Fig 1.
Two distinct mice populations separated by blood glucose.
(A) Blood glucose as a function of age shows an early and large difference. At right: example images of each mouse (22 weeks, black arrow). S1 Video shows their similar activity level. (B) Weight as a function of age shows hyperglycemic mice have minor weight loss. The light background highlights data with multiple mice as the last four points represented only the oldest pair. (C) Plot of 946 BG and weight measurements taken from 13 euglycemic and 14 hyperglycemic mice. Histograms along each axis show high overlap in weight and low overlap in BG level with the average and standard deviation reported for each.
Fig 2.
Imaging paradigm and examples from euglycemic and hyperglycemic mice.
Example ROI (outlined by dotted boxes) distributions from four different mice. (A-B) 20x magnification montages. Scale bar = 500 μm. (C-D) 55° FOV fluorescein angiography. Scale bar = 100 μm. (E) Representative ex vivo Z-Stacks for both glycemia groups organized by eccentricity (horizontal) and age (vertical). The number of male mice for each age range is in the left column, while the number of Z-Stacks for each age and eccentricity group is within each box. Scale bar = 25 μm. Light blue lines highlight the analysis bins corresponding to the in vivo hyperglycemic mouse data. Contrast adjustment for visualization purposes, original images available upon request. *One euglycemic mouse was analyzed using a montage from optic disc to ora serrata; it was included in the total but not in any of the eccentricity by age boxes.
Fig 3.
Vascular metrics were not significantly different between groups.
(A) Age and eccentricity matched examples from each glycemia group. The first row is a projection of the entire Z-Stack. Each row beneath is a projection of a subset of the Z-Stack corresponding to the superficial, SI, intermediate, ID, and deep layers. Scale bar = 25 μm (B) Vessel density, (C) vessel branching, (D) vessel length, and (E) tortuosity quantification. (B-E) are divided into layer total (left) layer breakdown (center), and a visual diagram created with Biorender.com (right). White asterisks denote nodes between vessel branches or segments. One extreme outlier in the euglycemic data was omitted for visualization purposes. We used Welch’s unpaired t-test for total and two-way ANOVAs with post-hoc Tukey’s for each vascular layer.
Table 1.
Summary of vascular metrics for euglycemic and hyperglycemic mice.
Fig 4.
Pericyte density is similar between euglycemic and hyperglycemic mice.
(A) Euglycemic or (B) hyperglycemic Z-Stack with colored squares highlighting locations of pericyte examples in each of the vascular layers. Scale bar = 25 μm. Enlarged images of the pericytes are arranged in increasing depth from left to right. Scale bar = 10 μm. (C) Pericyte density and (D) vessel length to pericyte soma ratio quantification. Welch’s unpaired t-test for all and two-way ANOVAs with post-hoc Tukey’s for each vascular layer. *P < 0.05.
Fig 5.
Interconnecting microvessels did not show significant changes.
(A) Euglycemic and hyperglycemic examples of each of the 5 categories of axial vessels. Images are aligned with a colored diagram on the left showing the vascular layers. Scale bar = 25 μm Beneath is the corresponding quantification of the number of axial vessels per area. Two-way ANOVAs with post-hoc Tukey’s used for each axial vessel type. (B) Plots of the relationship between BG severity and number of axial vessels for each category.
Table 2.
Densities of interconnecting microvessels.
Fig 6.
Blood glucose severity is a poor predictor of vascular changes.
(A) vessel density, (B) vessel branching, (C) vessel length (D) tortuosity, and (E) pericyte density correlations between the average value and average BG. Shown with linear regression lines. The columns from left to right: total; superficial, intermediate, and deep; SI and ID quantifications. Embedded in the y-axis are histograms of the number of mice to show the distribution of average values.
Fig 7.
Vessel structure was similar across 12 weeks of elevated blood glucose.
(A) Weekly BG measurements, yellow shading is placed over the time window imaged. (B) Projections of in vivo Z-Stacks from each vascular layer of one ROI at each time point. Far right column shows the SNT traces to highlight the matching features and variability between imaging sessions. The trace color darkens for each progressive time point and all three are aligned by their top right corner. White arrows highlight shared landmarks between sessions. (C) All hyperglycemic in vivo data showing all layers, locations, and time points. SNT traces are overlayed. Scale bar = 10 μm. Images were contrast adjusted for visualization purposes; original images available upon request. Graphs of 4 vascular metrics: (D) vessel density, (E) vessel branching, (F) vessel length, and (G) tortuosity. Each dot represents one of four locations within the same mouse.
Fig 8.
In vivo and ex vivo data showed good agreement.
(A) NIR, FA, and flat mount images of the same retina centered on the optic disc. Scale bar = 500 μm (B) Images from the three trilaminar layers at the same location in vivo (left) and ex vivo (right). Scale bar = 10 μm (C) Vessel density quantification where each location and time was represented in vivo (n = 12) and compared to each location ex vivo (n = 4). (D) Vessel length traces for one-to-one vessel length comparison. Due to FOV movement, values were averaged across all time points for in vivo locations (n = 4) and compared to the same location ex vivo (n = 4).