Transforming growth factor-β signaling pathway-associated genes SMAD2 and TGFBR2 are implicated in metabolic syndrome in a Taiwanese population

The transforming growth factor-β (TGF-β) signaling pathway and its relevant genes have been correlated with an increased risk of developing various hallmarks of metabolic syndrome (MetS). In this study, we assessed whether the TGF-β signaling pathway-associated genes of SMAD family member 2 (SMAD2), SMAD3, SMAD4, transforming growth factor beta 1 (TGFB1), TGFB2, TGFB3, transforming growth factor beta receptor 1 (TGFBR1), and TGFBR2 are associated with MetS and its individual components independently, through complex interactions, or both in a Taiwanese population. A total of 3,000 Taiwanese subjects from the Taiwan Biobank were assessed. Metabolic traits such as waist circumference, triglyceride, high-density lipoprotein cholesterol, systolic and diastolic blood pressure, and fasting glucose were measured. Our results showed a significant association of MetS with the two single nucleotide polymorphisms (SNPs) of SMAD2 rs11082639 and TGFBR2 rs3773651. The association of MetS with these SNPs remained significant after performing Bonferroni correction. Moreover, we identified the effect of SMAD2 rs11082639 on high waist circumference. We also found that an interaction between the SMAD2 rs11082639 and TGFBR2 rs3773651 SNPs influenced MetS. Our findings indicated that the TGF-β signaling pathway-associated genes of SMAD2 and TGFBR2 may contribute to the risk of MetS independently and through gene–gene interactions.


Results
describes the demographic and clinical characteristics of the study population. First, we investigated the association between MetS and 8 TGF-β signaling pathway-associated genes, namely SMAD2, SMAD3, SMAD4, TGFB1, TGFB2, TGFB3, TGFBR1, and TGFBR2 genes. Based on LD, we filtered SNPs and selected 141 tag SNPs (Supplementary Table S2). Among the 141 tag SNPs assessed in this study (Supplementary Table S3), there were 20 tag SNPs, among those SNPs present in the SMAD2, SMAD3, SMAD4, TGFB2, TGFB3, and TGFBR2 genes, which showed evidence of an association (P < 0.05) with MetS (Table 2).
Furthermore, as shown in Table 2, the association of two key SNPs, namely SMAD2 rs11082639 and TGFBR2 rs3773651, with MetS remained significant after applying Bonferroni correction (P < 0.05/(141 × 3) = 0.0001). As demonstrated in Table 2, for the SMAD2 rs11082639 SNP, an increased risk of MetS was observed among subjects with MetS and those without MetS after adjustment for covariates such as age and sex for genetic models, including the additive model (odds ratio [OR] = 1.66; 95% confidence interval [CI] = 1.32-2.08; P = 1.4 × 10 −5 ) and recessive model (OR = 2.82; 95% CI = 1.80-4.43; P = 6.6 × 10 −6 ). Similarly, for the TGFBR2 rs3773651 SNP, an increased risk of MetS was observed among the subjects after adjustment for covariates for genetic models,  Table S2). Table 3 shows the OR analysis of the association of two key SNPs, namely SMAD2 rs11082639 and TGFBR2 rs3773651, with the individual components of MetS: (a) high waist circumference vs. normal waist circumference; (b) high triglyceride vs. normal triglyceride; (c) low high-density lipoprotein (HDL) vs. normal HDL; (d) high blood pressure vs. normal blood pressure; and (e) high fasting glucose vs. normal fasting glucose. As shown in Table 3, for the SMAD2 rs11082639 SNP, an increased risk of high waist circumference was observed among the subjects after adjustment for covariates for genetic models, including the additive model  Table 2. Covariate-adjusted odds ratio analysis of the relationship between MetS and 20 tag SNPs in the TGF-β signaling pathway-associated genes of SMAD2, SMAD3, SMAD4, TGFB2, TGFB3, and TGFBR2 with evidence of an association (P < 0.05). A1 = minor allele, A2 = major allele, CI = confidence interval, MetS = metabolic syndrome, OR = odds ratio, TGF-β = transforming growth factor-β. Analysis was performed with adjustment for covariates including age and sex. P values of < 0.05 are shown in bold.  The effect of SMAD2 rs11082639 on high waist circumference remained significant after Bonferroni correction (P < 0.05/30 = 0.0017). In addition, we examined the association of the two implicated SNPs with MetS traits considered as continuous variables, including waist circumference, triglyceride, HDL, systolic blood pressure, diastolic blood pressure, and fasting glucose. As shown in Supplementary Table S4, the results suggest an association between SMAD2 rs11082639 and MetS traits such as waist circumference (P = 0.0048) or fasting glucose (P = 0.0027).
