And Beta-Cells in Human Islets of Langerhans

Diabetes Publish Ahead of Print, published online February 25, 2010

cell arrangement in human islets of Langerhans

Unique arrangement of alpha- and beta-cells in human islets of Langerhans

Short running title: cell arrangement in human islets of Langerhans

Domenico Bosco, Mathieu Armanet, Philippe Morel, Nadja Niclauss, Antonino Sgroi, Yannick D. Muller, Laurianne Giovannoni, Géraldine Parnaud and Thierry Berney

Cell Isolation and Transplantation Center, Department of Surgery, Geneva University Hospitals and University of Geneva Rue Michel-Servet 1, 1211 Geneva 4, Switzerland

Corresponding author: Domenico Bosco E-mail: [email protected]

Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org

Submitted 7 August 2009 and accepted 6 February 2010.

This is an uncopyedited electronic version of an article accepted for publication in Diabetes. The American Diabetes Association, publisher of Diabetes, is not responsible for any errors or omissions in this version of the manuscript or any version derived from it by third parties. The definitive publisher-authenticated version will be available in a future issue of Diabetes in print and online at http://diabetes.diabetesjournals.org.

Obective: It is generally admitted that the endocrine cell organisation in human islets is different to that of rodent islets. However, a clear description of human islet architecture has not yet been reported. The aim of this work is to describe our observations on the arrangement of human islet cells.

Research Design and Methods: Human pancreas specimens and isolated islets were processed for histology. Sections were analysed by fluorescence microscopy after immunostaining for islet hormones and endothelial cells.

Results: In small human islets (40-60 μm-diameter), beta-cells had a core position, alpha-cells a mantle position and vessels laid at their periphery. In bigger islets, alpha-cells had a similar mantle position but were found also along vessels that penetrate and branch inside the islets. As a consequence of this organisation, the ratio of beta-cells to alpha-cells was constantly higher in the core than in the mantle part of the islets, and decreased with increasing islet diameter. This core-mantle segregation of islet cells was also observed in type 2 diabetic donors but not in cultured isolated islets. Three-dimensional analysis revealed that islet cells were in fact organized into trilaminar epithelial plates, folded with different degrees of complexity and bordered by vessels on both sides. In epithelial plates, most beta-cells were located in a central position but frequently showed cytoplasmic extensions between outlying non-beta-cells.

Conclusions: Human islets have a unique architecture allowing all endocrine cells to be adjacent to blood vessels and favouring heterologous contacts between beta- and alpha-cells, while permitting homologous contacts between beta-cells.

2 cell arrangement in human islets of Langerhans

slets of Langerhans are micro-organs sometimes still presented with a simple core- located in the pancreas and composed by mantle architecture similar to that of rodent I at least 4 types of endocrine cells. The islets, many reports described decades ago alpha- and beta-cells are the most abundant human islets with a different cell organisation and also the most important in that they (1; 9-11). Pioneer works from Orci and Unger secrete hormones (glucagon and insulin, (1) depicted human islets with alpha- and respectively) crucial for glucose homeostasis. delta-cells located in the mantle and grouped The prevailing description of islet cell against capillary walls within the core of beta- composition and structure comes from studies cells. It has been also proposed that human performed in rats and mice. It is generally islets were subdivided into lobules or subunits accepted that endocrine cells are not comprising clusters of beta-cells surrounded randomly distributed into islets. In most by alpha-cells (9; 11) and that these lobules or rodents, beta-cells compose the core of the subunits were separated by vascularised islets and the non-beta-cells, including alpha-, connective tissue and non-beta-cells (9). delta- and PP-cells, form the mantle region. Grube et al. (10) proposed a different This unique architecture seems to have some organisation where endocrine cells were functional implications (1). For instance, in organised in a ribbon like manner rather than several murine models in which insulin in separated subunits. In their model beta- secretion is decreased, normal organisation of cells are located in the islet core and alpha- islet cells was found to be perturbed (2) so cells are arranged at the periphery and along that beta-cells were intermingled with non- intraislet capillaries. These views were beta-cells. In addition, in vitro experiments challenged by more recent publications showed that homologous contacts between rat claiming that endocrine cell types were beta-cells improved their function, as dispersed throughout the human islets (8; 12). heterologous contacts between beta- and non- A summary of this controversy has been beta-cells had not effect (3). This observation skilfully reviewed by Bonner-Weir and suggests that a core-mantle segregation of O’Brien, who furthermore described human islet cells is useful in favouring homologous islets as complex cell arrangements with contacts between beta-cells which in turn different profiles including cloverleaf pattern improves insulin secretion. The characteristic (6). islet architecture may also serve to facilitate The purpose of this work was to bring some interactions between the different islet clarifications to this controversy by describing hormones via interstitial or vascular routes (4; our observations on the distribution of alpha- 5). and beta-cells in human islets. The sparse works on the structure of human islets do not provide a clear description of RESEARCH DESIGN AND METHODS their cellular organisation. There is a Pancreas procurement. Human pancreata consensus on the different endocrine cell were harvested from adult heart-beating, types, which do not differ significantly brain-dead donors and designed to be between rodent and human islets, and on the processed for islet isolation and proportion of islet beta-cells that is lower in transplantation. For different motives humans compared to rodents (6-8). (prolonged ischemia time, suspicion of Controversies persist about the topographical tumours, etc) some pancreata were not arrangement of endocrine cells within human processed for islet isolation and small islets. Even though human islets are specimens were taken for histology.

