NEUROGENIN3+ (NEUROG3+) cells are considered to be pancreatic endocrine progenitors. Our current knowledge on the molecular program of NEUROG3+ cells in humans is largely extrapolated from studies in mice. We hypothesized that single-cell RNA-seq enables in-depth exploration of the rare NEUROG3+ cells directly in humans. We aligned four large single-cell RNA-seq datasets from postnatal human pancreas. Our integrated analysis revealed 10 NEUROG3+ epithelial cells from a total of 11,174 pancreatic cells. Noticeably, human NEUROG3+ cells clustered with mature pancreatic cells and epsilon cells displayed the highest frequency of NEUROG3 positivity. We confirmed the co-expression of NEUROG3 with endocrine markers and the high percentage of NEUROG3+ cells among epsilon cells at the protein level based on immunostaining on pancreatic tissue sections. We further identified unique genetic signatures of the NEUROG3+ cells. Regulatory network inference revealed novel transcription factors including Prospero homeobox protein 1 (PROX1) may act jointly with NEUROG3. As NEUROG3 plays a central role in endocrine differentiation, knowledge gained from our study will accelerate the development of beta cell regeneration therapies to treat diabetes.

Transplantation of pancreatic islets is an effective therapy for severe type 1 diabetes. As donor shortage is a major problem for this therapy, attempts have been made to produce a large number of pancreatic islets from human pluripotent stem cells (hPSCs). However, as the differentiation of hPSCs to pancreatic islets requires multiple and lengthy processes using various expensive cytokines, the process is variable, low efficiency and costly. Therefore, it would be beneficial if islet progenitors could be expanded. Neurogenin3 (NGN3)-expressing pancreatic endocrine progenitor (EP) cells derived from hPSCs exhibited the ability to differentiate into pancreatic islets while their cell cycle was arrested. By using a lentivirus vector, we introduced several growth-promoting genes into NGN3-expressing EP cells. We found that SV40LT expression induced proliferation of the EP cells but reduced the expression of endocrine lineage-commitment factors, NGN3, NEUROD1 and NKX2.2, resulting in the suppression of islet differentiation. By using the Cre-loxP system, we removed SV40LT after the expansion, leading to re-expression of endocrine-lineage commitment genes and differentiation into functional pancreatic islets. Thus, our findings will pave a way to generate a large quantity of functional pancreatic islets through the expansion of EP cells from hPSCs.

During pancreas development, Neurog3 positive endocrine progenitors are specified by Delta/Notch (D/N) mediated lateral inhibition in the growing ducts. During neurogenesis, genes that determine the transition from the proneural state to neuronal or glial lineages are oscillating before their expression is sustained. Although the basic gene regulatory network is very similar, cycling gene expression in pancreatic development was not investigated yet, and previous simulations of lateral inhibition in pancreas development excluded by design the possibility of oscillations. To explore this possibility, we developed a dynamic model of a growing duct that results in an oscillatory phase before the determination of endocrine progenitors by lateral inhibition. The basic network (D/N + Hes1 + Neurog3) shows scattered, stable Neurog3 expression after displaying transient expression. Furthermore, we included the Hes1 negative feedback as previously discussed in neurogenesis and show the consequences for Neurog3 expression in pancreatic duct development. Interestingly, a weakened HES1 action on the Hes1 promoter allows the coexistence of stable patterning and oscillations. In conclusion, cycling gene expression and lateral inhibition are not mutually exclusive. In this way, we argue for a unified mode of D/N mediated lateral inhibition in neurogenic and pancreatic progenitor specification.

The generation of therapeutic β-cells from human pluripotent stem cells relies on the identification of growth factors that faithfully mimic pancreatic β-cell development in vitro. In this context, the aim of the study was to determine the expression and function of the glial cell line derived neurotrophic factor receptor alpha 3 (GFRα3) and its ligand artemin (Artn) in islet cell development and function. GFRα3 and Artn expression were characterized by in situ hybridization, immunochemistry, and qRT-PCR. We used GFRα3-deficient mice to study GFRα3 function and generated transgenic mice overexpressing Artn in the embryonic pancreas to study Artn function. We found that GFRα3 is expressed at the surface of a subset of Ngn3-positive endocrine progenitors as well as of embryonic α- and β-cells, while Artn is found in the pancreatic mesenchyme. Adult β-cells lack GFRα3 but α-cells express the receptor. GFRα3 was also found in parasympathetic and sympathetic intra-islet neurons as well as in glial cells in the embryonic and adult pancreas. The loss of GFRα3 or overexpression of Artn has no impact on Ngn3 and islet cell formation and maintenance in the embryo. Islet organization and innervation as well as glucose homeostasis is normal in GFRα3-deficient mice suggesting functional redundancy.