Innervation of the pancreas in development and disease Free

Pancreatic developmental biology has been studied intensely with the purpose of developing treatments for pancreatic diseases, in particular diabetes. The molecular mechanisms underlying pancreatic cell fate choice in vivo have informed pluripotent stem cell (PSC) differentiation approaches in vitro. Supporting cells, such as endothelial and mesenchymal cells, play a crucial role in instructing epithelial morphogenesis, cell differentiation and function of the pancreas by secretion of signaling factors and extracellular matrix. In comparison, far less is known about the contribution of neuronal cells. This is surprising, given that the adult pancreas is densely innervated by sympathetic, parasympathetic and sensory branches from the autonomic nervous system (ANS) and contains intrapancreatic ganglia (Babic and Travagli, 2016). It is well established that brain–pancreas crosstalk controls adult function and there is a shared etiology between neurodegenerative diseases and diabetes (Yu et al., 2022). Innervation of the pancreas is therefore being viewed with increasing interest and several recent studies have used modern imaging techniques to document pancreatic innervation (Alvarsson et al., 2020; Bsharat et al., 2023; Campbell-Thompson et al., 2021; Croizier et al., 2016; Krivova et al., 2022; Yang et al., 2018, 2022), discovering patterns implying connections between innervation and endocrine differentiation and diabetes. In this Review, we give an overview of recent progress in mapping pancreatic innervation and understanding the interactions between pancreatic neurons and pancreatic epithelial morphogenesis and cell differentiation, drawing parallels with other branched organs. Pancreatic innervation has been investigated in several species (Alvarsson et al., 2020; Kirchgessner and Adlersberg, 1992; Kirchgessner et al., 1996; Rozman et al., 2002; Sheikh et al., 1988; Yang et al., 2018); we focus on studies in the mouse, comparing these with human data and supplementing with data from other species where relevant. We also discuss how these findings apply to human pancreatic function and in the development of diabetes. We believe unraveling the relationship between innervation and pancreatic development will be crucial for understanding the full etiology of pancreatic diseases such as diabetes. Additionally, such knowledge could help identify factors that increase the efficiency of human PSC differentiation protocols for beta cells and islet organoids.

The pancreas is an endoderm-derived abdominal gland located close to the stomach. The gland has two functionally distinct epithelial compartments: exocrine and endocrine (Fig. 1). The exocrine compartment comprises two major cell types: (1) acinar cells, which form discrete clusters (acini) surrounding the peripheral ends of the ducts and synthesize nutrient-digestive enzymes, and (2) ductal epithelial cells, which neutralize these enzymes by secretion of chloride and bicarbonate and form a tubular structure that transports the enzymes into the duodenum (Avolio et al., 2013). The endocrine compartment contains cell clusters known as the islets of Langerhans (hereafter simply denoted ‘islets’). Each islet comprises five different endocrine cell types – alpha, beta, delta, epsilon and PP cells – that regulate glucose homeostasis by producing and secreting peptide hormones (glucagon, insulin, somatostatin, ghrelin and pancreatic polypeptide) into the bloodstream (Avolio et al., 2013). Insulin and glucagon play an important role in blood glucose homeostasis, with insulin promoting glucose uptake and storage, and glucagon stimulating glycogenolysis and gluconeogenesis. Mature beta cells also secrete the neurotransmitters dopamine (Barrado et al., 2015), serotonin (Almaça et al., 2016) and gamma-aminobutyric acid (GABA) (Menegaz et al., 2019), which are believed to exert autocrine and paracrine regulation of insulin and glucagon secretion (Menegaz et al., 2019). Additionally, the pancreas contains neurons and mesenchymal, endothelial, lymphatic and immune cells. Most of these non-epithelial cells reside in the mesenchymal tissue surrounding the pancreatic buds during development (Glorieux et al., 2022). These features are conserved between rodents and humans.

