Local Dialogues Between the Endocrine and Exocrine Cells in the Pancreas Free

In this Perspective, we revisit long-standing controversies, that is, “an elephant in the room” in the field of islet biology and insulin secretion, and discuss how they could be resolved with a more integrative analysis of pancreas. We lay down a number of arguments for why we need to focus on common traits of the endocrine and exocrine pancreas and how understanding of their dialogue could change our current understanding of the metabolic signaling in the pancreas.

The prevailing model of islet blood supply was, to the best of our knowledge, first collectively proposed by Wharton (1) in his thesis for the degree of Master of Science in Medicine at the Mayo Foundation in 1932, which is entitled “The Blood Supply of the Pancreas, With Special Reference to That of the Islands of Langerhans.” The opening remarks are as follows, “This study was undertaken in order to clarify the various ideas concerning the relation of the blood supply to the glandular tissues of the pancreas.” The author evaluated a number of papers published between 1877 and 1930 to elaborate on the controversies at the time. Particularly regarding the intrapancreatic circulation, he categorized the publications into three models: model 1, arterial blood supply to the islet, wherein the islet was essentially an arterial formation with single or numerous afferent arteries which arose from the distributing intralobular arteries (2–7); model 2, venous blood supply to the islet, wherein intralobular vessels surrounded by islet cells were considered veins, since they were without membrana propria or adventitia (8); and model 3, others, wherein the vessels of the islets were dilated portions of the general capillary bed, with which they retained their connections (9–11). Specifically, Opie (9) described that a single afferent vessel did not enter the glomerular arrangement (i.e., the islet), as in the kidney, but that it was continuous with the capillary interacinar plexus. Hansemann (10) considered the islets as arising from the stroma with the capillaries but retaining their capillary connections with the surrounding capillary bed. Similarly, Pensa (11) reported the continuous capillary network between the islets and the rest of the pancreas. It appears that the third model (summed as “others”) indicates the integrated pancreatic blood flow between islets and exocrine tissues. However, Wharton’s interpretation of his own experiments was that blood supply to the islet is arterial, which is further discussed below. Nonetheless, Wharton’s model of the pancreatic blood supply has strongly influenced our views in the following decades. However, it is noteworthy that a number of controversies remain largely unresolved: the endocrine and exocrine pancreas being segregated or integrated; the presence of dedicated input and output blood vessels with a glomerulus-like capillary network in the islet; the general direction and speed of the blood flow in the pancreas; and local hormonal signaling.

Pancreatic islets have traditionally been viewed as distinct micro-organs, separate from the exocrine pancreas, with their own dedicated arteriole for blood supply and a venule for drainage (1,12–16). This implies one-way traffic with no continuous capillary network between islets and surrounding acinar cells and no local hormonal dialogue (Fig. 1Aa). How has this idea emerged in the first place? We speculate that the capillary network of the islet in 2D representations has been recognized as resembling that of the kidney glomerulus, but without considering anatomical, not to mention functional, implications of such organization. A specialized bundle of capillaries of the glomerulus are uniquely situated between two resistance arterioles that require a special support structure to maintain a very high blood flow in these essential capillary units and vast kidney filtration volume (17). Indeed, the glomerular capillaries are the only capillary beds in the body that are not surrounded by interstitial tissue. In fact, this observation dates back to at least the 1870s, when the terms “islets” and “glomeruli” were used interchangeably (1,12), and the islet has still been referred to as “glomerulus-like” in recent years (15).

Based on the resulting concept that the islet is an enclosed structure, three models of islet microcirculation were proposed: (model 1) mantle-to-core, (model 2) core-to-mantle, and (model 3) polar, regarding hormonal regulation between β-, α-, and δ-cells (18). At the international symposium held in 1995, after all, three groups of investigators agreed to disagree. Over a decade later, in 2008, Nyman et al. (19) carried out in vivo imaging of islet blood flow “to examine the relationship between blood flow and islet cell type arrangement by real-time in vivo imaging of intra-islet blood flow in mice.” The authors reported the observation of all three types of blood flow. Then, another decade later, in 2020, we revisited the issue using a similar experimental setting, but using DiI-labeled red blood cells (RBCs) with a hypothesis that if the islet would be an enclosed structure, we should observe a closed circulation of RBCs within an islet, rather than expecting to categorize the islet blood flow (e.g., three flow patterns) (20). The RBC flow was recorded by intravital imaging in the intact islets with surrounding exocrine tissues of MIP-GFP mice (21). Using image analysis software Imaris, individual RBCs were digitally captured, and their movements were tracked. Accumulated RBC flow traces provided islet capillary networks with the information on their speed and direction. We quantified the direction of blood flow at the interface as outflows and inflows. The ratio of the two directions was approximately 1:1. This suggests that the directionality is not a functional factor. Since the RBC movement is random, it implies that there would be no preferred paracrine effect within an islet.

