The Transplanted Fetal Endocrine Pancreas Undergoes an Inherent Sequential Differentiation Similar to That in the Native Pancreas: An Ultrastructural Study in the Pig-to-Mouse Model

Porcine pancreata were procured from Swedish land race animals purchased from a local breeder. A total of two adult animals, two newborn piglets (1 day old), and fetuses from pregnant sows (10–14 per litter) obtained at 70 and 90 days of gestation were included in the study. The animals were killed by means of a slaughtering mask and did not have any metabolic or infectious disease. Immediately after death of the animals, the pancreata were removed and fixed or placed on ice (fetal pancreas at 70 days of gestation) (see below). From fetal animals, the whole pancreata were used, whereas from the newborn pigs and the adult animals, only small pieces from randomly chosen parts of the gland were collected. All materials were kept chilled (4°C) during the 1-h transport to the laboratory. All tissue was collected in accordance with the Helsinki Declaration and the local ethics committee.

Fetal pigs served as donors for discordant xenogeneic implantation. Pregnant sows were killed at 70 days of gestation, and the fetuses were immediately collected from the uterus and placed on ice during transport to the laboratory. After aseptic removal, the pancreatic glands were minced into 1- to 2-mm³ fragments in cold Hanks’ solution and then treated with collagenase (∼10 mg/ml; Boehringer Mannheim, Mannheim, Germany) during vigorous shaking as previously described (25). The digested tissue was washed and explanted into culture dishes, allowing cellular attachment (Nunclon 90 mm; Nunc, Kamstrup, Denmark). The culture medium was RPMI-1640 (11.1 mmol/l glucose; Flow Laboratories, Irvine, UK) supplemented with 10% (vol/vol) human serum (The Blood Center, Huddinge Hospital, Huddinge, Sweden) and 10 mmol/l nicotinamide (Sigma Chemicals, St. Louis, MO). The culture dishes were kept at 37°C in 5% CO₂ in humidified air, and the culture medium was changed every second day. On day 4 of culture, most of the islet-like cell clusters (ICCs) were free-floating or could easily be detached by gently flushing the culture medium. All free-floating fragments (diameter <0.7 mm) were considered ICCs and were harvested without any further purification.

At the time of transplantation, 2 × 3 μl of fetal porcine ICCs (i.e., ∼300 ICCs) were implanted through an incision in the left renal capsule of avertin-anesthetized mice (26). All rodents had free access to tap water and pelleted food (Type R34; AB AnalyCen, Lidköping, Sweden) throughout the experimental period.

Animals were killed 16 or 41 days and 3.5 or 6 months after transplantation. The islet grafts, clearly visible as a whitish spot under the renal capsule, were excised with a margin of ∼3 mm. Half of the tissue was left for electron microscopical analysis (see below). The other half was stored in a histological transport medium (Histocon; Histolab, Betlehem Trading, Gothenburg, Sweden) at 4°C until snap-frozen in isopentane and subsequently stored at −70°C. Serial sections (6 μm thick) were cut in a cryostat (−20°C), air-dried, and then stored at −70°C. These tissue samples were routinely processed for light microscopical histology and stained with hematoxylin.

The native pancreatic tissues and the dissected grafts were cut into 1-mm³ cubes in a drop of the fixative; the isolated and cultured ICCs were mildly pelleted. Tissue samples for morphological analysis were routinely processed with fixation in 2% glutaraldehyde in 0.1 mol/l cacodylate buffer (pH 7.2), supplemented with 0.1 mol/l sucrose, postfixed in 1% OsO₄, dehydrated in ethanol, immersed in propylene oxide, and embedded in the epoxy resin Agar 100 (Agar Aids, Stansted, Essex, UK). Specimens for immunogold labeling were fixed at 4°C for 2 h in 4% paraformaldehyde/0.5% glutaraldehyde in cacodylate buffer. During subsequent rapid dehydration in 50–95% ethanol, the temperature was further lowered to −20°C. Infiltration, embedding, and photopolymerization of the samples in the hydrophilic low-temperature embedding medium Lowicryl K4M (Agar Aids) were also performed at −20°C. The procedure has been described in detail earlier (9,27,28 29). Semi-thin sections stained with toluidine blue were used to localize islets in the adult pancreata. Ultra-thin sections were cut with a diamond knife on a LKB Ultrotome no. V (LKB Prod. AB, Bromma, Sweden) and placed on polyvinyl formal–coated nickel grids.