Pancreas development becomes morphologically evident with the induction of two pancreatic buds emerging from the foregut endoderm through the expansion of multipotent progenitor cells at embryonic day (E) 9.0 in the mouse (Burlison et al., 2008; Jørgensen et al., 2007) and at Carnegie stage 13 in humans (Jennings et al., 2013). The induction of the pancreatic buds depends on signals from adjacent mesoderm-derived tissues (Deutsch et al., 2001; Lammert et al., 2001; Rossi et al., 2001; Yoshitomi and Zaret, 2004), constituting the first cellular crosstalk in pancreatic development. Next, a branched, tubular epithelial network is generated through the stratification and polarization of epithelial cells and through microlumen formation and fusion (Villasenor et al., 2010). Parallel to these morphological changes, differentiation of the three main pancreatic cell types occurs. Acinar cells differentiate in the extending tip epithelium of the branches (Schaffer et al., 2010; Zhou et al., 2007), whereas the trunk contains bipotential progenitors, which differentiate into endocrine and ductal cells (Fig. 1). In the latter stages of embryonic development and the first weeks of postnatal development, endocrine cells migrate inside the epithelium and coalesce into islets in close association with ducts (Nyeng et al., 2019; Sharon et al., 2019). For a detailed account of the molecular regulation of pancreatic differentiation and morphogenesis, we refer the reader to other reviews (Flasse et al., 2021; Pan and Wright, 2011; Shih et al., 2013). During development, the diverse cell types in the pancreas influence each other by extensive crosstalk, exerting distinct roles depending on the developmental stage. Classical tissue-recombination studies showed that signals from the mesenchymal tissue are necessary for maintaining pancreatic epithelial cells in an undifferentiated and proliferative stage during early murine development (Golosow and Grobstein, 1962; and several others). Subsequent genetic and co-culture experiments have addressed contributions from individual supporting cell types, including among others: mesenchymal cells (Cozzitorto et al., 2020; Landsman et al., 2011), endothelial cells (Lammert et al., 2001; Pierreux et al., 2010; Sand et al., 2011; Yoshitomi and Zaret, 2004), pericytes (Sakhneny et al., 2021) and macrophages (Banaei-Bouchareb et al., 2004; Migliorini et al., 2023 preprint). Understanding the crosstalk between pancreatic epithelial cells and mesenchymal and endothelial cells has informed the design of protocols that aim to differentiate human PSCs into insulin-producing beta cells as a potential treatment for diabetes (Seymour and Serup, 2019; Sneddon et al., 2018; Talavera-Adame et al., 2016). In comparison, relatively little is known of the interplay with neuronal cells.

The rich nerve supply to the pancreatic islets was noted when the islets were first discovered (Langerhans et al., 1937). Subsequently, multiple studies verified that the pancreas forms an intrinsic neural network that is innervated by the ANS through sympathetic, parasympathetic and sensory branches (Burris and Hebrok, 2007; Campbell-Thompson et al., 2021; Croizier et al., 2016; Krivova et al., 2022; Sharkey et al., 1984) (Fig. 2). The ANS is crucial for organ homeostasis and controls many functions in the body, including blood pressure, body temperature and cardiac output (Sheng and Zhu, 2018), upon stimulation from triggers such as starvation, eating and stress (Scheurink et al., 1996; Tahara et al., 2017). It is currently not known whether similar triggers exist during embryonic development. Autonomic neurons are highly dependent on guiding signals from their target organs during development (Glebova and Ginty, 2005). Nerve growth factor (NGF) is one such neurotrophic factor that is expressed by peripheral tissues and guides their innervation by the ANS (Aloe et al., 2012). Conversely, if and how autonomic innervation contributes to the development of target organs remains an open question. Owing to the intricate shape of neuronal networks and the heterogeneous cellular composition of the pancreas, it is difficult to map precisely the co-development of epithelial and neuronal cells. Slender neuronal cells and minuscule neuronal–epithelial connections are easily missed, especially in two-dimensional sections. Recent studies (Bsharat et al., 2023; Campbell-Thompson et al., 2021; Glorieux et al., 2022; Plank et al., 2011; Yang et al., 2022) have therefore used three-dimensional (3D) imaging techniques, such as confocal and light-sheet microscopy, on whole or partial organs. However, more 3D imaging studies are needed, especially at the embryonic stages. A further complication is the varying results obtained depending on which neuronal markers are used. For example, the so-called pan-neuronal marker NF200 does not label all tyrosine hydroxylase (TH)⁺ sympathetic neurons and vesicular acetylcholine transporter (VAChT)⁺ parasympathetic neurons in mouse pancreas (Alvarsson et al., 2020). Furthermore, some antigens, such as VAChT (Rodriguez-Diaz et al., 2011), which are considered neuronal markers, also label pancreatic endocrine cells.