3 cell arrangement in human islets of Langerhans

Specimens from 21 donors were used for glutamine, HEPES and 10 % fetal calf serum these analyses. Sixteen specimens were from (hereafter referred to as complete CMRL). non diabetic donors with a mean age of 40.1 ± Islets used for morphological analysis were 16.2 years (range: 15-69 years) and a mean cultured for a total of 36-48 h. For islet cell body mass index (BMI) of 25.2 ±2.9 (range: dispersion, cultured islets were rinsed in PBS 20-29). Five specimens were from 5 type 2 and incubated in Accutase (Sigma, St Louis, diabetic donors with a mean age of 57.8 ± 9.1 MO, USA) for about 9 min with occasional years (range: 42-65 years) and a BMI of 32.3 pipetting. Single cell suspensions (~90 % ± 10.0 (range: 26-50). Glycemia at the time of single cells) were rinsed with complete hospital admission was 8.9 ± 3.8 mmol/l CMRL and aliquots of 105 cells were (mean ± SD, n=12) for non diabetic donors incubated for 24 h at 37°C in non adherent and 13.9 ± 5.3 mmol/l (mean ± SD, n=5) for 60- mm-diameter Petri dishes containing 6 ml type 2 diabetic donors. Specimens from non complete CMRL. Cells were harvested and diabetic donors were obtained from the head attached within Cunningham chambers (15) (n=4), the body (n=2), the tail (n=5) and for immunofluorescence assessment. unspecified regions (9) of the pancreas (for 4 Histological sections of pancreata and pancreata, specimens from both the head and isolated islets. Pancreas samples were the tail were available). Specimens from type incubated for 24 h in (methanol free) 10 % 2 diabetic donors were obtained from the head formalin, dehydrated and embedded in (1), the body (2) and the tail (2) of the paraffin. Isolated islets were incubated 24 h in pancreas. Histopathological analysis of the methanol free 10 % formalin, then deposited diabetic specimens revealed no apparent at the bottom of flat-bottomed tubes, alteration of the islets. However, by double embedded in agar to immobilize them, immunofluorescence for insulin and amylin dehydrated and finally embedded in paraffin. (using a rabbit anti-amylin antibody All pancreata and islet samples were purchased from Progen Biotechnik GmbH, sectioned at 5 μm. For one pancreas sample, Heidelberg, Germany), several beta-cells 80 consecutive sections were performed. were found to be positive for amylin, in all 5 Attachment of islet cells within Cunningham specimens. In one specimen, extracellular chambers. Islet cells cultured 24 h in staining for amlylin was also observed. complete CMRL were harvested and injected Mouse pancreata were harvested from into poly-L-lysine-coated Cunningham C57/BL6 mice (Janvier Laboratories, Le chambers (15). After 1 h at 37°C, chambers Genest-Saint-Isle, France). To this end, mice were rinsed to remove unattached cells and were first anesthetised and then exsanguinated filled with methanol free 10 % formalin. After by aortic sectioning. Pancreata were rinsed in 20 min incubation at room temperature, they PBS before processing for histology. were rinsed with PBS and stored at 4°C Islet and islet cell isolations. Human islet before being used for immunofluorescence. isolations were performed as previously Immunofluorescence. Before described according to the Ricordi method immunofluorescence, sections were with local adaptations (13; 14). The use of deparaffinized and rehydrated with a series of human islets for research was approved by alcohol solutions of decreasing concentration. our local institutional ethical committee. After Sections and cells attached to Cunningham purification, islets were cultured overnight at chambers were then washed with phosphate 37°C, and at 25°C thereafter, in CMRL buffer saline (PBS). Antibody dilutions and medium containing 5.6 mmol/l glucose and rinsing steps were performed in PBS. supplemented with penicillin, streptomycin,

