The biology of MSCs has been broadly studied [1, 22, 61, 62] because of their therapeutic potential. Therefore, we decided to study the function of CD90, one main immunophenotypical marker of MSCs, in order to better understand its relationship with MSC morphology, proliferation, and differentiation. Here, we used MSCs isolated from three sources (dental pulp, adipose tissue, and amniotic fluid) to verify whether the effects caused by CD90 ablation would be source-specific. We used lentivirus-mediated CD90-shRNAi to stably reduce CD90 expression and further evaluate its function in MSCs. Here, we generated MSC lines transduced with lentivirus-carrying small hairpins (shRNA) targeting CD90. After the establishment of CD90-shRNA expressing MSCs (in shRNA CD90 MSCs), the reduction of CD90 expression was confirmed in immunophenotypical analysis using flow cytometry (Fig. 1). We subsequently evaluated the immunophenotypic profiles of modified MSCs, in addition to CD90. As expected, we found that these cells expressed the positive MSC markers CD29, CD73, and CD105, and did not express the following cell markers: CD14, CD31, CD34, CD45, CD106, and HLA-DR (Additional file 2: Figure S1). Surprisingly, we found that a knockdown in CD90 expression in all CD90-shRNAi MSCs obtained here led to a reduction in the CD44 and CD166 expression (Fig. 5).

The role of CD166 in MSCs has not been determined to date. However, CD44 (hyaluronan receptor) [63] is expressed by a large number of cells and is involved in cell adhesion, migration, and homing in MSCs [6466]. Furthermore, CD44 has been recognized as a stem cell marker for several types of cancer and is strongly linked to metastatic spread. It has been shown that the reduction of CD44 in cancer stem cells caused them to differentiate into non-cancer stem cells [67]. The receptor CD166 (activated leukocyte cell adhesion molecule, ALCAM) is a member of the immunoglobulin superfamily of cell adhesion molecules [68, 69] and is present in undifferentiated MSCs and other cell types [70]. Like CD44, CD166 has been shown to participate in tumour invasion [7173]. A recent study using liver cancer cell lines relates the close interaction between CD44 and CD166. The authors showed that a knockdown of CD166 inhibits the expression of CD44 via the NFkB pathway [74].

CD90 has also been identified as a candidate marker for adult stem cells. Few studies have shown a functional association between CD90 and CD44 or CD166 markers. Strikingly, the few data showing association come from cancer stem cell research: CD90, CD44, and CD166 are notably considered cancer stem cell markers [44, 70, 71, 75]. Our results showed that the knockdown of CD90 leads to a decrease in CD44 and CD166 expression, which could indicate a shift in the stemness state of MSCs towards a state more susceptible to differentiation.

CD90 has been linked to the spindle-shape of lung fibroblasts. Observing lung fibroblasts sorted on the basis of CD90 expression, Phipps and co-workers [76] affirmed that the lung CD90 fibroblast subpopulation showed a more polygonal shape than the spindle-shaped CD90+ fibroblasts. In contrast to these observations, in our study, a reduction in CD90 expression in shRNA CD90 MSCs and CD90-negative MSCs did not present altered morphology or proliferation rate when compared to control cells (Fig. 3). Here, we also demonstrated that a reduced expression of CD90 does not affect the immunosuppressive activity of MSCs on lymphocyte proliferation in vitro (Fig. 5), a very important therapeutic MSC property.

We carried out assays to investigate the differentiation of CD90-ablated MSCs into osteogenic and adipogenic lineages. In our differentiation assays, shRNA CD90 MSCs and CD90-negative MSCs showed a higher rate of adipogenic differentiation when compared to the controls (Fig. 8). In the same way, an enhanced osteogenic differentiation was observed in samples of shRNA CD90 MSCs and CD90-negative MSCs. Alizarin Red S staining showed that CD90-negative MSCs, the CD90 negative fraction of shRNA CD90 MSCs, accumulated more mineralized matrix than shRNA CD90 MSCs (Figs. 6 and 7). According to our results, the knockdown of CD90 expression in MSCs facilitates osteogenic and adipogenic differentiation.

Recently, Woeller and colleagues [77] showed that CD90 controls adipogenesis. They had previously observed that CD90-null mice gain weight at a faster rate, and that ectopic overexpression of CD90 blocked adipogenesis [77]. They also stated that, although pre-adipocyte fibroblasts expressed CD90, fat adipocytes presented almost undetectable CD90 levels. In agreement with this study, we also observed that a loss of CD90 expression in MSCs increased the production of adipogenic matrix in vitro. Based on their study, Woeller and colleagues [77] suggest that CD90 could be a new therapeutic target for obesity. However, our results indicate that this differentiation facilitation related to decreased CD90 expression is not only for adipogenic differentiation, since we observed the same facilitation for osteogenic differentiation. Our data indicate that the knockdown of CD90 seems to lower the stemness guard of MSCs, thereby enabling further differentiation when in the presence of the specific stimuli.

The finding that the level of CD90 regulates both MSC adipogenesis and osteogenesis is very interesting, because it is well accepted that differentiation stimuli usually cause an “inverse relationship” between adipogenic and osteogenic differentiation [78], although the molecular pathways that can converge into adipogenesis and osteogenesis have not been completely elucidated. Here, we demonstrated that the production of mineralized matrix directly correlates with the level of CD90 ablation: higher in the samples of CD90-negative MSCs than shRNA CD90 MSCs. It is unclear how CD90 can affect adipogenesis. However, it has been demonstrated that CD90 also regulates RhoGTPase activity in fibroblasts. Exogenous expression on CD90-non-expressing fibroblasts results in Rho GTPase activation [42]. CD90 participates in many signalling pathways, and it is becoming clear that, although CD90 has been recognized as a plain cell marker, it is also an important regulator of MSC signalling [79]. In order to accurately understand the effects of CD90 on all cis– and trans-signalling networks that it participates in, significant further studies are required. Improving our knowledge of these mechanisms may allow a better understanding of MSC stemness and differentiation.

An increasing number of studies have shown that MSCs from different sources display significantly diverse properties and characteristics that may impact on their future therapeutic applications. The capacity of differentiation may vary according to the cell source [7, 59, 60]. In agreement with previous reports [59, 8082], we observed that cells from the dental pulp tissue and aminiotic fluid produced a larger quantity of osteogenic matrix than cells from adipose tissue (Figs. 6 and 7). Despite the expected variance in the differentiation potential among MSCs from different tissues [83, 84], we confirm that a reduction in CD90 expression leads to a more efficient osteogenic differentiation, irrespective of the source.

CD90 is a GPI-anchored protein expressed in various cell types. In general, it appears to influence cell proliferation, differentiation, migration, and survival. The functions of CD90 are tissue- and cell-specific and, in the present work, we found that shRNA-induced knockdown in human MSCs increases the differentiation efficiency of these cells. Our group previously showed that CD90 expression could be used as an indicator to follow the differentiation commitment degree of MSCs. Immediately after the induction of differentiation, a progressive decrease in CD90 mRNA level correlates with the degree of differentiation observed [25]. It is important to reiterate that the ablation of CD90 expression did not result in a spontaneous differentiation. However, it facilitated MSC differentiation in the presence of inductors, indicating that CD90 may play an important role in maintaining the undifferentiated state of MSCs, perhaps by acting as an obstacle to be overcome during the early steps of cellular differentiation commitment.