Pancreatic innervation starts with the development of intrinsic neurons, which are believed to form the intrapancreatic ganglia (Li et al., 2019) (Fig. 3). These ganglia are diffusely distributed in the pancreatic parenchyma of the adult pancreas, located near nerve trunks and islets (Li et al., 2019). Their axons project to other intrapancreatic ganglia, to islets in the endocrine pancreas, and to ducts and acini (Li et al., 2019; Tang et al., 2018a,b) (Fig. 2). Intrapancreatic ganglia express markers of sympathetic, parasympathetic and sensory neurons, including noradrenaline (NA) (Yi and Love, 2005), acetylcholine (ACh) (Wang et al., 1999) and substance P (SP; tachykinin 1) (Shen et al., 2016) (Table 1), making it difficult to distinguish between extrinsic and intrapancreatic neurons. The exact role of these ganglia is far from understood, but they appear to directly stimulate insulin secretion (Sha et al., 2001; Stagner and Samols, 1985) and inhibit exocrine secretion (Jiang et al., 2011). In the adult mouse, these intrapancreatic ganglia integrate input from the sympathetic and parasympathetic neurons and make up a significant portion of the neurons that innervate the islets (Alvarsson et al., 2020). The development of pancreatic islet innervation is closely integrated with postnatal islet maturation; just days after birth, both sympathetic vesicular monoamine transporter (VMAT)⁺ and sensory calcitonin gene-related peptide (CGRP)⁺ neurons are observed adjacent to endocrine cells and these neurons coalesce (i.e. form an aggregate) with the endocrine cells to form islets in the first postnatal week (Burris and Hebrok, 2007). Recent single-cell expression analysis has shown that beta and alpha cells express several neurotransmitter receptors, including sympathetic and parasympathetic receptors (Bsharat et al., 2023). Notably, the transcription factor MAFB, which directs beta and alpha cell differentiation, is required for expression of the neurotransmitter receptor genes, suggesting that endocrine differentiation and neuro-endocrine crosstalk are closely linked (Bsharat et al., 2023).

Intrapancreatic ganglia are believed to derive from neural crest (NC) cells that invade the pancreas during development (Li et al., 2019) (Fig. 3). Lineage-tracing studies have demonstrated that NC cells form the intrinsic neural network of the lungs (Burns et al., 2008) and the gut (Le Douarin and Teillet, 1973; Li and Ngan, 2022). Genetic knockout (KO) studies suggest that NC cells also establish the intrinsic pancreatic neural network. The absence of NC cells in Phox2b (Nekrep et al., 2008) and Foxd3 KO mice (Plank et al., 2011) is associated with reduction/loss of intrapancreatic neurons and glia cells at embryonic stages. Lineage tracing and depletion of all NC cells in zebrafish via Sox10 KO (Yang et al., 2018) suggest that NC cells are responsible for pancreatic innervation in other species as well. However, lineage tracing and KO studies of longer duration and the use of more specific neuronal markers are required to determine whether the intrapancreatic ganglia are entirely NC derived.

The NC originates from the ectoderm, in the dorsal neural tube. It is a multipotent embryonic cell lineage giving rise to multiple cell types throughout the body (Bronner-Fraser and Fraser, 1988). The NC population, which is believed to form the intrinsic neural network of the pancreas and the gut, derives from the vagal neural crest in the hindbrain region (Hutchins et al., 2018) and is marked by expression of Phox2b and Wnt1 (Pattyn et al., 1999; Plank et al., 2011; Teng et al., 2008). These vagal neural crest cells go through a two-stage process whereby they first migrate through the foregut mesenchyme, and subsequently infiltrate the pancreatic buds. Apart from sharing the same entry point in the foregut, several protein expression studies have found that NC cells in the gut and the pancreas share common neuronal precursors (Tharakan et al., 1995; Kirchgessner and Liu, 2001; Kirchgessner and Pintar, 1991). Glial cell-derived neurotrophic factor (GDNF) plays an important role in the development of both the enteric nervous system and the intrinsic neural network in the pancreas, by regulating the migration, proliferation and differentiation of NC cells (Heuckeroth et al., 1998; Muñoz-Bravo et al., 2013; Sanchez et al., 1996; Young and Newgreen, 2001). Additionally, other factors, such as netrin 1, netrin 3 and DDC (deleted in colorectal cancer receptor), are involved in the migration of NC cells to the gut and pancreas (Jiang et al., 2003). Studies investigating what controls the NC choice of migratory pathway to the gut tube versus the pancreas are, however, missing.

Anatomical dissections have uncovered extensive networks of nerves traversing between the gut and the pancreas, predominantly serotonergic and cholinergic (Tiscornia, 1977), and anterograde studies have shown that neurons from the myenteric plexus of the gut innervate intrapancreatic ganglia, acinar cells and islets in guinea pigs and rats (Chandrasekharan and Srinivasan, 2007; Kirchgessner et al., 1996; Kirchgessner and Gershon, 1990, 1995). In addition, a variety of hormonal peptides in the gut, including gastrin-releasing-peptide (GRP) and SP have been shown to influence pancreatic exocrine secretion through neural pathways (Konturek et al., 2003). All these findings suggest that the enteric nervous system contributes to the innervation and function of the pancreas. When enteric neurons initiate innervation of the pancreas and how the function of the entero-pancreatic innervation is coordinated during pancreatic development has not been clarified.

