Frequency of common polymorphisms in Caveolin 1 (CAV1 ) gene in adults with high serum triglycerides from Colombian Caribbean Coast

Abstract Background: Caveolin 1 gene (CAV1) has been associated with insulin resistance, metabolic syndrome and hypertension in humans. Also, it has been related to high serum triglycerides in rodents, however there is little evidence of this relation in humans. Aim: To describe frequencies of common variations in CAV1 in adults with high serum triglycerides. Methods: A case-control study was carried out with adults from Colombian Caribbean Coast. A whole blood sample was employed to measure serum concentrations of triglycerides, glucose, total cholesterol and HDLc. Six common Single Nucleotide Polymorphism (SNP) in CAV1 were genotyped (rs926198, rs3779512, rs10270569, rs11773845, rs7804372 and rs1049337). Allelic and genotypic frequencies were determined by direct count and Hardy-Weinberg Equilibrium (HWE) was assessed. Case and control groups were compared with null-hypothesis tests. Results: A total of 220 cases and 220 controls were included. For rs3779512 an excess in homozygotes frequency was found within case group (40.4% (GG), 41.3% (GT) and 18.1% (TT); Fis=0.13, p=0.03). Another homozygotes excess among case group was found in rs7804372 (59.5% (TT), 32.3% (TA) and 8.2% (AA); Fis= 0.12, p= 0.04). In rs1049337, cases also showed an excess in homozygotes frequency (52.7% (CC), 35.0% (CT) and 12.3% (TT); Fis= 0.16, p= 0.01). Finally, for rs1049337 there were differences in genotype distribution between case and control groups (p <0.05). Conclusion: An increased frequency of homozygote genotypes was found in subjects with high serum triglycerides. These findings suggest that minor alleles for SNPs rs3779512, rs7804372 and rs1049337 might be associated to higher risk of hypertriglyceridemia.

CAVEOLIN-1 protein is indispensable for caveolae formation in plasma membrane of many cellular types, however adipose tissue expresses the highest amount of this protein in humans, rodents and other animals, suggesting that CAVEOLIN-1 and caveolae are involved in energy homeostasis 5,6 . Moreover, in vitro studies demonstrated that CAVEOLIN-1 expression is significantly increased during adipogenesis, also it has been saw that the presence of this protein is necessary for appropriated preadipocytes proliferation and differentiation 7,8 .
In adipocytes, CAVEOLIN-1 has been directly linked to insulin receptor as an enhancing factor for intrinsic phosphorylation 9-11 , and it has been associated with stability of the Glucose Transpoter 4 (GLUT 4) 7,12 . Additionally, CAVEOLIN-1 and caveolae are closely implicated in triglycerides metabolism through their association with lipid droplets conformation, and also caveolar domains in cell membrane are suspected to be sites for fatty acids entry and for triglycerides synthesis [13][14][15] .
Animal models with rodents have demonstrated that caveolin-1 is crucial for triglycerides metabolism in adipose tissue since it seems to be involved in lipid droplets distribution and lipid storage dynamics 16 . Therefore it has been found that CAV1 knockout individuals under an exposition to high fat diet are lean and more likely to develop hypertrygliceridemia as well as insulin resistance 16,17 .
In humans, common variations in CAV1 gene have been associated to metabolic disorders such as insulin resistance, high blood pressure and metabolic syndrome [18][19][20][21] . Initially two exonic variations in CAV1 were associated with high blood pressure and metabolic syndrome in a population from Southern Spain 18 . After that, an intronic SNP (rs926198) was associated with insulin resistance, and partially related to high blood pressure in a translational study that employed a human cohort with European ancestry 20 . Recently, the same intronic polymorphism was related to metabolic syndrome, type 2 diabetes and low HDL cholesterol in a human cohort with European ancestry, and these findings were replicated in an admixed Hispanic group, which represents the first evidence involving CAV1 with lipid alterations in humans 19 .
In spite of growing evidence linking CAV1 with lipid homeostasis, there are still little findings supporting an association with triglycerides metabolism in human groups. Triglycerides (or triacylglycerol), the most relevant molecule for energy storage, are mainly accumulated in adipose tissue, but also are constitutive components of blood plasma and serum 22 . High serum triglycerides (or hypertriglyceridemia) is a prevalent alteration among adult population that is generally associated with a highenergy unbalanced environment 23 . In developing countries, dyslipidemias and particularly high serum triglycerides are public health issues, which contribute to cardiovascular mortality among more vulnerable communities 24 .
Together with most of Latin American countries, Colombia has registered a permanent elevation in frequencies of hypertriglyceridemia with notable intra-national variations in a range of prevalence that has been found as low as 19.3% and as high as 43.9% 25,26 . In Cartagena de Indias, on the Colombian Caribbean Coast, high serum triglycerides were found in 38.2% of women 27 , and between 67.7% and 77.8% of the general population 28 , which represents one of the highest national frequencies.
Considering that Colombian Caribbean populations would require novel strategies to reduce the impact of high serum triglycerides on public wellness, it has been proposed that genetic factors might be a source of valuable data to design preventive interventions or therapeutic approaches. In this way, this study was aimed to describe frequencies of six SNPs in CAV1 in adults with high serum triglycerides from the Colombian Caribbean Coast.