4 cell arrangement in human islets of Langerhans

Incubations were done at room temperature, were also analysed by confocal microscopy except as indicated below. (Zeiss LSM510 Meta) configured for For double staining (insulin +glucagon, simultaneous analysis of red and green insulin + somatostatin or insulin + PP), slides fluorescences. Quantifications on isolated were treated 20 min with 0.5% Triton X100 cells were performed by microscope and 20 min with 0.1% BSA, at room observation. Quantifications on islet and temperature. Slides were then incubated 1 h pancreas sections were performed on digital with a mouse anti-glucagon antibody (1:3000; images. Analysis of red and green Sigma), a rabbit anti-somatostatin antibody fluorescence on digital images was performed (1:300; Dako, Carpinteria, CA, USA) or a using the offline MetaMorph imaging rabbit anti-PP antibody (1:300; Dako). After software for microscopy (Universal Imaging rinsing, slides were incubated 1 h with either Corporation, West Chester, PA). This an Alexa 488-conjugated anti-mouse antibody software was programmed to automatically (1:500; Invitrogen, Basel, Switzerland) or a quantify red and green stained areas within FITC-conjugated anti-rabbit antibody (1:400; defined regions of interest. Data were Jackson ImmunoResearch Laboratories, West expressed as mean ± SEM of n different Grove, PA, USA). Then, slides were rinsed experiments or islets. Differences between and incubated 1 h with a guinea-pig anti means were assessed either by the Student’s t- insulin antibody (1:1000; Dako) and test and when required by one-way ANOVA. successively 1 h with a Rhodamine- When ANOVA was applied, Scheffé’s least- conjugated anti-guinea-pig antibody (1:100, significant difference post-hoc analysis was Jackson ImmunoResearch Laboratories). For used to identify significant differences. triple staining (insulin, glucagon and CD34), deparaffinized sections were treated 10 min at RESULTS 37°C with a 0.1% trypsin solution, 20 min When pancreas sections were double stained with 0.5% Triton X100 and 20 min with 0.1% for insulin and glucagon, apparently small BSA, at room temperature, and incubated islets (40-60 μm-diameter) showed a successively 1 h with a mouse anti-CD34 segregated cell type distribution with alpha- antibody (1:50; Serotec, Oxford, UK), 1 h cells localized in the mantle and beta-cells in with a 488-Alexa conjugated anti mouse the core of the islets (Fig. 1A and G). In antibody (1:500; Invitrogen), 1 h with rabbit apparently bigger islets (Fig. 1A-C), alpha- anti-glucagon (1:50; Dako) and guinea-pig cells were found in the mantle as in 40-60 anti-insulin (1:1000; Dako) antibodies and 1 h m-diameter islets, and also along simple or with a Rhodamine-conjugated (1:750) anti- ramified “empty” areas (Fig. 1A-C) present rabbit and AMCA-conjugated anti-guinea-pig inside these islets. Islets with apparent antibodies (1:150; Jackson ImmunoResearch diameters of 50 to 100 μm generally Laboratories). displayed one simple empty area (Fig. 1A), Morphological analysis and quantifications. whereas islets with apparent diameters over Microscopic sections and islet cells were 100 μm displayed multiple and/or ramified analysed with an Axioskop microscope empty areas (Fig 1A-C). Labelling for CD34 (Zeiss, Feldbach, Germany) equipped with revealed that vessels surrounded the periphery UV illumination and filters for blue, red and of islets and localized inside these empty green fluorescences. Images were captured areas, hereafter referred to vascular channels. with an Axiocam color CCD camera (Zeiss) (Fig. 1H and I). Careful examination showed and recorded on computer through the that alpha-cells were always lining the AxioVision™ software (Zeiss). Isolated cells vascular channels and the mantle region of the

