Table 1.
Reliability of EPSPs and IPSPs.
Figure 1.
Paired recordings from visually identified L4 neurons.
A. An IR-DIC image of a coronal slice from area 17 of the cat cortex. Layer 4 was identified under low magnification (10×) as a dark stripe extending over the middle third portion of the cortex. Lower panel: at a higher magnification (60×) an excitatory neuron (left) and an inhibitory neuron (right) were recorded simultaneously with patch clamp pipettes. B, C. Biocytin stain of the same cell pair and its 3-D reconstruction, identifying them as a spiny stellate cell and a smooth basket cell. The cells are presented separately for clarity. D. The cell pair was reciprocally connected as is evident by the EPSP (red) and IPSP (blue) evoked in the basket and spiny stellate neurons, respectively. The presynaptic APs in the basket and stellate cells are colored blue and red respectively. Vertical scale bar is 0.4 mV for the EPSP/IPSP and 50 mV for the APs and horizontal scale bar is 20 ms.
Figure 2.
Reconstructed 3D morphology of layer 4 pairs.
A–E. illustrating 5 E–E pairs. A, B pairs of spiny stellate neurons. C, D, E pairs of star pyramidal neurons. Preynaptic excitatory cells are cyan colored and their axons colored black. Postsynaptic excitatory cells are colored blue and their axons grey. F–J illustrates 5 pairs of excitatory and inhibitory neurons connected via an E–I synapse. In F–I the presynaptic neuron was a spiny stellate, in J a star pyramid. The postsynaptic neurons in F–J are all basket cells. Presynaptic excitatory cells are colored blue and their axons black, postsynaptic basket cells are colored red and their axons green. K–O Pairs of basket cells contacting excitatory spiny stellate (K–M) or star pyramidal (N–O) cells. Note the dense local ramifications of the basket cells’ axons. Presynaptic basket cells are red and their axons green, postsynaptic stellate and star pyramidal cells are blue and their axons grey. For presentation purposes all dendrites and axons were drawn with the same line thickness. Cell pairs F and I were reciprocally connected.
Figure 3.
A. An AP evoked in the presynaptic inhibitory neuron and the simultaneously recorded IPSP (average of 15 traces, each) from the postsynaptic excitatory neuron at two membrane potentials. The IPSP was hyperpolarizing at a Vm of −50 mV but depolarizing at −75 mV. B. The IPSP amplitude plotted as function of the membrane potential (Vm). The data were fitted with a straight line (dashed) intercepting the x-axis (indicated by an arrow) at −65 mV. C. In 7 subsequent traces, IPSP amplitudes varied in response to single presynaptic APs, but no failures were observed. The postsynaptic Vm was held at −50 mV D. A histogram of the IPSP amplitudes (filled bars) was constructed from 60 traces and the background noise (empty bars). The IPSP histogram was fitted with a Gaussian distribution with a median at −0.53 mV (arrow) and a σ of 0.36 mV.
Figure 4.
Reliability of E–E and E–I synapses.
A. Fluctuations in the EPSP amplitudes at an E–E connection. The upper trace illustrates a presynaptic AP and the 7 subsequent traces show the response evoked in the excitatory postsynaptic cell. B. The EPSP amplitude histogram (filled bars) has a median at 0.24 mV and a σ of 0.152 mV. Black dots in A and B indicate spontaneously occurring EPSPs prior to stimulation. C. Fluctuations in the EPSP amplitudes at an example E–I connection. The upper trace illustrates a presynaptic AP and the 7 subsequent traces show the response evoked in the inhibitory postsynaptic cell. D. A broad distribution of the E–I EPSPs, with a median at 0.84 mV and a σ of 0.435 mV.
Figure 5.
Variance analysis of E–E, E–I and I–E synapses.
A–C. The averaged (±s.d.) PSP amplitude (A), CV (B) and percentage failures (C) were compared among the three connection types. The EPSP amplitudes of E–I connections were significantly larger than E–E EPSPs or I–E IPSPs (p<0.05). I–E connections had on average lower CVs and percentage failures, but these were not statistically significant. D. The average horizontal distances between the pre- and postsynaptic neurons were short for all connection types: E–E 47±14, n = 10; E–I 40±14, n = 8, I–E 27±5, n = 8. E–F. The CV and percentage failures of individual synapses were plotted against their average PSP amplitude. In, general, larger PSP amplitudes were correlated with lower CVs and percentage failures and smaller PSP amplitudes correlated with the higher CVs and percentage failure. Among medium-sized PSPs no obvious correlations with the CV or percentage failures were observed. G. PSP amplitudes were plotted against the horizontal distance between the pre- and postsynaptic neurons. No correlation could be observed between these parameters (cor. Coeff. = −0.08).
Figure 6.
A. Averaged PSPs from 3 individual connections illustrate the differences in synaptic kinetics of the various connection types. The presynaptic APs were aligned at their peak (depicted by the long dashed line). The short dashed lines on every PSP denote the time at which the PSP rose to 10%. The latency between the presynaptic AP peak and the 10% PSP was 0.8, 0.5 and 0.9 ms for the E–E, E–I and I–E synapses, respectively. The arrowheads point towards the 10% and 90% values of the PSP amplitude and the delay between them is defined as the rise time. Vertical scale bar represents potential and was 80 mV for all three APs and 0.13 mV, 0.85 mV and 0.2 mV in the E–E, E–I and I–E PSPs, respectively. B–E. E–I connections have significantly shorter latencies and faster rise times compared with E–E and I–E connections. All bar graphs represent the averaged values ±s.d.
