Regenerating 1 and 3b Gene Expression in the Pancreas of Type 2 Diabetic Goto-Kakizaki (GK) Rats

Regenerating (REG) proteins are associated with islet development, β-cell damage, diabetes and pancreatitis. Particularly, REG-1 and REG-3-beta are involved in cell growth/survival and/or inflammation and the Reg1 promoter contains interleukin-6 (IL-6)-responsive elements. We showed by transcriptome analysis that islets of Goto-Kakizaki (GK) rats, a model of spontaneous type 2 diabetes, overexpress Reg1, 3α, 3β and 3γ, vs Wistar islets. Goto-Kakizaki rat islets also exhibit increased cytokine/chemokine expression/release, particularly IL-6. Here we analyzed Reg1 and Reg3β expression and REG-1 immuno-localization in the GK rat pancreas in relationship with inflammation. Isolated pancreatic islets and acinar tissue from male adult Wistar and diabetic GK rats were used for quantitative RT-PCR analysis. REG-1 immunohistochemistry was performed on paraffin sections with a monoclonal anti-rat REG-1 antibody. Islet cytokine/chemokine release was measured after 48 h-culture. Islet macrophage-positive area was quantified on cryostat sections using anti-CD68 and major histocompatibility complex (MHC) class II antibodies. Pancreatic exocrine-to-endocrine Reg1 and Reg3β mRNA ratios were markedly increased in Wistar vs GK rats. Conversely, both genes were upregulated in isolated GK rat islets. These findings were unexpected, because Reg genes are expressed in the pancreatic acinar tissue. However, we observed REG-1 protein labeling in acinar peri-ductal tissue close to islets and around large, often disorganized, GK rat islets, which may retain acinar cells due to their irregular shape. These large islets also showed peri-islet macrophage infiltration and increased release of various cytokines/chemokines, particularly IL-6. Thus, IL-6 might potentially trigger acinar REG-1 expression and secretion in the vicinity of large diabetic GK rat islets. This increased acinar REG-1 expression might reflect an adaptive though unsuccessful response to deleterious microenvironment.


Introduction
The pathological roles of inflammatory-mediated mechanisms in type 2 diabetes are emerging from clinical and experimental studies [1]. Experiments in animal models can contribute to new knowledge of the molecular partners involved. We previously demonstrated pancreatic islet inflammation in the GK rat model of spontaneously occurring type 2 diabetes, as well as in other animal models of spontaneous and/or induced type 2 diabetes and in patients [2][3][4]. Indices of islet inflammation in GK rats were first demonstrated by a transcriptome (Affymetrix) analysis and immunohistochemistry. Upregulated expression of numerous inflammatory genes and increased numbers of macrophages were observed in 4-month-old GK rats, i.e., after 3 months of chronic mild hyperglycemia, vs age-matched Wistar controls [2][3][4]. Islets of younger (2-month-old) diabetic GK rats also exhibited high CCL2 (CC-chemokine ligand-2 or MCP-1, monocyte chemoattractant protein), CCL3 (CC-chemokine ligand-3 or MIP-1a, macrophage inflammatory protein-1a), CXCL-1 (CXC-chemokine ligand-1 or chemokine GRO1/KC, a murine interleukin-8 (IL-8) analog) and IL-6 expression and release. Treating in vitro GK rat islets or in vivo GK rats with the IL-1 receptor antagonist lowered islet mRNA levels and release of these cytokines/chemokines, and ameliorated in vivo glucose homeostasis parameters [4]. Therefore hyperglycemia/hyperlipidemia-induced islet IL-1 activity promotes cytokine/ chemokine expression, leading to recruitment of innate inflammatory cells, which alter both functional b-cell mass and insulin sensitivity [1][2][3][4]. However, concomitant increased expression of endogenous IL-1 receptor antagonist and IL-1 mRNA levels in GK rat islets suggests that GK rats are, to some extent, able to mount anti-inflammatory defense mechanisms [4].
