Dynamic properties of the EPSP discharge in the isolated frog labyrinth

The dynamic properties of the posterior afferent synapse were examined in the isolated frog labyrinth. The EPSPs generated at the cyto-neural junction were intracellularly recorded from single fibres of the posterior nerve in the presence of TTX (10-6 M) to prevent afferent spike discharge. Recordings were performed at rest and during long lasting (>1 min) sinusoidal rotations of the canal at constant frequency (0.1 Hz) and at different amplitude (20-220 deg/s) resulting in acceleratory peaks of increasing intensity (7.88-86.68 deg/s2). At rest, the EPSPs are discharged at a considerably high rate (70-400 EPSPs/s) and their frequency consistently increases during the excitatory stimulation (Rossi et al.1977). This fact leads to a high degree of EPSP temporal summation and overlapping which prevents the correct evaluation of their numbers by direct counting techniques. To obtain a reliable estimate of the EPSP amplitude and frequency, at rest and during rotation, the EPSPs were treated as "shot noise". Ideal shot noise is a fluctuating signal produced by the linear summation of uniform elementary events (shots) that occur at random and at a constant mean rate. Theoretical treatment of shot noise states that its power spectrum has the same shape as the spectrum of the shot waveform and that the mean, lambda1, and the variance, lambda2, of the noise are related to the mean shot rate, r, amplitude, h,and waveform w(t), by the equations lambda1.rh2integralew2(t)dt and lambda2rh2integralw2(t)dt (Campbell 1909). However, serious errors arise when these equations are applied to preparations in which: a) the membrane potential is affected by factors other than summation of the unitary events; b) the unitary events do not occur at a stationary rate; c) the unitary events do not sum linearly. At the frog neuromuscular junction, where the shots are the mEPPs, this type of analysis, performed during periods of intense neurosecretion, was based upon the higher semiinvariants of the noise according to the generai relation lambdan==rhnIn (Rice 1944) where In= integralf(w(t))ndt so that, in principle, any pair of semiinvariants can be used to compute r and h. If lambda2 (variance) and lambda3 (skewness) are used, then errors arising from slow, spurious changes in membrane potential are avoided (the parameter "mean" is not considered) and those arising from the non linear summation are greatly reduced (Segal et al. 1985). This method is used here to analyze the properties of the EPSP discharge at the frog cyto-neural junction.The moving average over a suitable time interval (5 s) of the variance and skewness of the noise allows us to measure changes in EPSP frequency during sinusoidal stimulation. The noise power spectrum shows that the unitary event waveform is well fitted by a dimensionless function w(t)=((betat)gamma/gamma(y+1))e-betat, where the parameter y (0.8-1.8) defines the shape of the event , and the parameter beta (800<=beta<=1800 Hz) is a time scale factor. The single shot amplitude remains fairly constant both at rest and during rotation (1.5-4 mV in the different units). During rotatory accelerations stronger than 7.88 deg/s2, the change in EPSP frequency is approximately sinusoidal, indicating that a close functional relationship exists between receptor response and mechanical stimulus. For accelerations of 7.88 deg/s2, the EPSP discharge was often unrelated to the sinusoidal stimulus. The EPSP emission was described by the following parameters: 1) the excitatory and inhibitory peak firing levels during rotation; 2) the mean rates of EPSPs released during the excitatory and inhibitory periods of the sinusoidal cycles. With regard to excitation, in some units a linear relationship was found between stimulus and response. In these linear units the gain was estimated by correlating the EPSP emission rate to the acceleratory peak values; in particular the peak EPSP rate and the average EPSP rate were considered. For the excitatory peak response the gain, computed from the slope of the regression lines and expressed as a percentage of the resting rate was 2.9-5.08/deg.s-2 for the first peak response and 2.25-4.51/deg.s-2 for the mean peak response over several cycles of sinusoidal stimulation. Similarly, the gain of the average EPSP rate during the excitatory phase of the sinusoidal cycle was 1.81-3.11/deg.s-2 for the first cycle and 1.38-2.72/ deg.s-2 over several cycles. This clearly shows that the gain of the whole response is always lower than that of the initial response, which indicates an adaptation of the EPSP discharge. Usually adaptation increases with acceleration and becomes evident during prolonged stimulation. During excitation other units exhibited a non linear intensity function. In these non-linear fibres the initial and the mean peak values of the response were related to the logarithm of acceleration and the slope of the corresponding regression lines was 26.06-155.98/deg.s-2 and 15.96-113.87/deg.s-2 (% of the resting level), respectively. Similarly, the average EPSP rate observed during the first sinusoidal cycle and over the entire duration of the stimulation period was proportional to the logarithm of acceleration giving a slope of 20.61-90.94/deg.s-2 and 10.7- 68.01/deg.s-2, respectively. Here it may be also noted that the slope of the whole response is always lower than that of the initial response. When the same parameters employed to quantify the excitatory response were used to define the EPSP properties during inhibition, it appears that linear and non linear behaviours were also present in the different units. In fact, in some units the decrease of the EPSP rate with respect to the resting level was linearly related to the stimulus intensity. The loss of EPSPs computed from the regression line slope and expressed as percentage of the resting level was 0.3-0.82/deg.s-2 and 0.3-0.72/deg.s-2 for the initial inhibitory peak and for the mean inhibitory peak, respectively. In the same units the average number of EPSPs lost during the inhibitory phase of the first sinusoidal cycle and that over the entire duration of the stimulating period was also linear with respect to acceleration, giving an EPSP loss of 0.35-0.97/deg.s-2 and 0.22-1.07/deg.s-2 , respectively. It is clear that during inhibition the initial and the whole response were similarly affected by acceleration; therefore, the inhibitory mechanism does not show any adaptation. As far as the non linear units were considered, it is clear that the first and the mean inhibitory peaks were proportional to the logarithm of acceleration, and regression analysis gives a slope of 24-25.73/deg.s-2 and 22.92-41.24/deg.s-2, respec tively. Similarly, when the average number of EPSPs lost was considered the slope proved to be 11.1-14.29/ deg.s-2 and 19.24-28.69/deg.s-2 for the first inhibitory cycle and for the whole response, respectively. By examining the properties of the EPSP discharge during rotation (accelerations larger than 30 deg/s2) it becomes evident that, whatever the parameter used to evaluate the magnitude of the response, a marked asymmetry around the resting level is present during excitation and inhibition. The excitatory response,in fact, is systematically larger than its inhibitory counterpart when the canal is subjected to equivalent stimuli of opposite polarity. It was also observed that the excitatory and inhibitory peak responses are usually in phase with the peak angular velocity, thus indicating that at 0.1 Hz the synaptic output from the semicircular canal is directly interpreted by central vestibular neurons as a measure of head velocity. The present results are in line with previous data on the afferent spike discharge (Rossi et al.1986) and suggest that many properties of the afferent response such as asymmetry, adaptation, linear or non linear intensity function are already present in the afferent pathway earlier than at the encoder, and thus are mainly due to the dynamic characteristics of the afferent synapse.