Dual origin, development, and fate of bovine pancreatic islets



Dr. med. Albert M. Ricken, Institute of Anatomy, Faculty of Medicine, University of Leipzig, Liebigstraße 13, DE-04103 Leipzig, Germany. T: + 49 341 9722048; F: + 49 341 9722009; E: albert.ricken@medizin.uni-leipzig.de


Endocrine cells are evident at an early stage in bovine pancreatic development when the pancreas still consists of primitive epithelial cords. At this stage, the endocrine cells are interspersed between the precursor cells destined to form the ductulo-acinar trees of later exocrine lobules. We here demonstrate that, in bovine fetuses of crown rump length ≥ 11 cm, the endocrine cells become increasingly segregated from the developing exocrine pancreas by assembly into two units that differ in histogenesis, architecture, and fate. Small numbers of ‘perilobular giant islets’ are distinguishable from larger numbers of ‘intralobular small islets’. The two types of islets arise in parallel from the ends of the ductal tree. Aside from differences in number, location, and size, the giant and small islets differ in cellular composition (predominantly insulin-synthesising cells vs. mixtures of endocrine cells), morphology (epithelial trabeculae with gyriform and rosette-like appearance vs. compact circular arrangements of endocrine cells), and in their relationships to intrapancreatic ganglia and nerves. A further difference becomes apparent during the antenatal period; while the ‘interlobular small islets’ persist in the pancreata of calves and adult cattle, the perilobular giant islets are subject to regression, characterised by involution of the parenchyma, extensive haemorrhage, leukocyte infiltration (myeloid and T-cells) and progressive fibrotic replacement. In conclusion, epithelial precursor cells of the ductolo-acinar tree may give rise to populations of pancreatic islets with different histomorphology, cellular composition and fates. This should be taken into account when using these cells for the generation of pancreatic islets for transplantation therapy.


The pancreas is a complex tissue and is still perhaps the least understood organ besides the brain in mammals (Hennig et al. 2004). In mammals, the pancreas develops from a dorsal evagination and one of two slightly more distal ventral buds of the primitive foregut endoderm (Slack, 1995; St-Onge et al. 2006). The two anlagen rapidly enlarge, invade the surrounding mesodermal mesenchyme, and develop an epithelial branching structure successively comprising solid cords, tubules and, finally, ducts with acini. Due to rotation of the developing stomach, the persisting ventral anlage moves dorsally and fuses with the dorsal anlage. This occurs at around embryonic day 13–14 in mice, day 41 in humans, 38–52 days post coitum (d.p.c.) in swine, and before 45 d.p.c. in cattle. As a result, the more ‘duodenal part’ of the definitive pancreas is usually derived from both anlagen, whereas the more ‘splenic part’ is derived from the dorsal anlage only.

Histologically, the development of the endocrine part of the pancreas is closely associated with the development of the exocrine part of the organ. From a temporal perspective, the exocrine part initiates and pursues its morphological differentiation programme before endocrine islets are morphologically distinct. The endocrine islets first become visible at the transition between the embryonic and fetal period, when the exocrine parts already consist of tubules and acini (O'Rahilly, 1983). This is in sharp contrast to endocrine cells being among the first cells on the scene during pancreatic organogenesis. Indeed, in many species, endocrine cells secreting pancreatic hormones are detectable in the epithelium of the foregut even before the ventral and dorsal anlage are apparent morphologically (Fujii, 1979; Carlsson et al. 2009). To date, five different endocrine cell types have been described that arise as a result of species-specific pancreatic temporal differentiation programmes (Rall et al. 1973; Carlsson et al. 2009). The endocrine cell types are thought to derive, together with ductal and acinar cells, from a common endodermal progenitor cell (Zhou et al. 2007; Desgraz & Herrera, 2009). The differentiation of the endocrine and exocrine cells then results from the complex sequential action of different secreted signals, and through activation of specific transcription factors (Bonal & Herrera, 2008; Mastracci & Sussel, 2012). The endocrine cells are initially dispersed as single cells, or small clusters of cells, between the exocrine precursor cells (Merkwitz et al. 2011), before becoming increasingly confined to morphologically tangible islets.

Endocrine islets within a single pancreas are not all identical (Steiner et al. 2010). In many species, intrapancreatic differences have been described in islet microarchitecture (cellular composition, topology, and vascular supply), and threshold for hormone responsiveness (Larsson et al. 1975; Orci et al. 1976; Baetens et al. 1979; Jorns et al. 1988; Redecker et al. 1992). These differences appear to be influenced partly by islet size (Grube & Weimann, 1985; Kaihoh et al. 1986; Huang et al. 2011) and by the origin of the cells forming the islets, in terms of whether they arose from the ventral or dorsal anlage (Orci et al. 1976; Baetens et al. 1979). ‘Perilobular giant islets’ and ‘intralobular small islets’, the focus of this investigation, represent a higher degree of qualitative diversity among islets. Although first observed by Laguesse (1896) in sheep more than a century ago, perilobular giant islets have been largely ignored, even in ruminants, where they were initially described. This neglect likely reflects a perceived unimportance, resulting from the fact that their existence beside intralobular small islets is widely considered to be restricted to a small number of species, e.g. ruminants (Bonner-Weir & Like, 1980) and limited to fetal and neonatal life (Titlbach et al. 1985).

