Early deficits in insulin secretion, beta cell mass and islet blood perfusion precede onset of autoimmune type 1 diabetes in BioBreeding rats

Aims/hypothesis Genetic studies show coupling of genes affecting beta cell function to type 1 diabetes, but hitherto no studies on whether beta cell dysfunction could precede insulitis and clinical onset of type 1 diabetes are available. Methods We used 40-day-old BioBreeding (BB) DRLyp/Lyp rats (a model of spontaneous autoimmune type 1 diabetes) and diabetes-resistant DRLyp/+ and DR+/+ littermates (controls) to investigate beta cell function in vivo, and insulin and glucagon secretion in vitro. Beta cell mass was assessed by optical projection tomography (OPT) and morphometry. Additionally, measurements of intra-islet blood flow were performed using microsphere injections. We also assessed immune cell infiltration, cytokine expression in islets (by immunohistochemistry and qPCR), as well as islet Glut2 expression and ATP/ADP ratio to determine effects on glucose uptake and metabolism in beta cells. Results DRLyp/Lyp rats were normoglycaemic and without traces of immune cell infiltrates. However, IVGTTs revealed a significant decrease in the acute insulin response to glucose compared with control rats (1685.3 ± 121.3 vs 633.3 ± 148.7; p < 0.0001). In agreement, insulin secretion was severely perturbed in isolated islets, and both first- and second-phase insulin release were lowered compared with control rats, while glucagon secretion was similar in both groups. Interestingly, after 5–7 days of culture of islets from DRLyp/Lyp rats in normal media, glucose-stimulated insulin secretion (GSIS) was improved; although, a significant decrease in GSIS was still evident compared with islets from control rats at this time (7393.9 ± 1593.7 vs 4416.8 ± 1230.5 pg islet−1 h−1; p < 0.0001). Compared with controls, OPT of whole pancreas from DRLyp/Lyp rats revealed significant reductions in medium (4.1 × 109 ± 9.5 × 107 vs 3.8 × 109 ± 5.8 × 107 μm3; p = 0.044) and small sized islets (1.6 × 109 ± 5.1 × 107 vs 1.4 × 109 ± 4.5 × 107 μm3; p = 0.035). Finally, we found lower intra-islet blood perfusion in vivo (113.1 ± 16.8 vs 76.9 ± 11.8 μl min−1 [g pancreas]−1; p = 0.023) and alterations in the beta cell ATP/ADP ratio in DRLyp/Lyp rats vs control rats. Conclusions/interpretation The present study identifies a deterioration of beta cell function and mass, and intra-islet blood flow that precedes insulitis and diabetes development in animals prone to autoimmune type 1 diabetes. These underlying changes in islet function may be previously unrecognised factors of importance in type 1 diabetes development. Electronic supplementary material The online version of this article (10.1007/s00125-017-4512-z) contains peer-reviewed but unedited supplementary material, which is available to authorised users.


Introduction
Type 1 diabetes is associated with the immune-mediated destruction of islet beta cells. Studies in human monozygotic twins, sharing identical genomes, demonstrate pairwise type 1 diabetes of 13-52%, suggesting that environmental and genetic causes may contribute similarly to the disease [1].
Research pertaining to the genetic contribution of type 1 diabetes have for the past decades focused on genetic loci implicated in regulation and selection of autoreactive T lymphocytes [2], although single nucleotide polymorphisms within the human insulin (INS) gene (mainly present in beta cells) remain one of the most important risk factors for the development of type 1 diabetes [3]. Recent studies have revealed that several candidate genes found in genome-wide association studies of type 1 diabetes susceptibility loci are expressed in beta cells and could thus influence beta cell function [4].
The BioBreeding (BB; LEW.1WR1) rat acts as a model of type 1 diabetes, whereby type 1 diabetes is suggested to originate from selective autoimmune destruction of beta cells [5]. As in humans, the major histocompatibility complex holds genetic factors that predict disease in this model [6,7]. This explains some, but not all, of the inherited predisposition to type 1 diabetes. In the inbred BB rat strain BBDRLyp/Lyp (herein referred to as DRLyp/Lyp), onset of type 1 diabetes is linked to lymphopaenia, which is caused by a frameshift mutation in the Gimap5 gene, while their littermates DRLyp/+ and DR+/+ are resistant to diabetes [8,9]. Loss of T cells because of lymphopaenia affects both CD4 + and CD8 + T cells, especially ART2.1 + T cells [5]. In fact, depletion of the ART2.1 + T cells in diabetes-resistant BB rats induces type 1 diabetes, suggesting that loss of regulatory T cells is associated with insulitis and type 1 diabetes [10].
