Layer III neurons control synchronized waves in the immature cerebral cortex

Abstract

Correlated spiking activity prevails in immature cortical networks and is believed to contribute to neuronal circuit maturation; however, its spatiotemporal organization is not fully understood. Using wide-field calcium imaging from acute whole-brain slices of rat pups on postnatal days 1-6, we found that correlated spikes were initiated in the anterior part of the lateral entorhinal cortex and propagated anteriorly to the frontal cortex and posteriorly to the medial entorhinal cortex, forming traveling waves that engaged almost the entire cortex. The waves were blocked by ionotropic glutamatergic receptor antagonists but not by GABAergic receptor antagonists. During wave events, glutamatergic and GABAergic synaptic inputs were balanced and induced UP state-like depolarization. Magnified monitoring with cellular resolution revealed that the layer III neurons were first activated when the waves were initiated. Consistent with this finding, layer III contained a larger number of neurons that were autonomously active, even under a blockade of synaptic transmission. During wave propagation, the layer III neurons constituted a leading front of the wave. The waves did not enter the parasubiculum; however, in some cases, they were reflected at the parasubicular border and propagated back in the opposite direction. During this reflection process, the layer III neurons in the medial entorhinal cortex maintained persistent activity. Thus, our data emphasize the role of layer III in early network behaviors and provide insight into the circuit mechanisms through which cerebral cortical networks maturate.

Figure 1.

Large-scale cortical synchronized waves in acute brain slices prepared from immature rats. A, Low-magnification time-lapse calcium imaging of a synchronized wave in a horizontal whole-brain slice. A posterior region of the cerebral cortex is the source of the synchronized wave. B, Simultaneous whole-cell patch-clamp recording (top) and calcium imaging (bottom) from the same cell, revealing that individual calcium activities reflected action potentials. C, Frequencies of synchronized waves before and after bath application of inhibitors [i.e., 1 μm TTX, 20 μm CNQX plus 50 μm AP5 (CNQX+AP5), 20 μm CNQX, 50 μm AP5, 100 μm picrotoxin (PTX), 100 μm 18β-glycyrrhetinic acid (GA), 100 μm carbenoxolone (CBX), 100 μm flufenamic acid (FFA, CAN channel blocker), and 30 μm riluzole (pNa channel blocker)]. The numbers in parentheses indicate the numbers of slices tested. Data are means ± SDs. *p < 0.05, **p < 0.001 versus control, Mann–Whitney U test.

Figure 2.

Wave initiation in LEC. A, Representative line-scan plot of the calcium signal across bilateral cerebral cortices. The calcium signal (right) measured along the red line shown in the left image indicates the LEC as an origin of synchronized waves. Yellow asterisks indicate the initiation sites (the mass centers of the earliest activated areas) of the second wave in the left plot. A, anterior; Ctx, cortex; HF, hippocampal formation; L, left, P, posterior. R, right. B, The initiation sites of individual waves are indicated by yellow asterisks, many of which were located in layer III of the LEC. C, Comparison of synchronized waves in three zones (1–3) of the EC before (top) and after (bottom) surgical dissection. 1, anterior LEC; 2, border area between the LEC and the MEC; 3, posterior MEC. The right traces indicate the changes in the fluorescence intensity of the corresponding zones. Every part of the cortex was capable of generating spontaneous activity. D, Wave frequency in intact slice and minislices. Although the wave frequency was lower in the minislices, the minislices that contained the original site of wave initiation (i.e., zone 1) exhibited waves more frequently than other minislices. Data are means ± SDs of five slices. *p < 0.001, Mann–Whitney's U test.

Figure 3.

Electrical stimulation-evoked synchronized waves. A, Four sites (1–4) of stimulation. B, Propagation of a spontaneously generated wave. C–F, Waves evoked by electrical stimulation (50 μs, 40 μA, 2 pulses) of the lateral end of the LEC (C, site 1), the border between the MEC and LEC (D, site 2), the medial end of the MEC (E, site 3), and the hippocampal CA1 stratum radiatum (F, site 4). Stimulation of all sites could induce synchronized waves that propagated beyond the EC. The same results were obtained from all six slices tested.

Figure 4.

Autonomous cells in LEC layer III. A, Typical raster plot of activity of individual cells (left) in an LEC network before (middle) and after (right) bath application of 20 μm CNQX, 50 μm AP5, and 100 μm picrotoxin (PTX). The wave initiation site was searched using a low-magnification objective (4×) and imaged under high magnification. B, Activation of the layer III cells preceded the activation of the other layers (p < 0.0001, Kolmogorov–Smirnoff test; n = 3–26 waves from 6 slices). C, Percentage of autonomously active cells to the total cells monitored in each layer. Data are shown as box plots *p < 0.05, **p < 0.01, Mann–Whitney U test, n = 5 slices. D, The mean durations of autonomous calcium events were longest in layer III cells. The pseudocolored map (right) indicates the mean event durations in individual cells of the same slice as A. Data are means ± SDs of 5 slices. **p < 0.01, Mann–Whitney U test.

