Exposure to bisphenol A, but not phthalates, increases spontaneous diabetes type 1 development in NOD mice

2.1. Mice and exposure conditions

120 female (randomized into four groups) and 60 male non diabetic NOD/ShiLtJ mice from Jackson Laboratory (Maine, USA) were used for breeding at 8 and 10 weeks of age, respectively. The exposure to BPA and/or phthalates via drinking water started at the time of mating of the mice and continued throughout the life time of the female offspring. The 4 exposure groups included: (1) negative control (water only), (2) phthalates: DEHP 1 mg/l, DBP 0.2 mg/l, BBP 10 mg/l and DiBP 20 mg/l, (3) BPA 1 mg/l, and (4) BPA 1 mg/l + DEHP 1 mg/l, DBP 0.2 mg/l, BBP 10 mg/l and DiBP 20 mg/l. The exposure level of each chemical was chosen at a dose corresponding to 3 × tolerable daily intake (TDI) for the respective chemical, since BPA at this exposure level has previously been shown to affect T1DM development in NOD mice [36].

Only female offspring were selected at the time of weaning, since insulitis and diabetes development is most prevalent in female mice [37]. The endocrine disruptors (EDs) were dissolved in deionized autoclaved water heated to 60 °C. Controls received similar water without EDs. BPA-free water bottles, made of 100% polyethylene terephthalate (PET) that does not require the use of phthalates or other softening additives (Innovive, San Diego, USA) were used and the water was changed once every week. Also cages were BPA-free and with minimal leakage of phthalates (100% PET polyethylene terephthalate plastic). The diet contained minimal levels of phytoestrogens (2919X, Harlan Laboratories, Indianapolis, USA) and the mice had free access to feed and water. The mice were exposed to a 12-h light/12-h dark cycle and 35–75% humidity. To keep the dams as the statistical unit for all data, female siblings from each dam (approximately 3–4 females/dam) were separated into different cages, and divided into 4 sub-groups; (i) 25 female offspring were used to monitor diabetes development by weekly measurements of blood glucose from 6 to 35 weeks of age, and blood samples were collected at 6 and 10 weeks of age, (ii and iii) two separate subgroups, each containing 8 offspring, were used for histological examination of pancreas, ex vivo assessment of phagocytic function of isolated peritoneal macrophages and isolation of splenocytes and pancreatic lymph node cells at 7 and 11 weeks of age, respectively, and (iv) a subgroup of 4 offspring was used for isolation of pancreatic islets at 8 weeks of age, for ex vivo assessment of cytokine induced apoptosis induction. All experiments were performed in conformity with the laws and regulations for experiments with live animals and were approved by the local representative of the Norwegian Animal Research Authority.

2.2. Blood glucose measurements

In the diabetes development sub-group, all mice were monitored for blood glucose levels every week from 6 to 35 weeks of age, whereas for the two histology sub-groups, blood glucose levels were only determined once before euthanizing at the age of 7 or 11 weeks of age. Blood glucose levels were determined in blood samples from the femoral vein using Accu-Check (Roche Diagnostics, GmbH Mannheim, Germany). Mice were considered diabetic after two consecutive measurements within 24 h with glucose levels above 13.9 mmol/l and were immediately euthanized.

2.3. Serum insulin and testosterone measurements

Since testosterone has been shown to protect against diabetes development in NOD mice [38] the serum testosterone levels were analyzed in blood samples from 10 weeks old NOD mice. Further, elevated insulin secretion may also be protective against diabetes in the NOD mouse and BPA has been shown to induce insulin secretion in beta cells [39], [45]. Therefore, the non-fasting serum insulin levels were also determined at week 10. Serum insulin and testosterone concentrations were determined with ELISA-kits from Mercodia (Uppsala, Sweden) and DRG Instruments GmbH (Marburg, Germany), respectively, according to the manufacturer's recommendations.