The immunogold labeling method has been described in detail previously (9,30,31). By using primary antibodies raised in different species of animals (Table 1), it was possible to construct two parallel but not interacting antibody chains. The sections were blocked for any unspecific binding by applying nonimmune serum before incubation overnight with one of the primary antisera. The second primary antiserum was added the following day and incubated for 2 h, and then the sections were carefully rinsed. The antibodies constituting the second layer were designed to bind only one of the two primary antibodies used and consisted of an antibody conjugated to either 5 or 15 nm colloidal gold (Table 1) (Amersham International, Amersham, UK). Upon single labeling, the sections were incubated with the primary antibody, rinsed, and incubated with the proper secondary antibody; otherwise, the procedure was the same as for double labeling. Finally, the sections were rinsed and contrasted with 4% uranyl acetate and Reynolds lead citrate. Tris-buffered solution (TBS; 0.05 mol/l Tris-HCl buffer at pH 7.2) supplemented with 0.1% bovine serum albumin (BSA; Sigma, St Louis, MO) was used to dilute the primary antibodies. TBS with 0.2% BSA was used for rinsing between the applications of the primary and secondary antibodies, and TBS (pH 8.2) with 1% BSA was used to dilute the secondary antibodies. The optimal dilution of each antibody was determined by serial dilution tests (Table 1). Due to sterical hindrance, the secondary antibodies conjugated to 15-nm gold particles label fewer binding sites than the 5-nm colloidal gold IgG conjugates. Therefore, each labeling experiment was performed so that each hormone combination was visualized with each of the two different gold marker sizes. Granule diameter is given as mean ± variation width, calculated from the largest granule diameter found in a minimum of 100 cells. The intensity of the immunoreactivity in different granules and cells was estimated after the relative amounts of respective gold markers found in different cells from the same section. Only the results of three to four successful labeling experiments in each combination were considered. Each experiment included technical controls.

According to information provided by the manufacturers, all antibodies have excellent specificity and no cross-reactivity. Controls were performed, both on the light-microscopical (paraffin-embedded) and electron-microscopical levels, by including known positive and negative specimens (adult exocrine and endocrine pancreas) and by omitting the primary antibody or replacing it with diluted serum from nonimmunized animals before application of the IgG-gold solution. The immunolabeling specificity was controlled in adsorption tests by adding excess amounts of each antigen in separate experiments (10–100 μg peptide ml diluted antibody). Each peptide antibody was tested for cross-reactivity with any of the other peptides examined. No such cross-reaction was observed.

The ultrastructure of the tissue was well preserved with easily identifiable cell organelles (Fig. 1). The low temperature process applied in the present study left ∼5% hydrated water in the tissue sections, and the overall mild treatment gave a better preserved morphology when compared with the marked condensation of cellular material, including granule matrix, obtained when applying a classic morphological processing protocol. However, the contrast within low temperature–processed tissue sections was inferior to osmium-fixed tissue sections. The reason to use a low temperature embedding procedure was to better preserve the immunoreactivity of the tissue.

The fetal endocrine cells grew in small nests (Fig. 1a) or were diffusely scattered among the exocrine tissue. They could be separated into four different types, dependent on the morphology of their secretory granules. 1) Cells with electron-dense spherical granules with a diameter of 270 ± 80 nm (Figs. 1a and2a and b). 2) Cells with translucent heteromorphous granules with a largest diameter of 340 ± 50 nm, referred to as “translucent intermediate” (Fig. 1a). Most of these granules were amorphous (Figs. 2c and e), and some had an irregular, crystalline core surrounded by a clear space (Fig. 2d). In tissue embedded for morphology, many of these granules showed a condensed core (Fig. 1a). 3) A third cell type had translucent heteromorphous granules with homogenous matrix and a largest diameter of 400 ± 80 nm (Figs. 1a and c). This type of granule is called “translucent large.” 4) The fourth cell type had translucent heteromorphous granules with a diameter of 270 ± 50 nm. These granules are referred to as “translucent small” (not shown).

