Transparent, biodegradable, mechanically robust, and surface-patterned silk films were evaluated for the effect of surface morphology on human corneal fibroblast (hCF) cell proliferation, orientation, and ECM deposition and alignment. A series of dimensionally different surface groove patterns were prepared from optically graded glass substrates followed by casting poly(dimethylsiloxane) (PDMS) replica molds. The features on the patterned silk films showed an array of asymmetric triangles and displayed 37–342 nm depths and 445–3 582 nm widths. hCF DNA content on all patterned films were not significantly different from that on flat silk films after 4 d in culture. However, the depth and width of the grooves influenced cell alignment, while the depth differences affected cell orientation; overall, deeper and narrower grooves induced more hCF orientation. Over 14 d in culture, cell layers and actin filament organization demonstrated that confluent hCFs and their cytoskeletal filaments were oriented along the direction of the silk film patterned groove axis. Collagen type V and proteoglycans (decorin and biglycan), important markers of corneal stromal tissue, were highly expressed with alignment. Understanding corneal stromal fibroblast responses to surface features on a protein-based biomaterial applicable in vivo for corneal repair potential suggests options to improve corneal tissue mimics. Further, the approaches provide fundamental biomaterial designs useful for bioengineering oriented tissue layers, an endemic feature in most biological tissue structures that lead to critical tissue functions.

During corneal development, neural crest derivatives from the periocular mesenchyme migrate into the cornea and differentiate into corneal fibroblasts. During this time, these cells interact with a variety of extracellular matrices for proper orientation and development. In the present studies, we have examined the interaction of beta(1) integrins on periocular mesenchyme cells (POM) and corneal fibroblasts (CF) with fibronectin and tenascin by perturbing the function of this integrin. POM and CF attached and spread to a much greater extent on fibronectin than on tenascin. An antibody against beta(1) integrin, CSAT, decreased spreading and attachment, and resulted in a lack of immuno-detectable beta(1) integrin in focal adhesions on fibronectin; few beta(1) positive focal adhesions were observed in cells grown on tenascin. An anti-sense retroviral construct decreased endogenous levels of beta(1) integrin protein, and caused decreased attachment and spreading as well as sparse, disorganized focal adhesions. These data indicate that in vitro, both POM and CF have beta(1) integrins that interact with fibronectin and allow them to attach and spread, while tenascin is anti-adhesive. Further studies using both of these experimental paradigms will clarify whether these interactions also occur in vivo.