In addition, the generalized multifactor dimensionality reduction (GMDR) analysis was used to assess the effects of interaction between two key SNPs, namely SMAD2 rs11082639 and TGFBR2 rs3773651, on MetS and its individual components, with adjustment for the covariates of age and sex. Table 4 summarizes the results of GMDR analysis for two-way gene-gene interaction models with covariate adjustment to assess the effects of the two SNPs on MetS. As shown in Table 4, the two-way model involving SMAD2 rs11082639 and TGFBR2 rs3773651 was significant (P < 0.001). The effects of these two-way models remained significant after Bonferroni correction (P < 0.05/6 = 0.008). This finding indicated that a potential interaction between SMAD2 and TGFBR2 influences MetS. However, no two-way gene-gene interaction model influencing the individual components of MetS was obtained in this study.
Furthermore, we utilized multivariable logistic regression analysis with adjustment for age and sex to assess the two-way SMAD2 rs11082639 and TGFBR2 rs3773651 interaction models selected by the GMDR method (Table 5). Our analysis revealed that subjects with the CC genotype of SMAD2 rs11082639 and the AA genotype of TGFBR2 rs3773651 had a 2.5-fold higher risk of MetS than those with the T allele of SMAD2 rs11082639 and the AA genotype of TGFBR2 rs3773651 (Table 5). Subjects with the CC genotype of SMAD2 rs11082639 and the G allele of TGFBR2 rs3773651 also had a 5.85-fold higher risk of MetS than those with the T allele of SMAD2 rs11082639 and the AA genotype of TGFBR2 rs3773651 (Table 5). Finally, subjects with the T allele of SMAD2 rs11082639 and the G allele of TGFBR2 rs3773651 had a 1.55-fold higher risk of MetS than those with the T allele of SMAD2 rs11082639 and the AA genotype of TGFBR2 rs3773651 (Table 5).
Finally, statistical power analysis revealed that the present study had 99.9% power to detect associations of SMAD2 rs11082639 and TGFBR2 rs3773651 with MetS among the subjects with MetS and those without MetS.

Discussion
To date, our association study is the first to examine whether 141 tag SNPs in 8 TGF-β signaling pathway-associated genes, namely SMAD2, SMAD3, SMAD4, TGFB1, TGFB2, TGFB3, TGFBR1, and TGFBR2 genes, are significantly associated with the risk of MetS and its individual components independently, through gene-gene interactions, or both among Taiwanese individuals. Here, we report for the first time that the SMAD2 and TGFBR2 genes may play a key role in the development of MetS in a Taiwanese population. Notably, the significant association of two key SNPs, namely SMAD2 rs11082639 and TGFBR2 rs3773651, with MetS remained significant after Bonferroni correction (P < 0.0001). In addition, our data revealed that interactions between the SMAD2 and TGFBR2 genes may contribute to the etiology of MetS. Finally, our data revealed that the SMAD2 rs11082639 SNP was associated with the individual component of MetS, namely high waist circumference.
In the present study, we found a positive association of MetS with five SNPs in the SMAD2 gene, particularly the rs11082639 SNP. We also detected an association of SMAD2 rs11082639 with the dichotomous categorical variable of high waist circumference and the continuous variables of waist circumference and fasting glucose. The SMAD2 gene encodes the SMAD2 protein, which is recruited and phosphorylated by TGF-β receptors in response to TGF-β signals 1,2 . SMAD2 forms a heteromeric SMAD complex with SMAD4, which is subsequently translocated to the cell nucleus. Yang et al. reported that in Smad2-silenced cells in mice, gene expression related to lipogenesis was suppressed and gene expression related to β-oxidation was increased when Smad2 was inactivated by using an animal model of nonalcoholic steatohepatitis, one of the hepatic manifestations of MetS 12 . Their results indicated that the Smad signaling pathway is crucial for modulating lipid metabolism and lipid accumulation in hepatocytes through the suppression of lipogenesis-related genes and the induction of β-oxidation-related genes that promote the development of nonalcoholic steatohepatitis 12 24 .
