Using Gene Expression to Predict the Secretome of Differentiating Human Preadipocytes

Objective: To characterize the secretome of differentiating human preadipocytes using global gene expression profiling.

Design: Gene expression was measured using microarrays at days 0, 1, 3, 5, 7 and 10 in primary preadipocytes undergoing adipogenesis (n=6 independent subjects). Predictive bioinformatic algorithms were employed to identify those differentially expressed genes that code for secreted proteins.

Measurements: Gene expression was assessed using microarrays and real-time reverse transcriptase PCR, bioinformatic predictive algorithms were used to identify the secretome of differentiating preadipocytes, and the secretion of the most significant candidates were confirmed at the protein level using western blots or ELISA tests. Gene expression was also assayed in the adipocyte and stroma vascular fraction (SVF) of obese subjects.

Results: Microarray analysis identified 33 genes whose expression significantly changed (false discovery rate of 1%) during adipogenesis and code for secreted proteins. Of these genes, 18 are novel candidate adipose tissue 'secretome' genes. Their relative gene expression in adipocyte and SVF of obese subjects revealed that most of these genes are more highly expressed in the SVF. A novel candidate, matrix gla protein (MGP), was upregulated (approximately 30-fold) during adipogenesis, second only to leptin (approximately 50-fold). MGP and another secretome candidate protein, inhibin beta B (INHBB), were detected in the secretion media of adipocytes isolated from adipose tissue explants.

Conclusions: Gene expression coupled with predictive bioinformatic algorithms has proved a valid and alternative approach to further define the adipocyte secretome. Many of the novel candidate secretome genes are components of the coagulation and fibrinolytic systems. MGP and INHBB represent new adipokines whose function in adipose tissue remains to be unravelled.