Cells can sense forces applied to them, but also the stiffness of their environment. These are two different phenomena, and here we investigate the mechanosensitivity of the 2nd kind: how the cell can measure an elastic modulus at a single point of adhesion-and how the cell can receive and interpret the chemical signal released from the sensor. Our model uses the example of large latent complex of TGF-β as a sensor. Stochastic theory gives the rate of breaking of latent complex, which initiates the signaling feedback loop after the active TGF-β release and leads to a change of cell phenotype driven by the α-smooth muscle actin. We investigate the dynamic and steady-state behaviors of the model, comparing them with experiments. In particular, we analyse the timescale of approach to the steady state, the stability of the non-linear dynamical system, and how the steady-state concentrations of the key markers vary depending on the elasticity of the substrate. We discover a crossover region for values of substrate elasticity closely corresponding to that of the fibroblast to myofibroblast transition. We suggest that the cell could actively vary the parameters of its dynamic feedback loop to 'choose' the position of the transition region and the range of substrate elasticity that it can detect. In this way, the theory offers the unifying mechanism for a variety of phenomena, such as the myofibroblast conversion in fibrosis of wounds and lungs and smooth muscle cell dysfunction in cardiac disease.