Ultrastructural analysis of ICCs prepared from the 70-day-old fetal porcine pancreas showed that the different cell types had almost the same morphology as found in the native pancreatic tissue described above (Figs. 1b and c). However, during the procedure of preparation, the cells were partly separated from each other and were either lined up in rows or loosely gathered in nests. Normal cell-to-cell contacts were often affected and clear spaces were interrupted by short cell-to-cell bridges without junctions; desmosomes or interdigitations were seen between neighboring cells (Figs. 1b and c).

1) Cells with dense, spherical granules colocalized glucagon, insulin, IAPP, and to a lesser degree, also somatostatin and PP in most of the granules (Figs. 2a and b). 2) In cells with translucent heteromorphous granules of intermediate size, the granules colocalized insulin, glucagon, and IAPP at moderate concentration and somatostatin and PP at weak concentration (Figs. 2c–e). 3) Cells with translucent large granules demonstrated somatostatin at an intense concentration and colocalization of glucagon and IAPP at weak concentration (Fig. 2f). 4) Cells with small translucent granules expressed PP exclusively.

1) Cells with dense granules demonstrated expression of glucagon at a high intensity and insulin, somatostatin, and PP at a weak intensity. No immunoreactivity to IAPP was found in these cells. 2) Cells with translucent intermediate granules revealed a weak concentration of insulin, somatostatin, and PP and a moderate concentration of glucagon and IAPP. 3) Cells with translucent large granules contained somatostatin at intense concentration, and the granules also colocalized insulin, glucagon, PP, and IAPP at weak concentration. 4) Cells with small translucent granules expressed PP exclusively.

1) In cells with dense granules, some of the granules had developed a translucent halo around a spherical dense core, but most granules were still amorphous. 2) Among the cells with translucent intermediate granules, an increased number of cells displayed an irregular, condensed granular core, similar to mature β-cell granules. 3) Cells with large translucent granules had, at this developmental stage, matured into δ-cell–like cells with granules more spherical in shape and equal in size when compared with earlier gestational stages. The diameter of these granules was 300 ± 50 nm. 4) Cells with small translucent granules had matured into PP-cell–like cells with small, spherical granules with a diameter of 230 ± 50 nm.

The endocrine cells constituting the ICC graft—prepared from the 70-day-old fetal porcine pancreas, cultured for 4 days, and removed 16 days after transplantation, i.e., a total fetal age of 90 days—were found in small islet-like nests. Necrotic parts could be seen in the nests. The plasma membranes of the endocrine cells were in close contact with each other at this stage, although the previously described open spaces between the adjacent cells could still be found. Most cells had an indistinct morphology with big cytoplasmic vacuoles (Fig. 1d). The individual cells were matured in parallel with what is described above for the native islet cells (Figs. 1d and 3a–d).

1) Cells with dense, spherical granules colocalized glucagon at intense concentration, insulin and IAPP colocalized at moderate concentration, and somatostatin and PP colocalized at weak concentration. 2) Cells with translucent intermediate granules displayed insulin at moderate concentration and glucagon, PP, and IAPP at weak concentration. 3) Cells with large translucent granules colocalized somatostatin at intense concentration and IAPP at weak concentration. 4) Cells with small translucent granules expressed only PP at intense concentration.

The maturation of the transplanted fetal cells mainly followed that of the native fetal pancreas. 1) Cells with dense granules expressed intense amounts of glucagon and weak amounts of insulin, somatostatin, and PP (Fig. 3a). No IAPP could, however, be detected. 2) Cells with translucent intermediate granules showed colocalization of insulin, glucagon, somatostatin, and IAPP at moderate concentration and PP at weak concentration (Figs. 3b and c). 3) Cells with translucent large granules displayed intense concentration of somatostatin, moderate concentration of glucagon, and weak concentration of IAPP (Fig. 3c). 4) Cells with small translucent PP-cell–like granules expressed intense amount of PP exclusively (Fig. 3d).

