Depolarizing GABA/glycine synaptic events switch from excitation to inhibition during frequency increases

By acting on their ionotropic chloride channel receptors, GABA and glycine represent the major inhibitory transmitters of the central nervous system. Nevertheless, in various brain structures, depolarizing GABAergic/glycinergic postsynaptic potentials (dGPSPs) lead to dual inhibitory (shunting) and excitatory components, the functional consequences of which remain poorly acknowledged. Indeed, the extent to which each component prevails during dGPSP is unclear. Understanding the mechanisms predicting the dGPSP outcome on neural network activity is therefore a major issue in neurobiology. By combining electrophysiological recordings of spinal embryonic mouse motoneurons and modelling study, we demonstrate that increasing the chloride conductance (gCl) favors inhibition either during a single dGPSP or during trains in which gCl summates. Finally, based on this summation mechanism, the excitatory effect of EPSPs is overcome by dGPSPs in a frequency-dependent manner. These results reveal an important mechanism by which dGPSPs protect against the overexcitation of neural excitatory circuits.


Supplementary Figure Legends
responses of an E13.5 MN to an isoguvacine puff at a series of holding potentials in voltageclamp mode. A2, Amplitude of the responses that were obtained in A1 against the holding potential E m . The slope of the regression line yields the peak conductance (g Clp ) as activated by the isoguvacine puff. This line intercepts the zero current at the E Cl value (-41 mV in the illustrated example). B, Original method that was developed to quantify the inhibitory and excitatory isoguvacine effects, as expressed by R GABA . This parameter is a ratio and is calculated as the percentage of spikes that were triggered by depolarizing current pulses during the isoguvacine puff divided by the percentage of spikes that were triggered during control trials preceding the puff (at least 4 series as illustrated). B1-B2, Series of 4 trials of depolarizing pulses preceded the puff with a fifth pulse 20 ms prior to puff onset (B1) or 20 ms following puff onset In an E17.5 MN, E Cl was measured before (A1) and after (A2) a series of isoguvacine puffs (1 Hz for 1 min). Following the repetitive activation of GABA A R, the dGPSP amplitude decreased from 4.7 mV to 2.0 mV (B1), whereas the E Cl shifted from -77 mV to -64 mV (B2). Figure 5: Fitting of the decay phase of the synaptic conductance. To calculate the steady-state conductance that was reached during dGPSP trains (see text for details), the decay phase was fitted to a single exponential decay curve. The value at t = 0 yields × !"# as used in the equation (1) providing the limit of summated conductances (g Peaklim ): (1) in which N is the dGPSP train frequency, τ is the time constant of dGPSP decay, and K is the coefficient that is used to assimilate the dGPSP to a single exponential decay curve (beginning from a theoretical initial value of × !"# ). Tables   Table 1 E13. 5

Recording and staining
Because a high R in of immature motoneurons may lead to erroneous values of the resting membrane potential (E Rest ), spike threshold (E Thr ) and E Cl , we used equation (2) as proposed by Tyzio and collaborators, which corrects for this inaccuracy 1 . (2) where E'm is the measured value, E 0 m is the compensated value, Rps is the seal resistance (always ≈ 10 GΩ), Rin is the input resistance of the recorded motoneuron, and Eps is the liquid junction potential between the bath and intra-pipette solutions (12.5 mV at E13.5 and 14.3 mV at E17.5). All of the potential values were then corrected using this procedure. However, [Cl -] i , which was calculated from E Cl values, was systematically below [Cl -] intra-pipette , thus suggesting that the potential values produced by the correction were slightly over-compensated.
because of the low HCO 3 conductance of GABA A R in embryonic spinal MNs 2 . Consequently, the E Cl was measured as the zero IGABAAR current crossing voltage. The slope of this I/V curve was an estimation of the g Clp (Suppl. Figure 3A2).

Electrical stimulation of the ventral funiculus
To activate local GABA-/glycinergic interneuronal projections produced on recorded E13.5 MNs, a bipolar stimulating electrode was displaced on the surface of the ipsilateral ventral

Stimulation and GABA A R activation
The injection of current pulses and the triggering of isoguvacine ejection were controlled with a programmable multichannel pulse generator (A.M.P.I. Master-8, Jerusalem, Israel). The threshold current pulse was adjusted to elicit 1 to 3 spikes in a series of 4 pulses that were spaced by 2 s (Suppl. Figure 3B). If this was the case (i.e., the current intensity was sufficiently close to the threshold), a fifth current pulse was generated during the isoguvacine puff. If four spikes were produced, the process was aborted, and the current pulse was slightly reduced before a new series trial was performed. If no spikes occurred of the four current pulses, the process was also aborted; the current pulse intensity was then slightly increased before a new trial was performed.
The series of 5 current pulses (0.5 Hz) was separated by a resting time of 20 s to avoid alterations in E Cl by the overactivation of GABA A R. To explore the inhibition/excitation occurring during the rising phase of the isoguvacine puff, the delay between the fifth current pulse and the onset of the isoguvacine puff was gradually modified (-20, 0, 20, 40, 60 and 80 ms) in successive assays. Each assay was repeated 4 to 10 times. The quantification of inhibition and excitation that were produced by the isoguvacine puff were then expressed by the ratio R GABA.
This parameter represents the percentage of spikes that were triggered by depolarizing current pulses along the isoguvacine puff divided by the percentage of spikes that were triggered before the puff (at least 4 series, as illustrated in Suppl. Figure 3B3). In these experiments, we did not analyze the inhibition/excitation occurring during the falling phase of the response to isoguvacine puff because this phase was not comparable to the falling phase of a physiological synaptic event. Indeed, the time constant of this repolarizing phase (281.7 ms at E13.5 and 317.4 ms at E17.5) was much slower than the respective capacitive time course (33.7 ms and 45.9 ms) (compare Suppl. Figures 2B1-2B3 and Suppl. Figures 2C1-2C3). This difference was ascribed to the wave of isoguvacine traveling on the surface of the recorded MN and progressively reaching the GABA A R sites, even during the repolarizing phase. Therefore, this repolarizing phase was not essentially capacitive in contrast to the falling phase of physiological GABA/glycine synaptic events (dGPSPs).