The fact that endocrine cells may assemble into more than one histologically distinct type of islets is intriguing! A detailed understanding of these two types of islets may prove to be relevant and important for non-ruminant mammals, including humans, where fetal islet tissue, with remarkable similarities to the perilobular giant islets noted in ruminants, has sporadically been described, and may potentially underlie islet cell adenomatosis observed in neonates with persistent hyperinsulinaemic hypoglycaemia of infancy (Shermeta et al. 1980), and hypoglycaemia following gastric bypass surgery in adults (Kaczirek & Niederle, 2004). Little is known of the embryological origins, histogeneses, and fates of these two islets, and the current study in the bovine pancreas was performed in an attempt to fill some of these gaps in our current knowledge. Our findings shed new light on this dual population of islets, and have important clinical relevance for therapies involving isolation of islet precursor cells.

Materials and methods


In total, the pancreata of 38 embryos and fetuses (see Table 1) with crown–rump lengths ranging from 0.9 to 90 cm, were histologically and immunohistochemically investigated. In addition, pancreata from three calves and three adult cattle were included for comparison. The embryos and fetuses were obtained from erroneously slaughtered pregnant cows at the local abattoir. Larger fetuses were immediately dissected, their pancreata excised en bloc from the retroperitonei and either snap-frozen in liquid nitrogen or fixed in 4% paraformaldehyde in 0.2 m sodium phosphate buffer (pH 7.4). Where appropriate, whole pancreata were divided into specimens representing the ‘duodenal part’ and ‘splenic part’. Embryos and smaller fetuses (< 10 cm) were immediately fixed en bloc or in pieces in paraformaldehyde. The paraformaldehyde-fixed specimens were embedded in paraffin wax and serially sectioned together with the frozen samples at 7 μm. A series of sections were prepared from each collected tissue specimen and a sequence of sections was stained with haematoxylin and eosin (H&E). Analysis of the H&E-stained sections was used to select adjacent sections for specific indirect immunostaining or Sirius red staining (Haber et al. 1999). Finally all sections were thoroughly analysed under a light microscope (Axioplan 2; Carl Zeiss, Jena, Germany) equipped with a computerised digital recording system (Progres C3 digital camera and progres capturepro software, both Jenoptik, Jena, Germany).

Table 1. Sampling collective
CRL in cm0.8–25132533455669≥ 81  
End of month12 3 4 5 6 7 89 + x  






Indirect immunostaining

Indirect immunostaining was carried out as reported previously (Hsu & Raine, 1981; Tsikolia et al. 2009; Merkwitz et al. 2011). In brief, paraffin sections were dewaxed in xylene and brought to distilled water by hydration in graded alcohols. A blocking step for endogenous peroxidase activity was performed on frozen and dewaxed paraffin sections. Additionally, antigen retrieval was performed on the dewaxed paraffin sections by microwaving the sections in 0.1 mm sodium citrate buffer (pH 6) or, for glial fibrillary acidic protein (GFAP) detection, by incubating the sections in 0.1% (w/v) proteinase K solution in tris-buffered saline (TBS)-CaCl2 buffer (pH 7.5) for 4 min at 21 °C. All sections were then rinsed then blocked for 30 min with normal serum, species-matched to the biotinylated anti-IgG secondary antibody. For GFAP detection, avidin and biotin solutions were included in this blocking step. Antisera and primary antibodies were diluted in antibody diluent buffer (10% bovine serum albumin in phosphate-buffered saline, PBS, or, in the case of GFAP detection, TBS plus 0.1% Tween 20) to ratios and concentrations as indicated in Table 2. Rinsed sections were incubated with antisera and primary antibodies in a moist chamber at 4 °C overnight. With the exception of GFAP detection, incubations with biotinylated anti-IgG secondary antibody and avidin : biotinylated enzyme complex were performed sequentially for 30 min at room temperature (Axxora, Lörrach). For GFAP, incubation with the biotinylated anti-IgG secondary antibody was performed for 1 h at room temperature, followed by a 30-min incubation with buffered aqueous ExtrAvidin-peroxidase solution (Sigma, Munich, Germany). Visualisation was achieved using the chromogen 3,3′-diaminobenzidine as a peroxidase substrate. Indirect immunostained sections were then rinsed, counter-stained with haematoxylin, and mounted with Roti-Histokitt (Carl Roth, Karlsruhe, Germany). The indirect immunostaining was analysed and photographed using the Axioplan 2 microscope and the digital recording system from Jenoptik.