At that point, we were still deeply influenced by the prevailing model of islet blood supply and modeled an alternative microcirculation with a feeding arteriole as shown in Fig. 1Ab, although the integration of pancreatic microvasculature was found in its entirety (Fig. 1B). We did capture such a perfect islet with a single arteriole running straight to the center of the islet, and thus we even made 3D prints (20). However, retrospectively, this islet was rather exceptional, and we were not able to find another one. Overall, it was perceivable that islets are not enclosed. The next question was whether unenclosed islets would require a dedicated arteriole.

In the diagram of Wharton’s model of islet blood supply, every islet in one lobule receives a dedicated arteriole blood supply resembling the lily of the valley (Fig. 2) (1). The schematic drawing was explained as “The intralobular arteries arise from the interlobular arterial plexus, and usually a single artery enters each lobule, where it divides to end in glomerular-like structures among the cells of the islands of Langerhans; thus, all of the blood in these larger lobules passes first to the islands of Langerhans.” This has been deduced based on large-scale image capture using the advanced experimental tools such as cardiac perfusion and even tissue clearing by the Spalteholz method, which are basically still powerful techniques used today.

The network structures, such as vasculature or neurons, are best studied in 3D with an ability to adjust the perspective angle. In a simple 2D section, sliced tubular structures appear as random dots and short lines. Fortunately, in recent years, confocal microscopy has become widely accessible, and a stack of high-resolution optical panels of the z axis can be readily obtained. Many islet studies use it; however, they typically present only a maximum projection rendering of such stacked images or a top-down view video, and it is not common to apply further options of 3D reconstruction of optical panels, surface rendering, and volume measurement. Using these tools, we set forth to see whether every islet would have a dedicated blood supply, specifically by examining the spatial relationship between arterioles and islets (22). The arteriole was identified by the existence of smooth muscle and its diameter, according to a well-referred textbook by Silverthorn (23). The most commonly used marker for smooth muscle is α-smooth muscle actin (α-SMA). Out of these three types of blood vessels that express α-SMA, the arteriole is much smaller than the artery and vein (the mean diameter: artery 4 mm, vein 5 mm, and arteriole 30 μm). The maximum value of the arteriole is known to be 100 μm (24). The expression pattern of α-SMA in the pancreas is shown in Fig. 3Aa (magenta) (22). This original fluorescence image is 3D surface rendered in Fig. 3Ab. To examine the arteriole contact of islets, the plain 3D reconstruction of the x-y axis is not enough. An example is shown in Fig. 3Ac: an arteriole (yellow) appears to feed an islet (cyan). However, from a different perspective angle, these two structures have no physical contact; the presence of a feeding arteriole has been an illusion (Fig. 3Ad). We reported this regrettable mistake (22) and would like to stress here again that, as we originally counted “such feeding arterioles from one angle by eye” in as much as 85% of islets (25), the same illusion may have deceived Wharton and a number of observers before and after him as well. In our recent article, we measured the shortest distance between arterioles and islets in 3D using Imaris software (22). Namely, if there is a contact, the distance is zero; if not, it assumes positive values. We carried out the quantification in six mammalian species: human, monkey, pig, rabbit, ferret, and mouse. While some arterioles passed by or traveled through islets, the majority of the islets (e.g., 79% for human islets) were not in contact with arterioles throughout the species examined (Table 1) (22). Overall, the arterioles emerge to feed the exocrine pancreas regionally within each lobe (Fig. 3B). Vascularizing the pancreas in this way may allow an entire downstream region of islets and acinar cells to be simultaneously exposed to changes in the blood levels of glucose, hormones, and other circulating factors.