When exactly the NC cells enter the pancreas is somewhat disputed, ranging from E10.0 to E12.5 (Burris and Hebrok, 2007; Kirchgessner and Adlersberg, 1992; Nekrep et al., 2008; Plank et al., 2011; Shimada et al., 2012). A series of elegant experiments using pancreatic rat explants with or without inclusion of associated foregut demonstrated that the pancreas has no internal neuronal precursor cells until E13.5 but is innervated by NC cells between E12.5 and E13.5 (equivalent to E11.5-E12.5 in the mouse) (Kirchgessner and Adlersberg, 1992). These findings were later confirmed by lineage tracing in the mouse, with Wnt1-Cre marking NC cells (Nekrep et al., 2008). However, a later Wnt1-Cre lineage-tracing study demonstrated that murine NC cells already make contact with the pancreatic mesenchyme at E10.0 (Plank et al., 2011). According to this study, these NC cells surround the pancreatic buds at E11.5 and intermingle with pancreatic cells at E15.5 (Plank et al., 2011) (Fig. 3). The discrepancy between the two studies may be due to the use of 3D imaging in the Plank study. After arrival, the NC cells can be seen close to pancreatic epithelial cells, where they downregulate the NC marker Phox2b and start neuronal differentiation (Nekrep et al., 2008). Interestingly, NC-derived cells tend to be enriched near endocrine cells, in particular alpha cells (within 20 µm) during late development (E18.5) (Shimada et al., 2012). In the adult pancreas, NC-derived cells are enriched at the periphery of islets, consistent with the regulatory role neurons have in pancreatic hormone secretion (Plank et al., 2011; Shimada et al., 2012). Several studies have demonstrated that signals from the pancreatic NC cells regulate endocrinogenesis (Nekrep et al., 2008; Plank et al., 2011). Loss of NC cells in a global Phox2b KO mouse model led to an increase in the number of beta cells and beta cell proliferation at E17.5 and an increase in expression of the beta cell differentiation factor Nkx2-2 at E15.5 and E17.5, indicating that NC cells negatively regulate beta cell proliferation and/or differentiation during embryonic development (Nekrep et al., 2008). No effect on other endocrine cell types was found (Nekrep et al., 2008). In Foxd3 KO mice, the absence of NC cells likewise led to increased Nkx2-2 expression and beta cell proliferation at embryonic stages, but was also associated with decreased beta cell maturation, as evidenced by decreased expression of the transcription factor genes Mafa and Pdx1 (Plank et al., 2011). In summary, it appears that NC-derived cells form most of the intrinsic pancreatic neural network and contribute to regulation of beta cell development by decreasing beta cell proliferation and/or differentiation and enhancing beta cell maturation.

The sympathetic innervation of the pancreas originates from the dorsal root ganglia in the thoraco-lumbar area of the spinal cord. Here, sympathetic preganglionic neurons send axons to the celiac and superior mesenteric ganglia, through the splanchnic nerves (Babic and Travagli, 2016). Sympathetic postganglionic fibers travel from the celiac and superior mesenteric ganglia to the pancreas, where they innervate intrapancreatic ganglia, islets, acini, ducts and vasculature (Fig. 2, Table 1). The main sympathetic neurotransmitter is NA, but sympathetic neurons also release galanin and neuropeptide Y (Babic and Travagli, 2016) (Table 1). In the mature pancreas, sympathetic neurons cause vasoconstriction of pancreatic blood vessels (Barlow et al., 1974) and regulate blood glucose homeostasis by inhibiting insulin secretion from beta cells (Gilliam et al., 2007; Malaisse et al., 1967) and upregulating glucagon release from alpha cells (Adeghate et al., 2000) to convert glycogen stores to blood glucose, as a response to stress and hypoglycemia (Table 1). For an in-depth review of neuronal regulation of pancreatic islet function, we refer the reader to other reviews (Hampton et al., 2022; Babic and Travagli, 2016). The significance of sympathetic innervation for the function of intrapancreatic ganglia, acini and ducts is not known.

The earliest presence of sympathetic neurons has been detected at E12.5 in mice (Burris and Hebrok, 2007) and at 8 weeks post-conception in humans (Krivova et al., 2022) based on immunohistochemical staining with the sympathetic markers VMAT2 (SLC18A2) and TH, respectively (Fig. 3). Ngf-deficient mice have reduced sympathetic innervation of the pancreas at E16.5 and shortly after birth (Glebova and Ginty, 2004), and overexpression of Ngf in beta cells leads to increased sympathetic innervation of the pancreatic islets in transgenic mice (Edwards et al., 1989), suggesting that NGF directs sympathetic innervation of the embryonic pancreas. Ngf is expressed by beta cells in the mature rat pancreas (Rosenbaum et al., 1998). NGF has also been detected in rat embryonic pancreas, but not in rat embryonic islets (Miralles et al., 1998). In the mouse embryo, Mafb⁺ pancreatic endocrine progenitor cells express Ngf (Bsharat et al., 2023). Further analysis of single-cell sequencing data from embryonic pancreas is needed to determine whether other pancreatic cell types express NGF.

