(368ax) Computer-Aided Insight into Disease Mechanisms and Inter-Relatedness Involved with Diabetic Kidney Disease for Exploring Disease Intervention Strategies

Research Interests: My research interests involve using systems biology modeling, pharmacometrics, and data-driven ML for applied biological sciences. I am interested in applying mathematical modeling to understand disease progression and inter-relatedness and advanced drug molecule discovery to enable the development of new therapeutics.

Current Work:

Diabetic kidney disease (DKD) is a leading cause of kidney failure worldwide. Aberrant glucose metabolism and oxidative stress are physiological disorders in diabetes that lead to immune response and inflammation in the glomerular cells in the kidney. In kidneys, hyperglycemia and inflammation induce intercellular junction disruption and enhance microvascular leakage. This leads to glomerular endothelial cell (GEC) activation and dysfunction in the early stages of DKD development and progression. GECs are essential to the structure and overall hydraulic resistance of the glomerular filtration barrier, yet little is known about how they are regulated and their role in disease. A deeper mechanistic analysis of the glucose-mediated inflammatory mechanisms in the kidney glomerulus can contribute to the understanding of GEC dysfunction in early-stage DKD.

Computational and large-language models allow the integration of experimental and literature evidence and disease mechanisms in systems biology. In this work, we study the crosstalk between macrophages and GECs in DKD through a protein-protein interaction network under systemic glucose, and inflammation. Using ML-aided text mining^[1,2], we also uncovered key concepts associated with DKD and their relatedness across biomedical literature. The network interactions were formulated using normalized-Hill type functions and logic-based ordinary differential equations^[3]. We performed a composite model analysis involving structure and identifiability analysis, global sensitivity analysis, parameter estimation, and uncertainty quantification of parameters^[4].

The proposed model identifies proteins and pathways that regulate GEC activation, dysfunction, and morphological changes in the early stages of DKD. The model provides insight into glucose-dependent fluctuations in junctional integrity, uncoupling of NO pathway, oxidative stress, and competitive VEGF receptor activation in GECs. Unlike ODE systems, this modeling approach overcomes the unavailability of mechanistic details and kinetic parameters. A lower uncertainty in predictions and estimated parameters confirmed that the model was consistent with the literature. The model predicts that reducing the strength of interactions activating NF-kappaB, VEGF-A, VEGF receptor 1, PLC-gamma, NO, and Ca reduced the perturbations in GEC morphology.

The proposed model showed key biomarkers involved and intervention strategies to reverse GEC activation and dysfunction in the early stages of DKD. This work supports the study of early-stage DKD especially when temporal measurements of structural changes in GEC are challenging to obtain through experiments. Ongoing work involves model-based hypothesis testing of the impact of dysregulated VEGF, NO, Ca, myosin-light chain kinases, and actin disassembly on the loss of GEC fenestrations in DKD. Future work may include relating these physiologic changes to clinical parameters (filtration rate, permeability) of DKD development and progression.

References: ^[1] Kirkpatrick, A et al., Big Data Cogn. Comput. (2022). ^[2] Patidar, K, et al., Int J Molec Sci (2024). ^[3] Kraeutler, MJ et al. BMC Syst Biol (2010). ^[4] Patidar, K and Ford Versypt, AN, biorxiv (2023).

Future Interests: I aim to increase my involvement in ML-based and mechanistic computational models and contribute to projects that support the drug discovery and development process. I am interested in applying my technical skills in quantitative systems pharmacology groups. I aim to contribute to projects that bridge the gap between pre-clinical and clinical studies through physiology-based mechanistic models. I have gained knowledge of several programming languages/software (Matlab, Python, R, Simbiology), and have experience working with physiology-based PK/TMDD, and ML/AI models. Through academic, research, and professional experience at a pharmaceutical company, I have acquired a broader perspective of integrating and translating my technical and computing skills towards various applications in biological and material sciences.

Keywords: Computational ODE Models; Machine Learning; Disease Intervention; Physiology-based Models.

Acknowledgments: This work was supported by the National Institutes of Health grant (R35GM133763) to A.N. Ford Versypt, the National Science Foundation CAREER grant (2133411) to A.N. Ford Versypt the National Science Foundation CAREER grant (1944247) to C.S. Mitchell, the National Institute of Health grant (R35GM152245) to C.S. Mitchell, the National Institute of Health grant (U19-AG056169) sub-awarded to C.S. Mitchell, and the Chan Zuckerberg Initiative grant (253558) to C.S. Mitchell.