MSC responses to variations in ECM stiffness have been studied previously in mice [8, 34], rats [35], and humans [36, 37]. The present study examined this phenomenon in hMSCs to provide an accurate theoretical basis for clinical treatment. hMSCs are often subjected to complex interactions to induce osteogenic differentiation, including chemical and physical stimuli [38]. Because it is difficult to provide factors in vivo at the high concentrations that they are often provided in vitro and in a localized manner, this work focused solely on the ability of ECM stiffness to affect hMSC differentiation.

Cell morphology differs based on ECM stiffness [39, 40]. Consistently we showed that hMSCs displayed an oval, adipocyte-like appearance when cultured on 13–16 kPa ECM, but they presented a polygonal, osteoblastic morphology on 62–68 kPa ECM (Fig. 2a). However, another report found no association between ECM stiffness and MSC morphology, although similar effects were observed for osteogenic marker expression [14]. Thus, the connection among ECM stiffness and hMSC differentiation requires further study—in particular, the surface molecular mechanisms that sensed matrix stiffness.

Integrin β1 is a key molecule involved in the cellular response to substrate stiffness and related effects on differentiation potential [14, 41]. Our results demonstrated that integrin β1 expression was unaffected by ECM stiffness, but localized to the cell surface or cytoplasm when cells were cultured on soft or stiff substrates, respectively; this is opposite to the findings of Du et al. [42]. Du et al. coated the gel coated with type 1 collagen but we used fibronectin. Because different integrins have corresponding ligand proteins, that have different effects on the differentiation of cells [43], we suggest that this difference in protein might have caused the distribution of integrin β1 to be different. Then, we explored the role of integrin α subunits in the process.

Integrin α5 plays an important role in osteogenic differentiation, consistent with results of the present study. In particular, our results suggest that integrin α5 protein was localized to the cell surface, indicating that α5 did recognize the external ligands on the cell surface. Gandavarapu et al. [44] and Hogrebe and Gooch [45] demonstrated previously that increasing the binding strength of integrin α5 to ECM by adding the peptide c (RRETAWA) and RGD fragments, and increasing the site density of integrin α5, can effectively induce osteogenesis in cells cultured on the stiff ECM substrate. Increased integrin α5 could improve osteogenic differentiation of hMSCs.

Integrin α5 interacts with several signaling molecules [46], including FAK and ERK, which play important regulatory roles in matrix-induced osteogenic differentiation and gene expression [47]. Similar to the results of previous studies, our results indicated that FAK and ERK expression was markedly increased on 62–68 kPa ECM, but was not affected by blockade of integrin α5. When hMSCs were cultured in osteogenic medium on tunable polyacrylamide hydrogels, ROCK, FAK, and ERK1/2 expression was altered upon knockdown of integrin α2 by siRNA [48]. These results are different from our results possibly because we only manipulated ECM stiffness to affect hMSC differentiation. Mechanism of osteogenic differentiation would be different with changing environment.

In comparison, expression of PI3K could regulate hMSC osteogenic differentiation. They were elevated during osteogenesis induced by dexamethasone or low-intensity ultrasound, and PI3K/Akt played a critical role in this process [34, 49]. Previous reports are consistent with our results showing increased p-Akt expression during matrix-induced osteogenic differentiation. In vitro, osteoclast supernatant can also regulate osteoblast proliferation and differentiation through PI3K/Akt [50]. However, a few studies have found that PI3K/Akt signaling inhibits osteogenic differentiation [51], and these studies have focused on targeted differentiation in cells such as the precursor osteoblast cell line MC3T3-E1. In this study, hMSCs were obtained from healthy subjects to explore the effect of Akt and p-Akt signaling on the osteogenic differentiation of hMSCs, and detected significant increases in p-Akt and osteogenic differentiation on the 62–68 kPa ECM. In summary, PI3K/Akt signaling is involved in hMSC osteogenic differentiation. Osteogenic differentiation is accompanied by the expression of Akt protein downstream of the signaling pathway. In our follow-up experiments, Akt and p-Akt were increased after we blocked integrin α5. This is counter to the previous reports. We believe that when our blocking antibody blocks integrin α5, the blocked protein site may trigger a signaling molecule that activates Akt, resulting in increased Akt expression. This confirms that Akt is not regulated by integrin α5 alone and that it may be under the control of other signaling proteins.

Previous studies have shown that Wnt signaling is responsive to matrix stiffness [52]. Microarray screening results have revealed a significant promotion of the canonical Wnt/β-catenin pathway by stiffer ECM, which was confirmed by Du et al. [53]. The Wnt/β-catenin pathway can control diverse cell behaviors including cell adhesion, migration, differentiation, and proliferation, behaviors which respond significantly to ECM stiffness [26]. Inhibition of Akt has been shown to only partially block the effect of ECM stiffness on the β-catenin pathway [53], indicating that Akt contributes to but is not required for this process. Our results showed that with inhibition of Akt, β-catenin levels did not change. The integrin-activated β-catenin pathway can promote the Wnt signal by ECM stiffness and the regulation of hMSC differentiation by forming a positive feedback loop [53]. Our results showed that the promotion of the canonical Wnt/β-catenin pathway was not dependent on stiffness per se, but was caused by the accumulation of β-catenin.

Wnt proteins transduce their signals through disheveled proteins to inhibit GSK-3β, leading to accumulation of cytosolic β-catenin [26, 54, 55]. Regarding the role of integrins in the regulation of Wnt signaling, inhibition of integrin α5 by a functional blocking antibody significantly increased the levels of phosphorylated GSK-3β, but not those of β-catenin. As an important downstream element of integrin signaling, the FAK/Akt pathway is well documented as a regulator of GSK-3β [56, 57]. We found that 62–68 kPa ECM could increase the expression of β-catenin and phosphorylated GSK-3β. Accumulation of β-catenin was not mediated by Akt activity, as an Akt inhibitor blocked the differences in the levels of β-catenin between the stiff and the soft ECMs.