Pancreatic β-cell heterogeneity in adult human islets and stem cell-derived islets

Abstract

Recent studies reported that pancreatic β-cells are heterogeneous in terms of their transcriptional profiles and their abilities for insulin secretion. Sub-populations of pancreatic β-cells have been identified based on the functionality and expression of specific surface markers. Under diabetes condition, β-cell identity is altered leading to different β-cell sub-populations. Furthermore, cell–cell contact between β-cells and other endocrine cells within the islet play an important role in regulating insulin secretion. This highlights the significance of generating a cell product derived from stem cells containing β-cells along with other major islet cells for treating patients with diabetes, instead of transplanting a purified population of β-cells. Another key question is how close in terms of heterogeneity are the islet cells derived from stem cells? In this review, we summarize the heterogeneity in islet cells of the adult pancreas and those generated from stem cells. In addition, we highlight the significance of this heterogeneity in health and disease conditions and how this can be used to design a stem cell-derived product for diabetes cell therapy.

Introduction

Pancreatic β-cells are glucose-regulated insulin-producing cells located in islets of Langerhans in the pancreas. These cells play an essential role in glucose metabolism and regulation. Reduction in β-cell function and/or mass can lead to alterations in glucose metabolism leading to abnormal elevation in blood glucose levels and eventually causing diabetes. Different therapeutic approaches for diabetes aim to maintain/restore the β-cell mass, hence function, including whole pancreas or pancreatic islet cell transplantation [1]. However, with islet transplantation being the ideal treatment for type 1 (T1D) and end-stages of type 2 diabetes (T2D), it remains difficult due to the limited number of donors and the need for the usage of strong immunosuppressant drugs [2]. Moreover, pancreatic islet cell transplantation requires at least 10,000 islet equivalents per kilogram of body weight which can be obtained from 2–3 donors approximately [3]. This also adds up to the limited resources for treating more patients with diabetes in need for such interventions. New research approaches aim to efficiently generate an unlimited source of functional pancreatic β-cells in vitro using human pluripotent stem cells (hPSCs) [4]. hPSCs are a valuable tool for generating different types of cells, including pancreatic islets and β-cells. As the structure of pancreatic islets is essential for proper cell-to-cell interaction and proper regulation of insulin secretion, multiple efforts focus on understanding the complex connective network between the different endocrine cells [5].

Recent studies have described that not all pancreatic β-cell populations are similar but heterogenous. To gain more insight, one should consider the factors that could contribute to β-cell heterogeneity. Hence, understanding the process of β-cell development, β-cell interaction with neighbouring cells, and β-cell response to stimulus can open the door to an explanation behind β-cell heterogeneity.

Islet structure and composition

The adult endocrine pancreas, islets of Langerhans, constitutes about 1–4% of the total pancreas [6]. Each islet contains hormone-producing cells that synthesize and release glucagon (GCG; α-cells), insulin (INS; β-cells), somatostatin (SST; δ-cells), pancreatic polypeptide (PPY; γ-cells) and ghrelin (GHRL; ε-cells) in a nutrient-dependent fashion [7, 8]. Pancreatic islets receive up to 20% of the total pancreatic blood supply, hence, they are defined as highly vascularized micro-organs [9]. In addition, studies have shown that the microvascular density of pancreatic islets is 5–10 times higher than that of surrounding exocrine tissues [10]. The increased islet vascularization helps in sensing the systemic blood glucose levels, which is essential for the proper accomplishment of the islets’ endocrine function and efficient hormone delivery to target tissues [9, 11]. It has been reported that islets are not simply disordered clusters of endocrine cells but are highly organized micro-organs with a species-specific three-dimensional architecture. This unique cellular organization allows pancreatic islets to effectively carry out their primary physiological function of responding to changes in metabolic demands and regulating optimal glucose levels in the bloodstream. Furthermore, physical, and electrical cell–cell coupling enable synchronous hormone secretion, and intra-islet paracrine signaling is directed and connected with the nervous system for feedback regulation [12,13,14].