2.4. Histological evaluation

For histological evaluation, pancreata were collected from 8 mice (at 7 and 11 weeks of age), fixed in formalin, embedded in paraffin and processed as described previously before haematoxylin and eosin staining [34], [35]. For each mouse 6 sections at different depth of the pancreas were examined and all islets present in the sections (10–15 islets/section) were graded for insulitis according to the area of an islet infiltrated by lymphocytes. 0% infiltration = grade 0, periinsulitis and up to 10% infiltration = grade 1, 10–49% infiltration = grade 2, 50–74% infiltration = grade 3 and 75–100% infiltration = grade 4, as previously illustrated [36]. For each section, an overall grade was assigned which corresponded to the highest grade detected in at least 3 islets. Then, the final grade for a pancreas/mouse was set to the highest grade determined for the 6 analyzed sections. The mean insulitis grade for each exposure group corresponds to the mean of the final pancreas grade for each mouse.

Sections of the formalin fixed pancreas were also stained over-night with antibodies towards Foxp3 (regulatory T-cells, eBioscience, San Diego, USA, 1:50), F4/80 (tissue resident macrophages, AbD Serotec, Oxford, UK, 1:50) and CD68 (invasive macrophages, AbD Serotec, 1:300), and active caspase-3 (apoptotic cells, Cell Signalling Technology, Beverly, MA, USA, 1:400), as previously described [36]. For each antibody staining and insulitis grade, the number of positive cells per islet was counted in two pancreatic sections per mouse. The counts per islet were compared between the exposure groups within each insulitis grade with a particular focus on grade 0 islets to investigate the effects of exposure on initial events in diabetes development.

2.5. Phagocytic function of peritoneal macrophages

To investigate if the exposures affected the macrophage functionality, the phagocytic activity of peritoneal macrophages was investigated by flow cytometry. Peritoneal macrophages were isolated from mice at 7 and 11 weeks of age (n = 8) by peritoneal lavage. The cell suspension was centrifuged at 250 × g for 10 min and erythrocytes were removed by dissolving the cell pellet in 0.2% NaCl on ice for 4 min. The cell concentration was determined and cells were seeded at 2 × 10⁶ cells/ml in 48 wells plates with RPMI cell culture medium. After 1 h of culture 85–95% of the attached cells are assumed to be macrophages [40]. Then the medium with unattached cells was discarded, replaced with fresh medium, and the cells were incubated over night. To assess the phagocytic function, FITC-labelled Zymosan particles were added at a ratio 20 particles per cell (In vitrogen Life Technologies, Carlsbad, CA, United States) incubated for 30 min at 37 °C, washed twice in PBS and incubated with Accutase (In vitrogen Life Technologies) for at least 15 min to remove particles from cell surfaces and to detach cells. The cells were washed twice with PBS, fixed in 0.2% paraformaldehyde in cell culture medium and analyzed for phagocytosis (FITC intensity per cell) by flow cytometry (LSR II, BD Bioscience, Franklin Lakes, NJ, USA).

2.6. Insulin secretion and NO production in isolated pancreatic islets

Pancreatic islets were isolated from 8 weeks old NOD mice (n = 5) as previously described [41] using injection of collagenase in the porta vein of euthanized animals before excision of the pancreas. Further dispersion of whole islets was performed in 37 °C collagenase for 17 min, after which islets were washed and thereafter hand-picked with pipette from a petri dish using a microscope. Glucose-induced insulin secretion was determined for 10 islets of equal size per mice as described previously [42]. Briefly, islets were incubated in glucose free medium in 48 wells plates for 30 min before a 60 min stimulation with either 6 or 16 mM glucose. Supernatants were harvested for insulin determination according to manufacturer's recommendation (ELISA, Mercodia) and islets were frozen for analysis of protein content.

NO production was measured using the Griess assay (Sigma–Aldrich) according to manufacturer's instructions.