Table 2. Antibodies (Ab) and staining procedures
Monoclonal antibody (mAb)/antiserum (As)CloneCatalogue &#TypeConcentration (mg mL−1)SourceDilution immunohistochemistry
Primary Ab/As anti- 
insulinK36AC10I2018mAb5.7Sigma-Aldrich (Munich, Germany)1 : 1000
GlucagonK79bB10G2654mAb11.7Sigma-Aldrich1 : 1000
Cytokeratin (pan)Lu-5DLN-10752mAb0.4Dianova (Hamburg, Germany)1 : 1000
CD4CC30MCA834mAb1.0Serotec (Düsseldorf, Germany)1 : 500
CD8CC63MCA837GmAb1.0Serotec1 : 500
CD18/11 BAQ30AmAb1.0VMDR Labordiagnostik (Leipzig, Germany)1 : 500
CD14VPM65MCA920mAb1.0Serotec1 : 100
CD34N26GiftmAb5.4Sakurai et al. (2006)1 : 400
CD45 CACTB51AmAb1.0VMDR Labordiagnostik1 : 500
Neurofilament 200 kDaRT97MAB5262mAb1.0Milipore (Schwalbach, Germany)1 : 4000
GFAPZ0334AS2.9Dako (Hamburg, Germany)1 : 1000
Secondary Ab
Goat anti-mouse IgG Ab (Vectastain)BA-92001.5Axxora (Lörrach, Germany)1 : 200
Goat anti-rabbit IgG Ab (Vectastain)BA-10001.5Axxora1 : 200
Goat anti-rabbit IgG, biotin-SP conjugateAP132BUnknownChemicon (Millipore)1 : 1500

Specificity of staining was controlled for by omitting the primary antibody, or replacing it with an equivalent amount of non-immune serum or immunoglobulin derived from unimmunised rabbits. As part of the immunohistochemical detection of islet hormones, pre-absorption control experiments were performed as described in detail elsewhere (Merkwitz et al. 2011). In brief, to demonstrate specificity of staining, primary antibodies were pre-incubated with a manifold excess of the corresponding target hormone before being applied to tissues. In the analysis of leucocyte subpopulations, parapancreatic lymph nodes served as internal control tissue.

Sirius Red collagen staining

Paraffin sections were dewaxed, rehydrated, and stained in Sirius Red F3b solution [0.1 g Direct Red 80 (Sigma-Aldrich) dissolved in 100 mL saturated aqueous solution of picric acid] for 1 h. Stained sections were washed in two changes of acidified water (10 mL acetic acid (glacial) to 22 mL of distilled water), dehydrated in increasing alcohol concentrations, cleared in xylene, and mounted with Roti C°-Histokitt.


The study includes pancreata of bovine embryos and fetuses from the end of the first month of gestation to term (Table 1), as revealed by comparing the measured crown–rump length (CRL) with the corresponding gestational ages reported in the literature (Rexroad et al. 1974; Bertolini et al. 2002).

Endocrine cells are present in the epithelium of very early bovine pancreata

In fetal pancreata, single or groups of pancreatic hormone-staining cells could be detected before ductulo-acinar structures and islets were clearly identifiable (Fig. 1). Numerous endocrine cells were found intermingled with the exocrine precursor cells in the epithelium of primitive tubular structures invading the mesenchyme at CRL = 0.9 cm (Fig. 1). Furthermore, glucagon and/or glucagon-like peptide 1 (GLP-1)-staining cells were present in the epithelium of the adjacent gut (Fig. 1F,G). Endocrine cells continued to contribute to tubules and primitive acini at subsequent stages (Fig. 1H–J) until, through progressive sorting, endocrine and exocrine cell types had become organised into distinct morphological compartments.

Figure 1.

Illustrates the topographical relationship between endocrine and exocrine pancreas cells during development. (A–J) Endocrine and exocrine pancreatic elements intermingled within primitive tubular structures during the first gestational month, before they become progressively segregated into distinct compartments. (A–C) Single or small groups of endocrine cells (arrows) can be identified, even when the developing pancreas consists only of primitive tubular structures. (D,E) Insulin (Ins) and glucagon/glucagon-like peptide1 (Glu and/or GLP-1) stained cells (arrows) are detectable within the epithelial lining of the primitive tubular structures. (F,G) Glu and/or GLP-1-stained cells (arrows) are also present along the epithelial lining of the intestinal wall, in contrast to Ins-stained cells. (H–J) Glu and/or GLP-1-stained cells still predominate in the developing pancreas at the end of the first gestational month. In summary, in the first month of bovine pancreatic development, endocrine cells are present but morphogenesis of a discrete endocrine compartment has not yet occurred. This situation changes at later developmental stages. (K–R) Endocrine cells are now located preferentially in two discrete endocrine units, accompanied by their progressive disappearance from exocrine tissue, which has now developed into a ductal tree with acini. ‘Perilobular giant islets’ (K, triangle, 6–7 months' gestation) can be distinguished from intralobular small islets (O, star, 9 months' gestation). (L–N) Ins-stained cells constitute the vast majority of cells in the perilobular giant islets; N shows an enlargement of the dashed line area in L. (O–Q) Glu and/or GLP-1-staining (arrows) can be observed in a more highly magnified ‘intralobular small islet’ (star).(R) Ins-stained cells (arrows) occur alongside the Glu and/or GLP-1 stained cells in the intralobular small islets, where they are mainly centrally located, yet still constitute the predominant cell type. (S,T) Divergent islet hormone staining for Ins (S) and Glu and/or GLP-1 (T) in a ‘perilobular giant (triangle)’ and an ‘intralobular small islet (arrow)’ within one section. (U,V) The purified peptide hormone has specifically bound the corresponding antibody during the pre-absorption (Preabs) control experiment, yielding negative immunostaining for Ins (U) and Glu and/or GLP-1 (V). Haematoxylin and eosin (HE) stain, and indirect immunostaining for Ins (B D,F,I,L,N,R,S) and Glu and/or GLP-1 (C,E,G,J,M,Q,T), respectively. (U,V) Pre-absorption (Preabs) control experiments. Scale bars as indicated.