Materials and Methods
A cross-sectional, case-control study was carried out in urban zone of Cartagena de Indias, an admixed population with a predominant European ancestry (60%) as a consequence of Spaniards colonization between XVI and XIX centuries 29 .
Non-sibling individuals from both genders, aging 18-80 years old were included. Subjects with primary endocrine alterations or previous diagnosis of genetic disease, eating behavior disorders, recent hospitalization, cancer, personal antecedents of surgical treatment for obesity, pregnant women, and breast-feeding mothers were excluded. All subjects were asked to participate, and their informed consent was required, following Universidad de Cartagena Ethics Committee recommendations.
Subjects were selected through a convenience sampling in work places and household clusters. For this procedure seven sampling points were included, which represent a socio-economic mixed population that is currently employed in diverse labors (e.g., management, civil construction, industrial production, security services, and teaching/learning). Unemployed populations were included through a random cluster sampling that was executed in five socio-demographic diverse areas.
Classification of cases and control groups was based on Joint Interim Statement (JIS) criteria for high serum triglycerides 30 . Therefore, cases were those subjects with serum triglycerides ≥150 mg/dL or clinical antecedents of high serum triglycerides, while subjects with serum concentrations below 150 mg/dL and without previous diagnosis of dyslipidemia were selected as controls.
Fasting blood sample were collected for measuring of serum glucose, cholesterol, triglycerides and HDL cholesterol. Colorimetric assays to determine serum concentrations were performed in UNIMOL laboratory at the University of Cartagena following standard protocols. Serum biochemical parameters were assessed in order to bring a wide description of metabolic status. Among of these variables only serum triglycerides was employed as selection criteria for cases or controls.
Socio-demographic variables were registered employing a validated survey previously applied to Caribbean urban communities 27,28 . Cases and controls groups were paired by sex and age according to this data.
For genetic analysis, 3 mL of whole blood were collected in sterile tubes containing EDTA as anticoagulant; these samples were refrigerated and transported to UNIMOL laboratory at the University of Cartagena for further procedures. Genomic DNA was isolated through the employment of a commercial kit (Promega Corp., Madison, USA) following principles published by others 31 , and then quantified through fluorometric method with Qubit 2.0 (Thermo Fisher Scientific Inc., Waltham, USA) using proper reagents.
For genotyping assay, prior to molecular procedures, common variants in CAV1 were selected employing data from HapMap project and Haploview 4.2 software (Broad Institute, Cambridge, USA) 4,32 . Genomic data from CEU population (Utah Residents with Northern and Western European Ancestry) was used as reference to select SNPs with Minor Allele Frequency (MAF) ≥0.25, and correlation coefficient (r 2 ) ≥0.8. According to these parameters, six polymorphisms were selected for further molecular analysis (Table 1).
Selected SNPs were genotyped through quantitative Polymerase Chain Reaction (qPCR), using specific TaqMan probes (Thermo Fisher Scientific Inc., Waltham, USA). Allelic discrimination was automatically performed with end-point fluorescent data, which was analyzed through StepOne Real-Time PCR Software (Thermo Fisher Scientific Inc., Waltham, USA).
Sociodemographic data, anthropometric measures and serum concentrations were described employing main tendency and frequency values. When appropriate, mean values were compared through Student's t test, while frequencies were compared with X 2 or Fisher's exact tests. Allelic and genotypic frequencies were determined by direct count, linkage disequilibrium was estimated employing Arlequin 3.5 33 , and Hardy-Weinberg Equilibrium (HWE) was assessed through F is values with Genetix 4.05 software. Differences in genotype distributions among cases and controls groups were assessed through null-hypothesis tests, as it has been described elsewhere 34 .

