A coordinated functional genomics program was implemented to identify secreted polypeptides with therapeutic applications in the treatment of diabetes. Secreted factors were predicted from a diverse expressed-sequence tags (EST) database, representing >1,000 cDNA libraries, using a combination of bioinformatic algorithms. Subsequently, approximately 8,000 human proteins were screened in high-throughput cell-based assays designed to monitor key physiological transitions known to be centrally involved in the physiology of type 2 diabetes. Bone morphogenetic protein-9 (BMP-9) gave a positive response in two independent assays: reducing phosphoenolpyruvate carboxykinase (PEPCK) expression in hepatocytes and activating Akt kinase in differentiated myotubes. Purified recombinant BMP-9 potently inhibited hepatic glucose production and activated expression of key enzymes of lipid metabolism. In freely fed diabetic mice, a single subcutaneous injection of BMP-9 reduced glycemia to near-normal levels, with maximal reduction observed 30 hours after treatment. BMP-9 represents the first hepatic factor shown to regulate blood glucose concentration. Using a combination of bioinformatic and high-throughput functional analyses, we have identified a factor that may be exploited for the treatment of diabetes.