Endocrine cells separate into two types of islets

Endocrine cells first appeared separate from the exocrine pancreatic portions, as distinct islet structures, at CRL = 11 cm. Two types of islet structure could be distinguished from this stage on. These islet types were designated perilobular giant islets (Fig. 1K–N), and intralobular small islets (Fig. 1O–R).

The endocrine cells in the perilobular giant islets were predominantly insulin-positive β-cells (cf. Fig. 1L with Fig. 1M). Only a very small proportion of cells were glucagon and/or GLP-1-positive α-cells. These cells were predominantly found at the periphery of the perilobular giant islets. The cellular composition of the perilobular giant islets did not differ with progressing development, and appeared uninfluenced by the site of islet origin within the pancreas (‘duodenal part’ or ‘splenic part’). A considerable number of endocrine cells in the intralobular small islets were glucagon and/or GLP-1-positive α-cells (cf. Fig. 1Q with Fig. 1R). A marked decrease in glucagon and/or GLP-1-positive α-cells, and a concomitant increase of insulin-positive β-cells, accompanied increasing developmental stages, regardless of intra-pancreatic location.

Both types of islets started to develop around the same time. ‘Perilobular giant islets’ were first observed in developing pancreata of fetuses with CRL = 11 cm, whereas the intralobular small islets were first detected in pancreata of fetuses with CRL = 15 cm. The intralobular small islets were always the more numerous of the two islet types. Emergence of intralobular small islets continued over a longer phase of development compared with the perilobular giant islets, all of which appeared to arise simultaneously. Both types of islets persisted to the antenatal stage of CRL = 90 cm; however, there were distinct differences regarding their fate thereafter. In pancreata of calves and adult cattle, perilobular giant islets appeared to have disappeared, and only individual or fused intralobular small islets could be observed.

Histogenesis of the perilobular giant islets

In this study, perilobular giant islets, were first clearly identifiable in pancreata of bovine fetuses at CRL = 11 cm, and remained present until term.

The perilobular giant islets most probably originated from the endodermal epithelium, and typically arose at the very ends of primitive ductulo-acinar complexes (Fig. 2A,B), contiguous with the connective tissue and fibrous septae at later stages. The perilobular giant islets appeared to be direct offshoots of the primitive ductulo-acinar structures (‘D’ in Fig. 2A–D). Occasionally, a trunk-like structure, composed of a single or double row of cuboidal epithelium, could be observed, and appeared to link the island with the neighbouring exocrine tissue (D in Fig. 2D). At all subsequent stages, perilobular giant islets were present only in small numbers. ‘Perilobular giant islets’ were large to giant, and irregular in shape in routine histological sections (Fig. 2E). ‘Perilobular giant islets’ were many times larger than the intralobular small islets, even when the latter coalesced (see below) to form larger complexes at later developmental stages. It was common to observe perilobular giant islets with diameters ranging from a few 100 μm in younger fetuses, to over 1 mm in older fetuses. The cells in the islets were arranged in single- or double-rowed anastomosing polymorphic cell bands, which were of rosette-like or gyriform appearance (Fig. 2F–H). The cells themselves were cylindrical in shape in early stages, but changed to a cuboidal shape at later stages (cf. Fig. 1I with Fig. 1F,H). The rich cytoplasm of the islet cells stained intensely red with eosin, and the nuclei were predominantly centrally or apically located. The spaces between the cell bands were infiltrated by a delicate capillary network (Fig. 2J).

Figure 2.

Provides insight into the morphogenetic, structural, and topographical properties of perilobular giant islets. (A, B) Early evidence of the existence of these islets (triangle) can be found in pancreata of ≥ 11 cm CRL fetuses (i.e., during the third and fourth month of gestation). The islets reveal a close topographical relationship to the ends of primitive ductulo (‘D’)-acinar (‘A’) structures. (C,D) When further developed, the islets (triangles) and the associated ductulo (‘D’)-acinar (‘A’) tree resemble vines bearing leaves and bunches of grapes due to the different sizes, shapes, and locations of the participating structures. (E) Fully developed perilobular giant islets (triangles) stand out on account of their tremendous size. (F–I) The islet cells organise into rosette-like (circle) and gyriform (arrowheads) patterns. The islets usually have a cylindrical shape (G,H), but can be cuboidal, particularly at later developmental stages (I). (J) A delicate capillary network supports the parenchymal cells. The central space of each epithelial rosette is occupied by a CD34-stained capillary (arrows). (K) Note that no immunohistochemical staining of the capillaries is observed when the endothelial specific anti-CD34 antibody (Sakurai et al. 2006) is replaced by a non-relevant antibody (arrow). (L–N) Nerve fibres (‘N’) connect the perilobular giant islets (triangle) to ganglia (‘G’) that lie conspicuously close (L) or even adjacent (M and N) to them. (O,P) An immature islet (triangle) developing in close proximity to neuronal structures (‘NS’). Indirect neurofilament (NF)-immunostaining allows differentiation between the neuronal elements and the immature endocrine tissue. (Q) NF-stained nerves fibres (‘N’) that extend over a longer distance to a ‘perilobular giant islet’ (triangle). (R,S) Here, the point of entry (R, star) and spread (S, arrowheads) of glial fibrillary acidic protein (GFAP)-immunoreactive structures are demonstrated in an islet. Haematoxylin and eosin (HE) stain, and indirect immunostaining for CD34 (J), CK (K), NF (P,Q), and GFAP (R,S), respectively. Scale bars as indicated.