1) In cells with dense spherical granules, the granules were α-cell–like. 2) In cells with translucent granules of intermediate size, the granules were β-cell–like. 3) Cells with translucent large granules had, at this stage, matured into δ-cells. Upon maturation, the somatostatin granules became more spherical in shape and uniform in size, i.e., the typical δ-cell granule morphology. The final diameter of these granules was 300 ± 50 nm. 4) Cells with translucent small granules had matured into PP cells with granules more equal in size and more spherical and with a final condensed diameter of 230 ± 25 nm.

The ICC graft—prepared from the 70-day-old fetal porcine pancreas, cultured for 4 days, transplanted, and removed 41 days after transplantation (a total age of 115 days)—was organized in islet-like nests with β-cells at the highest numbers. The different cell types were matured and had the same morphology as described above for the native pancreata.

1) Cells with α-cell–like granules coexpressed glucagon at intense concentration, insulin at weak concentration, and IAPP at moderate concentration. 2) Cells with β-cell–like granules displayed colocalization of insulin at intense concentration, glucagon at weak concentration, and IAPP at moderate concentration; no somatostatin or PP was present. 3) Cells with δ-cell granules showed coexpression of somatostatin at intense concentration and IAPP at weak concentration. 4) Cells with PP-cell granules expressed exclusively PP at an intense concentration.

1) Cells with α-cell–like granules expressed only glucagon. 2) Cells with β-cell–like granules colocalized insulin at intense concentration, somatostatin at weak concentration, and IAPP at moderate concentration. 3) Cells with δ-cell granules expressed intense amount of somatostatin and weak amount of IAPP. 4) Cells with PP-cell granules only expressed PP.

Mature islets were demonstrated only after birth, i.e., after 115 days of gestation. The number of insulin-producing cells increased during ontogeny and became the most numerous cell type, whereas the number of cells expressing glucagon, somatostatin, or PP decreased upon maturation. 1) The well-known morphological characteristics of porcine α-cell granules—spherical granules with an eccentric dense core surrounded by a less dense halo—were seen. Upon maturation, the granule diameter slightly decreased to 250 ± 50 nm. 2) Cells with translucent intermediate granules matured into β-cells with their typical morphology, i.e., round vesicular granules with a crystalline core of medium density and a clear space along the membrane. These granules condensed during development, and the final diameter was 270 ± 30 nm. 3) δ-cells and 4) PP cells displayed a mature phenotype already at 90 days of gestation and have been described above.

The ICC graft, prepared from 70-day-old fetal porcine pancreas and removed 3.5 or 6 months after implantation (total age 180 and 257 days) also showed an increased number of β-cells and a corresponding decrease of glucagon-, somatostatin-, and PP-producing cells. The different islet cell types were matured into α-, β-, δ-, and PP cells, with their typical morphology of the granules (Fig. 1e) identical to that found in the islet cells in the adult porcine pancreas described above. Between adjacent cells, cell-to-cell contacts, i.e., desmosomes, interdigitations, and junctions, were commonly seen. The spaces between adjacent cells observed in grafts with shorter observation periods were not seen. Morphological evidence of a matured revascularization process is shown in Fig. 1e, demonstrating a capillary with fenestrated endothelium. An exchange of substances by transport vesicles both in the islet graft cell and in the endothelium is demonstrated. Any necrotic parts were not seen. The β-cell granules in pig and mouse have different ultrastructural morphology. The β-cell granules in a mouse have a morphology with a round, electron-dense core (Fig. 1f), whereas the porcine β-cell has an irregular and crystalline granule core (Fig. 1e).

1) Cells with adult α-cell granules stored glucagon (Fig. 4b) and occasionally weak amounts of IAPP. 2) Cells with adult β-cell granules displayed an intense expression of insulin (Fig. 4a) and IAPP. 3) In cells with δ-cell granules, an intense expression of somatostatin and a weak expression of IAPP were demonstrated. 4) Cells with PP-cell granules expressed exclusively intense amounts of PP.

1) Cells with α-cell granules were only immunoreactive for glucagon. 2) Cells with β-cell granules displayed colocalization of insulin at high intensity, somatostatin at weak intensity, and IAPP at moderate intensity. 3) Cells with δ-cell granules stored intense amounts of somatostatin and weak amounts of IAPP. 4) Cells with PP-cell granules stored exclusively intense amounts of PP.