The architecture of mouse islets is characterized by a central, rounded core primarily composed of insulin-secreting β-cells. Meanwhile, the islet periphery is mainly occupied by α-cells that secrete glucagon, δ-cells that secrete somatostatin, and γ-cells that secrete polypeptide [15]. The β-cells located in the central core of the islet form rosette structures that are polarized around the blood capillaries [16]. Within these rosettes, β-cells have an apical domain located on the outer edge of the rosette, where primary cilia extend into the shared extracellular spaces. The basal domain of the β-cell is located adjacent to the blood vessel, where the insulin granules are located. The lateral domains between the edges of β-cells are enriched in glucose transporters and Ca2 + sensing machinery [16,17,18]. The architecture of human islets is characterized by a higher degree of complexity. There are several proposed models for the “stereotypical” human islet, each with its unique features [19,20,21]. The proportion of the primary endocrine cell types within the islet varies between mice and humans. In mice, approximately 75–80% of the islet cells are β-cells, with 20% α-cells, and less than 10% δ-cells. In contrast, the islets in humans have a lower proportion of β-cells, ranging from 55 to 70% and a higher percentage of α-cells, ranging from 30 to 45%, with less than 10% for both δ- and γ-cells [19, 20, 22,23,24]. The size of human islets can vary significantly, with a diameter range of approximately 50–500 μm and an average of 1500 cells per islet. Variations in the ratio of endocrine cells among individuals are reflected in differences in β-cell mass, [25, 26], while the distribution of islet cells across pancreatic regions exhibits only minor differences [27]. Differences in the methods used to study human islet architecture have resulted in various proposed models, and it is uncertain whether a stereotypical architecture exists in humans at all [11, 26]. However, research supports the existence of a common design principle between mice and humans, wherein homotypic interactions among the same type of endocrine cells are more crucial than heterotypic interactions between different types of endocrine cells [26, 28, 29]. Recent findings suggest that there are also conserved β-cells polarity domains in both species, but their functional roles in islets remain unclear and require further investigation [18, 30, 31].

Heterogeneity of pancreatic β-cells

Cumulative evidence showed the existence of heterogenous β-cell sub-populations. By implementing new biomarkers and using advanced single-cell analyses, the identification of the different β-cell sub-populations became more promising. Initially, β-cell heterogeneity has been suggested through the identification of various response levels to glucose observed by different INS biosynthesis and secretion patterns [32]. Additional β-cell heterogeneity aspects include differences in maturity and proliferative states, redox states, membrane potentials, and glucose transport [33]. Different studies have revealed β-cell heterogeneity in animal and human models. However, here, we only focus on the identified β-cell populations in humans.