5 cell arrangement in human islets of Langerhans

islets, and juxtaposed to endothelial cells ratios between subregions were amplified (not (Fig.1G-I). Interestingly, no CD34 staining shown), correlating with the observation that was observed inside the beta-cell bulk, the alpha-cells were lining the islet mantle suggesting that vessels did not penetrate the and vessels, as a single layer (~10 μm). The beta-cell bulk. Similar observation was made mean apparent diameter of all islets analysed for 40-60 μm-diameter islets. Since in this here was 154 ± 4 μm (mean ± SEM of 255 analysis apparently small islets could be islets from 21 different pancreata). tangential sections of bigger islets, we also When pancreatic islets from type 2 diabetic examined serial sections permitting to follow donors were similarly analysed, an equivalent several islets (n=8) through their whole pattern of islet cell distribution was observed thickness. Thus, after a triple (Fig. 2C). However, the range of values was immunofluorescence for insulin, glucagon lower in diabetic compared to control islets. and CD34, we confirmed that no vascular This last observation is in agreement with channel penetrated the core of 40-60 μm- published works showing a decrease of beta- diameter islets or the clusters of beta-cells in cell mass in type 2 diabetes (16-19). bigger islets. We then studied the profile of b/(a+b) values When double staining was performed for in cultured isolated islets. After enzymatic insulin and somatostatin or insulin and PP, we digestion, the islet vasculature is damaged and observed that delta- and PP-cells had roughly cell arrangement is expected to be perturbed the same topographical position as alpha-cells after culture. Section analysis of these islets (not shown). revealed that few structures resembling To confirm segregation of alpha- and beta- vascular channels were remaining. Total cells, areas labelled for glucagon (green=a) profile area of these structures was 8 times and insulin (red=b) were quantified in three lower in isolated islets compared to profile different subregions of each islet profile. area of vascular channels in islets in situ. These subregions were 1) the mantle defined Morphometric analysis indicated similar as a region of 20 μm deep that follows the b/(a+b) ratios between the core and mantle of external perimeter of islets, 2) the core cultured isolated islets (Fig. 2D). defined as the total islet area minus the 20 When analysis of control pancreatic islets was m-mantle area and 3) the vessel area performed according to their apparent defined as a region of 20 μm deep that diameter, b/(a+b) ratios in the mantle follows the contour of vessels (Fig. 2A). The remained constant (Fig. 3A), as ratios in the results expressed as b/(a+b) ratios were 0.71 core gradually decreased with the increasing for the core and 0.50 for the mantle and the islet apparent diameter (Fig. 3B). This vessel area of islets (Fig. 2B), showing that observation confirmed that apparently small beta-cells preferentially localized at the core islets (50-100 μm-diameter) had a core- compared to mantle and vessel area. mantle organisation close to that of rodent Considering that single cell section areas are islets (Fig. 2D), and that apparently bigger roughly similar between alpha- and beta-cells, islets had elevated numbers of non-beta-cells one can extrapolate that the core contains 3 in their core. times more beta-cells than alpha-cells, To gain an insight into the islet cell whereas the mantle and the region around organisation, we performed consecutive serial vascular channels contain roughly the same sections through pancreas specimens and a number of alpha- and beta-cells. When three-dimensional analysis of islets (Fig. 4). It subregions were selected with a 10 μm appeared evident that islet cells were not instead of a 20 μm rim, the differences in organised into discrete core-mantle subunits

6 cell arrangement in human islets of Langerhans

as previously described (9; 11). Rather, islet alpha-cells (Fig. 5G). When association cells were organised into a continuous between alpha- and beta-cells was assessed in structure, like a folded epithelial plate (board) cultured isolated islet cells, we observed that span in a 3D network throughout the many alpha-cells wrapped by beta-cells and whole islet. The epithelial plate, whose rarely the contrary (Fig. 6). Of all alpha-cells thickness did not vary much, had a triple- contacting a beta-cell, 38 ± 8 % (mean ± SEM laminated section with a central part of beta- of 3 experiments) were round and had a cells lined at both sides with alpha- and other perimeter almost completely wrapped by a islet cells. Moreover, vessels were juxtaposed beta-cell. This result revealed a unique at both sides of the epithelial plate structure, plasticity of beta-cells that were able to spread invaginating in and following its folds. In 54 around alpha-cells and suggested that this islets from 5 different pancreata, we measured characteristic was intrinsic to beta-cells and the distance between vascular channels, not dictated by some islet coercions, such as equivalent to the thickness of the trilaminar extracellular matrix or islet vasculature. plate. The value observed of 42 ±13 m The specimens used to analyze the pancreatic (mean ± SD) roughly fits the diameter of the islet architecture were from different regions smallest islets. of the pancreas, including the head, the body Alpha-cells, due to their position within the and the tail. We did not observed obvious epithelial plate, are exclusively aligned along variations in islet structures according to the vessels. Beta cells were also found aligned regional origin of the pancreas specimens. To directly along vessels in alternation with further investigate this point, we compared alpha cells. Where alpha cells were the most head and tail specimens from the same numerous, they formed like a fence between pancreas (n=4). Islets with the trilaminar plate endothelial cells and beta cells. In this case, structure were observed in both head and tail beta cells appeared to form a second layer of regions of all pancreata. cells over the first layer of alpha-cells. However a careful examination of these cells DISCUSSION revealed that they develop extensions that In this morphological analysis, we show that infiltrated between alpha-cells and progressed islet cells are organised into a trilaminar plate until the surface of endothelial cells (Fig 5A- comprising one layer of beta-cells sandwiched F). Confocal microscopic analysis showed between two alpha-cell-enriched layers. This that narrow strips of insulin staining between structure has a folded pattern and vessels alpha cells corresponded to true cytoplasmic circulate along both of its sides. These extensions and not to insulin released into observations are schematized in Figure 7. extracellular compartment. Indeed, confocal No evident vascular channel penetrates the analysis revealed that insulin staining between central layer of beta-cells inside the trilaminar alpha-cells had a granular pattern similar to plate, but many vascular channel profiles are that observed elsewhere inside beta-cells, and found adjacent to the alpha-cell-enriched colocalized with actin staining (see layers. Absence of CD34 staining within the Supplmental Figure in the online appendix beta-cell layer strongly suggest absence of which is available at endothelial cells. If some endothelial cells http://diabetes.diabetesjournals.org) As a unlabeled for CD34 were nevertheless consequence of this cell organisation, the present, they ought to form tiny and collapsed heterologous contacts between alpha- and vessels since no structures resembling typical beta-cells around vessels are most numerous vessels were observed. There is no doubt that than homologous contacts between beta- or most, if not all, alpha-cells are in direct