In close proximity to the perilobular giant islets, larger blood vessels and, in particular, neuronal elements were found (Fig. 2L–S). Nerves, demonstrating neurofilament and GFAP-like immunoreactivity were demonstrated linking perilobular giant islets with intrapancreatic ganglia (Fig. 2L,Q). In addition, ganglia were also found in immediate juxtaposition to the islets (Fig. 2M–P). In the ‘perilobular giant islet’ neuronal- and glia-like immunoreactivity continued to be detectable in the spaces between the endocrine cell bands, underscoring the tight neuronal control of the endocrine cells (Fig. 2R,S).

One striking feature of the perilobular giant islets was the presence of intra-insular cavities, in which extravasated red blood cells could be seen (see below).

Histogenesis of intralobular small islets

The intralobular small islets were closely associated with the exocrine pancreas, in which they became entirely embedded when lobules formed at later developmental stages. The intralobular small islets developed, in common with the acini, from tubular structures. Endocrine cells accumulated to form sub-islet-sized cell clusters at very early stages of ductulo-acinar development, as outlined above and seen in Fig. 1A–D. The endocrine cell clusters were predominantly found within the acini present at the sides and ends of the ductal tree (Fig. 3A–C), where they could be observed in proximity to the intercalated duct cells (arrow in Fig. 3C).

Figure 3.

Illustrates the properties of intralobular small islets, which develop within the ductulo-acinar tissue. (A,B) The intralobular small islets (stars) develop in parallel with the ‘intralobular giants islets’ (triangle) in pancreata of fetuses with CRL ≥ 15 cm (i.e. during the 3rd and 4th month of gestation). The ‘intralobular small islets’ are much smaller in diameter than the ‘perilobular giant islets’, but are present in greater numbers. (C,D) The small islets (star) arise within the distal segments of the developing ductulo (‘D’)-acinar (‘A’) tree. Note the intercalated duct cells in C (arrow). (E) The islet cells lack positivity for cytokeratin (CK), which facilitates their discrimination from CK-stained exocrine components of the ductulo (‘D’)-acinar (‘A’) tree. (F) As the intralobular islets grow in size, they give the impression of impinging on the exocrine cell-population in the acinar ends, which can be easily recognised by their granulated cytoplasm and light nuclei with prominent nucleoli (arrows). (G) A vertical section through a mature ‘intralobular small islet’ reveals connective tissue elements from the outside in continuity with a central vascular connective tissue core (arrowheads). (H–K) A horizontal section trough the middle of a mature ‘intralobular small islet’ reveals endocrine cells organised in concentric zones. A central vascular connective tissue core (1) is followed by an inner zone of radially arranged elongated cells (2), and an outer zone of smaller cells (3), which abut a rim of capillaries (arrowheads) and larger exocrine cells (arrows). (L,M) Indirect CK-immunostaining illustrates the persisting topographical association between the ‘intralobular small islet’ and the ductal tree (‘D’). (N) The islet depicted in M, and here immunostained for CD34, gives the impression of having arisen from the duct (‘D’), and migrated laterally to engulf the vascular connective tissue core (star), which is connected to the exterior. (O) The ‘intralobular small islet’ (stars) are displaced to the centre of the ductulo-acinar trees as the ductulo-acinar trees evolve further and develop into lobules. (P) The higher magnified view of the framed area in O demonstrates that the development of new ‘intralobular small islet’ at the distal ends of the lobules has ceased by the antenatal stage. Haematoxylin and eosin (HE) stain, and indirect immunostaining for CK (E,L,M) and CD34 (N), respectively. Scale bars as indicated.

At later developmental stages, the number of cells in these clusters frequently exceeded several-fold the number of intercalated and acinar cells in the acinus from which they had evolved (Fig. 3D–J). The endocrine cells appeared to physically displace these other cell types (arrows in Fig. 3F,J).

The larger clusters of endocrine cells now constituted a separate morphological entity within the ductulo-acinar tree, and were termed intralobular small islets. It is of particular note that these larger clusters of endocrine cells maintained physical contact with the cytokeratin-positive ductulo-acinar structures whence they apparently originated (Fig. 3E,L,M).

At later developmental stages, the larger clusters of endocrine cells occurred within acini more centrally located in the ductal tree. However, it was only near term, at CRL = 80 cm (Fig. 3O), that these islets gave the impression of being structurally discrete from the surrounding exocrine tissue. The intralobular small islets were now no longer constituents of the intercalated duct cells and acini in the periphery of the lobules (Fig. 3O,P); rather, they appeared in the central areas of the lobules, where they could be observed forming loose insular aggregates.

‘Intralobular small islets’ were always many times more abundant than perilobular giant islets. The intralobular small islets, were round or oval in shape, well defined (Fig. 3I,J), and did not exceed 100 μm in diameter, even in histological sections from older fetuses.