Early changes in beta cell function and blood glucose have not been elucidated in DRLyp/Lyp rats, although local changes in beta cells in inbred DRLyp/Lyp are reflected by production of eotaxin (an eosinophil and mast cell recruiting factor) in islets at about 40 days of age, before insulitis, hyperglycaemia and type 1 diabetes [11,12]. However, positive staining of infiltrating monocytes remains to be shown at this age [11]. Additionally, islets from 40-day-old DRLyp/Lyp animals express lower levels of genes involved in the metabolism of reactive oxygen species (ROS) [13] and are more sensitive to changes in redox balance [14]. Over time, such an inherent sensitivity could contribute to accumulation of the ROS that diminish beta cell function, rendering cells more sensitive to immune cell attack.
Islet function is also dependent on functional islet vasculature and blood flow. In fact, inflammatory changes in vascular endothelial cells, characterised by increased expression of surface receptors, facilitate immune cell extravasation into the inflamed tissue [15]. Additionally, islet vasculature plays a critical role in maintaining oxygen and nutrient supply to the islets [16] and poor intra-islet blood flow is associated with changes in acute insulin response to glucose in vivo [17]. Interestingly, venular defects were observed in islets from BB (DP-BB/Wor) rats [18]. This, in combination with an underlying beta cell defect, could impair beta cell function and promote insulitis and beta cell destruction.
Currently, evidence of changes in beta cell function prior to onset of type 1 diabetes is limited. Therefore, we set out to explore whether insufficient beta cell function, or changes in beta cell mass and intra-islet blood flow, precede type 1 diabetes using the DRLyp/Lyp rat as a disease model.

Methods
Animals The BB rat was originally derived from a Canadian colony of outbred Wistar rats (originating from the Ottawa Health Research Institute, University of Ottawa, Ottawa, ON, Canada) that spontaneously develop hyperglycaemia and ketoacidosis, characteristics of clinical onset of type 1 diabetes. Heterozygous BB DRLyp/+ rats were used to obtain congenic DRLyp/Lyp rats as previously described [9,19]. Briefly, the Lyp region from diabetes-prone BB rats was introgressed onto the diabetes-resistant BB rat and kept in sibling breeding for more than 50 generations by heterozygous breeders to yield 25% DRLyp/Lyp, 25% DR+/+ and 50% DRLyp/+ rats. All DRLyp/Lyp rats developed diabetes after transferring the entire colony from University of Washington, Seattle to Lund University (including the Clinical Research Centre in Malmö, Sweden), in 2008. Animals were bred/kept in a pathogen-free environment at the Clinical Research Centre in Malmö, Sweden. They were housed at 21-23°C (12 h light/dark cycle) and fed ad libidum. All experiments were approved by the Animal Ethical Committee in Uppsala and Lund. All animals used in experiments were 40 days old unless otherwise stated.
Genotyping Tail snips were obtained from rat pups between 25-30 days of age. DNA was isolated and genotyped based on microsatellite analysis, as previously described [9,20].
Blood flow measurements and islet morphometry DRLyp/Lyp (n = 11, 4M/7F) and control (n = 15, 6M/9F) rats were anaesthetised (i.p. injection of thiobutabarbital sodium; 120 mg/kg; Inactin; Sigma Aldrich) and placed on a heating pad to maintain body temperature. The trachea was detached and a polyethylene catheter was inserted to secure free airways. Catheters were inserted into the right ascending aorta and the left femoral artery. A pressure transducer was connected to the ascending aorta catheter. A blood sample was taken for blood glucose measurement (Freestyle Lite; Abbott, Calameda, CA, USA). When blood pressure had stabilised (10-15 min), animals were injected with 1.5 × 10 5 microspheres (diameter: 10 μm) (E-Z Trac Ultraspheres; Stason Labs, Irwin, CA, USA) into the ascending aorta and blood was collected as described [21]. Animals were then euthanised and the pancreas and adrenal glands were dissected, weighed, cut in pieces and placed between object glasses. Object glasses containing pancreatic tissue were freeze-thawed to visualise islets [21]. The percentage of islet volume was determined by a pointcounting [22], and the number of microspheres in the exocrine and endocrine pancreas, adrenal glands and reference sample was counted in a bright and dark field illumination microscope.
Immunofluorescence was examined in an epi-fluorescence microscope (Olympus, BX60, Tokyo, Japan). By changing filters, double staining was used to determine the location of the different secondary antibodies in one sample. Images were captured with a digital camera (Nikon DS-2Mv, Tokyo, Japan).