We reported that 4-month-old GK rat islets over-express Reg1, 3a, 3b and 3c vs Wistar islets [2]. The Reg gene family consists of five clustered genes mapped to a locus linked to altered insulin secretion [5] and encoding proteins associated with pancreatitis, diabetes, b-cell damage, b-cell replication and islet neogenesis [6][7][8][9][10]. In isolated rat islets, Reg1 mRNA levels are increased by glucose, amino acids, fetal serum or growth factors like insulin, growth hormone and platelet-derived growth factor [6]. In the rat acinar cell line AR4J2, Reg1 mRNA levels are significantly increased by IL-6, interferon-c (IFN-c) or tumor necrosis factora (TNF-a), but decreased by dexamethasone [6,11]. In the same cell line, Reg3b gene expression is induced by TNF-a or IFN-c and this effect is inhibited by dexamethasone [6]. Thus, several agents involved in glucose homeostasis (glucose, insulin, glucocorticoids and cytokines) are able to modulate Reg1 and Reg3b expression in different pancreatic cell types. Most of the Reg gene family products were first identified in the field of exocrine inflammatory diseases, as genes encoding 'pancreatitis associated proteins' (PAP) [6]. Most of the REG/PAP family proteins are secretory stress proteins, which are able to exert anti-inflammatory effects [6,12,13]. For example, antisense inhibition of all three PAP isoforms correlated with pancreatitis worsening [14]. In vitro also, PAP-1 inhibited rat macrophage activation by TNF-a [15].
Since REG-1 and REG-3-beta proteins are involved in cell growth/survival control and also exert anti-inflammatory effects [6,13], we analyzed their pancreatic expression in relationship with inflammation in the GK rat. This model is characterized from early life by reduced b-cell mass, not reflecting decreased bcell proliferation or increased b-cell death, but rather altered islet neogenesis [16]. We demonstrate the coordinated regulation of mediators of the inflammatory response and Reg gene expression at the GK rat islet level.

Ethics Statement
The study was performed in strict accordance with accepted standards of animal care established by the Centre National de la Recherche Scientifique. All the animal experimentation presented in the present manuscript has been performed at the animal facility of the UFR de Biologie de l'Université PARIS 7/Denis DIDEROT-Tour 33, 2, place Jussieu, 75251 Paris, France, agreement number A75-05-05. All animal experiments were done by investigators with agreement for animal experimentation (SC, nu B75-1571; PS, nu 75-1526; FHD, nu75-1530). Ethics committee approval was not necessary at the time of animal experiment. Every effort was made to minimize suffering.

Animals
All animal experiments were conducted on diabetic adult male GK rats and sex-and age-matched Wistar controls from the local colony of Paris-Diderot (ex-Paris VII) University (Paris, France). GK rats were bred together with Wistar control rats from which the GK strain derived after backcrossing of animals selected at the upper limit of normal distribution for glucose tolerance. Characteristics of GK rats have been previously described [17]. Rats were killed by decapitation and blood and pancreata collected for measurement of metabolic parameters, islet isolation for quantitative RT-PCR analysis and islet culture for measurement of cytokine/chemokine release, or pancreas immunohistochemistry.
For macrophage immunohistochemical study, 2-month-old Wistar and GK rat pancreas cryosections were incubated with mouse anti-rat CD68 or anti-Ia (MHC class II) antibodies (1:100 and 1:300, respectively, Serotec, Colmar, France) as previously described [2]. Staining was visualized using appropriate peroxidase-coupled secondary antibodies (Caltag, Cergy, France) and subsequent incubation with 3-amino-9-ethylcarbazole (Sigma). For each series of pancreas sections, one slide was stained only with the second antibody as a control for endogenous peroxidase activity and nonspecific antibody binding. Immunostained areas were visualized and quantified using an Olympus BX40 microscope [18].

