2.3.2. Results from Next-Generation Sequencing Studies

GWAS identified more than 100 genetic variants influencing BP [13]. However, the causal variants underlying the majority of genetic associations remained unknown. In recent years, three different NGS approaches have been proposed to study rare variants in hypertension and BP (Table 3).

The first approach is to check GWAS signals and describe novel associations by performing a re-sequencing of only a few genes previously indicated by GWAS. This approach, commonly called target re-sequencing, is cheaper and allows one to highlight the variations within the whole frequency spectrum in a precise genomic locus. The strategy was adopted by the CHARGE Consortium. In the frame of this consortium, the signals identified by precedent GWAS were re-sequenced with the aim of describing novel variations with large effects on several common diseases [74]. Concerning BP, within the CHARGE Targeted Sequencing Study, target re-sequencing of 4178 Europeans was performed on six BP genes identified by GWAS (*ATP2B1*, *CACNB2*, *CYP17A1*, *JAG1*, *PLEKHA7*, and *SH2B3*), however, neither common nor rare variants were consistently associated with the trait with large effect sizes, independently of the original GWAS signals [63].

Regarding hypertension, an association with rs3918226 in the *eNOS* gene promoter was described in the GWAS from Salvi et al. [75] (OR for minor allele T = 1.34 (95% CI, 1.25–1.44); *<sup>p</sup>* = 1.03 × <sup>10</sup>−14). In 2013, a 140 kb genomic area encompassing the *eNOS* gene was re-sequenced from the same group. The study identified 338 variants, including 61 novel variants, and rs3918226 still appeared as the SNP most closely associated with hypertension. Moreover, if compared with the C major allele, the T risk allele was associated with lower *eNOS* transcriptional activity when tested in HeLa cells [64].

A second approach is whole exome sequencing (WES) in which only the coding portions of the genome, (about 2%), estimated to harbor 85% of disease-causing mutations, are sequenced [76]. A WES study was performed on DNA samples from 17,956 individuals of European and African ancestries, included in the CHARGE, National Heart, Lung, and Blood Institute GO Exome Sequencing Project, Rotterdam Study, and the Erasmus Rucphen Family cohorts. These findings implicated the effect of the aggregation of 95 rare coding variants in *CLCN6* on decreasing BP levels of 3–4 mmHg, independently of the tagging SNP rs17367504 previously reported. The effect size described here was about four- to six-fold larger than previous common BP variants from GWAS [66].





*Int. J. Mol. Sci.* **2018**, *19*, 688

Genome-wide

(OR).

 Complex Trait Analysis (GCTA), Combined Association in the Presence of Linkage (CAPL), support vector machine (SVM), Unified Sparse Regression (USR), Odds Ratio

Two additional studies exploited WES data to focus on selected genes. Loss-of-function mutations in *SLC12A3*, *SLC12A1*, and *KCNJ1* genes, essential for normal renal NaCl reabsorption, cause Bartter's and Gitelman's syndromes. Their exons were screened to search for rare heterozygous variants within the Framingham Heart Study offspring cohort. Thirty different mutations were observed. The mean long-term SBP among mutation carriers was 6.3 mmHg lower than the mean of the cohort (*p* = 0.0009). For DBP, the mean effect was −3.4 mmHg (*p* = 0.003) [62]. Findings from previous GWAS indicated *ULK4* and *MAP4* genes, encoding, respectively, a Serine/Threonine-Protein Kinase and a non-neuronal microtubule-associated protein, as related to BP and hypertension [8,77]. Thirty-six rare haplotype blocks were found to be significantly associated with BP in *ULK4* gene, and ten in *MAP4* gene [61]. The study described above was conducted in the frame of the Genetic Analysis Workshops (GAWs). Since 1982, GAWs were held by a group of multidisciplinary scientists to deal with the role of genetics in complex diseases. For GAW18, GT2D-GENES Consortium and the San Antonio Family Heart Study provided data on the whole genome, systolic and diastolic BP, and related covariates in two Mexican American samples. In the GAW19, new data were included reaching a collection of WGS, WES, and gene expression data from 20 large families in addition to a set of 1943 unrelated subjects whose exome sequences were available. Simulated phenotypes were also included for each sample on the basis of the real sequence data [78]. Several papers have been published so far, mostly on methodological approaches (see the following paragraph "Statistical analysis of rare variants") to handle rare variations in relation to hypertension.

The third and most comprehensive NGS approach to examine the effect of rare variants is represented by WGS. Until now, to the best of our knowledge, only studies published within the GAWs analysed WGS data (Table 3) to search for genetic variations associated to hypertension, likely because a very large sample is needed to highlight rare variants, and this feature heavily affects the costs of the study. Three studies failed to identify significant associations after correction for multiple testing [68,69,72]. In the frame of the GAW18, Zhao et al. used novel sliding window approaches and a simulated dataset to analyse 142 unrelated individuals focusing on chromosome 3. The most significant windows fell into the known *MAP4* gene, considering both SBP and DBP. Other windows were reported within *SUMF* and *ARHGF3* genes in relation to DBP, and in *FLNB* and *BTD* for SBP [67]. Wang and Wei performed a gene-based genome-wide scan of 103 unrelated individuals to search for hypertension-associated genes. After using three different methods, only the *SETX* gene exhibited significant association. This gene consists of large intronic regions; indeed, most of the rare variants detected fall in intronic regions. The risk of hypertension, estimated after collapsing all the intronic variants, was 9.5 (OR = 9.5, 95% CI (3.43, 28.70); *<sup>p</sup>* = 8.8 × <sup>10</sup>−7) [65]. Other significant findings were reported within the *MACROD1/LRP16* locus [70], *ADCY5*, and *UBE2E2* genes [71], and in an additional 23 genes [73] using different statistical approaches.
