Revealing Early Spatial Patterns of Cellular Responsivity in Fiber-Reinforced Microenvironments

Abstract

Fiber-reinforcement approaches have been utilized to replace aligned tissues with engineered constructs after injury or surgical resection, strengthening soft biomaterial scaffolds and replicating anisotropic, load-bearing properties. However, most studies focus on the macroscale aspects of these scaffolds, rarely considering the cell-biomaterial interactions that govern remodeling and ECM organization towards aligned neo-tissues. Since initial cell-biomaterial responses within fiber-reinforced microenvironments likely influence long-term efficacy of repair and regeneration strategies, here we elucidate roles of spatial orientation, substrate stiffness, and matrix remodeling on early cell-fiber interactions. Bovine mesenchymal stromal cells (MSCs) were cultured in soft fibrin gels reinforced with a stiff 100-micron polyglycolide-co-caprolactone fiber. Gel stiffness and remodeling capacity were modulated by fibrinogen concentration and aprotinin treatment, respectively. MSCs were imaged at 3 days and evaluated for morphology, mechanoresponsiveness (nuclear YAP localization), and spatial features including distance and angle deviation from fiber. Within these constructs, morphological conformity decreased as a function of distance from fiber. However, these correlations were weak (R2 = 0.01043 for conformity and R2 = 0.05542 for nuclear YAP localization), illustrating cellular heterogeneity within fiber-enforced microenvironments. To better assess cell-fiber interactions, we applied machine-learning strategies to our heterogeneous dataset of cell shape and mechanoresponsive parameters. Principal component analysis (PCA) was used to project 23 input parameters (not including distance) onto 5 principal components (PCs), followed by Agglomerative Hierarchical Clustering (AHC) to classify cells into 3 groups. These clusters exhibited distinct levels of morpho-mechanoresponse (combination of morphological conformity and YAP signaling) and were classified as High Response (HR), Medium Response (MR), and Low Response (LR) clusters. Cluster distribution varied spatially, with most cells (61%) closest to the fiber (0 to 75 microns) belonging to the HR cluster, and most cells (55%) furthest from the fiber (225 to 300 microns) belonging to the LR cluster. Modulation of gel stiffness and fibrin remodeling showed differential effects for HR cells, with stiffness influencing the level of mechanoresponse, and remodeling capacity influencing the location of responding cells. Overall, clustering of individual cells in stiff-soft microenvironments revealed spatial trends in cellular responsivity not seen by evaluating individual cell parameters as a distance from fiber alone.