Background: Pluri-potent bone marrow stromal cells (MSCs) provide an attractive opportunity to generate unlimited glucose-responsive insulin-producing cells for the treatment of diabetes. We explored the potential for human MSCs (hMSCs) to be differentiated into glucose-responsive cells through a non-viral genetic reprogramming approach.
Methods and findings: Two HMSC lines were transfected with three genes: PDX-1, NeuroD1 and Ngn3 without subsequent selection, followed by differentiation induction in vitro and transplantation into diabetic mice. Human MSCs expressed mRNAs of the archetypal stem cell markers: Sox2, Oct4, Nanog and CD34, and the endocrine cell markers: PDX-1, NeuroD1, Ngn3, and Nkx6.1. Following gene transfection and differentiation induction, hMSCs expressed insulin in vitro, but were not glucose regulated. After transplantation, hMSCs differentiated further and approximately 12.5% of the grafted cells expressed insulin. The graft bearing kidneys contained mRNA of insulin and other key genes required for the functions of beta cells. Mice transplanted with manipulated hMSCs showed reduced blood glucose levels (from 18.9+/-0.75 to 7.63+/-1.63 mM). 13 of the 16 mice became normoglycaemic (6.9+/-0.64 mM), despite the failure to detect the expression of SUR1, a K(+)-ATP channel component required for regulation of insulin secretion.
Conclusions: Our data confirm that hMSCs can be induced to express insulin sufficient to reduce blood glucose in a diabetic mouse model. Our triple gene approach has created cells that seem less glucose responsive in vitro but which become more efficient after transplantation. The maturation process requires further study, particularly the in vivo factors influencing the differentiation, in order to scale up for clinical purposes.