The cells in the intralobular small islets were polyhedral, contained little cytoplasm, had small nuclei and, as a rule, displayed a higher nuclear–cytoplasmic ratio than the endocrine cells in the ‘perilobular large islands’ (cf. Fig. 2G,H with Fig. 3I,J). Depending on the orientation of the section, the cells of the intralobular small islets appeared irregularly or concentrically arranged in several zones around a central vascularised connective tissue core (Fig. 3I,J,N). The nerve supply to these islets was more discrete in comparison with the perilobular giant islets and manifested as a network of delicate nerve fibres.

Islet lesion and fate of the bovine pancreatic islets

Subtle evidence of regressive change could be detected in the perilobular giant islets even at the earliest stages studied (Fig. 4A,B). These changes essentially consisted of mild haemorrhagic foci. No similar lesions were apparent elsewhere in the developing pancreas.

Figure 4.

Dramatically describes the different fates of the two types of islets during the antenatal period. One striking feature of the perilobular giant islets is that they accrue intra-insular cisternae filled with red blood cells. (A,B) The cisternae are found even at the earliest stages of the development of these islets (A and B = 4th month of gestation) but increase dramatically in frequency and size over time. The cisternae may be caused by changes in the height of the epithelial parenchyma. Mechanical forces may be generated and transmitted to the delicate blood vessels in the connective tissue, which are then disrupted. (C,D) Higher magnification views of the cisternae clearly demonstrate the lower height of the epithelia directly lining the pools of haemorrhagic debris. (E) Indirect CD34-immunostaining demonstrates the presence of vascular elements (arrow) within the pool of red blood cells. (F–K) The accumulating structural changes (F) lead to moderate to severe lymphocytic infiltrates (G), progressive fibroses (H–J), and finally to total demise or involution of the islet parenchyma (K). (L,N) Of note, the immediately adjacent exocrine tissue may be haphazardly incorporated into the endocrine tissue (L, star), and now also subject to the effects of the degenerative changes (M, star). Note, in particular, the peribasal accumulation of leukocytes (arrows) in M. (N) The progressive replacement of the giant islet parenchyma by collagen fibres can be verified easily by Sirius Red staining. (O–Q) Illustrate how the rosette and gyriform islet parenchyma (star) is progressively replaced by areas of collagen (stained red). (R,S) In contrast to the perilobular giant islets, the ‘intralobular small islet’ rarely manifest regressive changes such as bleeding, leucocytic infiltration or fibrosis. The ‘small intralobular islets’ most often persist into postnatal life and likely represent the basis of adult islet tissue (star). (S) Note the central mass of vascular connective tissue (arrow) visible in a mature islet in the enlargement of R. Haematoxylin and eosin (HE), Sirius Red collagen stain (N–Q), and indirect immunostaining for CD34 (E), respectively. Scale bars as indicated.

Extravasations of red blood cells were found between the cell bands containing conspicuously flattened epithelial cells (Fig. 4C,D), and most likely resulted from leakage of fragile blood vessels in the supporting connective tissue sites between the cellular bands (Fig. 4E). Inflammatory infiltrates and replacement of the endocrine cells by collagen tissue were not a feature of these early haemorrhagic changes.

The haemorrhagic lesions intensified dramatically in the perilobular giant islets during antenatal stages of development (Fig. 4F). The lesions were now characterised by shrunken cellular bands, broadened interepithelial spaces, large red blood cell extravasates, and leucocytic infiltrates, and a replacement of islet tissue by collagen fibres (Fig. 4F–Q). In some instances, the regressive changes extended into neighbouring exocrine tissue, in particular where acinar cells were caught up and intermingled with the endocrine cells (Fig. 4L,M).

Whereas perilobular giant islets were subject to regression, the intralobular small islets persisted in the pancreata of calves and adult cattle (Fig. 4R,S).

Staining with antibodies specific for distinct leucocyte differentiation antigens revealed a ubiquitous presence of cells of the myeloid lineage in fetal pancreatic tissue between mid-gestation and term (Fig. 5A–E). Near term, the myeloid cells preferentially concentrated in the perilobular area with regressing giant islets and ganglia (Fig. 5F–I). In the associated lymphocytic infiltrates, large numbers of T lymphocytes were additionally seen that were primarily CD4-positive (Fig. 5H–K).

Figure 5.

Documents age related changes in leucocyte distribution and profile during later stages of pancreatic morphogenesis. (a–e) In pancreatic tissue of a fetuses with CRL 50 cm, leucocytes (arrows) are likewise present in the small islets/exocrine (star) and giant islets areas (triangle). (C,D) The cells stain positively for CD18/11 (C) and CD14 (D), suggesting leucocytes of the myeloid lineages. (D,E) In contrast, lymphocytes, stained with the T-cell markers are sparsely present at this stage. (F,G) In pancreatic tissue near term, e.g. of a fetus with CRL 80 cm, the small islets/exocrine area (star) appear emptied of leucocytic cells, whereas leucocytic cells spread in the giant islet area aside (triangle). Larger leucocytic infiltrates (dashed circle) are now also associated with the perilobular giant islets. The infiltrates house myeloid and predominantly CD4-positive cells. Note the ganglion (‘G’) and the island (triangle) adjacent to the large infiltrate (arrows) in (H–K). (I–O) Peripancreatic lymph nodes around the pancreatic parenchyma follicles serve as positive controls for indirect immunostaining of the leucocyte markers. Note the cell marker specific staining in the vicinity of a lymphoid follicle (‘LF’). Haematoxylin and eosin (HE) stain, and indirect immunostaining for CD18/11 (B,G,I); CD14 (C,M); CD4 (D,J,N) and CD 8 (E,K,O), respectively. Scale bars as indicated.