Islet Isolation and mRNA Analysis
Islets were isolated by pancreas digestion with collagenase (Boerhinger Mannheim, Mannheim, Germany), according to standard procedure. Subsequently, islets were purified using a continuous Histopaque (Sigma) gradient. Thereafter, all islets were collected and hand-picked under a stereomicroscope. Islets and acinar (islet-depleted) tissue was immediately frozen at 280uC. Total islet and acinar tissue RNA was extracted according to standard protocols using the RNeasy minikit from Qiagen (RNeasy Kit 74104, Hombrechtikon, Switzerland). Islet RNA yield was 26 mg/250 Wistar rat islets and 14 mg/140 GK rat islets. RNA was amplified (two cycles) and biotin-labeled cRNA probes were synthesized using Bioarray High Yield RNA transcript labeling reagents from Enzo Diagnostics (Microsynth, Balgach, Switzerland). Differential gene expression was confirmed by quantitative RT-PCR (RT: SuperScript II Reverse Transcriptase 11752-050, Invitrogen Life Sciences, Basel, Switzerland and PCR: qPCR Cyber Green kit: Eurogenetec RT-SN10-05, Seraing, Belgium). Quantification was performed after normalization to glyceraldehyde-3-phosphate dehydrogenase (Gapdh). Primer sequences were as follows:

Islet culture and Cytokine and Chemokine Determination
Adult Wistar and GK rat islets were plated on extracellular matrix-coated 3-cm dishes for 48 h (20 islets/2 ml of culture medium consisting of RPMI medium 1640 containing 11 mM glucose, 100 units/ml penicillin, 100 mg/ml streptomycin, 40 mg/ ml gentamycin, and 10% FCS), as previously described [4]. Then, conditioned media were removed and assayed for IL-6, CCL2, CCL3, CXCL1, using rat Luminex kits (Millipore). Released cytokines/chemokines were normalized to total islet protein content. Cytokines and chemokines were also measured in Wistar and GK rat serum samples.

Statistics
Data are presented as means 6 SEM. Statistical analyses used the Student's t-test for unpaired data. Correlations were assessed with non-parametric Spearman's rank correlation test, using Statview software. Significance was defined as p,0.05.

Pancreatic Endocrine vs Exocrine Reg1 and Reg3b Gene Expression
In a previous Affymetrix-based transcriptome study of 4-monthold GK vs Wistar islets, we found 4 Reg genes to be overexpressed: Reg1, Reg3a, Reg3b and Reg3c (by 12.4, 11.1, 45.5 and 9 fold increase, respectively) [2], whereas Reg IV [6,19] was not detected in GK rat islets due to the absence of the gene on the Affymetrix RG-U34A microarray. We selected Reg1, Reg3b and Amya1 (control for exocrine contamination) for further analysis by quantitative RT-PCR in pancreatic exocrine (acini and ducts) and endocrine (islets) tissues. As shown in Fig. 1A, pancreatic exocrine-to-endocrine ratios of Reg1 and Reg3b mRNA expression were about10 times higher in control Wistar than in diabetic GK rats. Conversely, GK vs Wistar rat islet Reg1 and Reg3b mRNA ratios ( Fig. 1B) were significantly increased, by 11.261.4 and 77.9616.3 fold, respectively (p,0.005). Because Reg1 and Reg3b are exclusively expressed in acinar tissue [6], we checked whether possible contamination of GK islets by acinar peri-islet tissue may explain elevated Reg expression: GK rat islets exhibited a statistically not significant 3.461.3-fold increase (n = 5) in Amya1 expression vs Wistar control islets (data not shown). Although not reaching significance, these data indicate some variability in GK islet purity among experiments, probably partially linked to the peculiar shape of these islets, as developed below.

Pancreas Immunohistochemistry for REG-1 Proteins
In order to compare REG protein localization in 4-month-old Wistar and GK pancreas, we used a monoclonal antibody against rat REG-1. Labeling of islet b-cells (insulin) and non b-cells (glucagon, somatostatin and pancreatic polypeptide), acinar tissue (a-amylase) and REG-1 was performed on serial pancreatic sections. As shown in Fig. 2A-B vs Fig. 2H-I and in Supporting Information ( Figure S1), endocrine hormone labeling highlights in GK pancreas the disorganized band non b-cell pattern induced by progressive islet fibrosis [2]. The evolution of GK vs Wistar islets at 1, 2, 3 and 4 months of age is also illustrated, in particular for: vascularization (anti-VWF, Figure S2) and/or fibrosis (antifibronectin, Figure S3) labeling and innate immune cell infiltration (anti-CD68 ( Figure S4) and anti-D53 ( Figure S5)). Comparison of a-amylase and REG-1 labeling in Wistar pancreas showed few slightly stained REG-1 + cells in the peri-islet exocrine tissue ( Fig. 2D vs Fig. 2C and Figure S1). By contrast, numerous large and markedly stained REG-1 + acinar cells were observed around GK islets (Fig. 2K vs Fig. 2J and Figure S1) and ducts close to these islets (Fig. 2N). Notably, REG-1 labeling could be observed inside the duct cavity. In addition, 2-month-old GK rats are characterized by greater islet size heterogeneity than those of Wistar rats and exhibit very small and large islets ( Figure S6). The latter (Fig. 2N) appeared to exhibit more peri-islet REG-1 + labeling than small ones (Fig. 2M).

