Integrative variants, haplotypes and diplotypes of the CAPN3 and FRMD5 genes and several environmental exposures associate with serum lipid variables

To determine whether the integrative variants, haplotypes and diplotypes of the calpain 3 (CAPN3) and the FERM domain containing 5 genes (FRMD5) and several environmental exposures are associated with an implication in lipid homeostasis, which are associated with cardiovascular risk. Genotyping of the CAPN3 rs4344713 and FRMD5 rs524908 was performed by Sanger sequencing in 1,640 subjects (Jing, 819 and Han, 821). Multivariate analyses of covariance models that adjusted by age, gender, body mass index (BMI), blood pressure and lifestyle (smoking and drinking), were constructed using variants, haplotypes and diplotypes of the CAPN3 rs4344713 and FRMD5 rs524908 as predictors and changes in lipid variables. Significant associations with low-density lipoprotein cholesterol and apolipoprotein (Apo) B were found. Linkage disequilibrium with each other showed the haplotype-phenotype associations with triglyceride and ApoA1. This study also suggested pleiotropic associations of the CAPN3-FRMD5 diplotypes with lipid variables. As potential confounders, diastolic blood pressure (DBP) and BMI were significantly associated with lipid variables. We conclude that integrative variants, haplotypes and diplotypes of the CAPN3 rs4344713 and FRMD5 rs524908, as well as DBP and BMI are associated with serum lipid variables in the Jing and Han populations.


Results
Sociodemographic and clinical characteristics. The sociodemographic and clinical characteristics of the Jing and Han participants are list in Table 1. As both groups were matched by age and gender, there were no statistically significant differences by age and gender. The mean value of body mass index (BMI) in the Jing ethnic minority was significantly higher than that in the Han population. However, the values of systolic blood pressure (SBP) and pulse pressure (PP) in the Jing group were significantly lower than those in the Han group. There were equivalent significant differences between the two ethnic groups for the social lifestyle. The percentage of light-moderate alcohol consumption was significantly higher in the Jing than in the Han individuals, while the % of severe cigarette smoking and severe alcohol consumption was significantly lower in the Jing than in the Han subjects. For serum lipid variables, the levels of total cholesterol (TC), TG and apolipoprotein (Apo) B were significantly higher and those of high-density lipoprotein cholesterol (HDL-C), ApoA1 and the ApoA1/ApoB ratio were significantly lower in the Jing than in the Han participants. However, no difference was noted in fasting serum low-density lipoprotein cholesterol (LDL-C) levels between the two ethnic groups. Genotypic and allelic frequencies. Regarding the analysis of gene frequencies in the CAPN3 and FRMD5, the prevalence of minor allelic, genotypic, haplotypic and diplotypic frequencies of the CAPN3 rs4344713 and FRMD5 rs524908 was significantly different between the Jing and Han populations ( Table 2). We found that the rare variant or minor allele frequency (MAF) of the CAPN3 rs4344713 and FRMD5 rs524908 were significantly higher in the Jing than in the Han subjects (2.93% vs. 1.58%, P = 0.009 and 21.31% vs. 17.78%, P = 0.011; respectively). The prevalence of the CAPN3 rs4344713CC and FRMD5 rs524908CC genotypes was also significantly higher in the Jing than in the Han ethnic groups (0.25% vs. 0.12%, P = 0.038 and 5.25% vs. 3.90%, P = 0.044; respectively). Two variants in the Jing and Han populations were in the Hardy-Weinberg equilibrium (HWE, P > 0.05 for all). There was weak linkage disequilibrium (LD) between the two SNPs (D′ = 0.834, r 2 = 0.075 in Jing, D′ = 0.761, r 2 = 0.049 in Han, P = 0.007). Pearson correlation analysis demonstrated that the CAPN3 rs4344713 and FRMD5 rs524908 SNPs were not totally independent from each other (Fig. 2). The frequencies of haplotype and diplotype are list in Table 2. Two haplotypes (rs4344713T-rs524908A and rs4344713T-rs524908C) and two diplotypes (rs4344713TT-rs524908AA and rs4344713TT-rs524908AC) with a frequency > 5%, called "common", were identified in the Jing and Han populations; respectively. We combined 1 haplotype (rs4344713C-rs524908C both in the Jing and Han) and 3 diplotypes (rs4344713TT-rs524908CC and rs4344713CT-rs524908AC in Han, and rs4344713CT-rs524908CC in Jing) with frequencies between 1% and 5% into one group, called "low-frequency". The rest of haplotypes and diplotypes with frequencies less than 1% into one group, called "rare". The haplotype of rs4344713T-rs524908A and diplotype of rs4344713TT-rs524908AA were the commonest, which accounted for over half of the % of each ethnic group. Table 3, the levels of LDL-C, ApoB and the ratio of ApoA1 to ApoB in the Jing population were significantly different between the CAPN3 rs4344713TT and CT/ CC or among the FRMD5 rs524908AA, AC and CC genotypes after the Bonferroni correction (P ≤ 0.022). There was no association of the two SNPs and serum lipid phenotypes in the Han population.