The endocrine cells with a biological age corresponding to 4.5 months after birth showed the same pattern of immunoreactivity as the transplants removed after 3.5 months, with the exception that no somatostatin expression could be found in the β-cell granules (Fig. 4c, d, and e).

Only solitary gold markers were found in the plastic, and unspecific background labeling was very low. The adsorption and cross-reactivity tests confirmed the specificity of the antibodies used in this study.

At the ultrastructural level, several sequential steps in differentiation of islet cells can be identified, based on the morphology and hormone content of the secretory granules of the cell examined (9,10). Applying this approach, we demonstrated that the ontogeny of the transplanted fetal endocrine pig pancreas follows the same sequential differentiation as the native pancreas, i.e., all four major hormone-producing cells seem to develop normally after transplantation when compared with the development in the native pig pancreas. The process seems to be under stringent control, apparently directly related to the biological age of the tissue, and independent not only of the new environment under the kidney capsule but also of the adult and xenogeneic milieu provided after transplantation to the athymic nude mouse. The results presented strongly suggest that the ontogeny of the fetal endocrine pancreas is determined by genetic information carried by the endocrine cells and their progenitor cells and not by the systemic or local environment.

The synthesis and intracellular localization of the putative islet hormone IAPP was also investigated. IAPP was produced by the pluripotent fetal endocrine cells as well as in the adult α-, β-, and δ-cell granules, confirming our previous findings (10,33). In the transplanted ICCs, IAPP expression was demonstrated in β- and δ-cells. However, no IAPP was found in granules of the α-cell lineage at any of the different developmental stages examined. The reason for this discrepancy is difficult to understand. However, in the adult human pancreas, α-cell granules only occasionally contain IAPP (10,33).

The presently applied model of endocrine pancreas differentiation involves transplantation of fetal porcine ICCs under the kidney capsule of normoglycemic athymic (nu/nu) mice. The culture technique used for the fetal porcine pancreas has previously been described (25). These ICCs resemble isolated islets but contain only 5–10% insulin-positive cells and display no or only a low degree of glucose-stimulated insulin release (25). The replicatory rate of the insulin-positive cells is within the range of that seen in adult islets; however, a marked growth of the transplanted ICCs is seen during the first 2 months after transplantation (26). The epithelioid cells, which do not react with any of the four major islet hormones but display a high replicatory activity, constitute approximately half of the total cell number within the grafts 1 month after transplantation. Their number decreases continuously, and this type of cell is rarely seen after 2 months. In parallel with the gradual disappearance of these epithelial cells, there is a marked increase in the proportion of insulin-positive cells (26). These findings are confirmed in the present study. Cell replication in fully differentiated endocrine cells was almost never found. In contrast, over the various biological ages examined, there was a sequential differentiation, with defined ultrastructural appearance, of all four major islet cell types from a pluripotent cell. Therefore, the major source of new β-cells is differentiation from the undifferentiated epithelioid cells rather than a high replicatory activity in a few differentiated β-cells within the transplanted ICCs.

Although the differentiation of the endocrine tissue of the pancreas can occur in the absence of mesenchyme, the development of the exocrine pancreas critically depends on mesenchymal signals (34). In light of recently presented data, it seems as if FGFs, particularly FGF-1, FGF-7, and FGF-10, represent such signals (34). The lack of a proper mesenchyme support in the presently applied model of differentiation could explain the exclusive development of the endocrine pancreas.

The amount of insulin in the granules of the cells constituting ICCs was reduced compared with that found in the native pancreas. This is most likely due to the culture condition applied. The ICCs were allowed to form during 4 days of culture in RPMI-1640. This culture medium contains 11.1 mmol/l glucose and stimulates insulin release as evidenced by insulin accumulation (25). However, there was a marked decrease in accumulation of insulin with time, i.e., the amount of insulin released during days 3 and 4 was only ∼50% of that released during days 1 and 2, indicating a depletion of insulin deposits over time in culture (25).