Islet Release and Circulating Levels of Cytokines/ chemokines
Islets from 2-month-old diabetic GK rats expressed and released more cytokines/chemokines, particularly IL-6, than those of control Wistar islets [4]. Because rat Reg1 promoter contains responsive elements for IL-6 [6], we compared islet cytokine/ chemokine release by small and large islets of 3-month-old rats (Fig. 3A). When data were expressed as mean of absolute values of cytokine/chemokine release, normalized to total islet protein recovered in the corresponding experiment ( Fig. 3B-E), large GK islets appear to release more IL-6, CCL2, CCL3 and CXCL1 than large Wistar islets, but the effect was significant for CCL3 (MIP-1a) only. However, these absolute values may vary from one experiment to another and we thus expressed the GK values relative to control Wistar values in each experiment (Fig. 3F): in this case, IL-6 release from large GK islets was also found to be significantly increased (x 4.760.9 for 3 different islet isolations, p, 0.02). Concerning circulating cytokines/chemokines, we already reported that while normoglycemic 1-week-old GK rats show higher cytokine/chemokine levels than age-matched Wistar controls [20], these levels dropped to similar low values at 2 months of age [4], as well as at 3 and 4 months of age (Table 1). The REG-1 protein was particularly abundant in the peri-islet area of large GK islets, which produced elevated levels of cytokines. Inflammatory macrophages exhibit the same localization from 2 months of age onwards [2], as illustrated for CD68 ( Fig. 4A) and MHC class II (Fig. 4B) and for CD68 ( Figure S4) and CD53 ( Figure S5). As particularly shown online, the peri-islet immune infiltration was not systematically continuous and regular. Thus, we looked for a possible relationship between islet size and extent of inflammatory cell infiltration. There was a significant correlation between both CD68 + and MHC II + surface areas and islet surface area in 2-month-old GK pancreas sections, r = 0.573, p,0.025 and r = 0.950, p = 0.003, respectively (Fig. 4C, D).

Discussion
Islets of diabetic GK rats are known to co-coordinately express 4 Reg genes at time of decreased b-cell replication, inflammation and fibrosis [2]. Here, we show by quantitative RT-PCR that the abundance of transcripts encoding REG-1 (also called pancreatic stone protein-1 (PSP) or lithostathine) and REG-3b (also called PAP, PAP-I, hepatocarcinoma-intestine-pancreas (HIP), REG-2 or peptide 23) appeared to be markedly and unexpectedly upregulated in GK rat islets vs control islets. We also demonstrate that acinar peri-islet REG-1 labeling characterizes large, often disorganized GK rat islets, which exhibit concomitant peri-islet macrophage infiltration and higher release of cytokine/chemokine CCL3 (MIP-1a) and particularly IL-6, a typical REG/Reg gene family inducer [6,21,22].