Relatively little is known regarding how sympathetic innervation impacts the developing pancreas, but a few studies have pointed to a role in regulation of endocrinogenesis (Table 1). Chemical (6-hydroxydopamine) and genetic (TrkA receptor Ntrk1 KO) ablation of sympathetic neurons in early postnatal mice perturbed the cluster-like appearance of islets and led to glucose intolerance 1 month after birth due to defects in insulin secretion (Borden et al., 2013). No defects in beta cell differentiation were observed, but beta cells were found to migrate towards sympathetic ganglia in vitro in a β-adrenergic receptor-dependent fashion, suggesting that the phenotype is due to defects in beta cell migration (Borden et al., 2013). In support of a role for sympathetic innervation in the regulation of endocrine differentiation, a study of the global KO mouse for TH (a rate-limiting enzyme in the production of catecholamines) found that the number of beta cells, and expression of the pro-endocrine factors Neurog3 and Ins1, were reduced at E13.5 (Vázquez et al., 2014). Treatment of wild-type mouse pancreas explants with exogenous dopamine increased beta cell numbers without affecting proliferation, suggesting that dopaminergic signaling increases beta cell differentiation (Vázquez et al., 2014). However, immature alpha and beta cells express TH (Teitelman et al., 1987; Vázquez et al., 2014) so it is unclear whether the effect of TH KO is due to signaling via sympathetic neurons or a direct effect on endocrine cells. These sympathectomy studies also only focused on endocrine cells and did not investigate whether there were any defects in general pancreatic morphogenesis and exocrine differentiation. This question has been explored in the salivary glands, which are also innervated by sympathetic and parasympathetic branches of the ANS (Proctor and Carpenter, 2007). Sympathectomy of the early postnatal rat mandibular salivary gland resulted in retarded development of acinar and ductal cells and reduced gland size (Srinivasan and Chang, 1977). It would therefore be interesting to investigate whether pancreatic morphogenesis and exocrine differentiation are likewise regulated by the sympathetic nervous system.

Parasympathetic innervation of the pancreas originates mostly from the parasympathetic preganglionic neurons in the dorsal motor nucleus of the vagus nerve and partly from the nucleus ambiguous in the brain stem (Fox and Powley, 1986). The majority of the parasympathetic preganglionic axons travel in the posterior vagal truncus (Tiscornia et al., 1976) and synapse on intrapancreatic ganglia, whereas parasympathetic postganglionic nerves present in the pancreas pass directly to the pancreatic islets and acini (Lkhagvasuren et al., 2021) (Fig. 2, Table 1). The main neurotransmitter released from parasympathetic neurons is ACh, but nitric oxide, vasoactive intestinal polypeptide and gastrin are also released (Hampton et al., 2022) (Table 1). Like sympathetic neurons, parasympathetic nerves are also involved in blood glucose homeostasis. Ingestion of food activates parasympathetic neurons, which stimulate secretion of digestive enzymes from acinar cells (Mussa and Verberne, 2008) and insulin secretion from beta cells (Mussa et al., 2011). As a result, glucose is removed from the blood and stored in the liver as glycogen and euglycemia is restored. Parasympathetic neurons also stimulate glucagon secretion from alpha cells in times of hypoglycemia (Berthoud and Powley, 1990) (Table 1). In the adult rodent pancreas, abolishing parasympathetic stimuli by vagus nerve transection decreases beta cell proliferation and increases acinar cell proliferation (Lausier et al., 2010). The significance of parasympathetic innervation for the function of intrapancreatic ganglia is not known, but we speculate that some of the parasympathetic effects on pancreatic epithelial cells may be mediated via the intrapancreatic ganglia. In conclusion, the parasympathetic system has a stimulatory effect on the endocrine and exocrine functions of the adult pancreas.

