FAM13A and POM121C are candidate genes for fasting insulin: functional follow-up analysis of a genome-wide association study

Abstract

Aims/hypothesis: By genome-wide association meta-analysis, 17 genetic loci associated with fasting serum insulin (FSI), a marker of systemic insulin resistance, have been identified. To define potential culprit genes in these loci, in a cross-sectional study we analysed white adipose tissue (WAT) expression of 120 genes in these loci in relation to systemic and adipose tissue variables, and functionally evaluated genes demonstrating genotype-specific expression in WAT (eQTLs).

Methods: Abdominal subcutaneous adipose tissue biopsies were obtained from 114 women. Basal lipolytic activity was measured as glycerol release from adipose tissue explants. Adipocytes were isolated and insulin-stimulated incorporation of radiolabelled glucose into lipids was used to quantify adipocyte insulin sensitivity. Small interfering RNA-mediated knockout in human mesenchymal stem cells was used for functional evaluation of genes.

Results: Adipose expression of 48 of the studied candidate genes associated significantly with FSI, whereas expression of 24, 17 and 2 genes, respectively, associated with adipocyte insulin sensitivity, lipolysis and/or WAT morphology (i.e. fat cell size relative to total body fat mass). Four genetic loci contained eQTLs. In one chromosome 4 locus (rs3822072), the FSI-increasing allele associated with lower FAM13A expression and FAM13A expression associated with a beneficial metabolic profile including decreased WAT lipolysis (regression coefficient, R = -0.50, p = 5.6 × 10^-7). Knockdown of FAM13A increased lipolysis by ~1.5-fold and the expression of LIPE (encoding hormone-sensitive lipase, a rate-limiting enzyme in lipolysis). At the chromosome 7 locus (rs1167800), the FSI-increasing allele associated with lower POM121C expression. Consistent with an insulin-sensitising function, POM121C expression associated with systemic insulin sensitivity (R = -0.22, p = 2.0 × 10^-2), adipocyte insulin sensitivity (R = 0.28, p = 3.4 × 10^-3) and adipose hyperplasia (R = -0.29, p = 2.6 × 10^-2). POM121C knockdown decreased expression of all adipocyte-specific markers by 25-50%, suggesting that POM121C is necessary for adipogenesis.

Conclusions/interpretation: Gene expression and adipocyte functional studies support the notion that FAM13A and POM121C control adipocyte lipolysis and adipogenesis, respectively, and might thereby be involved in genetic control of systemic insulin sensitivity.

Keywords: Genomics; Insulin sensitivity; Lipid metabolism.

Fig. 1

Relationship between FSI and BMI (a), adipose morphology (b), insulin-stimulated lipogenesis (c) and basal lipolysis (d). n = 114 women. Defined in the ‘WAT experiments’ section of the Methods. Linear regression was used in all analyses

Fig. 2

Gene expression of FAM13A, POM121C, SNRPC and UHRF1BP1 were monitored using quantitative RT-PCR during differentiation of hMSCs to adipocytes in vitro from start of differentiation (day 0) until day 12. POM121C (solid blue line), FAM13A (dotted black line), UHRF1BP1 (dashed black line) and SNRPC (dashed grey line). Results were analysed using the paired t test and are presented as relative fold change+SD vs day 0. *p < 0.05, **p < 0.01 and ***p < 0.001 vs day 0

Fig. 3

Expression of FAM13A (a), POM121C (b), SNRPC (c) and UHRF1BP1 (d) was knocked down using siRNA in hMSCs in vitro at day 4 of differentiation until day 7 and 12 of differentiation, upon which the expression of target and ADIPOQ, CEBPA, SLC2A4, LIPE and PPARG was monitored. We have performed three biological experiments with 3–4 technical replicates in each experiment; n = 11 technical replicates for NegC; n = 12 technical replicates for target genes. Results were analysed using the paired t test and are presented as relative fold change±SD vs negative control (NegC) at each time point during differentiation. Black bars, day 7; white bars, day 12. *p < 0.05, **p < 0.01 and ***p < 0.001 vs NegC

Fig. 4

Expression of FAM13A, POM121C, SNRPC and UHRF1BP1 was knocked down using siRNA in hMSCs in vitro and glycerol levels in conditional medium were evaluated. We have performed 3 biological experiments with 3–4 technical replicates in each experiment; n = 11 technical replicates for NegC; n = 12 technical replicates for target genes. Results were analysed using the paired t test and are presented as relative fold change±SD vs non-targeting siRNA pool (siNegC). Black bars, day 7; white bars, day 12. *p < 0.05 and ***p < 0.001 vs NegC