In agreement with the acinar localization of pancreatic Reg mRNA [12,23], islet-depleted exocrine-to-endocrine tissue ratio was elevated for both Reg1 and Reg3b genes in 4-month-old Wistar controls. These ratios were markedly lower in age-matched diabetic GK rats, probably because Reg1 and Reg3b were strongly overexpressed in diabetic GK rat islets. At least two non-exclusive reasons may account for Reg overexpression in diabetic GK rat islets. First, as mentioned above, despite careful islet isolation some Reg-containing acinar cells could be trapped and retained within GK islets due to their irregular fibrotic shape [2]. Second, overexpression of Reg1 and Reg3b in GK islets may arise from isolation stress procedure, as previously described for other genes [24], and/or deleterious type 2 diabetes-related islet microenvironment. Regarding the latter case, REG expression has been measured by microarray analysis in b-cell enriched tissue from type 2 diabetic vs non diabetic cadavers using laser capture microdissection in order to avoid islet isolation artifacts [25]. This study showed REG1a, REG1b, REG3a and REG3c upregulation without b-cell regeneration but with inflammatory gene upregulation, as observed in GK rat islets.
In the pancreas, Reg1 and Reg3 genes encode proteins that are normally expressed at very low levels. The expression of these proteins is strongly activated in acinar cells in response to many injuries, such as acute and chronic pancreatitis, hypoxia, toxins and diabetes among others [7,8,12,26,27]. In the Reg family genes, there is clear isoforms specificity in term of their response and biological effect [28]. It is well recognized that REG-1 protein plays a beneficial role in b-cell proliferation, survival and islet neogenesis [6]. In the surgical (90% depancreatized) rat model of diabetes, REG-1 administration improved insulin secretion [9]. In mice, overexpression or depletion of the Reg1 gene had beneficial and deleterious effects on glucose homeostasis, respectively [6,29]. During pancreatitis, the highest PSP/REG-1 expression was observed in acini surrounded by inflammatory cells [30]. A comparable phenomenon exists around diabetic GK rat islets and associated-ducts, which are in close contact with CD68 + and MHC class II + macrophages, neutrophils immature myeloid cells and cells immunoreactive for reactive oxygen species and lipid peroxidation [2,3,31]. A peri-islet REG-1 distribution has also been observed: 1) in Wistar rats (as single cells or groups of cells) [32], but it was very faint in our study; 2) in type 1 nonobese diabetic (NOD) mice and NOD-rat insulin promoter-1-human IFNb mice [33].
In GK rat pancreas, inflammatory cell levels increase from weaning onwards (1 month of age), reach a plateau between 2 and 3 months and decline thereafter to finally disappear almost completely around/in fibrotic islets [3]. Diabetic 2-month-old GK   rat islets also exhibited elevated expression and/or release of several cytokines/chemokines, which can be produced by all GK islet cell types, in particular macrophages, endothelial cells and endocrine islet cells [1,4,18]. Here, we showed that large GK islets release more cytokines/chemokines than their Wistar counterparts, in particular CCL3 (MIP-1a) and IL-6. Reg1 gene promoters contain IL-6-responsive elements [6,22]. Therefore, increased IL-6 release by large GK rat islets might be responsible for the marked acinar REG-1 expression in their vicinity. Because some proteins of the REG/PAP family exert anti-inflammatory effects [6,14,15], acinar peri-islet REG-1 expression in diabetic GK rats might represent an expected defense reaction against islet inflammation, as already observed concerning the upregulation of: 1) IL-1 receptor antagonist mRNA [4]; and 2) antioxidant/ cAMP/anti-apoptosis pathway, which confers b-cell protection by decreasing reactive oxygen species and apoptosis [31,34]. In addition, hypoxia stress was recently shown to stimulate the proliferation of several b-cell lines, with concomitant induction of IL-6, Reg family genes and hepatocyte growth factor (HGF) gene [22]. The use of siRNAs against rat Reg family genes attenuated hypoxia-induced b-cell proliferation. The authors conclude that hypoxia, by stimulating IL-6 expression, leads to overexpression of Reg family genes (stimulating b-cell proliferation) and HGF gene (inhibiting b-cell apoptosis), and consequently hypoxia increased b-cell number. Finally, IL-6 might indirectly act on b-cell adaption/renewal, via