7 cell arrangement in human islets of Langerhans contact with vasculature. With regard to beta- glucagon positively affected insulin release cells, many are intercalated between alpha- (20; 21). In this respect, we recently cells and clearly in contact with blood vessels. demonstrated that insulin secretion from Other beta-cells set on a layer of alpha-cells individual beta-cells was increased when they and display cytoplasmic extensions that run were in contact with alpha-cells (22). These between alpha cells to reach the vessel surface results combined with those showing that as well. Other beta-cells are more distant from heterologous contacts predominate in human vessels and doubts persist as to their ability to islets suggest that glucagon is a more potent reach the vasculature. A more meticulous 3D regulator of insulin secretion in human than in analysis could be able to answer this question. rodent islets. No endothelial cells were observed in the core Heterologous interactions between beta- and of 40-60 μm-diameter islets presenting a core- alpha-cells in human islets are not only mantle organisation, suggesting that vessels frequent but also unusual, since beta-cells do not penetrate these islets. But here again, most often wrap the neighbouring alpha-cells. beta-cell extensions intercalated between This behaviour of beta-cells towards alpha- peripheral alpha-cells were observed and it is cells was also observed in vitro in cultured possible that most beta-cells reach the isolated islet cells, demonstrating that it is surrounding vasculature by this way. innate to cells and not commanded by some Our description of human islet cell islet constraints. The molecular mechanisms organisation is innovative but does not that account for this phenomenon have not yet contradict the different views reported so far been studied. Most likely, cytoskeleton and about the human islet architecture (6; 8; 9; 11; adhesion molecules expressed by islet cells 12), including the diametrically opposed may play a role. A physiological relevance of opinions describing islets with either a this unique plasticity of beta-cells is possible, segregated or intermingled cell types. Indeed, especially considering that glucose-induced a clear segregation between alpha- and beta- insulin secretion in vitro is improved in cells was observed and this was particularly spreading beta-cells compared to round- true in small human islets where alpha-cells shaped beta-cells (23) and that cytoskeleton surrounded a core of beta-cells. In larger plays a role in insulin secretion (24). human islets, a kind of cell segregation was Cytoskeleton organisation could be modified also observed since alpha-cells were mostly in beta-cells wrapping alpha-cells and confined at the periphery of the trilaminar spreading beta-cells. It remains to be plate. From another point of view, human islet demonstrated whether this modification has a cells can be considered as intermingled, since direct impact on insulin secretion. we clearly showed that intercellular contacts Islets within pancreata of type 2 diabetic between alpha- and beta-cells predominate. donors have roughly the same structure as This is certainly a consequence of the islets from non diabetic donors. Indeed, we trilaminar organisation and cytoplasmic have found that the associations between extensions of beta-cells that intercalate beta- and alpha-cells were roughly similar in between alpha-cells. This is not the case in type 2 diabetic and non diabetic islets, rodent islets where the higher degree of cell suggesting that the beta-cell dysfunction segregation clearly favours beta- and alpha- observed in type 2 diabetes is not related to a cell homologous contacts. clear abnormality of islet cell organisation. The potential physiological significance of Certainly, more studies are required to heterologous contacts between alpha- and understand if a more subtle defect in cell beta-cells is suggested by works showing that organisation is involved. Our quantitative