Maturation and proliferative heterogeneity

Proper glucose response and INS secretion require mature β-cells that can synthesize sufficient amounts of INS upon glucose uptake. Immature β-cells are marked with higher proliferation and metabolic demands for the energy-consuming processes during cell proliferation [34]. Unlike immature β-cells, mature cells become more function-specific and less proliferative. Interestingly, the existence of both mature and immature β-cell populations has been discovered using single-cell analysis techniques. Wang et al. revealed three β-cell clusters (C1, C2, and C3) with different proliferative capabilities [34] (Table 1). C1 showed a quiescent cell state with no proliferation, while C2 and C3 revealed higher proliferative rates indicated by the expression of proliferation marker Ki67⁺ [34]. Interestingly, C1 showed high PDX1 and INS expression and is found to be increased with age but decreased with obesity. Proliferative C2 and C3 express a set of genes important for cell adhesion and migration including CD44, CD9, CD49F, and CYP26A1. Furthermore, C2 showed higher phosphorylation levels of PDGFRA, pERK1/2, pSTAT3, and pSTAT5. Regression models revealed that C2 negatively correlates with T2D, while C3 negatively correlates with age [34]. Several other studies support the presence of different β-cell populations depending on their maturity level. Szabat et al. identified two β-cell populations with distinct gene profiles: PDX1⁺/INS^low and PDX1^high/INS^high [35]. PDX1⁺/INS^low cell population exhibits immature β-cell characteristics, associated with upregulated developmental, proliferative, and apoptotic markers (Table 1). Since PDX1⁺/INS^low are immature cells, β-cell function associated genes are remarkably reduced and cells reflected progenitor-like phenotype with polyhormonal gene profile by expressing PPY, GCG, SST, and GHRL. Remarkably, the expression of calcium-modulated INS secretion regulators (CAMK2G and MAPK4) is increased. However, MAPK1, a regulator of INS secretion in response to glucose, is decreased in PDX1⁺/INS^low cell population [35]. Furthermore, a few diabetes-related genes are enriched in PDX1⁺/INS^low cells as well including DPP4, LEPR, and SIRT5. Syntaphilin (SNPH), an inhibitor of SNARE complex formation and exocytosis regulator, is also increased [35]. PDX1^high/INS^high population composed the majority of the β-cell population (78.3%) and showed mature β-cell characteristics with increased INS secretion. This cell population has high levels of mature β-cell function-related genes including GLUT2, MAPK1, CALML4, TIPRL, G6PC2, IAPP, and GCGR. Yet, SIRT1, a diabetes-related gene that has been shown to enhance INS resistance and diabetes, is found to be higher in PDX1^high/INS^high [35, 36]. Even though PDX1^high/INS^high are mature cells, development-related genes in TGFβ superfamily are also enriched such as BMP5, GREM1, and TGFBR3. Conversely, PDX1⁺/INS^low cells have increased expression of FSTL5 and BAMBI which are from the same TGFβ superfamily [35]. Nevertheless, mature cells display enriched maturity-related genes like GCGR, IAPP, and MAPK1 [35].

Table 1 Beta cell subpopulations identified in human pancreatic islets

Other research groups have identified β-cell heterogeneity according to surface markers expressed on β-cells. Dorrell and colleagues have categorized β-cells into four distinct populations based on two surface markers (CD9 and ST8SIA1) [37]. Interestingly, ST8SIA1⁻ cells, composing ~ 85% in healthy individuals, express high GLUT2 levels and show more mature phenotypes regarding β-cell functionality. By contrast, ST8SIA1⁺ cells show immature cell phenotypes and compose ~ 15% in healthy individuals. Additionally, immature cells had lower INS secretion profile and decreased GLUT2 expression [37]. Yet, ST8SIA1⁺ cells have increased K⁺ channel expression and a three-fold increase in f-channel HCN1 expression, a possible gene with INS secretion relevance [37]. Several transcriptional factors (TFs) are expressed in ST8SIA1⁺ as well, such as SIX3, RFX6, MAFB, and NEUROD1. SIX3 has been identified as an important TF in age-related β-cell maturation and expression levels positively correlate with INS secretion [37, 38]. Moreover, the ST8SIA1⁻ cell population is reduced in T2D, while the ST8SIA1⁺ cell population increases in the case of diabetes [37]. Recently, CD9 has been identified as a negative surface marker for functional glucose-responsive β-cells. hPSC-derived pancreatic β-cells, which are negative for CD9, have higher NKX6.1, MAFA, and C-peptide levels in comparison to CD9⁺ cells [39].

The hallmark of a mature β-cell involves a complex cellular identity with finely tuned coupling to the prevailing glucose level. Ultra-structurally, a β-cell is estimated to contain ~ 10,000 dense-core secretory granules with a clear peripheral mantle and the presence of fully processed proinsulin molecules in these dense-core granules depends on the maturity of the β-cell [40]. The concerted activity of key TFs determines the fate of endocrine cells; hence, mature β-cells are also distinct in terms of the expression of certain genes and TFs. Single-cell analyses that have revealed the transcriptional program of human pancreatic endocrine cells depicted genes such as PAX4, PDX1, MAFA, MAFB, DLK1, SIX2/3, ID1, IAPP, UCN3, and OLIG1 that are highly or exclusively expressed in human β-cells [143]. Taken together, it has been suggested that GHRL is a paracrine modulator of INS secretion and exert INS-static action on pancreatic β-cells via the inhibitory signaling of GHSR.