Results
A total of 440 subjects were included in the study (220 cases and 220 controls). Mean values for age, Body Mass Index (BMI), anthropometric and serum biochemical variables were represented in the Table 2. There were not statistically significant differences in sex distribution, age and anthropometric variables between both groups.
For polymorphism rs926198, minor allele (C) was found in 27.9% of cases and in 30.9% of control subjects (p= 0.34). In the group of cases, genotypes TT, TC and CC were found in 52.7%, 38.6% and 8.6% of subjects, respectively; while in controls these genotypes were found in 50.0%, 38.2% and 11.8%, respectively. Hardy-Weinberg equilibrium (HWE) was found in both groups with F is values in 0.04 (p= 0.31) and 0.10 (p= 0.07) for cases and controls, respectively (Table 3).
In rs10270569, when a change C/T was reported, minor allele (T) was found in 18.4% of cases and 18.9% of controls (p= 0.93). Homozygotes CC, heterozygotes CT and homozygotes TT represented 67.7%, 27.7% and 4.6% in the case group, respectively; and in the control group these proportions were 64.1%, 34.1% and 1.8%. HWE was estimated in the case group (F is = 0.07, p= 0.17), and also in the control group (F is = -0.11, p= 0.97 ) ( Table 3).
For SNP, rs11773845 minor allele (C) was found in 39.1% and 45.9% of cases and controls, respectively (p= 0.047). Among case subjects, proportions of homozygotes AA, heterozygotes AC and homozygotes CC were 40.0%, 41.8% and 18.2%, respectively. In the control group, genotypic distribution AA/AC/CC was 32.3%, 43.6% and 24.1%, in that order. In both groups F is was equal to 0.12 (k= 0.04), suggesting an excess in heterozygote frequency ( Table 3).
To assess differences in genotypic distribution between cases and controls, homozygotes for minor allele were compared in a recessive model. According to this approach, there were statistically significant differences between cases and controls for SNP rs1049337 (p <0.05) ( Table 3).

Discussion
The current research has contributed to the description of common variants in CAV1 gene in an admixed population from Caribbean Region, which represents an initial approach to cumulate evidence on genotype/phenotype relation involving SNPs in CAV1 and metabolic disorders that has been previously found in American-Hispanic population 19,20 . In the present study, marginal differences in allele and genoptype frequencies were found, suggesting that there are little statistical significance to determine a genetic association, however further researches with a meticulous association analysis should not be discharged. According to results from this work, when case and control groups were compared in the present study, the minor allele T in rs1049337 and the homozygotes TT were found to be more frequent in cases (Table 3). In a simple way these differences between groups suggest a tendency that relates CAV1 variations and hypertriglyceridemia in adult population, where the risk allele is acting as a recessive character. A significant Hardy-Weinberg disequilibrium was also found in the case group (F is = 0.16, p= 0.01), which indicates that an excess in heterozygotes frequency was found among these subjects, supporting findings of a recessive model that was described above. Other authors have previously analyzed this SNP in cohorts with obesity-related diseases, however a dominant model showed that there were not differences in genotype distribution between subjects with and without insulin resistance (p= 0.08) 20 . Another study applied a similar design to describe association between high serum triglycerides and rs926198, nevertheless there were not differences in allele or genotype frequencies 19 . Hence, this work brings novel evidence about this particular polymorphic variations and its behavior in groups with high serum triglycerides from an admixed population.
Since rs1049337 showed an interesting behavior, allele frequencies found in this study were compared to previous reports from reference populations in order to bring a deeper description that might be helpful in further analysis. In this regard, allele frequency for rs1049337 was similar to that reported by the HapMap project for CEU population (MAF: 27.6% vs 25.0%) ( Table 4). In spite of ethnic similarities with MEX (Mexican ancestry in Los Angeles, California), it was not possible to establish a comparison with current samples from Cartagena de Indias due to the lack of data for this SNP in HapMap Project database 35 .
In general, allelic frequencies and genotype distributions in this sample were similar to those found in groups with European ancestry living in North America (Table 4). In fact, findings from Cartagena de Indias were notably close to European frequencies even in SNPs with a high inter-population variation such as rs926198, rs3779512 and rs1049337, suggesting that Spaniard ancestry is a predominant component in contemporary population from Cartagena de Indias 36 .
Previous studies have described a three-hybrid admixed population in Cartagena de Indias, in which European ancestry represented more than a half of all lineages 29,37 , thus current genotype frequencies of CAV1 SNPs are in agreement with an elevated representation of Spanish ancestry. In this sense, it is possible that genetic association studies could be replicated employing simple fitting methods that require a conservative increase in size of study or sample size. However, following STREGA consortium recommendations 38 , genetic stratification in Cartagena de Indias remains a concerning issue that should be included as a confounder in weighted analytical procedures.
In spite of the historical contribution of African immigrants forced to come to the Colombian Caribbean Coast as slaves [39][40][41] , this recent demographic event showed little influence on CAV1 SNPs genetic distribution, considering that local allelic frequencies were distant to those found in reference African groups (Table 4).  These results suggest that sex-biased gene flow caused by social predominance of European groups during colonial period in Latin America and the Caribbean might be responsible for some of current population genetic patterns 42, 43 .
Although results from the present study have pointed out to a relationship between CAV1 polymorphisms and high serum triglycerides, there are some limitations that must be considered. First, this study was carried out on an admixed population where genetic stratification has been widely described, therefore a confounding effect must be discharged in further analysis 28,37 . Second, group sizes is relatively small for complete discrimination of genetic associations, thus this study was mostly restricted to descriptive and comparative observations.

Conclusion
Allelic and genotype frequencies described for SNPs in CAV1 in a sample from Cartagena de Indias were similar to those observed in groups with European ancestry. Additionally, differences in allelic and genotype distributions between cases and controls found in this study settled a precedent for incoming genetic association studies with a larger sample and higher performance in analytical procedures.