Statistical analysis Data are expressed as mean ± SEM. IVGTTs, AUC and acute insulin response to glucose (AIR Glucose ) were calculated as described [6,29,30]. Mann-Whitney non-parametrical testing was employed in all experiments, except for analysis of islet size (OPT), blood flow measurements, 1 h batch experiments, insulin content, qPCR and ATP/ADP measurements, which were analysed with Student's t tests, and plasma insulin levels, which were assessed using a two-way ANOVA. Statistical analyses were performed using GraphPad Prism 6 software (GraphPad Software, La Jolla, CA, USA). p < 0.05 was considered to be statistically significant. All experiments were performed and analysed in a randomised and blinded fashion when possible. Outliers were identified using Grubbs test for outliers.

Results
Diagnosis of diabetes DRLyp/Lyp and control (DRLyp/+ and DR+/+) rats were followed by daily blood glucose measurements until diagnosis of type 1 diabetes (Fig. 1a). Cumulative incidence revealed that all DRLyp/Lyp rats had developed diabetes by 80 days of age (Fig. 1b). Mean age at onset of type 1 diabetes was 60 days ranging from 47 to 80 days (Fig. 1d). Female rats developed diabetes earlier than males ( Fig. 1c; p = 0.004).
Serum insulin prior to type 1 diabetes onset Basal insulin levels were evaluated in DRLyp/Lyp and control rats over time. Despite normoglycaemia prior to onset of type 1 diabetes, insulin levels were lower at all time points in DRLyp/ Lyp rats and failed to increase with age compared with control rats (Fig. 2a; p = 0.0004).
In vivo insulin release is perturbed in DRLyp/Lyp rats In vivo glucose homeostasis and beta cell function were assessed with an IVGTT in DRLyp/Lyp rats. DRLyp/Lyp rats remained glucose tolerant (Fig. 2b). No difference in glucose clearance between groups was observed, also shown as AUC for glucose (Fig. 2d). However, DRLyp/Lyp rats secreted less insulin during the initial time points of the IVGTT vs controls (Fig.  2c) which was further highlighted by a reduction in AUC for insulin in DRLyp/Lyp rats ( Fig. 2e; 19466.9 ± 1060.2 vs 14310.8 ± 1454.2 pmol/l × min; p = 0.04) and a decrease in the AIR Glucose (Fig. 2f; 1685.3 ± 121.3 vs 633.3 ± 148.7; p < 0.0001).
Insulin secretion is decreased in islets from DRLyp/Lyp rats To assess differences in insulin release (as evident by the IVGTT) between DRLyp/Lyp and control rats, we characterised the dynamics of insulin secretion in vitro using a perifusion setup. Islets from DRLyp/Lyp and control rats were first subjected to a low concentration of glucose (2.8 mmol/l) (Fig. 3a). Basal insulin secretion was similar between the groups. When challenging islets with a stimulatory concentration of glucose (16.7 mmol/l) during a 40 min period, the amount of insulin secreted by islets from DRLyp/Lyp rats was reduced. Control rats responded robustly to elevated glucose concentrations ( Fig. 3b; control vs DR Lyp/Lyp AUC: 398.2 ± 53.8 vs 206.1 ± 21.6 pmol/l × min; p = 0.002). When islets were further challenged with 35 mmol/l KCl and 16.7 mmol/l glucose for 12 min, islets from DRLyp/Lyp rats continued to secrete less insulin than those from control rats ( Fig. 3c; control vs DR Lyp/Lyp AUC: 171.5 ± 18.8 vs 123.9 ± 14.9 pmol/l × min; p = 0.02). Insulin content, however, was similar in islets from DRLyp/Lyp and control rats (Fig. 3d).
Comparable results to those obtained in perifused islets were observed when islets were exposed to low (2.8 mmol/l) and high (16. DRLyp/Lyp rats vs control rats was evident (Fig. 3e). Glucagon secretion was similar in islets from both groups when exposed to low and high glucose concentrations (ESM Fig. 1a).
Previous work suggests that removing islets from an inflammatory milieu can restore GSIS [31]. Therefore, we cultured islets from DRLyp/Lyp and control rats for 5-7 days. Insulin secretion was measured after exposure to low (2.8 mmol/l) and high (16.7 mmol/l) glucose concentrations in a 1 h static incubation. Overall insulin secretion was improved, both in DRLyp/Lyp and control rat islets, but a significant decrease in GSIS was still evident in islets from DRLyp/ Lyp rats vs controls ( Fig. 3f; 4416.8 ± 1230.5 vs 7393.9 ± 1593.7 pg islet −1 h −1 ; p < 0.0001).
