Estimated and experimentally measured charge transfer
Because a single evoked IPSC lasted, in most of the recorded cells, longer than 30 ms (interstimulus interval used during HFS) (Fig. S1E), we asked whether the increased charge transfer observed post-lesion (Fig. 5E,F, main text) might be caused by an increased summation of single IPSCs with a prolonged decay time constant. To address this question we estimated the charge transfer that would result by a pure summation of single IPSCs evoked by the same HFS protocol. We first calculated the charge transfer of a single evoked IPSCs from 30 ms after the stimulation (time in which the second stimulus would occur during the HFS protocol) till the end of the postsynaptic event (Fig. S1E). This “residual” charge transfer was referred as tail charge transfer. In line with the prolonged decay time found post-lesion the tail charge transfer of single IPSCs was higher in neurons from lesion-treated rats (sham: 0.45 ± 0.08 pC, 34.15 ± 2.56 % of the total current, 17 cells from 3 animals; lesion: 0.86 ± 0.13 pC, 49.47 ± 2.49 % of the total current, 13 cells from 5 animals). Since a linear correlation was found between amplitude and charge transfer of single IPSC, as shown in Fig. S1D, we could compute an approximative estimation of the tail charge transfer of each IPSC measured during HFS in proportion with their mean amplitude. To estimate the charge transfer during the entire duration of the HFS protocol we finally summated the estimated tail charge transfer of the 40 consecutive eIPSCs measured during HFS (Fig. S2A). Remarkably, the estimated charge transfer was less than 50% of the value obtained experimentally both in sham and lesion-treated animals (sham 7.16 pC, 41.51 % of the experimentally measured value; lesion: 19.52 pC, 47.23 % of the experimentally measured value) (Fig. S2B).
Fig.S1 Illustration of the methods used to calculate the charge transfer of eIPSCc during HFS and to estimate the charge transfer based on single evoked IPSCs kinetics. A) Entire area defined by the eIPSCs (above the eIPSCs and below the baseline) during the high frequency stimulation (40 pulses, 33 Hz). B) Areas defined by each eIPSC. C) Area used to measure the charge transfer (in blue) obtained by subtracting the area shown in B from the entire area shown in A. D) Graph showing that peak amplitude and charge transfer of evoked IPSCs are linearly correlated. The graph contains data from 7 neurons from 4 animals. In each cell the extracellular stimulation intensity was varied (3 different levels of intensity were chosen) to produce IPSCs of different amplitude. E) Two representative IPSCs elicited by a single presynaptic stimulation from a sham (top) and a lesion-treated animal (bottom) are shown to illustrate the area from which the tail charge transfer was calculated (in blue). Note that 30 ms after the stimulation artifact corresponds to the time when the second stimulation would take place during the HFS protocol (C). At this time point the single eIPSCs are still not terminated.
Fig.S2 Estimated and experimentally measured charge transfer. A) Estimated cumulative tail charge transfer. Note that the value at the last (40th) stimulus corresponds to the estimated charge transfer. B) Comparison between estimated (estim.) and experimentally (exp.) measured charge transfer in sham and lesion-treated animals.
Influence of dendritic filtering on mIPSCs amplitude and kinetics
Because our whole cell recordings were performed from the soma of pyramidal neurons, inhibitory synapses located distal from the cell’s body will generate IPSCs with relatively slow kinetics and small amplitude (in comparison with synapses located close to the soma) due to dendritic filtering (Magee, Nat Rev Neurosci 2000). The prolonged rise and decay time post-lesion might therefore arise from a lesion-induced differential contribution of distal versus perisomatically located synapses to the recorded mIPSCs. This could be a plausible scenario since it is known that different subpopulations of interneurons (which might be differentially affected by the lesion) innervate distinct subcellular domains of principal cells (Markram et al., Nat Rev Neurosci 2004). Nonetheless the fact that the prolonged rise and decay time was not accompanied by a reduction in the mean mISPC amplitude (Fig. 2C, main text) already argued against this hypothesis. We additionally plotted the mean mIPSC amplitude versus the mean mIPSC rise time (or mIPSC decay time constant) in each recorded neuron (Fig. S3A,B). If the prolonged kinetics post-lesion were due to changes in dendritic filtering we should have observed two separated clusters of cells. Cells from sham-operated animals should be located down (fast rise/decay time) on the right side of the graph (big amplitude) while cells from lesion-treated rats should be located up (slow rise/decay time) on the left side (small amplitude). This separation was however not visible since the prolonged rise/decay time (more cells from lesioned slices in the upper part of the graphs) was not accompanied by a reduced IPSCs amplitude (cells from the two groups equally distributed through the x axes). This result indicates that changes in dendritic filtering unlikely contribute to the prolonged mIPSC kinetics observed post-lesion. Furthermore the negative slope, visible in Fig. S3A and B in both sham and lesioned animals, suggests an unchanged somato-dendritic distribution of functional synapses after lesion.
Fig.S3 Influence of dendritic filtering on mIPSCs amplitude and kinetics. Graphs showing the correlation between mIPSCs amplitude and A) mIPSCs rise time or B) mIPSCs decay time. Neurons from sham and lesion-treated animals are not separated in two different clusters suggesting that potential changes in dendritic filtering should not contribute to the prolonged mIPSCs kinetics post-lesion.
Fig.S4 Complete Western blot for two lesion-treated and two sham-operated animals showing the specificity of the antibodies anti-GAD67, GAD65 and α-tubulin. The marker at the left side of the membranes (precision plus protein standards, dual color, BIO-RAD) helps to identify the molecular mass in kDa. For GAD65 the lower, more intense band, corresponding to 65 kDa, was used for analysis.