The most unexpected finding of these recent analyses was that capillaries directly branched out from the trunk of arterioles. It is generally thought that a capillary network branches out at a terminal point of an arteriole (23,26). As we observed, when an islet is touching the arteriole shown in Fig. 4A, it can see changes of blood content through multiple capillaries at first hand (Fig. 4Ba). It may have an implication for the previous models that these “capillaries” branching out from an arteriole were considered as “small arterioles” (Fig. 4Bb) (12). Does such rare branching occur in nature? Yes, it does. Photographs in Fig. 4C were taken at the campus of the University of Chicago. The redbud tree is seemingly the only one that exhibits cauliflory in North America.

The important implication of our finding here is that while, in the prevailing models, individual islet response is more or less uniform within the lily-of-the-valley type of blood supply to islets in a lobule, our new model implies that there would be no stereotypical islet response as it has been taken for granted (Fig. 5). In fact, “stochasticity” and “heterogeneity” are the keywords for diabetes in many levels, from populations (race, sex and age), to individuals (pancreas regional difference in β-cell/islet mass) (27), to islets (islet cell composition) (28), to β-cells (expression of differential markers and function) (29–32). It may not be surprising that functional properties of individual islets would differ according to their microenvironments and stimulation contexts (33).

In conclusion, our findings suggest a new model for pancreatic microcirculation that allows a much more intimate integrated communication in the blood flow between the exocrine and endocrine pancreas than currently recognized.

One of the central questions posed in the past has been “why are the islets of Langerhans?” (34). Why do endocrine hormone–releasing islet cells exist as tiny cell collectives within the bulk of exocrine tissue and not as a separate homogenous organ or part of an endocrine organ with hundreds of thousands or millions of cells? Other endocrine systems have demonstrated the advantages of such organization and importance of portal systems. According to Henderson, the evolutionary pressure would be to have many small cell collectives (islets smaller than 100 μm) to increase the contact interface area between exocrine and endocrine cells within the local environment of the lobuli. An opposing argument could be the size distribution of the islets, with 10% of the largest islets accounting for 50% of the total islet volume (35). The question is whether these large islets account for 50% of the function. Experimental (36) as well as theoretical (37) work would rather suggest that exactly the opposite is the case, with smaller islets being more active. In such an organization, the autocrine and paracrine effects due to local diffusion or bulk flow would dominate systemic endocrine effects. High local insulin and glucagon concentration could be required to metabolically facilitate the activity of exocrine cells, and acinar and ductal cells could help stabilize local pH at high metabolic challenges (38). In fact, the absence of insulin has been reported to lead to pancreatic atrophy (39). In addition, it has been observed early on that both exocrine and endocrine tissue in the pancreas can be controlled by a number of common neural and hormonal factors, such as acetylcholine (40), secretin, gastrin, and pancreozymin (41,42). An integrated blood flow within a pancreatic lobe does provide an adequate scaffold for large-scale interactions between the tissues.

The recent advances in tissue processing, as well as image acquisition and processing, allowed a wider-angled view to answer a series of open questions regarding the local structure and density of pancreas vascularization, assessment of the blood flow dynamics, distribution of flow speeds within the islet, and blood flow directionality. These recent results show that a majority of pancreatic islets share a common capillary system with the surrounding exocrine tissue (20,22,25). This could also imply that a regional, lobular organization and interplay of both exocrine and endocrine cells, after being exposed to a specific blood supply, produce a pattern of endocrine hormone and digestive enzyme response that is specific for that exact topological region.

Changes in the quality and density of vascularization can have an effect on pancreas function (43), and it has been discussed how islet/pancreas blood flow plays a role in initiation and shaping of insulin secretion (44) and, furthermore, how such intensive vascularization may explain efficient delivery of insulin to all cells in the body (45).

Bonner-Weir and Orci (12) and Konstantinova and Lammert (46) claim the vascular density in islet cells to be 5- to 10-fold in comparison with surrounding exocrine tissue. This seems to be quite different from what we see, where, in the tissue surrounding the islets, the differences are twofold to threefold (47). Nevertheless, the previously observed 5- to 10-fold differences in the perfusion of the two tissue could be entirely due to the differences in capillary space filling of the tissue with larger-diameter acinar cells in comparison with the tissue with smaller-diameter islet cells (48). According to the Hagen-Poiseuille law, this increased capillary length adds to the resistance to blood flow, potentially resulting in a 5- to 10-times reduction in flow within the same tissue volume.

