3D quantification of changes in pancreatic islets in mouse models of diabetes type I and II

Abstract

Diabetes is characterized by rising levels of blood glucose and is often associated with a progressive loss of insulin-producing beta cells. Recent studies have demonstrated that it is possible to regenerate new beta cells through proliferation of existing beta cells or trans-differentiation of other cell types into beta cells, raising hope that diabetes can be cured through restoration of functional beta cell mass. Efficient quantification of beta cell mass and islet characteristics is needed to enhance drug discovery for diabetes. Here, we report a 3D quantitative imaging platform for unbiased evaluation of changes in islets in mouse models of type I and II diabetes. To determine whether the method can detect pharmacologically induced changes in beta cell volume, mice were treated for 14 days with either vehicle or the insulin receptor antagonist S961 (2.4 nmol/day) using osmotic minipumps. Mice treated with S961 displayed increased blood glucose and insulin levels. Light-sheet imaging of insulin and Ki67 (also known as Mki67)-immunostained pancreata revealed a 43% increase in beta cell volume and 21% increase in islet number. S961 treatment resulted in an increase in islets positive for the cell proliferation marker Ki67, suggesting that proliferation of existing beta cells underlies the expansion of total beta cell volume. Using light-sheet imaging of a non-obese diabetic mouse model of type I diabetes, we also characterized the infiltration of CD45 (also known as PTPRC)-labeled leukocytes in islets. At 14 weeks, 40% of the small islets, but more than 80% of large islets, showed leukocyte infiltration. These results demonstrate how quantitative light-sheet imaging can capture changes in individual islets to help pharmacological research in diabetes.

Keywords: Beta cells; Inflammation; Insulin; Light-sheet fluorescence microscopy; Tissue clearing.

Fig. 1.

Schematic representation of the study workflow. (A) C57BL/6JRj diet-induced obese (DIO) mice were implanted with minipumps containing either vehicle or the insulin receptor antagonist S961. Plasma glucose and insulin were measured at day 6 and day 13. After 2 weeks, the pancreata were isolated and stained with antibodies against insulin and Ki67. For scanning, the pancreas was embedded in agarose and cleared. The insulin channel was used as a reference to segment the beta cell volume, allowing quantification of the total number of insulin-positive islets, total beta cell volume and total number of Ki67-positive beta cells. (B) The entire pancreas was scanned using light-sheet fluorescence microscopy. The insulin channel was used to segment the beta cell volume (cyan), allowing quantification of numbers and volumes of insulin-positive beta cells. The Ki67 channel was used to count the number of proliferating beta cells. Scale bars: 1000 µm (low magnification); 200 µm (high magnification).

Fig. 2.

In vivo measurements in mice treated with either vehicle or S961. (A) Accumulated food intake during the 2-week study was similar in the two groups. (B) The S961 mice lost slightly more weight than the vehicle mice but the difference was only significant at day 14. (C) Oral glucose tolerance test identified significantly increased blood glucose in the S961 group compared to the vehicle group. (D) At day 13, the mean plasma glucose levels were 9.50±0.22 mmol/l in the vehicle group and 20.02±1.48 mmol/l in the S961 group. (E) The mean plasma insulin levels were 1450±96.1 pg/ml in the vehicle group and 34400±884 pg/ml in the S961 group. (F) Total pancreas weight did not differ significantly between the two groups. For statistical analysis, unpaired Student's t-tests were used. ***P<0.001, S961 compared to vehicle. Error bars represent s.e.m.

Fig. 3.

Increased beta cell volume and number of insulin-positive islets following S961 treatment. (A) Using the insulin signal, each islet was assigned a unique ID and a volume, making it possible to calculate the total number of insulin-positive islets and the total beta cell volume. (B) Total number of insulin-positive islets in vehicle- and S961-treated mice. (C) The same pancreas as shown in A, color coded for size distribution. The insulin-positive islets were allocated into four bins: small (yellow; 25-170×1000 µm³), medium (cyan; 170-1100×1000 µm³), large (magenta; 1100-7500×1000 µm³) and very large (red; 7500-50000×1000 µm³). (D) Total mean beta cell volume in the study groups. Individual data points are indicated. For statistical analysis, we used unpaired Student's t-test. ***P<0.001, S961 compared to vehicle. Error bars represent s.e.m. Scale bars: 500 µm.

Fig. 4.

S961 treatment response in different islet size categories: (A) Size distribution curve of islets. Size categories as indicated in the Fig. 3 legend. (B) Relative contribution of islets in different size categories to the total islet volume. Islets across all size categories contribute to the overall increase in total beta cell volume. (C) Quantification of the number of small (yellow), medium (cyan), large (magenta) and very large (red) insulin-positive islets in vehicle- and S961-treated mice. In all four bins, there is an increase in the number of insulin-positive islets following S961 treatment. (D) Quantification of the beta cell volume in the small (yellow), medium (cyan), large (magenta) and very large (red) islets in vehicle- and S961-treated mice. Individual data points are indicated. In C and D, both sets of data were separately investigated using 2×4 mixed ANOVA, and in follow-up tests on the treatment effect within each size category, the two measures were investigated in a multivariate manner (owing to correlation) by Hotelling’s T-squared test. *P<0.05, **P<0.01; S961 compared to vehicle. Error bars represent s.e.m.

Fig. 5.

Quantification of Ki67-positive beta cells. (A) 3D light-sheet microscopy image from a S961-treated mouse showing insulin (glow scale) and Ki67 (blue) staining. The Ki67 channel is shown separately on the right. (B) High-magnification section (20 µm) from the 3D image stack, showing insulin and Ki67 staining in an islet. (C) Total number of Ki67-positive islets is increased in the S961 group in comparison to the vehicle group. (D) Total number of Ki67-positive cells within all segmented islets is increased in the S961 group in comparison to the vehicle group. (E) Quantification of the number of small (yellow), medium (cyan), large (magenta) and very large (red) insulin-positive islets with Ki67 signal in vehicle- and S961-treated mice. Individual data points are indicated. For pairwise statistical analysis (in C-E), unpaired Student's t-test was used. *P<0.05, **P<0.01 (P=0.0022 in E); S961 compared to vehicle. In E, mixed ANOVA was applied prior to the pairwise statistics. Error bars represent s.e.m. Scale bars: 400 µm (A); 100 µm (B).

Fig. 6.

3D imaging of islet inflammation in NOD mice. (A) 3D light-sheet microscopy image of an area in 14-week-old NOD mouse pancreas, stained for insulin (glow scale) and CD45 (blue). Section from the 3D image stack, demonstrating insulin (glow scale) and CD45 (blue) staining in the 14-week-old NOD mouse. (B) High-magnification image of an individual islet from the NOD mouse, demonstrating insulin signal (glow scale) and leukocyte (CD45; blue) infiltration within the islet and accumulation at the islet periphery. (C) Overview image of a pancreas head, with islet inflammation segmentation data in glow scale overlaid with islet insulin segmentation data. Higher prevalence of CD45⁺ cells within an islet results in brighter islet color. (D) Quantification of the number of inflamed small (yellow), medium (cyan), large (magenta) and very large (red) insulin-positive islets. Islets with at least 5% volume of CD45⁺ cells were quantified. Individual data points from different mice are indicated in different colors. Error bars represent s.e.m. Scale bars: 200 µm (A); 100 µm (B).