Unveiling the Metabolic and Morphological Complexity of Mitochondria: The Pivotal Role of Mitochondrial Sirtuins and High- Resolution Transmission Electron Microscopy Images Analysis

Abstract

Kidney plays a critical role in oxygen sensing and energy metabolism, making the study of oxygen supply and hypoxia a central focus in kidney disease research. Due to its structural and functional characteristics, the kidney is particularly reliant on the VHL-HIF pathway for oxygen regulation. Mutations in the VHL gene lead to persistent activation of HIF, even under normoxic conditions, which increases susceptibility to tumor development and metabolic reprogramming. Mitochondria are central to these processes, serving as the primary site of energy metabolism. NAD metabolites are essential for ATP generation via the mitochondrial electron transport chain (ETC), with mitochondrial sirtuins serving as key regulators of metabolic homeostasis. Among them, SIRT3 is a major mitochondrial NAD⁺-dependent deacylase that modulates critical enzymes involved in fatty acid β-oxidation, the tricarboxylic acid (TCA) cycle, and the ETC. Despite its importance in mitochondrial function, the role of SIRT3 in cancer aggressiveness remains underexplored. Here, we demonstrate that SIRT3 promotes fatty acid β-oxidation through interactions with key metabolic proteins, including ECH1, PDK2, ACADVL, ACAA2, CRAT, ACAT1, ECI1, ECI2, and HSD17B8, as validated by integrative bioinformatics and biochemical analyses. These metabolic reprogramming shifts energy metabolism to promote fatty acid utilization over glucose oxidation, providing an energy-efficient adaptation to metabolic stress, such as nutrient deprivation in the tumor microenvironment. Furthermore, we show that hypoxic, VHL-deficient 786-O cells exhibit increased reliance on lipid metabolism, highlighting SIRT3 as a key regulator of mitochondrial energy metabolism in renal cancer carcinoma. Mitochondrial morphology, which is closely tied to function, plays a critical role in understanding cellular bioenergetics and pathology. In multinucleated muscle cells, efficient energy supply depends on the optimized spatial distribution of mitochondria, ensuring effective communication and interaction between mitochondria and nuclei. The multinucleated nature of muscle cells, however, presents unique challenges for mitochondrial segmentation. The complex and dynamic arrangement of mitochondria within these cells makes traditional segmentation methods difficult and often imprecise. To address this, we developed a novel deep learning framework combining probabilistic interactive segmentation with automated quantification. This approach reduced analysis time by 90% while maintaining high accuracy, evaluated on Lucchi++ benchmark datasets and real-world TEM images from mdx mouse models of Duchenne muscular dystrophy. The pipeline effectively identified pathological differences in mitochondrial morphology, offering a scalable, reproducible tool for high- throughput mitochondrial analysis and insights into disease mechanisms.