Islet cell transplantation is increasingly considered as a new generation therapy for insulin-dependent diabetes. It is however limited by the availability of donor islets. Differentiation of human iPS cells into islet-like progeny would offer an alternative approach, ensuring an unlimited supply for islet-like cells. Here, under feeder-free conditions, human iPS clones were successfully guided by a pancreatogenic cocktail - enriched with ILV/GLP-1 - into islet-like cells which expressed pancreas-specific transcription factors and islet-specific hormones. Importantly, derived islet-like cells secreted C-peptide in response to glucose stimulation. The present study therefore establishes an iPS-based feeder-free protocol for derivation of functional hormone-producing progeny, expanding the armamentarium for islet-like cell generation.
Human stem cell clones are typically maintained on non-human feeder cells impeding translation of human cell therapy protocols1
. Indeed, non-human feeder cells express undefined factors, which potentially compromise stem cell differentiation. Moreover, feeder cells have been reported to constitutively produce infectious endogenous retroviruses, often preventing immediate clinical application. To circumvent this limitation, we here generated feeder-free conditions using Matrigel-coated plates and an optimized serum-free medium, supplementing HEScGRO with 25% of mTeSR1. This previously untested approach achieved feeder-free/serum-free conditions, nurturing iPS cell expansion and maintaining the undifferentiated state long-term. In this way, we were able to sustain multiple iPS clones in feeder/serum-free milieu for >1 year without loosing salient structural and genetic traits of pluripotency.
It is established that the initial step to achieve successful generation of islet-like cells from a pluripotent cell source requires differentiation into definitive endoderm32,33
. Stimulation of mouse or human ES cells with activin A, Nodal and/or Wnts leads to endoderm induction18,20,34
, and was here implemented with human iPS cells. Specifically, we generated definitive endoderm cells by stimulating feeder-free human iPS cells with activin A and Wnt3a in the presence of low serum. Induction of definitive endoderm was particularly efficient, such that 70–90% of derived cells expressed the definitive endoderm marker SOX17. Similar results were observed across distinct iPS clones, derived from foreskin or cardiac fibroblasts, demonstrating that the applied approach offers a robust means for definitive endoderm induction.
Further lineage specification requires conversion of definitive endoderm into pancreatic endoderm35
. A recent study has identified a small molecule ILV, a protein kinase C activator, which strongly directs human ES cell-derived definitive endoderm into pancreatic endoderm25
. Pancreatic endoderm induction by ILV alone has been shown to generate approximately 53% PDX1-positive cells, in contrast to 25% PDX1-positive cells achievable with standard inductors, FGF10, RA and CYC32
. Here, using human iPS-derived definitive endoderm, we demonstrated robust induction of pancreatic endoderm markers, PDX1, NGN3 and NEUROD1, only after inclusion of ILV to the FGF10, RA and CYC cocktail, underscoring the feasibility of generating pancreatic endoderm cells from human iPS cells through ILV-mediated guidance. It has been proposed that ILV may share an RA-dependent signaling pathway25
, allowing ultimately to streamline the process of efficient pancreatic endoderm derivation.
Human ES cells after guided differentiation to pancreatic hormone-expressing cells generate up to 12% of insulin and synaptophysin expressing cells20
. However, these cells show minimal glucose-responsive C-peptide secretion and do not maintain expression of key beta cell markers, NKX
6.1 and PDX120
. Foreskin-derived iPS cells or iPS cells from patients with type 1 diabetes (T1D) were also recently differentiated into islet-like clusters in vitro29,30
, yet generation of glucose-responsive insulin-producing cells from human iPS cells remains inefficient. GLP-1 was shown to convert intestinal epithelial cells into functional insulin-producing cells36
, and could promote differentiation of mouse bone marrow13
or human embryonic37
stem cells into insulin-producing cells. Inclusion of GLP-1 in our differentiation cocktail enhanced the generation of insulin-producing cells from iPS-derived pancreatic endoderm cells. Derived islet-like cells expressed transcripts of PDX1
, recognized beta cell-specific markers. Co-expression of insulin and PDX1, a characteristic of pancreatic beta cells38
, was evident in iPS-derived insulin-producing cells by immunofluorescence. In contrast to previous reports which demonstrated limited PDX1 expression in ES cell-derived insulin-producing cells20
, we found prevalent insulin/PDX1-double positive cells after differentiation. Moreover, these islet-like cells secreted C-peptide in response to glucose stimulation. These observations indicate that the ILV/GLP-1-containing pancreatogenic cocktail offers proficient means to specify human iPS cells into functional beta-like cells.