Autocrine and paracrine action of β-cells and functional heterogeneity

Endocrine cells within the pancreatic islets secrete a wide range of diffusible chemical messengers, the autocrine or paracrine factors, which via binding to cognate receptors, are able to evoke biological effects in the neighbouring cells. Experimental studies have shown that by synchronizing their secretory activity in response to electrical coupling, autocrine or paracrine signaling, islet β-cells exploit several pathways of cell-to-cell communication [144]. However, human β-cells show complex and heterogeneous electrophysiological responses to many factors including ion channel antagonists [145]. This is probably due to the variability in the number of specific channels between β-cells that has been modelled to generate variable oscillation patterns and responses thus affecting the regularity of membrane potential bursting and [Ca²⁺] oscillations [146,147,148]. One of the paracrine mediators of human β-cell is mediated by glucagon (released from α-cell) signaling, which activates human β-cell G protein-coupled receptors (GPCR), including glucagon receptor (GCGR), and glucagon-like peptide 1 receptor (GLP-1R) [114, 116]. Both glucagon and GLP-1 increase intracellular cAMP levels and elicit synchronous intracellular Ca2 + oscillation respectively in human β-cell [116, 149]. By utilizing optogenetics, based on the signature of Ca²⁺ dynamics, discrete functional subpopulations of β-cells have been identified [14]. Therefore, it is likely that these subpopulations of β-cells will show functional heterogeneity in controlling coordinated electrical regulation and electrical dynamics in response to paracrine actions of glucagon or GLP-1.

Hub cell dynamics

The idea of having pancreatic β-cells residing freely without clustering has been recently studied. Several studies have suggested that the role of these scattered β-cells (hub cells) is to regulate INS expression while coordinating islet oscillatory in a pacemaker-like manner (Fig. 2). A previous study employed large-scale functional cell map** and optogenetics and demonstrated that ~ 1 to 10% of the β-cell population (i.e. hub cells) controls the synchronized and rhythmic action of pancreatic islets [150]. Hub cells have been found to be highly connected to other β-cells and robustly respond to high glucose levels compared to other “follower” β-cell subpopulations (Fig. 2). Hub cells have low expression of PDX1 and NKX6.1 pancreatic markers with increased expression of glucokinase (GCK), an important protein in the glycolysis pathway, but with less INS expression (~ 50%) [150, 151]. In addition, examination of the mitochondrial activity in these hub cells showed no difference in mitochondrial number, but increased activity is observed using tetramethylrhodamine ethyl ester (TMRE) imaging. The increased expression of GCK and mitochondrial activity in hub cells may suggest a mechanism of action for these cells for their high level of respondence to glucose levels. Furthermore, β-cells with low INS levels have been found to have high ATP levels with normal glucose sensing and cell survival [151]. This may suggest that hub cells conserve their energy by decreasing the ATP-consuming INS production process [151]. However, hub cells subjected to glucotoxicity, lipotoxicity or pro-inflammatory conditions, as in diabetes, fail to properly respond to the high glucose levels marking them as metabolically fragile cells, which eventually can lead to ER stress and cell dysfunction [150].

Fig. 2

Schematic representation of pancreatic β-cell hub and follower cells dynamics. Hub cells characterized by the high expression of GCK, possibly due to lower PDX1 and NKX6.1 expression levels compared to follower cells. Hence, hub cells can respond faster to increased glucose levels by metabolizing glucose into pyruvate resulting in more ATP synthesis. The increase in ATP levels causes the closure of ATP-dependent potassium pumps, hence lowering intracellular positive charges, which consequently results in the opening of voltage-gated calcium channels (VGCC) and increases Ca²⁺ uptake by the cells. Later, the taken Ca²⁺ by hub cells can be transferred to follower cells through gap junctions causing stimulation of these cells, with high INS content, to release their stored INS. Hence, hub cells sense the changes in glucose levels more rapidly and take action to serve as ‘pace-maker’ cells

