A spatial topology atlas of type I myofibers in porcine skeletal muscle reveals local slow-oxidative niches associated with meat quality traits
TL;DR
The results suggest that skeletal muscle should be modeled as a spatially organized tissue system rather than a simple mixture of fiber types.
Abstract
Skeletal muscle metabolism and meat quality are usually interpreted through myofiber type composition, fiber size, and histochemical state. Whether the spatial topology of myofibers constitutes an independent phenotype layer remains poorly quantified. Here, using succinate dehydrogenase (SDH)-stained cross-sectional images from multiple pig breeds and more than 40 skeletal muscle sites, we established a multiscale spatial phenotyping framework for type I myofibers. To enable scalable instance-level quantification, we developed an improved Mask R-CNN segmentation workflow, trained with SAM-assisted and manually corrected annotations, and optimized for dense, small, irregular, and variably stained muscle fibers. This framework captures singlefiber morphology and staining, local contact, connected-cluster architecture, diffusion-ring neighborhoods, and global tissue configuration. Type I myofiber abundance was separable from spatial organization capacity: some muscles formed stronger-than-expected local connectivity and clustering despite modest type I fiber abundance. Spatial transcriptomic integration further showed that slow-oxidative genes and functional modules were nonrandomly organized and formed local high-expression domains in type I fiber-enriched, high-contact neighborhoods. Across muscle sites and breeds, skeletal muscles displayed distinguishable type I fiber spatial topology identities, which were remodeled by breed background in a muscle-specific manner. In samples with matched meat-quality measurements, spatial topology features improved the explanation of pH, meat color, conductivity, and intramuscular fat. These results suggest that skeletal muscle should be modeled as a spatially organized tissue system rather than a simple mixture of fiber types. Highlights A multiscale spatial phenotyping framework was established for porcine type I myofibers. Type I fiber topology organization are separable from abundance and affects muscle fiber amplification and gene expression. Breed background remodels type I fiber spatial topology in a muscle-specific manner. Spatial topology features improve explanation of pH, meat color, conductivity, and IMF The improved deep learning algorithm supports more efficient muscle fiber segmentation and phenotype extraction.