PASSIVE PROPERTIES OF ALL OF THE COMPARTMENTS
The intra-compartmental potential, E, is described by the following differential equation: where Ileak is the passive leakage current and described by the following equation: and E leak and G leak are the equilibrium potential and the conductance of the leak current, respectively. In the present simulations, E leak was used to impose an E Rest on the simulated neuron. G leak = 1/R m ; with R m , the specific membrane resistance was set to 25,390 Ω·cm 2 in the E13.5 neuron model and 17,350 Ω.cm 2 in the E17.5 neuron model. The resultant R in was 900 MΩ and 140 MΩ in the E13.5 and E17.5 neuron models, respectively. These values correspond to the averaged R in measured from actual E13.5 and E17.5 MNs. E m is the membrane potential.
Icore is the axial current to neighboring compartments summed over all of the neighbors: The parameter Gcore (in S) denotes the core conductance from the compartment with respect to the neighboring compartment, as follows: where diam and l are the diameter and length of the compartment (in cm), respectively, and R a is the specific resistance of the axoplasm (in Ω·cm). All of the computations were performed assuming a specific axoplasmic resistance, R a , of 100 Ω·cm.
Ich and Isyn in Equation (3) represent intrinsic and synaptic currents, respectively. The intracellular current injection can be modeled by adding the current to the compartment.
c m is the capacitance (µF) of each compartment and was calculated as follows: where area is the membrane surface of the compartment in cm 2 and C m is the specific membrane capacitance (set to 1 µF·cm -2 ).

SYNAPTIC INPUTS
The kinetics of the PSPs were described by the alpha function conductance and defined as follows: where i = nanoamps, g = microSiemens, and g = 0 for t < onset. g = gmax × (t -onset)/tau ×

exp(-(t -onset -tau)/tau) (20) for t > onset
where gmax is the maximum conductance for the considered postsynaptic receptor channel (gmax = g Clp ). Em is the membrane potential, and E Cl is the equilibrium potential for the ion that permeates through the channel (Clion). This value was adjusted according to the physiological measurements of E Cl . The alpha function has the property that the maximum value is gmax and occurs at The time constant of IPSPs (tau) was set to 20 ms according to physiological values that were measured during the patch clamp recording experiments.
To compare the effects of the simulated dGPSPs in the E13.5 and E17.5 neuron models, we estimated the g Clp that was required to evoke similar depolarizing events in both of the models (Suppl. Figure 1A) for E Cl = -42 mV and E Rest = -75 mV. The peak amplitude of the depolarizing response to various simulated g Clp was measured in both of the models (Suppl. Figure 1A1-2), and the corresponding curves were plotted against g Clp (Suppl. Figure 1A3). This analysis revealed that the g Clp that was required in an E17.5 neuron model was 5.5 times larger than the g Clp that was required in an E13.5 neuron model to elicit similar peak amplitudes.
The time course of g Cl during a representative dGPSP as simulated in an E13.5 neuron model with g Clp = 3.5 nS (Suppl. Figure 1B) indicates that g Clp is reached at t = 20 ms before the dGPSP peak at t = 45 ms. To estimate the summation of the currents underlying the dGPSPs trains, the falling phase of dGPSPs was fitted with a single exponential decay that was used to estimate the conductance at t = 0 ms (Suppl. Figure 5). This onset value ( × !"# ) was used in the equation from Le Bon-Jego et al. (2004). In these experiments, K was estimated to 2.4.

IETC MAP COLOR CODE
To accurately represent the excitatory and inhibitory effects of dGPSPs in the IETC maps, we used the following color code: 0% effects were coded in white; excitatory effects were coded in red, and inhibitory effects were coded in blue.
Excitatory effects from 0 to 10% were coded in a color gradation between white and red; excitatory effects from 10 to 20% were coded in a color gradation between red and black; and excitatory effects from 20 to 30% were coded in a color gradation between black and red. These two latter gradations were repeated several times according to the maximum effect.
The inhibitory effects from 0 to 10% were coded with a color gradation between white and blue; inhibitory effects from 10 to 20% were coded with a color gradation between blue and black; and inhibitory effects from 20 to 30% were coded with a color gradation between black and blue. These two latter gradations were repeated several times according to the maximum effect.