Our current study, including 38 bovine pancreata from fetuses spanning a period from early gestation to term, reveals novel findings pertaining to the origin, histogenesis, and fate of endocrine pancreatic tissues. Our data thereby validate, enrich, and draw attention to largely ignored literature pointing to two populations of islets in cattle that arise almost in parallel, but differ (as summarised in Table 3) in origin, histogenesis, size, cellular composition, neighbouring environment, and fate, suggesting they likely represent two distinct entities and not outcomes of divergent developmental steps (Aron, 1931; Bonner-Weir & Like, 1980). Unlike the genesis of perilobular giant islets, the provenance, histology, and topographic location of the intralobular small islets resembles more closely the concept of classical acino-insular islet development. Most importantly, intralobular small islets appear to persist in the pancreata of calves and adult cattle, whereas the population of perilobular giant islets begins to dissipate at late fetal stages. We therefore propose that the term ‘perilobular giant islets’, or ‘islets of Laguesse’, should be used in future to distinguish this type of islet from the ‘intralobular small islets’, or ‘islets of Langerhans’.

Table 3. Characteristics of the two populations of islets in cattle
Perilobular giant isletsIntralobular small islets
Few, arising synchronously in a single episodeMany, arising in several waves
Located in the septae, in close proximity to ganglia and nervesEmbedded within the exocrine tissue
Large in sizeSmall in size
Characterised by epithelial trabeculae with gyriform and rosette-like appearancesCharacterised by concentric, dense epithelial arrangements
Comprised almost entirely of cytoplasm-rich, weakly immunoreactive β-cells with eccentric nucleiContain a significant portion of other islet hormone-bearing cells besides small-sized, strongly immunoreactive β-cells
Transitory, with perinatal involutionPermanent, with perinatal persistence

Differences in the histogenesis of the two types of fetal bovine pancreatic islets

It has become the prevailing dogma (Cleaver & Melton, 2005) that single or small groups of pancreatic endocrine cells arise within the ducts and acini at early development stages but very soon disseminate from the exocrine elements, forming small interstitial cell clusters and, ultimately, classical islets. This contrasts with our data from cattle.

At early stages in their development, bovine perilobular giant islets appear to be in direct peri-ductulo-acinar contact; the bovine intralobular small islets even originate within the exocrine ends of the peripheral ductal tree and retain, for a significant period of time, close relationships with the tubulo-acinar elements whence they arose. Hence, our findings provide morphological support for the existence of intimate topographical and functional associations between pancreatic ducts and islet tissue (Van Suylichem et al. 1992; Bertelli & Bendayan, 2005). It seems implicit that distinct mechanisms and regulatory factors must be involved in the development of two populations of islets from one and the same tubulo-acinar tree in cattle and other ruminants (Bonner-Weir & Like, 1980; Titlbach et al. 1985; Lucini et al. 1998). Thus far, factors that account for differences in origin, growth, morphology, cellular composition, and fate of the two populations of islets remain a mystery, and further studies are clearly required to elucidate the developmental mechanisms involved.

Differences in the fate of the two types of fetal bovine pancreatic islets

In contrast to Bonner-Weir & Like (1980), we observed regressive changes in the prenatal perilobular giant islets of cattle similar to those previously described in lambs (Titlbach et al. 1985). Thus, the perilobular giant islets likely are physiologically important mainly in the fetal period of ruminants. The demise of the perilobular giant islets before birth certainly contributes to the relative decrease of the β-cell mass during fetal and early post-natal life in ruminants (Grossner, 1967), and probably to the substantial perinatal remodelling of the β-cell mass in other species (Scaglia et al. 1997; Kassem et al. 2000). Involution of the perilobular giant islets seems to be a physiological rather than a pathological process, and may be necessary to prevent neonatal hyperinsulinaemic hypoglycaemia (Jaffe et al. 1980; Shermeta et al. 1980; Titlbach et al. 1985). Haemorrhagic islets and/or peliosis-like vascular changes are sporadically reported in man, in the pancreata of stillborn fetuses and infants dying during early post-natal life (Emery & Bury, 1964), and in pancreata showing focal adenomatous hyperplasia of islet cells (Kovacs et al. 1986; Rumilla et al. 2009). An inverse relationship between ‘haemorrhagic perilobular giant islet incidence’ and post-natal age in cattle is therefore intriguing. Oestradiol appears to attenuate the development of spontaneous haemorrhage, inflammation, and fibrosis in pancreatic islets of adult rats (Seemayer et al. 1986; Imaoka et al. 2007, 2009), suggesting that perinatal changes in hormonal milieu may, together with changes in nutritional status and carbohydrate metabolism, account for the lesions observed in ruminants (Bonner-Weir & Like, 1980; Titlbach et al. 1985).

