Systemic Biology Approach Linking Type 2 Diabetes and Depression in a Rio Grande Valley Family Study Population

Background: Type 2 diabetes and depression are major public health concerns that disproportionally impact Mexican Americans. Prevalence of T2D and depression have been estimated as high as 30% and 40%, respectively in Mexican American populations along the Texas-Mexico border. While the interaction between these two phenotypes is well documented, the underlying genetic basis of this interaction remains unclear. Our aims for this project were to identify genes co-expressed in type 2 diabetes and depression to dissect the interaction correlations between both conditions by means of a functional gene expression correlation network. Methods: A total of 528 Mexican American participants from the Rio Grande Valley family study were clinically evaluated and assessed for obesity, diabetes, hypertension, hyperlipidemia, and BDI (depression). Blood samples were collected, mRNA was purified and later sequenced to estimate mRNA abundance. All gene expression vectors were tested for association using a linear mixed model in the form where Ω is the phenotypic covariance matrix of each gene, is the total phenotypic variance, andeecm, respectively representing the proportion attributed to the residual additive effect of polygenes and environmental random effects. We used a likelihood ratio test comparing two models with and without the addition of T2D and BDI score as covariates. The individual association statistics between genes and T2D/depression covariates were used to create a Fischer combined statistics in the form: . This distribution can be modeled as a chi-square distribution with four degrees of freedom. We extracted the expression vectors of 110 significant genes with Fischer combined pvalues < 0.0001 and used their respective expression as an input to calculate a pair-wise gene expression Pearson correlation matrix. We identified 107 significant gene expression correlations in our gene set (Pearson correlation > [0.8]). We defined the k connectivity variable of each gene, and all gene-gene interactions were visualized using Cytoscape visualization tool. Results: Eight genome-wide significant association hits between T2D and gene expression information were identified, the most significant being CDK18 (p.value < 1.82 * 10-7). No genome-wide significant associations between depression and gene expression information were identified. Fisher combined analysis showed the genes with strongest association to both diabetes and depression to be GRP160 (p.value < 1.8*10-7), CDK18 (p.value < 1.1*10-6), CDHR1 (p.value = 5.2*10-6), LBR (p.value < 6.8*10-6), MACF1 (p.value < 8.4*10-6), NRCAM (p.value < 8.9*10-6), and PPP1R13B (p.value < 9.2*10-6). After correlation matrix calculation, the 6 genes with the highest connectivity within the gene network were SDCBP, GCA, EGLN1, CMTM6, HSD17B11, and RNF14. Conclusions: Our genetic association model, paired with Fisher combined statistics and Pearson correlation, highlights a significant gene-gene interaction between genes associated with T2D and depression. Our results explore the network gene expression between diabetes and depression and helped identify key genes that anchor the network of genes associated with both T2D and depression. Further research must be done to validate the study, as well as performing proteomics for these gene products to assess their metabolic function in relation to both diabetes and depression.