Moreover after the initial bleaching of the fluorescent mole

Moreover, after the initial bleaching of the fluorescent molecules in the targeted cell exposed to laser illumination, the neighboring reservoir Docetaxel synthesis are expected to give in return new fluorescent ones if the cells are coupled together by gap junction channels. This supposes that the level of fluorescence in the reservoir cells should decrease as fluorescence comes back in the bleached cell (see Delèze et al., 2001). This feature was qualitatively observed in gap-FRAP in SR101-loaded astrocytes from acute slices. Indeed, in astrocytes close to the bleached cell, fluorescence significantly decreased with time reaching 86 ± 6% (n = 9) of fluorescence, at t = 20 min, compared to its level before illumination, while in astrocytes located far from the targeted cell, the change in fluorescence level was not detectable 99 ± 1% (n = 16) (Fig. 4E). In the classical gap-FRAP method initially described between two coupled cells, there is a direct relationship between the fluorescence recovery in the bleached cell and the donor cell (Delèze et al., 2001). In our case, the situation is more complex since we are dealing with a bleached cell and a network of donor cells; also the fluorescence recovery of the bleached cell is not completed after 1250 s. Thus the rule of fluorescence loss in donor cells does not follow this direct relationship initially observed for a pair of cultured cells. This explains why the kinetics of fluorescence in donor cells continues to decrease while the fluorescence of the bleached cell reaches a plateau. In the presence of CBX, as expected, no change in the fluorescence was observed in astrocytes whether located close or far from the bleached cell (97 ± 2%, n = 6 and 98 ± 2%, n = 5, respectively) (Fig. 5F) while the rate of fluorescence recovery in this astrocyte was greatly reduced as described above (Fig. 5C). When applied to Cx30 KO and Cx43 KO, the estimates from our mathematical model for the intercellular coupling strength G were clearly nonzero but less than 1.0 (G = 0.27 and 0.50, respectively). This suggests that intercellular GJC is severely impaired in Cx30 KO and Cx43 KO hippocampal astrocytes as previously reported (see Rouach et al., 2008). This is in strong contrast with the value estimated in control conditions above, G = 1.0. The interpretation of G = 1.0 in the model is that the rate of intercellular diffusion of SR101 is not significantly different from its intracellular diffusion rate. This suggests that in control condition, the targeted astrocyte and its coupled neighbors can be considered as a unique cell from the point of view of SR101 diffusion, this is directly due to the well-known very high level of GJC in hippocampal astrocytes. Indeed, astroglial networking can involve about 80 to more than 100 astrocytes when biocytin is used as an intercellular tracer (Blomstrand et al., 2004; Xu et al., 2010; Strohschein et al., 2011). In contrast, in Cx30 KO and Cx43 KO, reduction of GJC strength would partially disrupt the astroglial networking. We tested this idea using non-linear fitting of the gap-FRAP curves. In the case of an isolated astrocyte, with no coupling to any other cell, one expects fluorescence recovery to exhibit 1-component exponential kinetics, since fluorescence recovers from a single, intracellular source of SR101. However, when the astrocyte is coupled to other cells, fluorescence recovers from two sources (intracellular SR101 and the SR101 initially present in the coupled cells), and one expects a 2-component kinetics with a fast component due to the intracellular SR101 diffusion and a slower one due to the intercellular SR101 diffusion. We thus fitted each gap-FRAP curves with 1-component or 2-component exponential functions and used Akaike Information Criterion to compare the quality of the fits (see Methods). Fig. 6A shows the fraction of cells in each condition for which the best model was found to be 1- component (white) or the 2-component (black) kinetics. With CBX, we found that the best model is 1-component kinetics for most of the cells, in agreement with our hypothesis above. Marginal improvements of the 2-component fits are not enough to compensate for the larger number of fitting parameters (Fig. 6B). We interpret the corresponding time constant of the single exponential (referred to as the “time scale” below) Kcbx = 0.00147 s−1 as the time scale for intracellular diffusion of SR101 (see Table 1). In control conditions however, the 2-component scenario is the best fit for all our experiments (Fig. 6C), again as expected. A minor fraction of recovery comes from a fast component with estimated time scale Kctrl1 = 0.00362 s−1 (Table 1), whereas a larger recovery fraction is due to a component that is almost tenfold slower (Kctlr2 = 0.00049 s−1). Considering the fraction of recovery associated with each component, we associate the fast component with intracellular diffusion and the slower one with intercellular diffusion. Comparing the values of the time scales between CTRL and CBX is difficult because they result from different fitting models. However, when we compare Kctrl1 and Kcbx, the decrease due to CBX is similar to the reduction of the apparent diffusion coefficient suggested by our mathematical model above (Kctrl1/Kcbx = 2.5 whereas Dctrl/Dcbx = 2.2). We therefore conclude that this analysis confirms the decrease of apparent diffusion kinetics with CBX.