Reversal of fibrosis by TGF-β3 in a 3D in vitro model.

Corneal scarring following moderate to severe injury is inevitable. Despite significant advancements in the field, current treatments following these types of injuries are limited, and often, the visual recovery is poor. One of the problems and limitations is that corneal wound healing is a complex process, involving corneal cells, extracellular matrix components and growth factors. Therefore, further understanding is required, along with new treatments and techniques to reduce or prevent corneal scarring following injury. Two isoforms of transforming growth factor-beta (TGF-β), TGF-β1 and -β3 (T1 and T3, respectively), are associated with corneal wound healing. T1 has been shown to drive the corneal keratocytes to differentiate into myofibroblasts; whereas, T3 has been found to inhibit fibrotic markers. In the current study, we examined whether the fibrotic characteristics expressed by human corneal fibroblasts (HCF) in our 3-dimensional (3D) construct following T1 stimulation could be reversed by introducing T3 to the in vitro system. To do this, HCF were isolated and cultured in 10% serum, and when they reached confluence, the cells were stimulated with a stable Vitamin C (VitC) derivative for 4 weeks, which allowed them to secrete a self-assembled matrix. Three conditions were tested: (1) CONTROL: 10% serum (S) only, (2) T1: 10%S + T1, or (3) Rescue: 10%S + T1 for two weeks and then switched to 10%S + T3 for another two weeks. At the end of 4 weeks, the constructs were processed for analysis by indirect-immunofluorescence (IF) and transmission electron microscopy (TEM). Different collagens that are normally present in healthy corneas in vivo, such as Type I and V, as well as Type III, which is a fibrotic indicator, were examined. In addition, we examined smooth muscle actin (SMA), a marker of myofibroblasts, and thrombospondin-1 (TSP-1), a multifunctional matrix protein known to activate the latent complex of TGF-β and appear upon wounding in vivo. Our data showed high expression of collagens type I and V under all conditions throughout the 3D constructs; however, type III and SMA expression were higher in the constructs that were stimulated with T1 and reduced to almost nothing in the Rescue samples. A similar pattern was seen with TSP-1, where TSP-1 expression following "rescue" was decreased considerably. Overall, this data is in agreement with our previous observations that T3 has a significant non-fibrotic effect on HCFs, and presents a novel model for the "rescue" of both cellular and matrix fibrotic components with a single growth factor.