Interestingly, other subpopulations of β-cells exhibiting high functional connectivity, referred to as “wave-initiators” or “leaders” as stated in Sterk et al. [152] and “first responders” as explained in Kravet et al. [153], have been recently identified. It has been suggested that hub, wave-initiator, and first responder cells are characteristic components of islets; however, they exhibit distinct Ca²⁺ signaling characteristics and do not appear to have overlap** functions [152, 153]. The first responders lead the first phase Ca²⁺ response, are more excitable and crucial for mobilizing β-cells to increase Ca²⁺ immediately after glucose stimulation [153]. It has been found that during the transition from low to high glucose levels, the first responder cells play a crucial role in recruiting and coordinating Ca²⁺ activity within a specific time frame, as different cells lead distinct phases of the Ca²⁺ response to glucose (first responders for the first phase, and leader cells for the second phase) [153]. Previous studies have linked leader (wave origin) cells to the regulation of second-phase Ca²⁺ dynamics [152], in which hub cells are associated with maintaining elevated and coordinated Ca²⁺ [14, 154]. This suggests that the heterogeneity controlling second-phase Ca²⁺ is distinct from that of first-phase Ca²⁺.

There has been significant debate about the function of hub cells and other subpopulations and their effect on islet function. The identification and categorization of distinct β-cell subgroups, evaluation of their operational features, and understanding of how the group of cells work together to regulate intercellular calcium activity and insulin release within an islet are highly engaging areas of research that have captured significant interest within the islet biology field. Therefore, in the years to come, we anticipate an increase in research aimed at revealing the distinct contributions of these populations towards the coordinated β-cell activity under different conditions.

Heterogeneity of stem cell-derived islet cells

Generation of stem cell-derived islet organoids: recent progress

Through utilizing hPSCs, different pancreatic differentiation protocols have been designed to mimic human embryonic development in vitro (Fig. 3). Understanding the sequential process of pancreatic development serves as the foundational template for protocols’ design. An initial step in pancreatic differentiation is to generate highly efficient definitive endoderm (DE) cells to be then directed towards hepatic or pancreatic tissues [155, 156]. Afterwards, activation of signaling pathways promoting pancreatic differentiation takes place while inhibiting hepatic-promoting signaling pathways [157]. To generate efficient DE, most protocols include Activin A in the first 3–5 days of differentiation to activate the Activin-Nodal signaling pathway [157, 158]. Generated DE cells are then introduced to FGF10, FGF7 or keratinocyte growth factor (KGF) and retinoic acid (RA) for FGF and RA signaling activation while inhibiting Sonic Hedgehog (SHH) signaling pathway through the addition of KAAD-Cyclopamine or SANT-1 for PDX1 induction [159,160,161]. An essential step then follows where hepatic lineage differentiation is prevented by inhibiting the BMP signaling pathway through BMP inhibitors such as Dorsomorphin or Noggin [162, 163]. Upon PDX1 induction, nicotinamide and EGF are added to induce NKX6.1 expression to generate PDX1⁺/NKX6.1⁺ pancreatic progenitors (PPs) [159, 160]. Following that, PPs are then differentiated into endocrine progenitors (EPs) expressing NEUROG3 through the addition of γ-secretase inhibitors or DAPT to inhibit the Notch signaling pathway, and β-cellulin and ALK5i to block the TGF-β signaling pathway [164]. To induce INS and MAFA production, high glucose media supplemented with Triiodothyronine (T₃), a thyroid hormone, as well as ALK5i is added to the generated EPs to prevent their transition from epithelial to mesenchymal cells [165, 166]. In addition, NKX6.1 expression is found to be exclusively restricted to β-cells where it participates in their development, INS-containing vesicle formation, INS secretory function, and maintains their β-cell identity [167, 168]. As a result, the efficient generation of PDX1 and NKX6.1 co-expressing PPs and later generation of NKX6.1⁺/INS⁺ β-cells are crucial for the generation of mature, functional, mono-hormonal β-cells. Hence, the process of in vitro pancreatic β-cell differentiation utilizes small molecules and growth factors targeting specific pathways to either suppress or enhance them [4].

