#2863 SEX-, AGING AND DIABETES-RELATED ALTERATIONS IN GLOMERULAR DIMENSIONS AND PODOCYTE DENSITIES USING DEEP-LEARNING QUANTIFICATION

Abstract

Abstract Background and Aims Kidney function, as well as its morphology, changes markedly with age and disorders such as diabetes. This process is associated with structural and functional alterations in cortical and juxtamedullary glomeruli. Currently, data on differences in cortical and juxtamedullary glomeruli associated with sex, age, genetic factors, and diabetes are limited. In this study, we investigated the abundance and morphometry of podocytes and glomeruli in mice of different ages and sex’, and suffering from diabetes or not using a deep-learning based analysis of immuno-stained kidney sections. Methods Male and female non-diabetic C57BL/6J mice and diabetes type II (db/db) mice were sacrificed at different time points: 4, 10, 20, 30, 34, 40 weeks and 6, 24 weeks, respectively. Subsequently, kidneys were extracted, embedded in paraffin, cut into sections, stained, and imaged for histological analysis. We used immunhistochemistry staining with Wilms Tumor 1 (WT1) antibody to specifically stain podocyte nuclei. Both manual and deep-learning based image segmentation were performed to analyze abundance and morphometry of podocytes and glomeruli. In total, 4134 glomerular structures were detected, and morphometry of glomeruli and podocytes was extracted and analyzed by an automated algorithm. Results Our study aimed to investigate aging- and diabetes-related differences in cortical and juxtamedullary glomeruli with respect to podocyte abundance and loss. Using a customized deep learning algorithm, podocyte nuclei could be quantified with comparable quality as a time-tedious manual analysis, which takes approximately 1 minute per glomerulus. Extracted morphometric features showed that juxtamedullary glomeruli had a larger cross-sectional area than cortical glomeruli (Figure 1a). For both cortical and juxtamedullary glomeruli the cross-sectional area slightly increased on average with aging (Figure 1b). As expected, podocyte endowment in cortical glomeruli was lower than in juxtamedullary glomeruli with podocyte density being higher (Figure 2a, 2b). Over life-time the number of podocytes and podocyte density per glomerulus slightly decreased both in cortical and juxtamedullary glomeruli (Figure 2b, 3b). Interestingly, female cortical glomeruli were on average smaller and had a higher podocyte density compared to males (Figure 1c, 3c). However, podocyte numbers did not differ between male and female (Figure 2c). Finally, we found that 24 weeks old db/db mice presented with glomerular hypertrophy in contrast to non-diabetic C57BL/6J mice of the same age (Figure 1d). Db/db mice lost podocytes from 6 weeks to 24 weeks of age with a decreased podocyte density. Surprisingly, podocyte loss occurred to a lower extent in non-diabetic mice (Figure 2d, 3d). Conclusion During aging and early diabetic disease both podocyte loss and glomerular hypertrophy occur. Similar changes occurred in juxtamedullary and cortical glomeruli in both sex’. Hyperfiltration might explain the pronounced extent in glomerular area in diabetic mice and should represent an increase in filtration surface to handle diabetes-related hyperfiltration. Less podocyte loss in diabetic mice at 24 weeks age compared to C57BL/6J mice might relate to the different mouse model and still moderate podocyte stress during the early stage of diabetic kidney disease.