Final questions address how an integral view of the pancreas advances our understanding of pancreas function. What would be the end objective of the regulation of the pancreatic blood flow? Does blood flow need to be coupled and adjusted to nutritional flow to support the process of insulin exocytosis or postexocytotic distribution of insulin? Alternatively, is there blood delivery beyond the nutritional, delivering plasma nutrient signature for the islet and islet network sensory function, securing an adequate metabolic response with the hormone release pattern? The current views would support mostly nutritional and distributional function.

What about the role for a perfusion system to support the sensory role, or, in other words, is there a nonnutritional role for the pancreas perfusion? From a physiological point of view, the vascularization of pancreas tissue is vast. Arterioles, rather than preferentially perfusing islets in a glomerulus-like feeding style, would feed the larger tissue volumes of pancreatic lobes.

Additionally, in our view, it is not a particular advantage to focus on the direct blood flow from one cell type to another within the islets. As mentioned above, the majority of islets do not exceed 200 μm (28) and contain capillaries with different velocities of blood flow (20), which imposes constraints on the directed communication between the cells in a spatially limited volume, particularly under stimulated conditions (49). These velocities would allow intralobular distribution ranges, with significant endocrine physiological, as well as pathophysiological, effects in the exocrine tissue (50). With nonbranched capillaries with high blood flow velocities, the paracrine effects would possibly propagate several hundred micrometers away from the release site in the first second after the release, which would typically be way outside of an islet. On the other hand, as the speed of perfusion in branched capillaries is comparable to that of diffusion, this removes the issue of directionality.

Another important question is why are there pancreatic lobes and are they, in effect, basic units that warrant the regional heterogeneity regarding the quality and timing of the sensory input? Our experimental evidence supports the idea that the parameters observed in the pancreas optimize the sensory activity of the endocrine pancreas. Some of the more obvious parameters worth mentioning are 1) the abovementioned limited and evolutionary well-conserved islet size, 2) cell-cell interactions, 3) versatility of capillary networks (fast nonbranched, slow-branched, trunk branching), and 4) islet-islet interactions (innervation) that have, to date, not been studied. Within a relatively small islet, β-cells are exposed to a concentration gradient of nutrients, and, with their preferentially weak interactions, they function as a sparse cell collective, significantly improving the cell collective and the overall sensory function (37,51). Similarly, to individual β-cells in an islet, islets within a network are not cross-correlated, which can further improve the sensory experience and optimize insulin release on an even larger scale (52). The importance of the nutrient gradients for the islet function has been acknowledged previously and has been experimentally addressed (53), although not under in vivo conditions. The exact nature and the topology of the metabolic code oscillations in the pancreas have also not been studied yet. We suggest that the oscillations can be associated with the sparse nature of the islet network to produce an alternating pattern of insulin release to maintain optimal sensitivity of target cells to metabolic hormones (54). The overall pancreatic arterial input is in series rather than parallel, introducing delays in blood delivery to individual islets and creating conditions where different sensory modalities are presented to a network of islets.

In a very timely fashion, an interesting debate over the opposite aspects of intraislet blood flow directionality was published in the same issue of Diabetes: 1) Weir and Bonner-Weir (55) claimed it is strictly from β-cells to α-cells and 2) Caicedo et al. (56) validated cross talk among islet cells. Here, we point out that both groups only referred to our study published in 2020 (20) but did not yet update the model with our findings published in 2023 (22), where we demonstrated the existence of islets without a so-called “feeding arteriole” throughout the six species examined. This excludes preferential directional β-cell–to–α-cell blood flow. Furthermore, an intriguing clinical study from Mayo Clinic has demonstrated the α-cell–to–β-cell cross talk by blocking glucagon-like peptide 1 receptor with infusion of exendin(9-39) (57).

In summary, the recently described organization of vascularization with mostly islets having no preferential perfusion in comparison with the neighboring tissue (22) lays a valuable foundation to improve our understanding of the additional layers of pancreas function, the vivid dialogue between different cell types involved, and common perils in a disease context.

Funding. The study is supported by National Institutes of Health DK117192, DK127786, and DK020595 to the University of Chicago Diabetes Research and Training Center (Physiology Core) and a gift from the Kovler Family Foundation to M.H., the Austrian Science Fund/Fonds zur Förderung der Wissenschaftlichen Forschung (bilateral grants I3562-B27 and I4319-B30), and National Institutes of Health DK127236 to M.S.R.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.H. is responsible for the work as a whole, including the study design, access to data, and the decision to submit and publish the manuscript. M.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. M.S.R. co-wrote the manuscript and contributed to the submission and revision process.