Il1b, Ifng and Tnf-α expression in islets isolated from DRLyp/ Lyp rats Next we determined expression of cytokines in islets isolated from DRLyp/Lyp and control rats. RNA was extracted either immediately after isolation or after culturing islets for 5-7 days. Il1b was present at similar levels in islets just after isolation (ESM Fig. 1b). However, Tnf-α and Ifng were undetectable. When islets where cultured over a 5-7 day period, detectable levels of all cytokines were present (ESM Fig. 1c) but did not differ between groups.
Islet blood perfusion To determine if reduced insulin secretion in vivo was associated with microcirculatory changes [17,32], we measured islet blood perfusion. Mean arterial blood pressure was recorded in animals prior to blood flow measurements with no significant difference between the two groups (data not shown).
Small and medium sized islets are less common in the pancreas of DRLyp/Lyp rats To understand whether the observed perturbation in insulin secretion in vivo was accompanied by differences in beta cell mass, we performed OPT on the whole pancreas from DRLyp/Lyp and control rats. Overall, beta cell mass did not differ between groups (Fig. 5a). However, there was a reduction in small (1.4 × 10 9 ± 4.5 × 10 7 vs 1.6 × 10 9 ± 5.1 × 10 7 μm 3 ; p = 0.035) and medium sized islets (3.8 × 10 9 ± 5.8 × 10 7 vs 4.1 × 10 9 ± 9.5 × 10 7 μm 3 ; p = 0.044) in the DRLyp/Lyp rats vs control rats (Fig. 5b). Representative images from the OPT of splenic, duodenal and gastric pancreatic lobes from a heterozygote DRLyp/+ rat and a DRLyp/Lyp rat (Fig. 6) present size determination by colour coding. Islets were stained with insulin: red depicts large islets, yellow depicts medium sized islets and white depicts small islets. Additionally, we employed a morphometrical method to assess islet mass in our model [22]. We found no decrease in overall islet mass in the DRLyp/Lyp rats compared with controls (ESM Fig. 1d).
Islet morphology and CD3 + cells are similar in DRLyp/Lyp and control rats To determine changes in islet morphology in DRLyp/Lyp rats, we performed insulin and glucagon staining. Islets in pancreatic sections from both DRLyp/Lyp and control rats displayed normal islet architecture (core of beta cells surrounded by alpha cells ; Fig. 8a,c). To confirm previous findings that 40-day-old DRLyp/Lyp rats do not present immune cell infiltration, we performed staining using a CD3 + specific antibody combined with a nuclear DAPI. As expected, staining was sparse, but similar in DRLyp/Lyp and control animals (Fig. 8b,d).

Discussion
The present study demonstrates that GSIS is perturbed in the DRLyp/Lyp rat as compared with diabetes-resistant littermates. The secretory defect was accompanied by significant reductions in the number of medium and small sized islets, and reduced intra-islet blood flow. Notably, these isletspecific derangements were observed at 40 days of age before hyperglycaemia, insulitis and onset of type 1 diabetes. Type 1 diabetes is associated with the immune-mediated destruction of beta cells, resulting in insulin deficiency. Recent advances have highlighted genetic and functional changes within the beta cell as part of type 1 diabetes pathology [4,29], suggesting that beta cells may have an inherent sensitivity that possibly makes them susceptible to autoimmune attack. We observed a significant reduction in insulin secretion both in vivo and in vitro in isolated islets from DRLyp/Lyp rats. Indeed, a previous study showed that non-inbred BB rats (BB/ Hagedorn; a model where lymphopenia is not present) displayed diminished release of insulin during stimulation with 20 mmol/l glucose in perfused whole pancreas at 50 days of age (before onset of type 1 diabetes) [35]. Similar observations have been made in islets from NOD mice, where insulin secretion immediately after isolation was perturbed (due to insulitis). However, culture of islets from NOD mice over a 5-7 day period improved insulin secretion significantly [31]. Indeed, islets from DRLyp/Lyp rats displayed an improved response to glucose after a culturing period; however, a secretory defect was still evident. Similarly, islets removed from people with new-onset type 1 diabetes show improved GSIS after culture [36]. It is noteworthy, however, that GSIS could not be fully restored in all individuals. A major difference between those studies and ours is that insulitis is not present in 40-day-old DRLyp/Lyp rats. Islets from 40-day-old DRLyp/ Lyp rats show reduced expression of the complement inhibitor protein CD59. CD59 is pivotal for normal beta cell exocytosis [37], suggesting that beta cell exocytosis is compromised in DRLyp/Lyp rats. This corresponds to our perifusion data, where islets from DRLyp/Lyp rats display an improved response to 35 mmol/l KCl, suggesting that insulin is not lost, rather that exocytosis is compromised. A previous study highlighted similar findings where non-metabolic secretagogues elicit insulin release in prediabetic conditions and in type 1 diabetes [38]. Additionally, insulin content is not altered in isolated islets from 40-day-old DRLyp/Lyp rats, which further supports this notion.