Table 2.
EPSP/IPSP kinetics.
Figure 7.
Compartmental model simulations of synaptic kinetics.
Electrotonic attenuation of the synaptic current along the dendrites en route to the soma, results in smaller and slower somatic EPSPs. A, E. A single synaptic contact with a conductance of 3.8 nS was simulated on (A) a spiny stellate or (E) a basket cell dendrite at the location where a putative contact was identified (arrows in B and F, respectively). The decay time constant (τ2) of the conductance in E was faster than in A (1 ms vs. 2.3 ms). The simulated EPSCs and the resulting EPSPs at the synapse are drawn in a thin line. The same EPSPs measured in the soma are drawn with a thick line. In (A) the somatic potential trace was multiplied (×30) to allow direct comparison of the kinetics. The 10% and 90% EPSP amplitudes are marked by the blue lines. Note that the more distal contact simulated in A is more strongly delayed, slowed and attenuated than the proximal contact simulated in E. B–H. A single synaptic activation was simulated successively in all compartments of the spiny stellate (B–E) and smooth basket (F–H) neurons. In each compartment, the latency of the EPSP arrival at the soma and the somatic 10%–90% EPSP rise times were analyzed and compared with the experimentally observed values. The somatic and dendritic compartments were then colored according to the following categories: Black: simulated values lower than the smallest experimentally observed latency or rise time. Grey: simulated values larger than the largest experimentally observed latency or rise time. Blue: simulated values within the range between the smallest and the largest observed values. Red: Mean EPSP latency (average±s.d.). Yellow: Mean EPSP rise time (average±s.d.). Green: Overlapping compartments in which both rise-times and latencies were within the average range. B, F. Latency plots. C, G, Rise time plots. D, H, overlap plots of latencies and rise times. The scale bar is 100 µm. I–L. Simulations were repeated in 5 spiny stellate cells and 1 star pyramid (I, J) and 4 basket cells(K, L) neurons, each presented by a different symbol. I, K, Somatic EPSP latencies plotted against the dendritic distance from the soma for all spiny neurons (I) and basket cell (K) dendrites. J, L, Plots of the rise times at different dendritic distances from the soma. The dashed lines represent the experimental values of the corresponding average+s.d. Red symbols represent the overlap compartments for the latency and rise times in each neuron. Spines in A–D, I–J, were incorporated by scaling Cm and Rm in the dendrites. A faster conductance (as in E) was simulated in F–H, K–L.
Figure 8.
Contact sites seen at LM between the presynaptic axon and the postsynaptic dendrites were determined for all reconstructed pairs. A–C illustrates examples of an E–E, E–I and an I–E pair. The presynaptic soma of the spiny stellate cell is colored cyan and the axon is black. The basket cell’s soma and dendrites are colored red and the axon green. Postsynaptic spiny stellate cell is colored blue. For clarity, axons were trimmed to a sphere of about 400 µm diameter around the soma, but complete dendritic trees are presented. Putative contacts are marked by yellow-filled circles. Scale bar is 200 µm. D. Dendrograms of excitatory cells (somata indicated by blue circle) and inhibitory cells (somata indicated by red circle). Position of contacts on the dendrites are indicated by cyan (E–E) and red (E–I) rectangles. I–E contacts are indicated by black circles. Open arrowheads indicate three apical dendrites clipped at 300 µm E. The averaged distance of E–I contacts was significantly lower than E–E and I–E (P<0.05).
Figure 9.
Effects of fast E–I recruitment on the activity and tuning curves of a model recurrent L4 network.
Three models were used to simulate activity in the recurrent network. All models included excitatory (E) and inhibitory (I) L4 neurons and feedforward thalamic (T) input synapses and were identical in all but the recurrent excitatory synapses. Excitatory and inhibitory synapses were drawn with arrows and filled circles, respectively. A1. The feedforward model in which E–E and E–I synapses were silenced. A2. The second model incorporating recurrent E–E and E–I synapses with identical kinetics and latencies. A3. The third model is structurally identical to A2 but E–I synapses have faster kinetics and shorter latencies. Models 1, 2 and 3 and all the results are colored black, blue and red, respectively. A–D, Responses of all three models to the same feedforward synaptic input (Fin = 29.5 Hz). B. Raster plots of the excitatory (E) and inhibitory (I) spike times, dashed line marks stimulus start. C. Peri-stimulus histogram of the firing rates of the excitatory (solid lines) and inhibitory (dashed lines) neurons binned at 10 ms intervals. D. Mean excitatory (positive to 0) and inhibitory (negative to 0) currents to the excitatory L4 neurons. Current is given in arbitrary units. The inset shows the mean inhibitory current between t = 40 and 80 ms from simulation start. The arrowheads point towards the earlier onset and faster rise of the current in model 3 compared to model 2. Scale bars are 10 ms and 0.03 I. E–F. Tuning curves of the excitatory neurons. E. Mean firing rates as a function of the stimulus parameter α, note the many-folds amplification of the responses in models 2 (blue) and 3 (red) in respect to model 1(black). F. The mean firing rates normalized to the maximal response as a function of α. The lines represent the best Gaussian fits to the normalized responses. The grey line is the normalized input frequency. Note the narrowing of the tuning curve in model 3.