Two cell types, adipocytes and osteoblasts; one common cellular ancestor, human mesenchymal stem cells (hMSCs). The distinguishing factors? Cell shape, cytoskeletal tension and RhoA–ROCK signalling, according to Christopher Chen's group, who report their findings in Developmental Cell.

Confirming previous reports, adipocyte differentiation was favoured when hMSCs were plated at a high density and in adipogenic culture medium, whereas a low plating density favoured osteoblastic commitment in the presence of osteogenic factors. Many cellular characteristics are affected by changes in cell densities — for example, increasing cell density decreases spreading and cell–substrate adhesion. So Chen and colleagues explored the effect of cell shape on stem-cell commitment. They generated 'islands' of fibronectin to control the spreading of hMSCs. hMSCs that were plated on small islands were more rounded, as they couldn't spread, and gave rise to adipocytes, whereas those seeded onto large islands could spread, and they generated osteoblasts.

Because cellular shape and cytoskeletal dynamics are intimately linked, the authors examined whether the actin cytoskeleton was somehow involved in the shape-induced commitment process. They disrupted the actin cytoskeleton using an inhibitor of ROCK, Y27632, to inhibit myosin-generated cytoskeletal tension. hMSCs that were seeded at low density, and therefore expected to become osteoblasts, instead expressed adipocyte-specific markers when treated with Y27632, which implies that the actomyosin cytoskeleton influences hMSC commitment.

Could there be a connection, therefore, between cell spreading, cytoskeletal tension and cell fate? RhoA regulates cytoskeletal tension in other cell types and could be a candidate for transducing signals that are induced by cell-shape changes in hMSCs to changes that affect cell fate. To test this, the authors plated cells out at low or high densities (favouring osteogenesis and adipogenesis, respectively) in osteogenic or adipogenic media and measured levels of active RhoA. RhoA activity was highest in low-density cultures in osteogenic media and correlated with increased cell spreading and ROCK activity.

These results led the authors to investigate whether RhoA could directly influence cell-fate choices in hMSCs. In culture medium lacking any differentiating factors, hMSCs adenovirally infected with a constitutively active form of RhoA became osteoblasts, whereas those infected with dominant-negative RhoA formed adipocytes. Conversely, dominant-negative RhoA blocked osteogenesis that was induced by placing the cells in osteogenic media, and encouraged these cells to adopt an adipogenic fate; and constitutively active RhoA inhibited adipocyte formation in cells that were given adipogenic media and redirected them to become osteoblasts. So RhoA can replace soluble-factor signalling, but this depends on its ability to affect the cytoskeleton — disrupting the actin cytoskeleton by inhibiting ROCK or myosin abrogated the ability of constitutively active RhoA to induce osteogenesis.

But manipulating RhoA signalling did not similarly redirect hMSC commitment in response to cell shape. Cells that were plated on small islands, which were therefore round and expected to become adipocytes, could not be persuaded to become osteoblasts when infected with constitutively active RhoA, whereas adipogenesis was blocked simply by inducing cell spreading, even when dominant-negative RhoA was expressed. So in this case, both cell shape and RhoA activity are required to specify hMSC cell fate. By contrast, constitutively activated ROCK seemed to be able to control the fate of the cells regardless of their shape, which implies that ROCK functions downstream of both soluble factors and cell shape in regulating hMSC commitment.