the stimulation of pancreatic a-cell glucagon products [35]. Acinar peri-islet localization of REG3b protein has been described in mice [23,28]. Beta-cell specific overexpression of Reg3b driven by rat insulin promoter-1 protected mice from hyperglycemia in the streptozotocin-induced model of type 1 diabetes [36]. In this study, 2 genes encoding ''protective'' molecules were upregulated: the acute responsive nuclear protein, transcriptional regulator, 1 (NUPR-1/P8) and osteopontin (or SPP-1, secreted phosphoprotein-1. A similar concomitant overexpression of genes encoding REG-3beta, NUPR-1 and osteopontin was observed in type 2 diabetic GK rat islets [2]. Two other less investigated genes of the Reg3 family (Reg3a and Reg3c), also present in rat acinar tissue [6,28], were found to be upregulated in our GK rat islet transcriptome study [2]. Next generation sequencing of the GK and Wistar-Kyoto genomes identified non synonymous polymorphisms in the genes encoding REG-3-alpha (A103C, T35P in exon 3 and C275A, T92K in exon 4) and REG-3-gamma (A478T, N160Y in exon 6), when compared to the rat reference genome (Brown Norway) (Gauguier, unpublished). Since DNA variants are conserved in both GK and Wistar-Kyoto rats, it is unlikely that the amino acid changes account for altered function of these proteins in GK islets. However, GK-specific polymorphisms detected in promoter, intronic and intergenic regions of the genes may explain their differential islet expression. Of note, REG-3c/PAPIII is a macrophage chemoattractant involved in nerve regeneration [37], like REG-1a protein, which plays a role in neurite outgrowth [38]. Thus, Reg proteins might be part of the relationship between innervation and islet neogenesis [39], particularly at work during diabetes pathophysiology.
To conclude, REG/PAP proteins might exert at least two roles during pancreas/islet injury of pancreatitis/diabetes: 1) downregulation of the inflammatory reaction; 2) b-cell functional rescue/ replacement. The latter phenomenon involves several mecha- nisms, including b-cell proliferation, ductal islet neogenesis and possibly transdifferentiation from acinar tissue [40]. Overexpression of REG/PAP proteins might result from increased IL-6, but also TNFa, IFNc and IL-8 secretion [6,11,23]. In this context, ductal cells like macrophages are known to produce 2 angiogenic factors, which are crucial in b-cell development, IL-8 and vascular endothelial growth factor (VEGF) [41][42][43]. GK rats present deficient peripheral vascularization, linked to VEGF anomalies [44,45] and altered islet vascularization and neogenesis from early life [16,20]. All these anomalies, together with the polymorphisms identified in Reg3a and/or Reg3c might contribute to maladaptive GK rat b-cell rescue/replacement. Finally, several clinical data may be related to the scenario, in particular: 1) the fact that PSP/ Reg1A has been suggested as a potential serum marker to detect increased b-cell apoptosis, or its therapeutic response, thus might assist the classification for maturity onset diabetes in the young (MODY) [46]; 2) the presence of REG antibodies in both human and/or rodent type 1 and 2 diabetes, which may represent a maladapted attempt for islet neogenesis [47,48]. Therefore, a better understanding of REG/PAP protein involvement in b-cell physiology and inflammation might lead to new therapeutic tools for type 2 diabetes. Figure S1 Immunohistochemistry for endocrine and exocrine pancreatic cells in 4-month-old Wistar and Goto-Kakizaki (GK) rat pancreas. Additional examples of immunohistochemical data for the protein encoded by regenerating gene-1 (REG-1) and endocrine and exocrine pancreatic cells in 4-month-old control Wistar and diabetic GK male rat pancreas. Serial staining (brown) for: insulin (b-cell marker), glucagon+somatostatin (SS)+pancreatic polypeptide (PP) cocktail (non-b cell markers), REG-1 and aamylase (acinar cell marker). For REG-1 labeling, we used the monoclonal anti-rat REG-1 antibody from Hiroshi Okamoto (Japan). For antibody dilutions and REG-1 negative pancreas section controls, see methods and Fig. 2 of the article, respectively. Endocrine hormone labeling of GK pancreas highlights the disorganized band non b-cell pattern induced by progressive islet fibrosis, as illustrated in the following supplementary figures. While few slightly REG-1 + cells are usually present in the peri-islet exocrine tissue of Wistar pancreas, more numerous, large and markedly stained REG-1 + acinar