For the two KO models CX and CX

For the two KO models (CX30 and CX43), fluorescence recovery is almost always best fitted with a 2-component exponential (Fig. 6A and D). Statistical analysis of the parameters (Table 1) shows that in both cases the two time scales and the fraction of recovery associated with the fast intracellular component are not different from control conditions. The only statistically significant difference is the fraction of recovered fluorescence associated with the slow intercellular component (Amplitude2) that is between 60 and 70% of the control value. We note that this decay is of the same magnitude as the decrease of the intercellular coupling strength parameter G in the KO astrocytes, obtained with our mathematical model above (between 27% and 50%). Therefore, our analysis suggests that CX30KO and CX43KO astrocytes are coupled to a lower number (or a lower volume) of astrocytes, as intuitively expected. The results with the Double KO CX30/CX43 are more complex. One half of our measured cells are best fitted with 1-component kinetics and the rest with 2-component kinetics (Fig. 6A). The cells for which 1-component kinetics is the best exhibit the same kinetic parameters (recovery time scale and amplitude) as the CBX cells (Table 1). However, the Double KO cells that are best fitted with 2-component kinetics display kinetic parameters that share common features with single KOs: compared to control, their fast intracellular kinetics are unchanged while the fraction of recovered fluorescence contributed by the slow intercellular component is much smaller. However, in addition, the time scale of the slow intercellular component of the Double KO is significantly larger, which is not the case for single KOs. Therefore, our analysis of the Double KO mutants indicates SR101 recovery kinetics that are intermediate between CBX and the single KOs.
Discussion We here report how the gap-FRAP technique, so far mainly used in cultured cell systems, can be adapted to assess gap junction-mediated communication in astrocytes studied in acute hippocampal slices from the adult mouse. Its accuracy to detect changes in GJC was demonstrated by using pharmacological as well as genetic tools to reduce or abolish GJC in astrocytes. In addition, we carried out a data fitting and model comparison that validated experimental data and interpretation. The development of this approach provides an easy going technique that allows assessing the level of GJC in astrocytes at late ages, a feature that was so far difficult to achieve and was time consuming when using classical dye coupling experiments in adult acute cannabinoid receptors slices. However, as it has been reported (Schnell et al., 2012) the active uptake of SR101 by astrocytes depends on brain areas; such feature should thus be taken into account for GJC measurement by gap-FRAP technique in regions other than the hippocampus. Up-to-now the sole work reported to use gap-FRAP technique in acute slices from aged brain was performed using CDCF (dicarboxy-dichlorofluorescein diacetate), a gap junction channel- and cell membrane-permeable dye that becomes membrane-impermeable after de-esterification (Cotrina et al., 2001). This study reported that while the expression of Cx43 and Cx30 is rather stable, according to age, GJC tends to decrease. However CDCF, based on its permeability properties, is expected to enter all cells, which is certainly a handicap in a tissue like the brain composed of a variety of cellular populations including neurons, glial and endothelial cells. Thus, although in adult mouse the main cell population connected by gap junctions is the astrocytes, it cannot be assumed that CDCF recovery after bleaching is only mediated by GJC between these cells. We overpassed this problem by using SR101, a dye that is specifically up-taken by astrocytes thanks to an active transport (Schnell et al., 2012) and passes through gap junction channels (Nimmerjahn et al., 2004). We verified that this was the case in 9 month-old hippocampal slices and thus considered to have in hand a method in which solely astrocytes were loaded. We have also been careful to avoid the secondary staining of oligodendrocytes that starts to be detected after 40 min (Hagos and Hulsmann, 2016; Hill and Grutzendler, 2014). Indeed, this astrocyte-to-oligodendrocyte coupling occurs through heterotypic gap junctions forming a panglial network that is, however, minor in the hippocampus compared to other brain region like the thalamus (Griemsmann et al., 2015). In our experimental condition, the use of the Aldh1L1-eGFP mouse indicated that more than 93% of the SR101 loaded cells expressed eGFP and thus were identified as astrocytes. Moreover, it has also been reported that low concentration (1 μM) of SR101 induces a direct effect on pyramidal neuron membrane structures, leading to a reduction in action potential firing threshold, and a long-term increase in neuronal excitability and synaptic efficacy (Kang et al., 2010; Garaschuk, 2013). So, we do not exclude that in our experiments such SR101-induced effects on neuronal activity occurred. However, we have previously shown that in the CA1 region of the hippocampus, dye coupling (tested with sulforhodamine B and biocytin) is not controlled by changes in neuronal activity including epileptic-like discharges or TTX application (Rouach et al., 2008). Moreover,this technical approach of GJC is mainly proposed to compare different situations (animal models of brain diseases or injuries) so we consider that if the bleaching process induces cellular changes, these changes are supposed to be similar in both models, allowing to compare their level of GJC. For instance, we have recently used such technique to compare the level of coupling in wild type mice and in APP/PS1 mice (a murine model of Alzheimer’s disease) (see Yi et al., 2017). Consequently, based on these considerations we consider that the gap-FRAP technique provides a trustable assessment of GJC in adult hippocampal slices and can be used to determine whether GJC is affected in different pathological situations as recently applied in a murine model of Alzheimer’s disease (Yi et al., 2016).