In the present study, we found an association of MetS with four SNPs in the TGFBR2 gene, particularly the rs3773651 SNP. The TGFBR2 gene encodes the TGF-β type II receptor, which is a member of the Ser/Thr protein kinase family and is a key mediator of TGF-β signaling transduction 1,2 . Several association studies have identified that the TGFBR2 gene is associated with CAD 17,20 , congenital heart defects 18,19 , and various cancers 25 , which are some of the hallmarks of MetS. By analyzing glucose and insulin tolerance test results in an animal model of nonalcoholic steatohepatitis, Yang et al. demonstrated that silencing Tgfbr2 may contribute to MetS manifestations, including weight gain and insulin resistance; their finding indicated that Tgfbr2 is a potent mediator in the development of hepatic steatosis, hepatocyte death, inflammation, and fibrosis 12 .
Furthermore, we inferred the epistatic effects between the SMAD2 and TGFBR2 genes on MetS by using the GMDR approach. To the best of our knowledge, no other study has evaluated the interactions between these genes. In addition to determining the statistical significance of the interaction, we examined the potential biological mechanism underlying the interaction. The functional relevance of the interactive effects of SMAD2 and TGFBR2 on MetS remains to be elucidated. The SMAD2 and TGFBR2 genes encode the SMAD2 protein and TGF-β type II receptor, respectively, which are two core components of the TGF-β signaling pathway. Understanding the nature and extent of cross-talk between these two core components of the TGF-β signaling pathway is a critical future research direction. Our current understanding of the TGF-β signaling pathway is as follows 1,2 : First, the TGF-β ligand binds to the TGF-β type II receptor at the plasma membrane, resulting in the formation of the complex of the TGF-β type I receptor with the TGF-β type II receptor. Subsequently, the TGF-β type II receptor phosphorylates the TGF-β type I receptor. In turn, the activated TGF-β type I receptor phosphorylates SMAD2 and SMAD3 proteins. Finally, the phosphorylated SMAD2 and SMAD3 proteins form a complex with the SMAD4 protein, which is subsequently translocated into the nucleus for regulating the expression of specific target genes. Thus, regarding potential explanations for the biological effects of synergy between SMAD2 and TGFBR2, we speculate that the SMAD2 and TGFBR2 gene products may participate in a common pathogenic pathway, that is the TGF-β signaling pathway, leading to MetS. This study has some limitations. The main weakness is that our findings require much more research to verify if the observations are replicated in diversified ethnic populations [26][27][28] . Moreover, in this study, a subject with MetS could be mistakenly classified as a subject without MetS because no records on the medications prescribed for treatment of dyslipidaemia, hypertension or diabetes were available. In the Taiwan Biobank, individuals were asked in a questionnaire whether a doctor had ever told them they had certain diseases such as hyperlipidemia, hypertension or diabetes. Thus, we further removed control subjects with a self-reported diagnosis of hyperlipidemia, hypertension or diabetes and then investigated the association between MetS and two key SNPs, namely SMAD2 rs11082639 and TGFBR2 rs3773651. The association of these two implicated SNPs with MetS remained significant after applying Bonferroni correction (Supplementary Table S5). Based on the candidate gene approach, the current findings are considered as only preliminary owing to the absence of supporting evidence from larger hypothesis-free GWAS studies 23,24 . Additional prospective clinical trials including other ethnic groups and conducting GWAS studies are warranted to thoroughly evaluate the association and interactions of the investigated genes with MetS and its individual components [29][30][31] .
In conclusion, we conducted an extensive analysis of the association and interactions of the TGF-β signaling pathway-associated genes with MetS and its individual components in Taiwanese subjects. Our findings indicated that the SMAD2 and TGFBR2 genes may affect the prevalence of MetS independently and through complex gene-gene interactions. Furthermore, the TGF-β signaling pathway-associated genes may be associated with the components of MetS. These findings contribute to accumulating evidence supporting that the TGF-β signaling pathway influences MetS. Further investigation with larger sample sizes is essential to provide more insights into the role of the TGF-β signaling pathway-associated genes investigated in this study.

Materials and Methods
Study population. This study included Taiwanese subjects from the Taiwan Biobank. This biobank collected specimens and associated data from the general Taiwanese population with no history of cancer through  [32][33][34][35][36][37][38][39] . Recruitment centers encompass regional and municipal hospitals, where advertisements were posted to recruit potential participants and incentives, such as free general health examinations and travel fee reimbursement, were offered to the participants 32 . This biobank is mainly funded by the Taiwanese government and aims to provide researchers with opportunities for collaboration to facilitate public health-related research concerning local common chronic diseases 32 . The study cohort consisted of 3,000 subjects. Individuals who could perform activities of daily living, were aged 30-70 years, and were self-reported as being of Taiwanese Han Chinese ancestry were included in this study 33 . Individuals with a history of cancer or nonresidents of Taiwan were excluded 33 . Ethical approval for this study was granted by the Institutional Review Board of the Taiwan Biobank before conducting the study. Each subject signed an approved informed consent form. All experiments were performed in accordance with relevant guidelines and regulations.