8 cell arrangement in human islets of Langerhans

analysis demonstrated that beta-cell mass integrins, in islet cells (23; 25). In rodents, relative to alpha-cell mass was substantially dissociated islet cells readily reaggregate in decreased in type 2 diabetic islets. These culture and after few days are able to form results are in agreement with some previous pseudoislets with a core-mantle organisation reports showing a decrease of the absolute similar to that of native islets. This indicates beta-cell mass or of the volume of islets that in rodents, information that decides of the relative to the volume of pancreas in type 2 islet architecture is foremost provided by islet diabetic subjects (16-19). cells themselves (26; 27). Our observations We have shown that the unique organisation suggest that islet cells by themselves are not of islet cells observed within human pancreata responsible for the unique cellular is not maintained in cultured isolated islets. organisation of human islets. In humans, the This is not surprising considering the insults more complex organisation of islets could sustained by islets during the isolation involve more complex mechanisms. The role procedure, including disruption of of vessels and extracellular matrix vascularisation, enzymatic digestion of components in the maintenance of human extracellular matrix and mechanical shaking islet architecture remains to be investigated. of the tissues. Islet morphology could be also Furthermore, it will be particularly interesting affected by culture. Indeed intercellular to determine whether islet morphology is contacts are under the control of molecular restored after transplantation and whether the events influenced by environmental factors. unique arrangement between alpha- and beta- Finally, the handling, fixation and paraffin cells is critical for the function of transplanted embedding of islets may also affect the islets. cellular structure of the islets. It would be interesting to investigate if islet architecture is ACKNOWLEDGEMENTS maintained in freshly isolated islets and what This work was supported by grants from the is the real effect of the culture on islet Swiss National Science Foundation (3200BO- morphology. Many components of the 120376). Human islets were obtained thanks medium could affect islet morphology in to a grant from the Juvenile Diabetes vitro. Among these components, there are the Research Foundation (31-2008-416). We insulin secretagogues, such as glucose, that thank Corinne Sinigaglia and David Matthey- have been shown to affect expression of Doret for their excellent technical assistance. adhesion molecules, including E-cadherin and

9 cell arrangement in human islets of Langerhans

REFERENCES 1. Orci L, Unger RH: Functional subdivision of islets of Langerhans and possible role of D cells. Lancet 2:1243-1244, 1975 2. Gannon M, Ray MK, Van Zee K, Rausa F, Costa RH, Wright CV: Persistent expression of HNF6 in islet endocrine cells causes disrupted islet architecture and loss of beta cell function. Development 127:2883-2895, 2000 3. Bosco D, Orci L, Meda P: Homologous but not heterologous contact increases the insulin secretion of individual pancreatic B-cells. Exp Cell Res 184:72-80, 1989 4. Samols E, Stagner JI: Intra-islet regulation. Am J Med 85:31-35, 1988 5. Samols E, Stagner JI: Islet somatostatin--microvascular, paracrine, and pulsatile regulation. Metabolism 39:55-60, 1990 6. Bonner-Weir S, O'Brien TD: Islets in type 2 diabetes: in honor of Dr. Robert C. Turner. Diabetes 57:2899-2904, 2008 7. Bosco D, Meda P, Morel P, Matthey-Doret D, Caille D, Toso C, Buhler LH, Berney T: Expression and secretion of alpha1-proteinase inhibitor are regulated by proinflammatory cytokines in human pancreatic islet cells. Diabetologia 48:1523-1533, 2005 8. Cabrera O, Berman DM, Kenyon NS, Ricordi C, Berggren PO, Caicedo A: The unique cytoarchitecture of human pancreatic islets has implications for islet cell function. Proc Natl Acad Sci U S A 103:2334-2339, 2006 9. Erlandsen SL, Hegre OD, Parsons JA, McEvoy RC, Elde RP: Pancreatic islet cell hormones distribution of cell types in the islet and evidence for the presence of somatostatin and gastrin within the D cell. J Histochem Cytochem 24:883-897, 1976 10. Grube D, Deckert I, Speck PT, Wagner HJ: Immunohistochemistry and microanatomy of the islets of Langrhans. Biomed Res 4 (Suppl.):25-36, 1983 11. Orci L: The microanatomy of the islets of Langerhans. Metabolism 25:1303-1313, 1976 12. Brissova M, Fowler MJ, Nicholson WE, Chu A, Hirshberg B, Harlan DM, Powers AC: Assessment of human pancreatic islet architecture and composition by laser scanning confocal microscopy. J Histochem Cytochem 53:1087-1097, 2005 13. Bucher P, Mathe Z, Morel P, Bosco D, Andres A, Kurfuest M, Friedrich O, Raemsch- Guenther N, Buhler LH, Berney T: Assessment of a novel two-component enzyme preparation for human islet isolation and transplantation. Transplantation 79:91-97, 2005 14. Ricordi C, Lacy PE, Finke EH, Olack BJ, Scharp DW: Automated method for isolation of human pancreatic islets. Diabetes 37:413-420, 1988 15. Cunningham AJ, Szenberg A: Further improvements in the plaque technique for detecting single antibody-forming cells. Immunology 14:599-600, 1968 16. Butler AE, Janson J, Bonner-Weir S, Ritzel R, Rizza RA, Butler PC: Beta-cell deficit and increased beta-cell apoptosis in humans with type 2 diabetes. Diabetes 52:102-110, 2003 17. Kloppel G, Lohr M, Habich K, Oberholzer M, Heitz PU: Islet pathology and the pathogenesis of type 1 and type 2 diabetes mellitus revisited. Surv Synth Pathol Res 4:110-125, 1985 18. Saito K, Yaginuma N, Takahashi T: Differential volumetry of A, B and D cells in the pancreatic islets of diabetic and nondiabetic subjects. Tohoku J Exp Med 129:273-283, 1979 19. Sakuraba H, Mizukami H, Yagihashi N, Wada R, Hanyu C, Yagihashi S: Reduced beta-cell mass and expression of oxidative stress-related DNA damage in the islet of Japanese Type II diabetic patients. Diabetologia 45:85-96, 2002