Diplotypes and serum lipid phenotypes. As shown in Table 5, the levels of TG (rs4344713TT-rs524908AA, rs4344713TT-rs524908CC and rs4344713CT/CC-rs524908AA), HDL-C (rs4344713TT-rs524908CC), LDL-C (rs4344713CT/CC-rs524908CC), ApoA1 (rs4344713TT-rs524908AA), ApoB (rs4344713TT-rs524908AA) and ApoA1/ApoB (rs4344713TT-rs524908AA, rs4344713TT-rs524908AC and rs4344713TT-rs524908CC) in Jing; and TG (rs4344713CT/CC-rs524908AA) and ApoA1 (rs4344713TT-rs524908CC) in Han were significantly different between the diplotype carriers and non-carriers Genotypes, haplotypes and diplotypes and serum lipid phenotypes. Table 6 shows the magnitude and direction of correlation between serum lipid levels and integrative genotypes, haplotypes and diplotypes in   the two populations. Many of the examining variants showed significant correlation with serum lipid levels in multiple linear regression analysis; although, these variants did not show significant association with serum lipid levels in the analysis of covariance (ANCOVA). Furthermore, the significantly differences in the LDL-C level change per increment of CAPN3 genotype was Beta/Se 0.059/0.046 (P = 0.016), as well as HDL-C, ApoA1, ApoB levels and the ApoA1/ApoB ratio change per increment of FRMD5 genotype were − 0.359/0.150 (P = 0.017); − 0.063/0.010 (P = 0.009); 0.079/0.010 (P = 0.001) and − 0.096/0.015 (P < 0.001) in the Jing participants, respectively. In addition, the significantly differently in the HDL-C change per increment of their haplotype was Beta/ Se 0.535/0.233 (P = 0.023), as well as marginal significantly differently in the ApoA1 change per increment of their haplotype was Beta/Se 0.159/0.075 (P = 0.033) and the ApoA1/ApoB ratio change per increment of their diplotype was Beta/Se 0.414/0.201 (P = 0.040). However, we did not obtain any significant association (having taken correction for multiple comparisons into account) with any of these biomarkers in the Han population of this ethnic-stratified analysis. The novel algorithm to combine haplotype cluster-and diplotype-based analyses is useful for identifying more precise and distinct signals over single-locus tests. Fig. 3, Pearson correlation analysis showed that the integrative variants, haplotype and diplotype evidence connects the CAPN3 rs4344713 and FRMD5 rs524908 to lipid variables. Several environmental exposures such as age, gender, cigarette smoking, alcohol consumption and traditional cardiovascular risk factors such as BMI and blood pressure levels were also correlated with serum lipid phenotypic parameters of both ethnic groups.

Discussion
In the present study, we first confirmed that (i) not only the CAPN3 rs4344713 CT/CC genotype and the FRMD5 rs524908 CC genotype, but also the rs4344713C-rs524908C haplotype and the rs4344713CT/CC-rs524908CC diplotype were good proxies for higher serum LDL-C and ApoB levels in the Jing but not in the Han populations; (ii) the rs4344713T-rs524908A, rs4344713T-rs524908C and rs4344713C-rs524908A haplotypes were associated with TG and ApoA1 in the Jing particiapnts, as well as associated with TG in the Han subjects; and (iii) the results of our study also suggest a pleiotropic association of the CAPN3-FRMD5 diplotypes with lipid variables in the both Jing and Han populations. In summary, the data presented here suggest differences in lipid variables between the two populations might partially attribute to CAPN3 rs4344713 and FRMD5 rs524908 variants, while this heterogeneity needs to be confirmed in other populations in order to better assess its important. Moreover,   to combine haplotype cluster-and diplotype-based analyses is useful for identifying more precise and distinct signals over single-locus tests. A couples of years ago, Schunkert et al. 8 found that the individuals with CAPN3 rs24212710G allele were positively associated with plasma TG levels (n = 153,909, P = 1.7 × 10 −11 ) and increased the risk of CVD (n = 79,267, P = 5.3 × 10 −3 ) in the European population. Likewise, in the same European population's large-scale association study, the FRMD5 rs2929282A allele carriers had higher TG levels (n = 83,616, P = 2 × 10 −9 ) and risk of CVD (n = 81,446, P = 2.8 × 10 −3 ) than the rs2929282A allele non-carriers. However, no association was seen between the FRMD5 rs2929282 and BMI (n = 122,284, P = 1.1 × 10 −2 ). In the present study, we found that the CAPN3 rs4344713CT/CC genotype and the FRMD5 rs524908CC genotype were more frequent in the Jing than in the Han populations. The subjects with the CAPN3 rs4344713CT/CC genotype and the FRMD5 rs524908CC genotype tend to have higher serum LDL-C and ApoB levels in the Jing ethnic minority. However, no such association     