Metabolic syndrome. Measurements of metabolic traits including waist circumference, triglyceride, HDL cholesterol, systolic and diastolic blood pressure, and fasting glucose were obtained when subjects underwent general health examinations [32][33][34] . MetS was defined according to the International Diabetes Federation definition. Individual were considered to have MetS if they had central obesity (defined as waist circumference of ≥90 cm in male subjects and ≥80 cm in female subjects) and two or more of the following four components: (1) triglycerides ≥1.7 mmol/L; (2) HDL cholesterol <1.03 mmol/L in male subjects and <1.29 mmol/L in female subjects; (3) systolic blood pressure ≥130 mmHg or diastolic blood pressure ≥85 mmHg; and (4) fasting plasma glucose ≥5.6 mmol/L 40 . Two measurements of blood pressure were taken in both arms at least 10-15 minutes apart in the sitting position. These measurements were averaged to obtain the final blood pressure used in this study.
Genotyping. DNA was isolated from blood samples using a QIAamp DNA blood kit following the manufacturer's instructions (Qiagen, Valencia, CA, USA). The quality of the isolated genomic DNA was evaluated using agarose gel electrophoresis, and the quantity was determined using spectrophotometry 41 . SNP genotyping was performed using custom Taiwan Biobank chips and was achieved using the Axiom Genome-Wide Array Plate System (Affymetrix, Santa Clara, CA, USA). To efficiently obtain maximal genetic information from the samples of subjects with Taiwanese Han Chinese ancestry, the custom Taiwan Biobank chips were designed using SNPs with minor allele frequencies (MAFs) ≥5% on the Axiom Genome-Wide CHB 1 Array (Affymetrix, Inc.), using SNPs in exons with MAFs >10% on the Human Exome BeadChip (Illumina, Inc., San Diego, CA, USA), and using SNPs previously reported in ancestry information panels, cancer studies, and pharmacogenetic studies 33 .
In this study, the SNP panel covered 261 SNPs from the following eight TGF-β signaling pathway-associated genes: SMAD2, SMAD3, SMAD4, TGFB1, TGFB2, TGFB3, TGFBR1, and TGFBR2 (Supplementary Table S1). Fifteen SNPs were excluded from further analysis due to failure to achieve the Hardy-Weinberg equilibrium (P < 0.05) or due to a genotyping call rate of <0.95. The genotyping results, including MAFs, P values for the Hardy-Weinberg equilibrium, and genotyping call rates, are shown in Supplementary Table S1. In addition, tag SNPs were identified using PLINK 42 , and an LD value (r²) of 0.8 was used as a threshold.
Statistical analysis. Categorical data were evaluated using the χ 2 test. The Student's t test was used to compare the difference in the means calculated from two continuous variables. To evaluate the association of the investigated SNPs with MetS, we conducted a logistic regression analysis to estimate the ORs and their 95% CIs, adjusting for covariates including age and sex 43 . Furthermore, we evaluated the association of the investigated SNPs with individual components of MetS by using logistic regression analysis, adjusting for age and sex 44 . We assessed whether the genotype frequencies were in the Hardy-Weinberg equilibrium by using the χ 2 goodness-of-fit test with 1 degree of freedom (i.e., the number of genotypes subtracted from the number of alleles). Multiple testing was adjusted using Bonferroni correction. The criterion for significance was set at P < 0.05 for all tests. Data are presented as mean ± standard deviation.
We employed the GMDR method to investigate gene-gene interactions 45 . We tested two-way interactions using 10-fold cross-validation. GMDR software provides some output parameters, including the testing accuracy and empirical P values, to assess each selected interaction. Moreover, we used age and sex as covariates for genegene interaction models in our interaction analyses. Permutation testing provides empirical P values of prediction accuracy as a benchmark based on 1,000 shuffles. To correct for multiple testing, we applied a conservative Bonferroni correction factor for the number of tests employed in the GMDR analysis.
Based on the effect sizes in this study, the power to detect significant associations was evaluated using QUANTO software (http://biostats.usc.edu/Quanto.html).
Data Availability. The data that support the findings of this study are available from the Taiwan Biobank but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. To apply for access to these third party data please contact the Taiwan Biobank.