In prediabetic NOD mice, beta cell dysfunction is suggested to occur as a consequence of early immune cell infiltration and activation of inflammatory cascades [39]. However, the DRLyp/Lyp rats do not display any major infiltration by mononuclear cells until a few days prior to clinical onset of type 1 diabetes [13]. We confirmed this, and islets from DRLyp/Lyp rats did not show increased infiltration of CD3 + cells in pancreatic sections. Moreover, we were unable to detect elevated expression of Il1b, Ifng and Tnf-α in islets from DRLyp/Lyp rats; cytokines that could be indicative of early immune processes within the islets [40,41].
Beta cell mass is tightly regulated during fetal life, a time point representing a critical window when the appropriate number of beta cells are set in place [42]. A potential weakness in the present study is that we have not investigated neonatal beta cell growth and postnatal expansion of beta cells in our model. It may very well be that DRLyp/Lyp rats are born with a reduced number of beta cells, or fail to expand their beta cell mass during postnatal stages. We observe significant reductions in small and medium sized islets in DRLyp/Lyp compared with control rats, albeit overall islet mass was not changed. A previous study shows that smaller islets contain more insulin per islet volume in situ and secrete insulin more efficiently in vitro [43]. In addition, large islets may be subjected to both hyperplasia and hypoxia [44], resulting in impaired beta cell function. Thus, loss of small and medium sized islets may very well impact insulin secretion. Additionally, OPT has an advantage over more conventional methods, since it can give information on spatial position and volume of individual insulin-expressing islets throughout the pancreas, with high resolution and the opportunity to categorise islets by size [23].
Another important factor influencing beta cell function is nutritional blood status and islet blood flow. This could be considered as the main avenue by which beta cells are kept informed of the body's nutritional state [45]. We observed reduced intra-islet blood flow in DRLyp/Lyp rats. The importance of this finding for development of type 1 diabetes remains to be determined, but in general lower blood perfusion in islets could compromise beta cell function through hypoxia or limited dispersal of insulin into the systemic circulation [17,32]. Moreover, decreased blood flow decreases shear stress, which increases the tendency for leucocyte adhesion in venules even in the absence of additional activators [46]. This could promote islet immune cell infiltration. Indeed, a previous study showed a venular defect in a related rat strain (BB/Wor rat), which supports our findings [18]. Currently, any relationship between blood flow changes and lymphopenia in DRLyp/Lyp rats remains unknown. High basal islet blood flow in diabetes-resistant (and/or wild-type) animals is to a large extent mediated by locally generated nitric oxide from endothelial cells and inhibiting this system decreases blood perfusion [47]. It is noteworthy that studies on islet endothelial cells from young normoglycaemic diabetesprone and diabetes-resistant BB rats have shown that diabetes-prone rats exhibit considerably lower endothelial cell nitric oxide synthase activity than diabetes-resistant rats [48].
Insulin release is to a large extent dependent on mitochondrial metabolism of glucose and the resulting increase in intracellular ratio of ATP/ADP [33]. Glucose uptake into beta cells is the initial step in GSIS. In rodents this is mediated by GLUT2 [49]. Mice lacking Glut2 lose the first phase of insulin secretion [34]. Thus, both the ATP/ADP ratio and Glut2 expression could influence GSIS in DRLyp/Lyp rats. We observed no changes in Glut2 expression. Intriguingly, however, ATP/ADP levels were elevated in islets from DRLyp/Lyp rats, which could signify a compensatory mechanism as mitochondria are striving to maintain a sufficient ATP/ADP ratio and coupling factors to ensure sufficient insulin release. It may also suggest that the secretory deficiency lies distal of ATP generation (i.e. depolarisation of the plasma membrane/Ca 2+ influx or exocytosis). Clearly, more intense research efforts are required in this area.
In summary, our results show that DRLyp/Lyp rats display a secretory defect prior to autoimmune onset of type 1 diabetes. This is manifested by perturbations in insulin secretion in vivo and in vitro, partial loss of beta cell mass and reduced intra-islet blood flow; all of which are factors that influence beta cell function. These changes may be of importance for the development of type 1 diabetes.