Continued was observed in the Han population. The reason for this discrepancy among these studies is not fully understood. It might result from the difference in the genetic background, ethnic LD pattern and/or environmental factors. On inter-locus interaction analyses, we confirmed the CAPN3 rs4344713 T > C variant in low LD with the FRMD5 rs524908 A > C. At least, they are not statistically independent. In single locus analysis, only LDL-C, ApoB levels and the ApoA1/ApoB ratio were associated with the detected variants. However, in haplotype cluster-and diplotype-based analyses, significant associations with pleiotropic lipid variables were found in the CAPN3-FRMD5 haplotype and diplotype models. These findings indicate that a potential gene-gene interaction might exist between the CAPN3 rs4344713 and FRMD5 rs524908 variants. Unfortunately, no previous study has investigated the inter-locus interaction between these variants, and therefore we cannot make comparisons with our results. Although, a statistically significant variant-, haplotype-and diplotype-based association was noted in this study, the biological mechanism underlying these genes and their interactions is still yet to be defined.
When assessing the association of several environmental exposures with lipid variables, we showed that diastolic blood pressure (DBP) and BMI were strongly associated with serum lipid levels in the both Jing and Han populations. A novel extension to the phenome-wide association study (pheWAS) approach 20 , using Bonferroni corrections, permutation testing and estimates of the false discovery rate to consider the strength of results given the number of tests performed demonstrated strongly associated outcomes included DBP and BMI with lipid profiles. They also found novel evidence of effects of BMI on a global self-worth score. Previous studies have reported not only higher DBP 21 but also both overweight and obesity 22,23 are associated with an increased risk of CVD death and all-cause mortality. Furthermore, achieving and maintaining a healthy weight has a favorable effect on metabolic risk factors (blood pressure, lipids) and low CVD risk. Whatever, we considered these variables as potential confounders and performed a crude statistical analysis and a multivariate analysis adjusting for these variables.
CAPN3 encodes CAPN3, a heterodimer consisting of a large and a small subunit, is a major intracellular protease 24 , although its function has not been well established. Mutations in this gene 25 are associated with lipid metabolism. Alternate promoters and alternative splicing result in multiple transcript variants encoding different isoforms and some variants are ubiquitously expressed. Likewise, FRMD5 encodes FERM-containing protein FRMD5, a p120-catenin interacting protein, regulates lipid metabolic progression 26 . Our study is the first one carried out in a south Chinese population and the first to find a significant association between this integrative variants, haplotype and diplotypes of the CAPN3 and FRMD5 with lipid variables. After adjustment for several potential confounders, this association remained significant.
Sampling bias cannot be ignored. In our statistics, the sample is collected in previous specimen library that some members of the intended population are less likely to be included than others. It results in a biased sample, a non-random sample 27 of a population (non-human or human factors) in which all individuals, or instances, were not equally likely to have been selected 28 . If this is not accounted for, results can be erroneously attributed to the   phenomenon under study rather than to the method of sampling. As an ascertainment bias sometimes classified as the inevitable type of bias. Our analyses benefited from a strong a priori hypothesis, a good sample size for a sub-ethnic study, and a healthy population, making generalizations to other samples less problematic than with clinical samples. However, there are some limitations: first, the association in the present study should be tested in more racially diverse populations. Second, many of our conclusions regarding associations did not survive the FDR correction for multiple testing, so should be considered preliminary. Third, it is not clear why some variants in LD with each other did not show the same genotype-phenotype associations. Moreover, observational analyses may not provide robust evidence of causality, as they are susceptible to confounding, reverse causation and measurement error. The last but not the least, the selected variants may not the accurate and precise hits because tagging SNPs selection bias is inevitable.
