Artificial intelligence-enabled analysis of electron micrographs to assess mitochondrial morphology in triple-negative breast cancer

Abstract

Advancements in transmission electron microscopy (TEM) have enabled in-depth studies of biological specimens, offering new avenues to large-scale imaging experiments with subcellular resolution. Mitochondrial morphology is of growing interest in cancer biology due to its crucial role in regulating the multi-faceted functions of mitochondria. We and others have established the crucial role of mitochondria in triple-negative breast cancer (TNBC), an aggressive subtype of breast cancer with limited therapeutic options. Building upon our previous work demonstrating the functional role of mitochondrial morphology dynamics in the metabolic adaptations and survival of chemotherapy-refractory TNBC cells, we sought to extend those findings to analysis of transmission electron micrographs. Here, we present a novel U-Net artificial intelligence (AI) model for automatic annotation and assessment of mitochondrial morphology and feature quantification. Our model was trained on 11,039 manually annotated mitochondria across 125 micrographs derived from a variety of orthotopic patient-derived xenograft (PDX) mouse model tumors and adherent cell cultures. The model achieves an F1 score of 0.85 on test micrographs at the pixel level. To validate the ability of our model to detect expected mitochondrial morphology changes, we utilized micrographs from mouse primary skeletal muscle cells genetically modified to lack Dynamin-related protein 1 (Drp1), a key mitochondrial fission protein. We subjected in vitro and in vivo TNBC models to conventional chemotherapy treatments commonly used for clinical management of TNBC, including doxorubicin, carboplatin, paclitaxel, and docetaxel. We found substantial within-sample heterogeneity of mitochondrial morphology in both in vitro and in vivo. In four of five PDX models, in vivo treatment with DTX elicited significant alteration in mitochondrial elongation and/or area. We went on to compare mammary tumors and matched lung metastases in a highly metastatic PDX model of TNBC, revealing altered mitochondrial elongation in metastatic lesions compared to their matched primary mammary tumor. The successful application of our AI model provides a framework for high-throughput quantitative analysis of mitochondrial morphology and enables future studies investigating how mitochondrial structural changes relate to chemotherapy response and mechanism of action. Our publicly available manually curated electron micrograph dataset serves as a unique resource for developing, benchmarking, and applying computational models, while further advancing investigations into mitochondrial morphology in breast cancer. This study provides proof of concept that mitochondrial structural remodeling is an additional layer of cellular reprogramming accompanying therapeutic resistance in TNBC that merits further investigation.