bulgarians, adds to the evidence from other studies for no association of this variant with oncogenesis.

The rs203462 in AKAP10 (A-kinase anchoring protein 10) is a missense variant designated as risk factor for breast cancer (VDA score = 0.010) by DisGeNet. Our WES data show that this variant is carried with high frequency in Bulgarian population, and no difference in allele frequencies between the centenarian and young individuals, indicating that the clinical significance of this variant in Bulgarian population is benign.

A study[19] from 2007 has shown that this is functional genetic variant, located in the kinase-binding domain of the A-kinase anchoring protein, and is associated with familial breast cancer. However, in line with our results, a study on a Polish population finds no significant differences in genotype or allele distribution of this variant between nonagenarians and newborns.[20]

The rs486907 in RNASEL (ribonuclease L) is a missense variant, included in DisGeNet as risk factor for prostate cancer (VDA score = 0.100). Our WES data show that this variant is with high and similar frequency in centenarian/young groups, indicating this variant not to be cancer risk factor.

Alvarez-Cubero et al [21] suggest the role of this variant as prognostic marker and predictor of aggressiveness and progression of prostate cancer.[21] Analysis of Ashkenazi Jewish descent individuals shows its contribution in early onset and familial forms of prostate cancer.[22]

Other researchers Shea et al [23] and Robbins et al [24] find no evidence of association between this R462Q polymorphism and prostate cancer risk in case/control analysis of Afro-Caribbeans and Afro-Americans. These results support our proposal for reclassification of this polymorphism from risk factor to benign.

Conclusion

Literature evidence for pathogenic/risk variants is often contradictory and based on single gene sequencing in case–control studies. Single nucleotide polymorphic variants in such studies that have high population frequencies might rather indicate the presence of unknown nearby risk alleles. The limitations of such study design are overcome by the introduction of WGS and WES technologies. Using WES data from Bulgarian centenarians and young individuals, 4 variants could be reclassified from pathogenic/risk factors to benign based on their high minor allele population frequencies and presence in centenarians.