Abstract: TH-PO030
Artificial Intelligence (AI) Reveals Dynamic Morphological Shifts in Parietal Epithelial Cells after Acute Tubular Injury
Session Information
- Augmented Intelligence for Prediction and Image Analysis
October 24, 2024 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Augmented Intelligence, Digital Health, and Data Science
- 300 Augmented Intelligence, Digital Health, and Data Science
Authors
- Yin, Mengmeng, Vanderbilt University Medical Center, Nashville, Tennessee, United States
- Zhao, Oliver Sihua, Vanderbilt University Medical Center, Nashville, Tennessee, United States
- Wang, Yu, Vanderbilt University Medical Center, Nashville, Tennessee, United States
- Zhao, Shilin, Vanderbilt University Medical Center, Nashville, Tennessee, United States
- Deng, Ruining, Vanderbilt University, Nashville, Tennessee, United States
- Huo, Yuankai, Vanderbilt University Medical Center, Nashville, Tennessee, United States
- Zhong, Jianyong, Vanderbilt University Medical Center, Nashville, Tennessee, United States
- Yang, Haichun, Vanderbilt University Medical Center, Nashville, Tennessee, United States
- Fogo, Agnes B., Vanderbilt University Medical Center, Nashville, Tennessee, United States
Background
We previously found that pre-existing acute tubular injury (ATI) sensitizes glomeruli to subsequent injury. Through spatial transcriptomics, we identified key ligand-receptor pairs that could contribute to this sensitivity, with ligands from the proximal tubules (PTs) and their corresponding receptors on adjacent parietal epithelial cells (PECs). In this study, we utilized artificial intelligence (AI) to evaluate the morphological changes in PECs and their spatial transcriptomics profiles following ATI.
Methods
Wild type (WT) and transgenic mice (n=7/group) expressing diphtheria toxin (DT) receptor in PTs were injected with DT at week 0 and week 1 to induce ATI and sacrificed at week 7. Periodic acid-Schiff stained kidney sections from normal mice were utilized to train an AI model. Spatial transcriptomics were conducted using the NanoString GeoMx DSP platform.
Results
We utilized fluorescent staining of claudin-1 as a PEC biomarker to train an AI model for classification and segmentation of PECs. This allowed for the quantification of 246 morphological profiles for each PEC nucleus. In mice with ATI, PECs nuclei from DT transgenic mice showed increased eccentricity and a greater proportion of staining intensity at the nucleus periphery, as well as reduced form factor of roundness and texture entropy compared to WT mice, suggesting a flatter nuclear morphology with reduced complexity in DT PECs nuclei. Furthermore, 153 differentially expressed genes were detected in PEC nuclei of DT vs WT mice, with enrichments in several pathways, including actin cytoskeleton regulation and PI3K-Akt-mTOR signaling. RPS6 gene expression, a gene involved in cell size regulation, was significantly increased in PECs nuclei of DT compared to WT. In contrast, PECs immunostaining of p-RSK1, a key enzyme involved in RPS6 phosphorylation, was decreased in DT compared to WT mice, suggesting decreased activity.
Conclusion
AI reveals dynamic morphological shifts in PECs following ATI, with flatter, less round nuclei and more chromatin staining at the nuclear periphery. These altered nuclear properties could represent activation states, potentially mediated by decreased activation of the RSK1-RPS pathway. These findings have significant implications for PEC-podocyte interactions and glomerular injury responses after ATI.
Funding
- NIDDK Support