Replacement of Exendin-4 (120 nM) with GLP-1 (55 nM) in the final differentiation step notably improved the differentiation of iPS-derived pancreatic endoderm into glucose-responsive, insulin-producing cells. Exendin-4, a 39 amino acid peptide secreted by the salivary gland of a venomous lizard, is a GLP-1 analog which binds and activates the GLP-1 receptor (GLP-1R). Intriguingly, Exendin-4 and GLP-1 bind GLP-1R with similar affinity39
but in different modes; Exendin-4 relies largely on the N-domain for interaction, while GLP-1 utilized the core domain for GLP-1R binding40
. We speculate that iPS-derived endoderm cells express GLP-1R with differential post-translational modification, affecting the binding of GLP-1R with Exendin-4, but not GLP-1. Alternatively, binding of GLP-1, but not Exendin-4, to GLP-1R on iPS-derived endoderm cells may facilitate activation of a critical intracellular signaling pathway necessary for successful endoderm differentiation into glucose-responsive islet-like cells.
Of note, human iPS clones demonstrated here no notable differences in morphology, pluripotency marker expression, and/or ability to differentiate into cells of the three germ layers. Although iPS clones were efficiently differentiated into definitive endoderm, they exhibited varying propensity to generate glucose-responsive insulin-producing cells. Moreover, we noted that high levels of C-peptide remained in glucose-stimulated cells (data not shown), suggesting an immature insulin secreting system in iPS-derived islet-like cells. In fact, it was notable that while certain clones displayed vigorous response to glucose with marked secretion of insulin, other clones failed to significantly secrete C-peptide even though features of insulin-expressing cells were documented. This diversity in response cannot be ascribed to variations in the genetic or tissue background of iPS clones, as variations were found among iPS clones derived from the same tissue origin. Clonal differences in differentiation propensities have been reported among pluripotent stem cells, which could be due to varying degrees of reprogramming, different copy numbers of integrated pluripotency genes, or different levels of pluripotency gene silencing/reactivation in individual clones29,30,41
. Further analysis using multiple iPS clones and improved islet differentiation efficiency will be necessary to establish the mechanisms underlying unique functional property of derived lineage-specified cell progeny in vitro
and in vivo
Retroviral or lentiviral vector integration has risks associated with insertional mutagenesis. Use of oncogenic c-MYC during reprogramming is also problematic for clinical application of the resulting iPS cells, as sustained c-MYC expression can increase the risk of tumor formation in iPS-derived cells in vivo
. Indeed, use of c-Myc has been associated with increased tumorigenicity in chimeric mice established with iPS cells42
. Additionally, reactivation or sustained expression of pluripotency genes in iPS-derived islet-like cells could result in spontaneous reprogramming of iPS progeny, leading to the potential risk of teratoma formation upon transplantation. Indeed, residual expression of pluripotency-associated genes, such as c-MYC, NANOG, hTERT and GDF3, was evident in differentiating iPS cells (), which is likely due to sustained vector transgene expression. Determining the levels of vector-derived and endogenous pluripotency gene expression in the different clones before and after differentiation will be necessary to understand the influence of integrating vectors on the diversity in differentiation propensities among iPS clones. In order to increase the safety of iPS-based cell therapy, it will be critical to generate iPS cells without integrating vectors and continuous c-MYC expression. Derivation of iPS cells with transient expression from non-integrating vectors43–45
will solve both problems, as the resulting iPS cells carry no genomic modifications. To further minimize the risk of teratoma formation, it will be necessary to develop further strategies, such as sorting out fully differentiated cell populations by flow cytometry, eliminating residual pluripotent cells by a suicide gene vector or encapsulating the iPS-derived cells before transplantation.”
In summary, we demonstrate that under feeder-free conditions an ILV/GLP-1-enriched pancreatogenic cocktail allows derivation of glucose-responsive, insulin-producing cells from human iPS cells, providing a platform for islet-like cell generation. Autologous iPS derivation and iPS differentiation into insulin-producing cells would allow modeling of patient-specific disease pathogenesis, and ultimately lead to personalized approaches for T1D cell therapy with iPS-derived islet-like cells.