We observe a general colonisation of bovine pancreata by myeloid cells at later fetal stages, and an accumulation of similar cells, probably accompanied by (mainly CD4-positive) T cells, in the regressing perilobular giant islets at term. It may well be that the regressing islets and nearby neuronal elements release antigens that are processed and presented by myeloid cells in the pancreatic tissue and draining lymph nodes (Mei et al. 2002; Razavi et al. 2006; Tsui et al. 2008). Whether CD4-regulatory T cells (Tregs) are present as well, and help prevent autoimmune activation during regression (Li et al. 2012), is not known. Further data are required to demonstrate a physiological auto-protective role, rather than a pathological auto-reactive role, of the perinatal accumulation of immune cells (Jansen et al. 1993; Korpershoek et al. 2004; Razavi et al. 2006).

Functional significance of the two types of fetal bovine pancreatic islets

It has been speculated that the presence of two types of islets permits perinatal changes in digestion/carbohydrate metabolism in ruminants (Bonner-Weir & Like, 1980; Titlbach et al. 1985). Differences in the release of insulin in response to glucose in islets of the ventral and dorsal anlagen of rats support this hypothesis (Baetens et al. 1979). The two types of islets in ruminants may harbour different β-cell populations, with more immature, glucose-independent, autonomous insulin-secreting cells, better suited to the endocrine demands of prenatal life, predominant in the larger islets (Shermeta et al. 1980; Titlbach et al. 1985; Huang et al. 2011; Sempoux et al. 2011).

The disparate relationships of the two populations of islets to the exocrine tissue and, in addition, to neuronal structures of the autonomic and sensory nervous system (Baltazar et al. 2000; Myojin et al. 2000) suggest the presence of effective insulo-acinar communication of the intralobular small islets, and intense insulo-neuronal communication of the perilobular giant islets. Glial fibrillary acidic protein immunoreactivity is reported in ganglia of the digestive system (Jessen et al. 1984) and is particular helpful in tracking their neurons (Krammer & Kuhnel, 1993).

Different types of islets in other species

It is well documented for several mammalian species that the arrangement of endocrine cells within the pancreas is not uniform (Steiner et al. 2010). In addition to being contained within islets, endocrine cells are found scattered in the exocrine parts of the fetal and adult pancreas (Kaihoh et al. 1986; Bouwens & Pipeleers, 1998; Lucini et al. 1998). Furthermore, the composition and topology of endocrine cells within islets can differ to the extent that more than one islet type can be defined (Baetens et al. 1979). For example, ‘primary neuro-insular complexes in the perilobular connective tissue’ and ‘secondary acino-insular complexes in the exocrine lobules’ have been described in man (Liu & Potter, 1962) and other species (Fujita, 1959; Watanabe et al. 1989; Persson-Sjogren, 2001). In addition, insular size has been reported to be a further discriminator of islet microarchitecture and function in rodents (Huang et al. 2011) and man (Kaihoh et al. 1986), where fundamental differences appear to exist between the microarchitecture of small and large islets in the pancreatic tail (Grube & Weimann, 1985).

The occurrence of ‘primary islets’ and/or their failure to regress has been implicated in neonatal hyperinsulinaemic hypoglycaemia (Jaffe et al. 1980, 1982; Shermeta et al. 1980; Titlbach et al. 1985). There are intriguing similarities between the description of ‘primary islets’ and ‘perilobular giant islets’, and this draws us to speculate that the two are related. In addition, it has not escaped our notice that the hypertrophic islets described in adult-onset nesidioblastosis bear a striking resemblance to perilobular giant islets (Service et al. 2005; Rumilla et al. 2009; Cui et al. 2011).

Concluding remarks

Our study highlights that, in the pancreata of cattle, two distinct types of islets originate from the ductal tree and differ markedly in their histogenesis, organ integration, architecture, cellular composition, fate and, probably, function. The fact that two types of islets most likely exist in the fetal pancreata of other mammalian species merits further attention and must be considered if meaningful conclusions are to be drawn from molecular and biochemical assessments of islet formation and function during fetal life. Islet type-specific markers should be sought to clarify interspecies correlates between islet types. The existence of discrete populations of islets, and probably β-cells, should be taken into account where fetal pancreatic islets are isolated for in vitro pancreatic research (Redecker et al. 1992) and where pancreatic (Bonner-Weir et al. 2000) or bilary tissue (Nagaya et al. 2009) serves as a source of islet progenitor cells for further in vitro differentiation (Puri & Hebrok, 2010; Desgraz et al. 2011).


We most cordially thank Angela Ehrlich for giving us innumerable rides to the abattoir in Altenburg, Germany. We gratefully thank Dr Jana Pürzel and her meat hygiene and inspection team for the great help on the site. Special thanks go to our colleagues at the Institute of Anatomy in Leipzig for their insightful discussions, comments, and technical help. Paul Lochhead is funded by a fellowship from the Chief Scientist Office of the Scottish Government.

Author contributions

A.R. designed research, assembled the data, and drafted the manuscript. C.M., J.B., M.M.-S., M.S., and R.G. were involved in data acquisition, analysis, interpretation, and presentation. P.L. revised the drafted manuscript critically and thoroughly, and brought it to its present form. All authors read and approved the final version.