cells are observed around GK islets after 3 months of hyperglycemia. (TIF) Figure S2 Evolution of pancreatic islet vascularization in Wistar and GK rats as a function of age. Two factors known to be produced by endothelial cells, von Willebrand factor (VWF, this figure) and fibronectin ( Figure S3), show the normal organization of islet vascularization (brown staining) in 1, 2, 3 and 4-month-old Wistar controls and its progressive disorganization in age-matched diabetic GK rats. While VWF + islet vascularization appears to be similar at 1 month of age (around weaning and onset of hyperglycemia), thereafter GK islet VWF + vascularization becomes hypertrophied, as illustrated here at 2 and 3 months of age. As previously published [1], VWF and fibronectin lesions progress similarly before islet invasion by fibronectin and other extracellular matrix proteins, as shown here at 4 months of age and in Figure  S3. Islet fibrosis leads to endocrine-cell disappearance. Rabbit anti-human VWF (DakoCytomation), dilution (1:100) [1]. The bordure of islets is defined by the yellow dashed line. Ducts are indicated by black arrows. (TIF) Figure S3 Progression of pancreatic islet fibrosis in Wistar and GK rats as a function of age. In Wistar rats, fibronectin labeling (brown) offers the same pattern of islet vascularization from 1 to 3 months of age. In the GK rat pancreas, fibronectin labeling is first limited to islet endothelial cells as for Wistar rats. Then, islets are invaded by fibronectin and other extracellular matrix proteins with progressive disappearance of endocrine cells. Rabbit anti-rat fibronectin (Novotec), dilution (1:40) [1]. The bordure of islets is defined by the yellow dashed line. Ducts are indicated by black arrows. (TIF) Figure S4 Progression of CD68 inflammatory cell infiltration in the pancreas of Wistar and GK rats as a function of age. CD68 labeling (brown) shows the presence of a few positive cells in Wistar rat pancreas, particularly in the vicinity of islets and/or ducts, and also dispersed in the exocrine tissue. In diabetic GK rats, these CD68 + cells are more numerous, particularly at 2 and 3 months of age, and are mainly located in peri-islet and peri-ductal areas, as previously published [1][2][3][4]. Mouse anti-rat CD68 (Serotec), dilution (1:100) [1]. The bordure of islets is defined by the yellow dashed line. Ducts are indicated by black arrows. (TIF) Figure S5 Progression of islet CD53 + cell infiltration in Wistar controls and GK rats as a function of age. The gene CD53 codes for cluster of differentiation 53, a broadly expressed leukocyte surface antigen [5]. CD53 is known to complex with integrins and cellular components involved in cell-cell and cell-matrix interactions and it plays a substantial role during inflammation [6]. In Wistar control pancreas, some CD53 + cells (brown) may be present in the peri-islet area or, sometimes, form a patch of cells at the islet-ductal junction. In the GK rat pancreas, the CD53 + infiltration around islets and ducts may be particularly large, discontinuous and irregular, as illustrated here at 2 and 3 months of age. Mouse anti-rat CD53, dilution (1/30): (Serotec) [1]. The bordure of islets is defined by the yellow dashed line. Ducts are indicated by black arrows. (TIF) Figure S6 Presence of small and large islets in 2-month-old GK rats, e.g. after 1 month of hyperglycemia, but not in age-matched Wistar controls. Data are taken from an experiment aimed at measuring the number of CD68 + macrophages per islet in both groups of rats [2]. Pancreas sections were selected from 9 different animals in each group and islet were classified as a function of increasing surface. Because islet size increases with differentiation, it may be suggested that the presence of small islets in diabetic GK rats reflects an attempt for neogenesis. By contrast, large islets may correspond to fibrotic islets, as shown in Figure S3 (fibronectin labeling). Green rectangles are indicative of the groups of small and large islets used for in vitro cyto/chemokine measurement in Wistar and GK rats. Finally, 2 classes of medium islets appear to exist in the GK pancreas. The first class (dotted arrow) shows a lower mean islet surface than the second class but no difference between Wistar and GK rats, by contrast to the second class (solid arrow), where the mean islet surface value is larger in GK than in Wistar rat pancreas sections. The second class probably reflects the progression of GK rat islets to fibrosis. (TIF)