The human pancreas comprises exocrine and endocrine tissue [1]. Endocrine cells are arranged into pancreatic islets, or islets of Langerhans, which include insulin-producing beta cells, glucagon-producing alpha cells, delta cells producing somatostatin, pancreatic polypeptide cells (PP cells) producing pancreatic polypeptide, and ghrelin-producing epsilon cells ^[2][3][2,3]. All types of cells work together in a complicated manner to maintain metabolic homeostasis. Further, their interactions play an important role in disease [4].

Diabetes mellitus affects millions of people all over the world. The cause of type 1 diabetes (T1D) is the loss of beta cells due to autoimmune response and inflammation [5]. Type 2 diabetes (T2D) is a metabolic disorder that develops because of beta cell dysfunction and is often characterized by insulin resistance [6]. Although major improvements in insulin delivery in type 1 diabetic patients have been achieved by using insulin pumps, glucose monitors, and an artificial pancreas, these methods can pose risks of hypo- and hyperglycemia, reduction of glycated hemoglobin, and other health complications in the case of comorbidities ^{[7][8][9][10]}[7,8,9,10]. Allotransplantation of cadaveric islets is another approach to diabetes therapy. Transplantation followed by immunosuppression can result in long-term insulin independence ^[11][12][11,12]. However, this method is limited due to the lack of suitable donors. A variety of pharmacological treatment options are available for T2D, including glucagon-like peptide 1 (GLP1) receptor agonists and sodium–glucose cotransporter-2 (SGLT2) inhibitors, which help against hyperglycemia, despite having several drawbacks [13]. Such therapy for both T1D and T2D is still not available to some people and is an economic burden to healthcare systems ^[14][15][14,15].

Due to a pressing worldwide situation with diabetes, ideas to use direct differentiation from embryonic stem cells (ESCs) and pluripotent stem cells (PSCs) to produce beta cells have surfaced [16]. Stem cells are thought to be an ideal source of all cell types including pancreatic beta cells. However, beta cells are highly specialized; because of that, differentiation of mature insulin-secreting cells is extremely complex.

The methods of in vivo differentiation of beta cells mimicking the genetic and functional profiles of native cells have come a long way. One of the approaches is transdifferentiation, or cell reprogramming. Usually, the choice of the cell source for reprogramming is based on a common background and an availability of the cells. For pancreatic beta cells, those are the gallbladder [17], liver ^[18][19][20][18,19,20], exocrine ^[21][22][21,22] and other endocrine pancreatic cell types [23]. Reprogramming is achieved by overexpression of the genes involved in pancreas development and characterizing mature beta cells such as pancreatic and duodenal homeobox 1 (PDX1), neurogenin-3 (NGN3), member of the Maf family of transcription factors MAFA, paired box protein PAX6, paired box protein PAX4, and other transcriptional factors ^[24][25][26][24,25,26]. The aforementioned protocols have their advantages; however, they all pose the same question of how to achieve maximum transdifferentiation efficiency and minimize heterogeneity.

In this case, it seems more promising to use undifferentiated cells because it could be easier to direct them towards the beta cell identity [27]. Direct differentiation methods of acquiring mature beta cells from hESCs (human ESCs) and hPSCs (human PSCs) have been modified and improved. First, protocols for hESC-derived definitive endoderm have been developed [16]. This approach as the more recent ones involves supplementing the culture media with small molecules and growth factors such as activin A and fibroblast growth factor 10 (FGF10) that can potentiate hPSCs and hESCs towards functional beta cells ^[16][28][16,28]. The next step is differentiation of PDX1⁺ pancreatic progenitors from hESCs [29]. With further research, it has become possible to differentiate hPSCs into insulin-producing cells [30] and adapt the previous protocols using 3D clusters to create conditions more similar to the in vivo ones [31].

Methods to produce hESC- and hPSC-derived beta cells have been developing in parallel ^[32][33][32,33]. A protocol proposed by Rezania et al. [32] included seven stages of hESC differentiation. Each one imitated a stage in pancreas embryogenesis from the definitive endoderm to maturating beta cells. Cells at the last stage were characterized by the expression of MAFA, homeobox protein NKX6.1, insulin (INS), and the absence of glucagon (GCG), as well as glucose-stimulated insulin secretion (GSIS) [32]. Pagliuca et al. [33] implemented a similar method for hPSCs and generated functional insulin-secreting cells. Differentiated beta cells transplanted in streptozotocin (STZ)-induced diabetic mice [32] or NOD-Rag1^null IL2rg^null Ins2^Akita (NRG-Akita) mice [33] returned animals to normoglycemia. However, a close analysis of cells at the final stage of both protocols showed that although beta-like cells had comparable expression profiles and functional similarities to native cells, such as GSIS in vitro, there were major differences in insulin granule morphology, key transcription factor expression was lower, and GSIS dynamics were far from identical. Moreover, the cell population was heterogeneous as polyhormonal cells were present ^[32][33][32,33].

The protocols published over the past few years seem to face the same complications such as low differentiation efficiency, presence of polyhormonal cells, low expression rates or the lack of some beta cell markers, and, most importantly, functional immaturity of cells ^{[32][33][34][35]}[32,33,34,35]. Some of these difficulties have been almost dealt with in more recent research [36]. However, it is still not possible to overcome all of them (Table 1).

It is worth mentioning that the search for progenitor cells in the adult pancreas is still ongoing, with a prospective that these cells could be used as another source for regenerative medicine. One example is a cell population that was identified in mice which expresses protein C receptor (Procr⁺ cells) [37]. It has been reported that adult murine islets contain Procr⁺ cells, which can differentiate into alpha, beta, delta, and PP cells. Pancreatic islet organoids derived from Procr⁺ cells can reverse hyperglycemia in diabetic mice.

However, the question of whether a population with the same properties and differentiation capacity exists in humans remains open. Interestingly, the expression of human endothelial protein C receptor (hEPCR) in donor murine islets improved the transplantation outcome [38]. Due to this fact, it might be perspective to further study protein C receptor functions. Despite the promising results, such findings should be addressed with caution because reports of progenitor populations in the adult pancreas have surfaced before, and the main drawback of such reports is that progenitor cells have an ability for clonogenic expansion only in vitro, while not showing the same properties in vivo [39]. The ways to generate beta cells are summarized in Table 1.

Direct differentiation can be used to generate insulin-secreting cells from human induced pluripotent stem cells (hiPSC) from patients with diabetes. T1D patient-derived hiPSCs serve as cell models for the anti-diabetic treatment effect, a search tool for new possible drugs, and for the assessment of different negative effects that could promote the onset of diabetes [40]. The fact that beta cells that were produced by Milman et al. [40] did not have any significant differences from stem cell-derived beta cell from non-diabetic donors can be a point of interest for the following research. Ones again, this can be the proof that diabetes is caused by not just one reason such as a mutation, but by several collective factors [41].

The progress that has been made in generating beta cells in vitro especially from human stem cells provides an opportunity for their application in regenerative medicine, studying beta cell identity, maturation, functions, as well as being tools for investigating disease and models for possible drug screening.