This study constructed a bioengineered multi-lamellar corneal stroma-like tissue in vitro by using PDLSCs as seeded cells, patterned silk membrane as scaffold, and substance P (SP) as a supplemental factor. Human PDLSCs were isolated and characterized, and could be efficiently differentiated into keratocytes with keratocyte differentiation medium. SP had no effect on the keratocyte differentiation, but promoted the expression of corneal stroma-related collagens (COL I, COL III, COL V, and COL VI). Patterned silk membranes guided cell alignment and supported corneal stroma ECM deposition (COL I, COL III, COL V, COL VI, LUM, and KERA) when seeded PDLSCs were cultured in keratocyte differentiation medium supplemented with SP. Based on this, multi-lamellar corneal stroma-like tissue was successfully constructed in vitro which could be beneficial for corneal tissue engineering and future corneal regeneration studies.

PDLSCs could be a potential cell source for keratocyte differentiation and corneal tissue engineering. Based on our results, PDLSCs can be efficiently differentiated into keratocytes with keratocyte differentiation medium. In addition, PDLSCs have several inherent advantages when compared to stem cells from other non-corneal tissues. Firstly, PDLSCs, as well as DPSCs, have the same neural crest origin as keratocytes in development, making them similar in proteoglycan secretion, and potentially beneficial for cell type transition between each other [9]. DPSCs have been reported capable of differentiating into keratocytes and to generate corneal stromal-like constructs [11]. However, to our knowledge, no similar evaluation has been previously carried out for PDLSCs. Secondly, PDLSCs are derived from periodontal ligaments. Tissues of ligaments/tendons have a high similarity to corneal stroma, both regarding physiological and pathological conditions. Normally, they both consist of aligned dense collagens, with COL I as the main component [26, 27]. Genome-wide gene expression pattern comparison has shown notable similarities in gene expression between tendons and corneas, especially in ECM collagens (collagen type I, III, V, and VI) and proteoglycan (lumican, decorin, and biglycan) [28]. In addition, both ligaments/tendons and corneas are subjected to physiological mechanical stress. When injured or degenerated, ligaments/tendons and corneal stroma often heal with scar formation, losing their well-organized collagen structures, with more cell accumulation and increased production of collagen type III relative to type I [2931]. Thirdly, PDLSCs share the advantages of other dental stem cells, such as accessibility, high proliferation, and immunomodulatory properties [9, 32], which is promising for future clinical applications.

SP plays important roles in the cornea. Increased expression of TAC1 (the gene coding for SP) and its receptor TACR1 was found during the differentiation of PDLSCs towards keratocytes, but not in primary keratocytes (Fig. 2f and g). A promotion effect of SP on collagen expression instead of keratocyte markers was subsequently confirmed in our study. Interestingly, global genome-wide gene expression analysis by Wu et al. showed higher expression of ECM-related genes (including collagen types I, III, and VI) in the immature (postnatal day 10) mouse cornea as compared with adult mouse cornea [28], which indicates the demand for high collagen expression during corneal development. In our current model, upregulated SP and its receptor during the differentiation process (similar to the developmental process in vivo) contributed to the increased collagen expression. Furthermore, primary keratocytes in culture, like the mature adult cornea in vivo, do not need a high expression of collagens, which is consistent with the observed low expression of SP and its receptor. The promotion effect of SP on corneal stroma-related collagen expression is meaningful for in-vitro construction of three-dimensional corneal stroma tissue, with higher success rates and less time required. A previous report from our group also showed that SP enhanced collagen remodeling and expression of collagen type III in primary tendon cells, or tenocytes [33]. However, the effect of SP on collagen expression seems different in other cell types. It has been reported that SP inhibits collagen synthesis of rat myocardial fibroblasts [34] and human lung fibroblasts [35]. More work needs to be done in the future to elucidate the differences observed in different cell types.

Silk has advantages in mechanical strength, biocompatibility, and controllable biodegradability. Therefore, it has been widely used in multiple tissue engineering applications, including skin, bone, cartilage, tendon, cornea, and so forth [3638]. Based on the well-organized collagen structures of normal cornea, aligned silk scaffolds have been designed and found to be efficient in supporting keratocyte proliferation, arrangement, and ECM deposition [3942]. Nevertheless, most of these studies were only evaluated in a traditional two-dimensional model with merely one single silk membrane or silk film [10]. Since corneal stroma is a three-dimensional orthogonally aligned structure, it is important to construct bioengineered corneal stroma tissue with multi-layered aligned silk membranes to closely mimic the in-vivo corneal microenvironment. In 2009, Lawrence and collaborators reported a multi-layered film construct that was assembled with seven layers of porous/flat silk films with keratocytes seeded on each layer [39]. One year later, they improved this concept of three-dimensional constructs of corneal stroma with stacked arginine-glycine-aspartate (RGD)-coupled porous/patterned silk films [40]. The same group reported early this year that they had prepared three-dimensional functional corneal stromal tissue by orthogonally stacking aligned silk films seeded with human corneal stromal stem cells (hCSSCs) cultured for 9 weeks [43]. Our present study fabricated bioengineered human corneal stroma tissue in a similar way by stacking orthogonally aligned patterned silk membranes. With the advantages of PDLSCs, and the promotion effect of SP on the expression of collagens, our construct supports cell growth and new tissue formation between every two silk membranes over a relatively short culture period (18 days). High expression of the main collagen types found in normal human corneal stroma (COL I and especially COL V) was found in this bioengineered human corneal stroma tissue. The critical proteoglycans, LUM and KERA, were also expressed in the constructs. This new three-dimensional bioengineered human corneal stroma tissue model improves current corneal tissue engineering and shows a potential for future clinical applications. Meanwhile, by closely mimicking the microenvironment of in-vivo corneal stroma, this model is useful for evaluating cell behavior and function in vitro.