Small numbers of parasympathetic fibers can be detected in the embryonic mouse pancreas at E12.5 based on immunohistochemical staining with the parasympathetic marker VAChT (Croizier et al., 2016) (Fig. 3). These fibers are located close to immature beta cells. Around E15.5, the density of the parasympathetic fibers is increased markedly, and at E18.5 parasympathetic axons appear to synapse with the endocrine cells (Croizier et al., 2016). Live-imaging studies in zebrafish likewise revealed close interactions between early endocrine cells and neurons, and parasympathetic ablation selectively decreased pancreatic delta cell numbers (Yang et al., 2018) (Table 1). Although the role of parasympathetic innervation in the embryonic mouse pancreas has not yet been directly investigated to our knowledge, a study on the effect of leptin during embryonic development suggests that long-term defects in glucose homeostasis after leptin injection are due to reduced parasympathetic innervation of the islets (Croizier et al., 2016), potentially linking endocrinogenesis and parasympathetic innervation. In mouse salivary gland, parasympathetic innervation maintains progenitor cells in an undifferentiated stage during organogenesis (Knox et al., 2010). Loss of parasympathetic innervation also impacts salivary gland tubulogenesis, leading to decreased epithelial branching and disrupted ductal lumen formation in an ex vivo explant model (Nedvetsky et al., 2014). This latter finding is of particular interest, because multiple studies have observed that pancreatic tubulogenesis is crucial for differentiation of pancreatic endocrine cells, and, in particular, beta cells (Kesavan et al., 2009; Löf-Öhlin et al., 2017; Nyeng et al., 2019). We therefore speculate that any potential effect of the parasympathetic system on pancreatic tubulogenesis could indirectly affect endocrinogenesis.

Sensory innervation of the pancreas is divided into two pathways: spinal afferent and vagal afferent (Hampton et al., 2022). Pancreatic spinal sensory neurons are located in the dorsal root ganglia in the thoraco-lumbar area and their axons meet the splanchnic nerves and celiac plexus, where they travel to the pancreas. Pancreatic vagal sensory neurons originate in the nodose ganglia, and travel in the vagus nerve to reach the pancreas (Lkhagvasuren et al., 2021) (Fig. 2 and Table 1). Neurotransmitters released by sensory neurons include SP and CGRP (Table 1). Several studies have found that sensory neurons play a role in disease-associated pain (Wick et al., 2006a,b). Whether sensory neurons have a role in pancreatic functions under normal conditions is controversial, as ablation of pancreatic sensory neurons has been shown to either reduce (Karlsson et al., 1992) or not affect (Jaworek et al., 1997; Van De Wall et al., 2005) glucagon secretion, and increase (Karlsson et al., 1994), decrease (Karlsson et al., 1994; Van De Wall et al., 2005) or not affect (Karlsson et al., 1992) insulin secretion. Immunohistochemical staining with the sensory marker CGRP has demonstrated that sensory neurons are present in the embryonic mouse pancreas at E12.5, and that sensory fibers are present from E15.5 (Burris and Hebrok, 2007) (Fig. 3). In the early postnatal stages, sensory neurons can mostly be found at the periphery of the islets, in close contact with the endocrine cells (Burris and Hebrok, 2007). If and how sensory innervation influences the morphogenesis and endocrinogenesis of the pancreas has, to our knowledge, not been investigated.

There are some notable differences between the mature pancreatic islets of humans and rodents. Human islets contain proportionally fewer beta cells and more alpha cells compared with murine islets (Cabrera et al., 2006). They also have different cytoarchitecture. In humans, alpha, beta and delta cells are intermingled throughout the islets, whereas murine islets are more organized, with alpha and delta cells at the periphery and beta cells in the core (Cabrera et al., 2006). However, fetal human pancreatic islets are formed with a more murine-like cytoarchitecture, with alpha cells at the edge and beta cells in the core (Jeon et al., 2009; Piper et al., 2004).