2.7. Isolation of beta-cells and islet macrophages, analysis of cytokine induced apoptosis

To investigate the cellular sensitivity to cytokine-induced apoptosis [43], 60 isolated pancreatic islets from 4 offspring per exposure group were incubated for 48 h and stimulated with 50 U/ml rm-IL-1β, 1000 U/ml rm-INFγ and 1000 U/ml rm-TNFα. The islets were then moved to 1 ml calcium free HEPES-based beta-cell isolation solution and incubated for 12 min at 37 °C, before 2 ml RPMI medium (5% FCS) was added and the islets were dispersed by harsh pipetting. The cell suspension were washed and resuspended in binding buffer (BD Biosciences, Franklin Lakes, NJ, USA) stained for PE-annexin V (apoptotic cells, BD Biosciences), PI 7-AAD (necrotic cells, BD Biosciences) and APC-F4/80 (macrophages, eBioscience). Cells were fixed and permeabilized for intracellular pacific blue-insulin (beta cells, Cell Signalling Technology, Beverly, MA, USA) staining. Flow cytometry was applied to determine the percentage of apoptotic and necrotic beta cells and apoptotic macrophages originating from the pancreatic islets challenged with cytokines in vitro.

2.8. Cytokine release in ex vivo stimulated splenocytes and pancreatic lymphocytes

Splenocytes and pancreatic lymph node cells were isolated at 7 and 11 weeks of age (n = 8) as previously described [34], [35]. Subsequently, cytokine secretion upon ex vivo stimulation for 48 h with LPS or ConA was determined for all exposure groups in splenocyte cell cultures and with LPS in pancreatic lymph node cell and islet cultures, to examine possible effects on systemic immune functions, performed as previously described [34], [35], [44]. Stimulation with LPS activates B-cells and macrophages (10 μg/ml, Sigma–Aldrich, GmbH Steinheim, Germany), while ConA stimulation for 48 h activates T-cells (50 μg/ml, Sigma–Aldrich). The cytokine levels (INFγ, TNFα, IL-1β, IL-2, IL-4, IL-6, IL-10, IL-13 and IL-17) in the supernatants were determined by cytometric bead based array (FlexSet, BD Bioscience) analyzed on a LSRII flow cytometer (BD Bioscience).

2.9. Statistical analysis

Data are presented as group means ± standard errors of the mean (SEM). Diabetes incidence data were analyzed by Cox regression analysis, while all other data sets were analyzed by one way analysis of variance (ANOVA). The Holm–Sidiak post hoc test was performed to evaluate significant differences between the groups. For all analyses, p-values <0.05 were considered statistically significant.

3.1. Diabetes development

In the evaluation of diabetes development, mice with sustained elevated blood glucose levels, at 13.9 mmol/l or above, at two measurements within 24 h were considered diabetic. The spontaneous diabetes development started from 12 to 14 weeks of age in NOD mice for all exposure groups (Fig. 1A). The diabetes incidence in the BPA treated group seemed to be increased from 25 weeks of age compared to the control group. However, cox regression analysis did not confirm statistical differences between the exposure groups. Phthalate exposure or the combination of BPA and phthalates did not affect diabetes development to the same extent as BPA alone. In the histological evaluation of insulitis, there were no significant differences in the mean insulitis grade between the exposure groups at 7 or 11 weeks of age (Fig. 1B). However, the percentage of mice with a final insulitis grade of 4 at 11 weeks of age were 62.5% and 75% in the BPA and the combined BPA and phthalate exposure groups, respectively, compared to 37.5% in the control group (Fig. 1B), suggesting an accelerated insulitis development in these two groups. The number of offspring (in average about 8 per dam), sex distribution (mean 50% females), body weight (mean 11, 23 and 29 g at age 3, 11 and 20 weeks respectively) or water intake at 10 and 20 weeks of age (typically 3 and 3.5 ml per animal per day, respectively) were not affected by any of the exposures. Likewise, the non-fasting serum levels of insulin or testosterone did not differ between the exposure groups (data not shown).