10 cell arrangement in human islets of Langerhans

20. Crockford PM, Porte D, Jr., Wood FC, Jr., Williams RH: Effect of glucagon on serum insulin, plasma glucose and free fatty acids in man. Metabolism 15:114-122, 1966 21. Huypens P, Ling Z, Pipeleers D, Schuit F: Glucagon receptors on human islet cells contribute to glucose competence of insulin release. Diabetologia 43:1012-1019, 2000 22. Wojtusciszyn A, Armanet M, Morel P, Berney T, Bosco D: Insulin secretion from human beta cells is heterogeneous and dependent on cell-to-cell contacts. Diabetologia 51:1843-1852, 2008 23. Bosco D, Meda P, Halban PA, Rouiller DG: Importance of cell-matrix interactions in rat islet beta-cell secretion in vitro: role of alpha6beta1 integrin. Diabetes 49:233-243, 2000 24. Tomas A, Yermen B, Min L, Pessin JE, Halban PA: Regulation of pancreatic beta-cell insulin secretion by actin cytoskeleton remodelling: role of gelsolin and cooperation with the MAPK signalling pathway. J Cell Sci 119:2156-2167, 2006 25. Bosco D, Rouiller DG, Halban PA: Differential expression of E-cadherin at the surface of rat beta-cells as a marker of functional heterogeneity. J Endocrinol 194:21-29, 2007 26. Cirulli V, Halban PA, Rouiller DG: Tumor necrosis factor-alpha modifies adhesion properties of rat islet B cells. J Clin Invest 91:1868-1876, 1993 27. Rouiller DG, Cirulli V, Halban PA: Uvomorulin mediates calcium-dependent aggregation of islet cells, whereas calcium-independent cell adhesion molecules distinguish between islet cell types. Dev Biol 148:233-242, 1991

Figure legends

Fig. 1. Organisation of alpha- and beta-cells in human pancreatic islets. Sections of human pancreata with islets of different sizes were either double-labelled for insulin (red) and glucagon (green) (A-C) or triple-labelled for insulin (blue), glucagon (red) and CD34 (green) (G-I). D-F and J-L: Outlines of endocrine tissue labelled for insulin and glucagon were drawn. Except for the 40-60 μm-diameter islets, all islets displayed one or several unstained empty areas (vascular channels) at their core. Most glucagon-expressing cells were located around vascular channels and at the mantle of islets, independently of their size. Insulin-expressing cells seemed clustered into discrete ovoid areas surrounded by alpha-cells. In triple labelled sections (G-I), vascular channels displayed staining for CD34 indicating that they contained vessels. Islets shown are representative of at least 200 islets observed on sections from 16 different pancreata. Scale bars, 50 μm. Fig. 2. Distribution of alpha- and beta-cells according to different defined islet subregions. A: Pancreatic islet labelled by immunofluorescence for insulin (red) and glucagon (green); rims (white lines) delimiting mantle and core subregions and one vascular channel area are shown; scale bar, 10 m. After immunofluorescence, areas labelled for insulin (b) and glucagon (a) were measured and results expressed as b/(a+b) ratio. This ratio was calculated for areas measured in the different subregions (mantle, core and vessels) and in the whole islets (whole). Analyses were performed on histological sections from non diabetic human (B), type 2 diabetic human (C) and control mouse (E) pancreata and cultured isolated human islets (D). Columns are means ± SEM. B: n=193 islets from 16 pancreata, C: n= 54 islets from 5 pancreata, D: n=20 islets from 2 pancreas, E=50 islets from 1 pancreas. ND = not determined. Fig. 3. Distribution of alpha- and beta-cells according to islet apparent size. Islets from human pancreas sections analysed in Fig. 2 are shown here according to their apparent diameter. After