Fig. 3

Overview of in vitro hPSC-derived pancreatic β-cell differentiation stages. hPSCs are differentiated into β-cells using a stepwise protocol of 7 stages each marked by the expression of specific transcriptional factors. A Some pancreatic differentiation protocols use the 2D culturing system from day 0 until the mature β-cell stage such as Hogrebe et al. [212]. B Other differentiation protocols use a 2D culturing system during the first 4 stages and then shift to 3D system during late stages (stages 5–7) to recapitulate the in vivo islet environment and allow better cell–cell contact such as Rezania et al. protocol [166]. C There are several protocols that start pancreatic differentiation as a 3D system from day 0 moving towards the last stages such as Melton’s protocols [166, 174]. hPSCs human pluripotent stem cells, DE definitive endoderm, PGT primitive gut tube, PF; posterior foregut, PPs pancreatic progenitors, EPs endocrine progenitors

Multiple studies have shown that the generation of functional mono-hormonal pancreatic β-cells is mainly derived from PDX1 and NKX6.1 co-expressing PPs, which currently, can be generated with high efficiency (~ 80–90% of PDX1⁺/NKX6.1⁺ cells) [159, 169]. Constant modifications and protocol enhancements in pancreatic differentiation have led us to the generation of efficient functional pancreatic β-cells ~ 30–50% co-expressing C-PEP and NKX6.1 [165, 166, 170,171,182]. Interestingly, slender T2D patients were found to have higher numbers of poly hormonal cells [183] that could play a role in stress-adaptation and maintaining normal glucose levels. Characterization of these poly hormonal cells revealed that these cells secrete INS and GCG upon non-glucose associated membrane depolarization and may be fated towards α-cells in vivo [182, 184,185,186]. The natural occurrence of these poly hormonal cells suggests that they might have a role in postnatal pancreatic β-cell maturation and function or other important roles that are yet to be discovered.

Pancreatic progenitor and endocrine progenitor subpopulations derived from hPSCs

During embryo development, multipotent PP cells expressing the TFs, PDX1, NKX6.1, FOXA2, SOX9, HNF6, and PTF1A, differentiate into exocrine (acinar and duct) and endocrine (islets of Langerhans) pancreas [168]. PDX1 and NKX6.1 specifically, are known to be crucial factors for generating functional, mono-hormonal INS-secreting β-cells [2]. Therefore, hPSC differentiation protocols aim to generate efficient PP cells co-expressing PDX1 and NKX6.1. However, variations in PDX1 and NKX6.1 expression patterns have been observed when applying the same differentiation protocol on eight different human embryonic stem cell (hESC)/ human induced pluripotent stem cell (hiPSC) lines [2, 159]. This can be considered one of the factors resulting in mature β-cells heterogeneity that must be considered when designing transplantation protocols.

PDX1 is the earliest TF to be produced during pancreas development [187]. Interestingly, as PPs further differentiate, PDX1 expression becomes restricted to β-cells. Although PDX1 is essential for β-cell differentiation, generated PDX1⁺/NKX6.1⁻ PPs differentiate into GCG-secreting cells or non-functional poly hormonal β-cells that co-express INS and other islet hormones. These polyhormonal β-cells cannot be used as an option for transplantation therapy as they lack the hallmark function of β-cells [2, 5]. Multiple studies have reported the generation of GCG-secreting α-cells upon in vivo transplantation of hESC-derived poly hormonal cells [176, 184, 185, 188, 189]. Interestingly, a novel PDX1⁻/NKX6.1⁺ PP population has been recently identified and is able to generate functional glucose-responsive β-cells [168, 190].

Several in vitro pancreatic differentiation protocols suggest the generation of functional β-cells from PDX1⁺/NKX6.1⁺ PP populations that later express NEUROG3 [185, 186]. Studies have shownheterogeneity at EP stage marked by the difference in NEUROG3 expression where NEUROG3^– EPs lacked PDX1 expression but co-expressed CDX2 [191]. Another study further identified subclusters within NEUROG3⁺ cells: EPs (39.1%), poly hormonal endocrine cells (Endo; 42%), duct cells (6%), liver cells (8.2%), and an unknown cell type cluster (4.7%) [192]. The identified EP cluster had three different sub-clusters where they expressed different profiles with EP1 expressing the highest NEUROG3 levels, EP2 expressing TPH1 and FEV, and EP3 expressing GAST. In addition, the authors identified three endocrine clusters targeted towards different cell fates. Endo1 expressed ERO1B and SLC30A8 indicating β-cell fate, Endo2 expressed GCG, PEMT, and IRX2 indicating α-cell fate, while Endo3 expressed SST and HHEX suggesting δ-cell fate [192]. These results show the further complexity in endocrine cell populations and further heterogeneity can be observed as cells differentiate towards end-stages of pancreatic differentiation.

