Dynamical properties

Analysis of the proposed dynamical systems shows that the time to convergence can widely depend on the initial condition. Convergence can be relatively very slow when initial conditions are near a threshold around which the system converges to two or more different outcomes. This is for example the case when 2 or more "switch elements" are at roughly equal concentrations, higher than that of others (the more elements in competition, the slower the competition becomes). It is interesting to note that slow effects are observed in induced-transdifferentiation experiments, and in cell fusion experiments.

• Fibroblasts reprogrammed to T-cell-like cells need to be incubated for many days before they acquire detectable T-cell characteristics (Håkelien et al., 2002). This may be due to the fact that fibroblast master genes are expressed at a high level, and the counter-acting T-cell master genes, introduced by permeabilisation of the membranes, are also present at a high concentration. An effect of the relative levels of cytoplasmic factor concentrations could be tested by incubation in T-cell and fibroblast cytoplasmic extracts, mixed at different ratios. Further investigation of master networks could involve incubation of cells in cytoplasmic extracts of 3 or more cells-types (or transient misexpression, at controlled levels, of antagonistic master genes).
• In hepatoma-fibroblasts hybrids, extinction of albumin production can take days (Mével-Ninio and Weiss, 1981). Most interestingly, some hybrids show reexpression of albumin after extinction. These two outcomes can be accounted for by the models proposed above: when two antagonistic "switch elements" are coexpressed at a high level (which probably corresponds to the fusion experiments, as upon fusion the protein contents of the cells, which are of different phenotypes, are mixed), it is possible for the system to revert to a state where all switch elements are turned off (total extinction), or for the two switch elements to decrease to a low level, before the trajectory of one of them picks up and goes back to a high state (extinction followed by reexpression).
• Activation of the myogenic phenotype also takes place on the scale of days, when muscle cells are fused to various other cell types, a delay which was suggested not to be linked to DNA duplication (Blau et al., 1985; see Blau and Blakely, 1999, for an extensive review).

Also, it could be that the progressive upregulation of differentiation-related genes observed during hematopoietic development is a cell-autonomous consequence of the slow dynamics of a switch network.