11 cell arrangement in human islets of Langerhans

immunofluorescence, insulin and glucagon labelled areas were measured and results expressed as b/(a+b) ratio, where a and b were representing glucagon and insulin labelled areas, respectively. A: b/(a+b) ratio calculated for areas measured in the 20 μm mantle part of islets. B: b/(a+b) ratio calculated for areas measured in the core of the islets. Columns are means ± SEM. n=39 for islets with 50-100 μm diameter, 60 for islets with 100-150 μm diameter, 58 for islets with 150-200 μm diameter, 24 for islets with 200-250 μm diameter, 12 for islets with 250-300 m diameter. Fig. 4. Three-dimensional (3D) analysis of human pancreatic islets. Consecutive sections through an entire pancreatic islet labelled for insulin (red) and glucagon (green). Images show that insulin (red) and glucagon (green) stained cells are organised into continuous 3D networks that span throughout the entire islet. Sections at both ends show islet profiles with an apparent similar core-mantle structure to that of 40-60 m-diameter islets. These consecutive images also reveal that vascular channels were in fact continuous ramified structures that were connected in places with the surrounding islet tissue (arrowheads). Image series is representative of 15 different pancreatic islets analyzed from one pancreas. Scale bar, 50 μm. Fig. 5. Unique association between alpha- and beta-cells in pancreatic islets. Pancreatic islets labelled for glucagon (green) and insulin (red) are shown at low magnification (A and D). Boxed areas are shown at higher magnification in B and E. Alpha- and beta-cells lining vascular channels in B are depicted in C. Alpha- and beta-cells lining the islet mantle in E are depicted in F. Arrow heads point cytoplasmic extensions of beta-cells that span over alpha-cells. Scale bars, 10 m in A and D, and 5 m in B and E. Scale bars, 10 μm. Heterologous contacts between beta- and alpha-cells (beta-alpha) and homologous contacts between alpha-cells (alpha-alpha) and beta-cells (beta-beta) around empty areas were scored. Their relative frequencies are shown in G. Columns are means ± SEM, n = 52 islets from 10 pancreata. Fig. 6. Unique association between alpha- and beta-cells in cultured islet cells. Human islet cells were isolated and cultured for 24 h. After a double immunofluorescence for insulin (red) and glucagon (green), islet cells were analysed by confocal microscopy. A-C: Images showing a cell pair composed of one alpha-cell (A) surrounded by one beta-cell (B); the merged image is shown in C. The cell pair showed here is representative of cell pairs observed in different human cell preparations from at least 10 different pancreata. D: One series of consecutive merged images of a cell pair composed of one alpha-cell (green) surrounded by a beta-cell (red). Scale bars, 10 m. E: All heterologous contacts between alpha- and beta-cells were scored according to their type of association: a beta-cell wrapping an alpha-cell (beta wrapping alpha), neutral apposition between alpha- and beta-cells (alpha-beta) and an alpha-cell wrapping a beta-cell (alpha wrapping beta). Results are shown as relative frequencies and columns are means ± SEM of 5 islet cell preparations from 5 different pancreata. From all heterogeneous contacts between alpha- and beta-cells, the percentage of alpha cells whose profile was round and perimeter almost completely wrapped by a beta-cell as in D was 38 ± 8 (mean ± SEM of 3 experiments). Fig. 7. Model of endocrine cell and vessel organisation in human islets. A: Alpha-cells (green) and beta-cells (red) are organised into a thick folded plate lined at both sides with vessels (blue). Alpha-cells are mostly at the periphery of the plate and in close contact with vessels. Beta-cells occupy a more central part of the plate and most of them develop cytoplasmic extension that run between alpha-cells and reach the surface of vessels. B: The plate with adjacent vessels is folded so that it forms an islet.

12 cell arrangement in human islets of Langerhans

Figure 1

A B C

D E F

G H I

J K L

13 cell arrangement in human islets of Langerhans

Figure 2 A

BC 0.75 0.75

0.50 0.50 b/(a+b) ratio b/(a+b) b/(a+b) ratio b/(a+b)

0.25 0.25 whole mantle core vessels whole mantle core vessels

D E 0.75 1.00

0.50 0.75 b/(a+b) ratio b/(a+b) ratio b/(a+b) ND ND 0.25 0.50 whole mantle core vessels whole mantle core vessels

14 cell arrangement in human islets of Langerhans

Figure 3 A

0.8

0.7

0.6

0.5 b/(a+b) ratio

0.4 0.35 50-100 100-150 150-200 200-250 250-300 islet diameter (μm) B

0.8

0.7

0.6

0.5 b/(a+b) ratio

0.4 0.35 50-100 100-150 150-200 200-250 250-300

islet diameter (μm)

15 cell arrangement in human islets of Langerhans

Figure 4

16 cell arrangement in human islets of Langerhans

Figure 5 A D

B E

C F

G 80

Frequency (%) 20

0 beta-alpha alpha-alpha beta-beta

17 cell arrangement in human islets of Langerhans

Figure 6 AB C

D E 60

20 Frequency (%) 0 beta alpha-beta alpha wrapping wrapping alpha beta

18 cell arrangement in human islets of Langerhans

Figure 7