Maturation of a glucose-induced insulin release is observed after transplantation, and the ICC grafts cure diabetic recipients after 4–6 weeks. Again, it seems as though the functional maturation of the fetal endocrine pancreas coincides with the biological age of the tissue. Also, both porcine and human β-cells retain their inherent ability to regulate glucose homeostasis after transplantation, i.e., transplanted human or porcine β-cells cause a lowering of the blood glucose levels to that normally obtained in humans and pigs, respectively (26,34). Compared with the milieu in the native pancreas, the transplanted ICCs in the present study were challenged with a relative hyperglycemia during the first 4–6 weeks after transplantation, after which the ICC graft lowered the blood glucose to values normally encountered in pigs. This slight hyperglycemia may enhance maturation of the endocrine pancreas (35,36; that is, at a total biological age of 115 days, we found monohormonal insulin- and glucagon-containing cells in the transplanted but not in the native pancreas. It is noteworthy that the transplanted pig cells retained the species-specific ultrastructural characteristics of their β-cell granule, i.e., the porcine β-cells had granules with an irregular and crystalline core, even long-term after transplantation, whereas the β-cell granule in the recipient mouse had a morphology with a round, electron-dense core.

Marked growth of the ICC grafts after transplantation was noted. Previous studies have shown that the DNA content of the grafts increases two to three times, whereas the insulin content increases 100-fold, resulting in a graft with a total insulin content corresponding to that of 6–8 mouse pancreata (26). These findings are in marked contrast to the results obtained after transplantation of syngeneic whole pancreata to normal rats (37). In these rats, the total β-cell volume doubled at the time of transplantation; however, with time, there was a marked reduction of the total β-cell volume to that normally encountered in rats. Interestingly, both the native and the transplanted pancreata contained only half of the normal β-cell volume found in pancreata removed from untreated control animals. This demonstrates that there is a tight regulation of the total β-cell volume in the adult pancreas by some systemic factor(s). This negative feedback is apparently not operating on the differentiation process of the transplanted fetal endocrine pancreas, suggesting that the differentiation of the fetal pancreas is regulated by factors inherent to the endocrine cells and their progenitor cells and not by the systemic or local environment.

Only few experimental models have convincingly demonstrated islet neogenesis in the adult pancreas (38,39,40,41,42). Common to these models is that they all induce proliferation and differentiation of ductal cells with the concomitant formation of nests of islet cells budding from the ductules. These nests of immature endocrine cells subsequently develop into new islets. Also, the development of the fetal endocrine pancreas has been described to occur in a similar pattern (43,44). The endocrine cells are first dispersed as single cells or in small clusters in the ductal epithelium. Subsequently, the developing islets lose their connection with the duct and form both bipolar and mantle islets. This neoformation of islets is a slow and continuous process, and the adult distribution and cellular composition of the endocrine pancreas is not attained until after birth (44,45). Also, in the presently applied experimental model, the epithelioid cells constituting the ICCs formed nests or duct-like structures after transplantation and the endocrine cells were frequently seen in these duct-like structures. Together, these studies indicate that islet neogenesis emanates from endocrine precursors present in the pancreatic ducts, and when proliferation of this ductal system occurs, as in the developing or regenerating pancreas, islet neogenesis is also induced.

It is intriguing to speculate that the ontogeny of the fetal endocrine pancreas seems to be regulated within the endocrine cells and related to chronological age of the tissue and not by factors outside the islets. If this is correct, it will have implications for the search for factors regulating the differentiation of the fetal endocrine pancreas. We hypothesize, from the results in the present study, that exogenous factors will have only limited effects on the differentiation process. Finally, the findings will have implications for the fetal endocrine pancreas as a source for transplantation in type 1 diabetes, because the tissue will most likely differentiate normally, irrespective of the new environment in the recipient.

This study was supported by grants from the Swedish Medical Research Council (nos. 06P-11813, 12X-6817, and 16X-12219), Sven Jerring Foundation, Göran Gustafsson Foundation, the Nordic Insulin Fund, the Torsten and Ragnar Söderbergs Foundation, the Swedish Diabetes Association, Barndiabetesfonden, the Swedish Society of Medicine, Clas Groschinsky Foundation, the Juvenile Diabetes Foundation International, and the Knut and Alice Wallenberg Foundation.