The innervation of the mature pancreas has been studied intensively in both mouse and human (Alvarsson et al., 2020; Rodriguez-Diaz et al., 2011; Tang et al., 2018a), but only a few studies have investigated innervation during embryonic development (Burris and Hebrok, 2007; Croizier et al., 2016; Krivova et al., 2022; Shimada et al., 2012). A handful of studies have directly compared adult mouse and human pancreatic innervation (Alvarsson et al., 2020; Rodriguez-Diaz et al., 2011; Tang et al., 2018a). They found that the endocrine compartment is more innervated than the exocrine in the adult mouse, whereas the two compartments are more equally innervated in humans (Alvarsson et al., 2020; Rodriguez-Diaz et al., 2011). Although all studies agree that rodent islets are innervated by both sympathetic and parasympathetic branches (Berthoud and Powley, 1990; Gilliam et al., 2007; Malaisse et al., 1967), the extent of human islet innervation has been disputed. Rodriguez-Diaz and colleagues reported that innervation of human islets is very limited compared with that of rodents (Rodriguez-Diaz et al., 2011). They based this conclusion on immunostaining data from human and murine tissue sections showing that human islets are mostly innervated by sympathetic neurons that rarely contact the endocrine cells, preferentially contacting the blood vessels instead. They observed only a few parasympathetic axons inside the islets. In contrast to these results, several recent 3D studies have revealed several similarities in innervation of human and mouse islets (Alvarsson et al., 2020; Campbell-Thompson et al., 2021; Tang et al., 2018a). Based on whole-tissue imaging of expression of NF200, Alvarsson and colleagues found that only a fraction of islets in both humans and rodents are in close association with neurons. Innervated islets were significantly larger than non-innervated islets in both species, further supporting the notion that neurons play a role in islet development and maturation (Alvarsson et al., 2020). However, human islets were less densely innervated than rodent islets, and human intrapancreatic ganglia were larger than the rodent ganglia (Alvarsson et al., 2020). 3D imaging of TH⁺ and VAChT⁺ fibers has also confirmed that human islets, like mouse islets, are densely innervated by sympathetic neurons (Campbell-Thompson et al., 2021; Tang et al., 2018a) and also have considerable parasympathetic innervation (Tang et al., 2018a). These results imply that neurotransmitters from both sympathetic and parasympathetic neurons reach endocrine cells directly. Although only a fraction of the beta cell population is directly innervated (∼9% in mouse and 4% in human), the coupling of beta cells through gap junctions (Serre-Beinier et al., 2009) permits the neuronal signal to reach multiple beta cells indirectly. The differing results between studies can likely be ascribed to the difficulty in assessing neuronal connections in 2D sections as opposed to 3D tissue, the considerable regional differences within each pancreas (Alvarsson et al., 2020), and variability between human pancreatic samples. Overall, given that multiple 3D studies agree that human islets are more densely innervated than previously assumed, it seems likely that human and murine pancreatic innervation could be more similar than previously thought, emphasizing the relevance of rodent models to human.

Diabetes is a global medical and societal problem (Lin et al., 2020). The disease can be categorized into type I and type II diabetes (T1D, T2D). T1D results from autoimmune-mediated destruction of beta cells (Gillespie, 2006), whereas T2D is a multifactorial disease characterized by insulin resistance in the target tissues leading to relative insulin deficiency and beta cell exhaustion (Chatterjee et al., 2017). In both types, lack of functional beta cells leads to low insulin secretion and hyperglycemia. One well-established complication of diabetes is damage to the autonomic nerves, and people with diabetes that have associated autonomic dysfunction have poorer glycemic control compared with those without nerve complications (Liao et al., 1995; Singh et al., 2000). In addition, there is accumulating evidence linking diabetes to neurodegenerative diseases, including Parkinson's disease (Yu et al., 2022) and dementia (Umegaki, 2012), and both type T1D and T2D are associated with decreased performance in several neuropsychological functions (Roriz-Filho et al., 2009), indicating that diabetes influences the central nervous system. Loss of local sympathetic islet innervation has been documented in several animal models of T1D (Mei et al., 2002; Persson-Sjögren et al., 2005; Taborsky et al., 2009; Winer et al., 2003) and in humans with T1D (Mundinger et al., 2016). In contrast, loss of sympathetic neurons has not been observed in people with T2D (Alvarsson et al., 2020), and gestational diabetes is associated with increased sympathetic activity (Plows et al., 2018; Reyes et al., 2020). Parasympathetic innervation is reduced in the exocrine, but not in the endocrine, pancreatic compartment of people with recently diagnosed T1D (Lundberg et al., 2017). Whether this reduction in parasympathetic axon density occurred before or after the onset of diabetes was not investigated. In conclusion, loss of sympathetic islet innervation and reduction of parasympathetic exocrine innervation is exclusively associated with early stages of T1D. Based on these findings and the influence of the ANS on alpha and beta cell function, it appears reasonable to argue that loss or reduction of pancreas sympathetic innervation in the early stages of T1D aggravates hormone secretion defects and contributes to the progression of diabetes. It is established that autonomic dysfunction can be a complication of diabetes, but it has also been suggested that defects in pancreatic innervation may increase the risk of developing diabetes. An underlying neuronal autoimmunity has been established in the non-obese diabetes (NOD) mouse model as well as in humans with T1D, indicating that diabetic autoimmune targeting of glia surrounding the islets contributes to early diabetic inflammation (Tsui et al., 2008). Sensory innervation seems to promote islet inflammation: Removal of transient receptor potential cation channel subfamily V member 1 (TRPV1)⁺ sensory neurons from islets in adult NOD mice results in protection from diabetes despite loss of self-tolerance (Razavi et al., 2006). Ablation of sensory nerves has further been shown to improve glucose-stimulated insulin release and glucose tolerance in a murine model for T2D (Zucker diabetic fatty rats) (Gram et al., 2007; Moesgaard et al., 2005) suggesting that sensory innervation may contribute to insulin resistance and T2D. Two epidemiological studies have found that people with impaired autonomic function have increased risk of developing T2D (Carnethon et al., 2003a,b). As described earlier, two rodent studies have shown that sympathetic neurons influence the formation of pancreatic islets (Borden et al., 2013; Bsharat et al., 2023) affecting the degree of glucose tolerance later in life (Borden et al., 2013). This points to a possible involvement of the ANS during prenatal or early postnatal pancreas development. More research is needed to understand how the different pancreatic endocrine cells are regulated by the ANS during development and to elucidate a possible contribution of the ANS in the pathogenesis of diabetes.