3.2. Cellular infiltration in the pancreatic islets

Prior to insulitis (i.e. in islets with insulitis grade 0), the number of F4/80+ tissue resident macrophages was significantly reduced in the BPA exposed group at 7 weeks of age compared to control (Fig. 2). The same trend was observed at 11 weeks of age, although not statistically significant. In contrast, exposure to phthalates or the combination of BPA and phthalates, did not reduce the number of tissue resident macrophages significantly compared to control prior to insulitis at 7 or at 11 weeks of age. That is, all of the exposure groups except BPA alone having about 5 tissue resident macrophages per section of islet at grade 0 at 7 weeks of age (Fig. 2A). However, the number of apoptotic cells, evaluated by the number of active caspase-3 positive cells, was increased both in the BPA and the combined BPA and phthalates exposure groups (Fig. 3) compared to the control at 11 weeks of age. A similar trend was observed after 7 weeks of age after BPA exposure, with increased apoptotic cells in pancreatic islets at grade 0, although not significantly different from control (Fig. 3). Phthalate exposure alone did not affect the number of apoptotic cells in pancreatic islets at any time point, nor was the number of regulatory T-cells and CD68+ macrophages affected by any of the environmental chemical exposures (data not shown). The apoptotic cells observed in the islets after BPA exposure were identified in an earlier study to be alpha cells, beta cells and tissue resident macrophages [34], [35].

3.3. Phagocytic function of peritoneal macrophages

BPA exposure alone decreased the relative number of phagocytic macrophages with about 30% compared to control levels at 7 weeks of age (Fig. 4), suggesting impaired macrophage function. In contrast, exposure to phthalates alone or in combination with BPA did not affect the phagocytic function in peritoneal macrophages significantly (Fig. 4). There was a similar trend, however not statistically significant at 11 weeks of age (Fig. 4).

3.4. Insulin secretion and NO production in pancreatic islets

There was no difference between the exposure groups in the glucose induced insulin secretion or in the NO production in isolated pancreatic islets (data not shown).

3.5. Apoptosis sensitivity in isolated beta cells and macrophages

The percentage of apoptotic beta cells after in vitro cytokine stimulation was lower in mice from the phthalate exposure group at 8 weeks of age compared to the control group, suggesting a decreased sensitivity to cytokine-induced apoptosis in these cells (Fig. 5A). A similar but non-significant trend was observed for the combined BPA and phthalate exposures. For macrophages, the percentage of apoptotic cells was increased in the BPA exposure group compared to the control, suggesting increased sensitivity to cytokine-induced apoptosis in macrophages (Fig. 5B).

3.6. Cytokine release from splenocytes, pancreatic lymph node cells and pancreatic islets

Possible systemic immunomodulating effects of BPA and phthalate exposure in NOD mice were examined by ex vivo stimulation of splenocytes and pancreatic lymph node cells with LPS (activating B-cells and macrophages) and ConA (activating T-cells). ConA stimulation of splenocytes ex vivo resulted in increased release of IL-6 in the phthalate exposure group, and reduced release of IL-10, IL-4 and TNFα in both the BPA and the combined BPA and phthalate exposure groups compared to control (Fig. 6A). Exposure to BPA alone or in combination with phthalates generally decreased the LPS-induced in vitro secretion of TNFα, IL-6, IL-10 and IFNγ in splenocytes, with the largest effects for the combined exposure group (Fig. 6B). Levels of IL-4 were below the detection level after LPS stimulation in splenocytes.

Ex vivo LPS stimulated pancreatic lymph node cells also showed a non-significant trend for decreased levels of TNFα and IL-6 secretion in the BPA and combined BPA and phthalate exposure group (Fig. 6A), whereas IL-4 and IFNγ levels were below the detection limit. Overall, exposure to BPA alone or in combination with phthalates induced systemic effects in the NOD mice in terms of altered functionality of splenocytes and pancreatic lymph node cells. There were no significant differences in LPS-induced cytokine secretion from isolated pancreatic islets, except for a non-significant trend of decreased LPS-induced IL-10 secretion after combined BPA and phthalate exposure (Fig. 6D).