In conclusion, integrative variants, haplotypes and diplotypes of the CAPN3 rs4344713 and FRMD5 rs524908, as well as DBP and BMI associated with lipid variables in the Jing and Han populations. These associations may contribute to future novel dyslipidemia diagnosis biomarkers and pharmacogenomics therapy targets. Subjects. The Jing participants in the current study were mainly recruited from three islands: Wanwei, Wutou and Shanxin. Eligibility criteria were (1) ≥ 18 years of age; (2) willing to participate in epidemiologic investigation, biochemical measurement and genetic analysis of this project; (3) their birthplaces and those of their parents, grandparents, and great-grandparents clustered in the abovementioned three islands; (4) their ethnicities and those of their parents, grandparents, and great-grandparents were the same one (Jing); (5) myocardial enzymes, aspartate aminotransferase, alanine aminotransferase, and creatinine results within normal range; and (6) normal electrocardiography (ECG) results. Exclusion criteria were (1) with history of coronary artery disease, stroke, diabetes, hyper-or hypo-thyroids, and chronic hepatic and renal disease; (2) use of lipid lower drugs, including prescription, over-the-counter and nutraceuticals; and (3) other medications known to affect lipid profiles, such as insulin use. After granting informed consent, participants underwent a baseline-screening visit and a fasting blood draw. A total of 1800 Jing subjects were asked to participate in the study and 1674 Jing subjects actually participated 17 . The response rate was 93% in Jing. A total of 52 persons (3.1%) with a history and/or evidence of other diseases were excluded from the Jing samples. The sampling of the Han population was also done in the same place, by the same method. Importantly, the participants of this study were randomly selected from our previous stratified randomized samples, just included 818 Jing subjects and 822 Han participants.

Materials and Methods
Biochemical analyses. Protocols for measuring lipid variables have been previously described 29 . All lipid variables were measured in our Clinical Science Experiment Center. The levels of fasting serum TC, TG, HDL-C and LDL-C in the samples were determined by enzymatic methods with commercially available kits. Fasting serum ApoA1 and ApoB levels were assessed by the immuneturbidimetric immunoassay, respectively.
Tagging SNPs selection. We selected two SNPs in the CAPN3/FRMD5 with the following steps: (i) CAPN3 gene clusters, which were selected from previous GWAS associated with lipid-metabolism. FRMD5 gene clusters are found to be close to CAPN3 gene clusters and associated with serum lipid level especially TG. (ii) Tagging SNPs, which were established by Haploview (Broad Institute of MIT and Harvard, USA, version 4.2) and functional SNPs predicted to lead to serum lipid changes from current version of online resource (1000 Genome Project Database). (iii) CAPN3 rs2412710 and FRMD5 rs2929282, which were selected by the block-based approach. This strategy is enable by the correlations between tagging SNPs as manifested as LD. Although classic is not goal of tagging SNP selection, innovative tagging SNPs selection bias is inevitable.
Variant genotyping. Genomic DNA was extracted from leucocytes of venous blood using the phenol-chloroform method. Genotyping was performed in each population separately with PCR and Sanger sequencing. The PCR products of the samples were sequenced with a sequencer ABI Prism 3100 Genetic Analyzer (Applied Biosystems, International Equipment Trading Ltd., Vernon Hills, IL, USA) in Shanghai Sangon Biological Engineering Technology & Services Co. Ltd., Shanghai, China 30 .
Statistical analyses. The statistical analyses were performed with the statistical software SPSS 21.0 (SPSS Inc., Chicago, IL, USA). Quantitative variables were presented as the mean ± SD for those, that are normally distributed, whereas the medians and interquartile ranges for TG, which is not normally distributed. General characteristics between the two groups were compared by the ANCOVA. The distributions of the genotype, allele, haplotype and diplotype between the two groups were analyzed by the chi-squared test; The HWE, Pair-wise LD, frequencies of haplotype and diplotype interaction comprising the variants were calculated using Haploview (version 4.2; Broad Institute of MIT and Harvard) and PLINK (version 1.9; Broad Institute of MIT and Harvard) software. The association of the genotypes, haplotypes and diplotypes with lipid variables was tested by the Univariant. As 2 loci are not totally independent from each other, we determined these variants represented 1.86 independent loci using Matrix Spectral Decomposition. For all analyses, a Bonferroni test adjusted P-value less than 0.025 was considered statistically significant, and the P-value between 0.025-0.05 was considered marginal statistically significant. Generalized linear models were used to assess the association of the genotypes (common homozygote genotype = 1, heterozygote genotype = 2, rare homozygote genotype = 3), alleles (the minor allele non-carrier = 1, the minor allele carrier = 2), haplotypes (the haplotype non-carrier = 1, the haplotype carrier = 2) and diplotypes (the diplotype non-carrier = 1, the diplotype carrier = 2) with lipid variables. The model of age, gender, BMI, waist circumference, systolic blood pressure, DBP, pulse pressure, cigarette smoking, alcohol consumption and fasting blood glucose level were adjusted for the statistical analysis. The heart-map of inter-locus models was measured by R software (version 3.3.0).