As hPSC-derived pancreatic differentiation protocols can generate heterogenous cell populations in which some could even be committed towards non-pancreatic lineages, several studies aimed to purify PPs directed towards β-cells. Previous studies reported that PPs can be purified using cell surface markers. Glycoprotein 2 (GP2), CD142, and CD24 have been found to be specifically expressed in PPs that can generate functional INS-secreting cells [186, 193,194,195,196]; therefore, those markers have been used to purify PDX1⁺ cells at the PP stage. On the other hand, NEUROG3-expressing cells can be isolated from the heterogenous population at the EP stage using specific surface markers, including SUSD2, CD318, CD133, CD200, and CD49f [186, 197]. This indicates that pure NEUROG3⁺ cells can be obtained for further differentiation into pancreatic islet cells. In 2018, a study compared in vivo and in vitro EP cells (stage 5) post-sorting based on E-cadherin, CD142 and SUSD2 surface markers which were previously described to discriminate between progenitors fated towards β-cells from those fated towards α- or γ-cells [196, 197]. In vitro EPs have increased levels of RFX6 and CDX2 that might reflect a mixed pancreas-duodenum fated populations or early PPs. On the other hand, human fetal EPs express high NKX6.1 and DLK1 levels. In addition, GPM6A, CDKN1A, and BMP5 have been found to be sporadically expressed in the in vitro-generated EPs [198].

Subpopulations of hPSC-derived islet cells

Recently, multiple single-cell transcriptome studies performed on human islet cells and hPSC-derived islets highlighted the unique genetic signatures and intra-islet heterogeneity levels. By tackling these studies, it is suggested that the early stages of pancreatic development are widely consistent with homogenous profiles. On the other hand, during late pancreatic developmental stages where endocrine commitment takes place, this uniformity starts to be lost resulting in different end-stage cells with different genetic profiles. Although limited studies focused on generated pancreatic β-cell populations derived from hPSCs have been found, multiple single-cell analysis studies have identified and supported that islet cells heterogeneity starts from pancreatic and EP stages. These studies also indicated that hPSC-derived islets generated in vitro, differ from adult human islets where cell heterogeneity depends on the differentiation protocol used.

One of the important studies that tackled hPSC-derived β-cell heterogeneity has been done by Veres et al. [174]. The study reported that cells’ heterogeneity starts to appear from stage 4 and continued onwards. At end-stages, β-cells expressing β-cell markers such as INS, NKX6.1, and ISL1, α-like cells expressing GCG, ARX, IRX2, including INS, and a small subset of SST⁺/HHEX⁺/ISL1⁺ resembling δ-cell population have been identified [174]. They also detected a new population of enterochromaffin cells, located within the intestinal epithelium, marked by the expression of TPH1 and SLC18A1, which are normally not present in fetal or adult islets [174]. Several study groups have identified such off-target populations during hPSC-derived islet cell differentiation [174, 192, 199, 200]. Therefore, β-cell purification methods using surface markers can be used to isolate the functional mono-hormonal β-cell fraction from the total heterogenous hPSC-derived islet populations. Recent studies have identified CD49a as a specific cell surface marker for the functional INS⁺ cell fraction, which then can be re-aggregated to form 3D islet organoid structures [174, 201]. Nevertheless, the isolation of β-cells from the generated islet organoids would cause disturbance of the islet architecture that is required to mimic the in vivo environment. Reports have shown that disturbed islets perform less when compared with intact islet cells during GSIS [202,203,204]. Yet, islet cell fractionation and then re-aggregation of isolated β-cells enhances their functional maturation and response to glucose [

Availability of data and materials

No data availability.