Discovery and translation of a target engagement marker for AMP-activated protein kinase (AMPK).

BACKGROUND: Activation of the AMP-activated protein kinase (AMPK) is an attractive approach for the treatment of type 2 diabetes. AMPK activation reduces glucose levels in animal models of type 2 diabetes by increasing glucose uptake in skeletal muscles and reducing hepatic glucose production. Furthermore, AMPK activation ameliorates hepatic steatosis in animal models. For the clinical development of AMPK activators it is essential to have a reliable target engagement marker for appropriate dose finding and to support proof of clinical principle. While the activation of AMPK by quantification of the phosphorylation of AMPK at Thr172 in target tissues can be assessed pre-clinically, this is not feasible in clinical studies. Therefore, we attempted to identify and translate a peripheral target engagement biomarker downstream of AMPK activation for clinical use in blood samples.

METHODS: For pharmacological activation of AMPK, two AMPK activators were synthesized (compound 1 and 2). A compound with structural similarities but no pharmacological effect on AMPK phosphorylation was synthesized as negative control (compound 3). Whole blood from healthy volunteers was incubated with an AMPK activator for up to 6 hours and mRNA sequencing was performed. Additionally, human PBMCs were isolated to evaluate Thr172-phosphorylation of AMPK in Western blots. In order to enable identification of translatable biomarker candidates, blood samples from HanWistar rats treated for two weeks with an AMPK activator were also subjected to mRNA sequencing. Furthermore, concentration-response curves for four biomarker candidates were recorded in human blood samples using Nanostring nCounter technology. Finally, ZDF rats were treated with increasing doses of compound 2 for five weeks to investigate the glucose-lowering efficacy. To investigate changes of mRNA expression of two selected biomarker candidates in this ZDF rat study, qRT-PCR was performed.

RESULTS: Pharmacological activation of AMPK in human PBMCs revealed an increase in Thr172-phosphorylation of AMPK, confirming target engagement in these blood cells. RNA sequencing of human blood samples identified 608 deregulated genes after AMPK activation. Additionally, AMPK activation led to deregulation of 367 genes in whole blood from HanWistar rats which mapped to the respective human genes. 22 genes out of the intersection of genes deregulated in both species are proposed as potential translatable target engagement biomarker candidates. The most prominent genes were transmembrane glycoprotein NMB (GPNMB, osteoactivin), calcium-binding protein A9 (S100A9), peptidoglycan recognition protein (PGLYRP1) and Ras homolog gene family, member B (RHOB). Specificity for AMPK was shown by testing inactive compound 3 in HanWistar rats. The exposure-effect relationship for GPNMB was investigated in a subchronic study in diabetic ZDF rats. GPNMB showed a dose-dependent up-regulation both acutely and after subchronic dosing. GPNMB up-regulation correlated with an increased Thr172-phosphorylation of AMPK in liver and quadriceps muscle in rats.

CONCLUSION: GPNMB has been identified as a translatable target engagement biomarker for use in clinical studies.