Stem cell-based therapies for diabetes have been a focus for the pancreas and diabetes field for more than a decade and some studies have now moved into clinical trials (Siehler et al., 2021). Early strategies leveraged knowledge of pancreas development to establish protocols for differentiation of beta cell populations (Pagliuca et al., 2014; Rezania et al., 2014; and several others). Because of difficulties in obtaining functional beta cells in vitro by this approach, it has been suggested that making islet-like organoids would be a more successful approach (Siehler et al., 2021). In support of such a strategy, several lines of evidence show that pancreatic progenitor cells and immature beta cells seem to require contact with other cell types in order to thrive; for example, culture of isolated pancreatic progenitor cells in vitro is challenging (Grapin-Botton, 2016), whereas entire pancreatic organs can be excised from mouse embryos and grown successfully in vitro. An organoid can be defined as ‘a 3D structure derived from either pluripotent stem cells or adult stem/progenitor cells, in which cells spontaneously self-organize into properly differentiated functional cell types, and which recapitulates at least some function of the organ’ (Huch et al., 2017). Attempts at making pancreatic organoids from embryonic mouse pancreatic progenitor cells or differentiated human PSCs have shown that these cell types can form all epithelial components of the pancreas (Greggio et al., 2013). However, they lack vessels and neurons, and are only partly functional (Greggio et al., 2013). By contrast, in vivo maturation of human PSC-derived immature beta cells typically generates highly vascularized islet organoids that resemble native islets both morphologically and functionally (Augsornworawat et al., 2020; Rezania et al., 2014). Whether these organoids contain endogenous neurons has not been investigated. Other protocols for making islet organoids have focused on addition of endothelial-derived ligands and have demonstrated significantly improved function, highlighting the importance of endothelial cell signaling (Aghazadeh et al., 2021; Yoshihara et al., 2020). To our knowledge, no studies have yet attempted to incorporate neurons or neuronal ligands into pancreatic or islet organoids. However, co-transplantation of isolated human islets with post-migratory NC stem cells increased beta cell proliferation and enhanced neural and vascular function of the engraftment in one study (Grapensparr et al., 2015). Based on this study, and given the functional requirement of neurons for pancreatic function reviewed here, incorporation of neuronal or NC cells into organoids may be a promising strategy.

It is well established that the pancreas has an intrinsic neuronal network and is densely innervated by sympathetic, parasympathetic and sensory neurons during development. We also know that the ANS regulates adult pancreatic function. Although the extent of human pancreatic innervation has been debated, recent 3D analysis has revealed that human islets are more innervated than previously believed. Combined with epidemiological and histological studies linking diabetes with both pancreatic innervation defects and neurodegenerative diseases, this points to an important role of innervation for pancreatic function. Despite this, only a handful of modern studies have investigated pancreatic innervation during embryonic development, and it remains an open question whether pancreatic neurons have a significant instructive role, as seen for the vasculature and mesenchyme. It would be of particular interest to map embryonic innervation in 3D and analyze heterologous interactions between neurons and other pancreatic cell populations. Such data could address if and how the differentiation and morphology of neuronal networks, vasculature and epithelial compartments are coordinated. Recent developments in four-dimensional imaging of pancreas development ex vivo (Nyeng et al., 2019), pancreatic organoid cultures (Greggio et al., 2013), light sheet microscopy (Glorieux et al., 2022), mouse transgenic reporter strains, spatial transcriptomics (Olaniru et al., 2023), single-cell-based analyses of pancreatic cell populations (Yu et al., 2019) and deep learning-based methods for analyzing imaging (Arnavaz et al., 2022) and gene expression data (Wang et al., 2022), have expanded the tool-box of the developmental biologist. These new methods should be used to further explore pancreatic innervation and determine the mechanistic interactions between endocrine and neuronal cells. This will hopefully deepen our understanding of the neuronal contribution to pancreatic development and disease.

We apologize to those scientists whose work has not been cited owing to space limitations.