Figure 5.

Wave propagation with a leading front in layer III. A, Confocal image of an OGB1-loaded slice (left) and the locations of 567 cells (right) in the LEC. A, anterior; D, deep; P, posterior; S, superficial. B, Spatiotemporal patterns of multineuronal activity during wave propagation. Each dot indicates the onset of each calcium activity. The data were obtained from the same slice as A. Forward, waves propagating in the posterior direction; Reverse, reflected waves propagating in the anterior direction. C, Top, Cell maps representing the activity onset in propagating waves #1, #2, and #4 shown in B. The onset timing is indicated in a pseudocolored scale. Bottom left, Onset timings of individual cells. The colors indicate the cortical layers. Bottom right, Summarized box plots of the onset timings of neurons in different layers. D, Comparison of the activity latency of individual cells between pairs of wave #1 versus #2, #2 versus #4, and #4 versus #1. Each dot indicates a single cell. The colors represent the layers of cells as shown in C. The positive linear correlations indicate that the order of cell activation was largely preserved across waves. E, The distribution of Spearman's rank correlation coefficients of the onset timings between all possible pairs of 76 waves in five slices.

Figure 6.

Balanced synaptic inputs during wave propagation. A, Ratios of cells commonly activated in 2–4 different wave events. The evoked waves were generated by field stimulation at intensities of 40 μA through a stimulation electrode placed on the lateral end of the LEC. Even pairs of spontaneous and evoked waves shared higher proportions of coparticipants compared with randomly shuffled surrogates (broken line). n = 26 waves from three slices. B, Maps of active cells of spontaneous and stimulation-evoked waves. The cells that participated in all monitored events were defined as core cells. The core populations shared by the spontaneous and evoked waves are indicated in the map on the left. A, anterior; D, deep; P, posterior; S, superficial. C, Voltage-clamp recordings (bottom) from an MEC layer III neuron (red asterisk) during spontaneous and LEC stimulation-evoked waves (top). Waves were elicited by LEC stimulation at intensities of 10–80 μA. The holding potential was −73 mV. Similar barrages of synaptic inputs were observed during spontaneous and evoked waves. Small synaptic inputs preceded the barrage inputs, indicating the existence of feedforward input. D, Representative traces (left) of synaptic currents recorded at −73 and 0 mV. E, Time evolution of glutamatergic and GABAergic synaptic inputs. Both conductances were positively correlated during a wave (right). The data were obtained from three neurons from three slices, and different colored traces represent different cells. F, Correlated synaptic currents between adjacent pairs of EC neurons. Cross-correlations between the neuron pairs were calculated for glutamatergic and/or GABAergic synaptic inputs without (0 time lag) and with time alignment at the current peak (peak correlation). Data are means ± SDs of 29, 13, and 14 pairs. Representative raw traces are shown in the left inset. Scale bars, 40 pA for glutamatergic and 100 pA for GABAergic inputs. G, GABA-induced calcium activity in EC neurons. Raw calcium traces shown by 13 representative neurons in the left photograph are presented. Arrowheads indicate the times of puff application of 200 μm GABA.

Figure 7.

Termination and reflection of synchronized waves in the MEC. A, Example time-lapse images of a synchronized wave without reflection. The wave was initiated at the lateral end of the LEC and propagated bidirectionally in a tangential direction. The posterior wave terminated at the border between the MEC and the parasubiculum (PS), with a small wave entering the subiculum (S). HF, hippocampal formation; PrC, perirhinal cortex. The wave fronts are indicated by arrows. B, The wave in A is plotted in the xyt form. Fluorescence changes >3% were detected and reconstructed along the time (vertical) axis. C, Example of a synchronized wave with reflection. The data were obtained from the same slice as A. The wave that reached the border between the MEC and the parasubiculum changed the direction of propagation and generated a reflected wave. D, xyt plot of the wave in C. The transparent red area indicates the reflected wave. E, The initiation sites (the mass centers of the earliest activated areas) of individual waves in the same slice are compared between waves without reflection (green) and with reflection (magenta). F, The initiation sites of all nonreflected and reflected waves observed in five slices are superimposed after spatial alignment to the rhinal sulcus. G, Simultaneous monitoring of somatic calcium signals and field potentials. Two wave events (#1 and #2) were magnified in two right panels. Top, Spontaneous field potential. Bottom, Reconstructed neuronal activity.

Figure 8.

Dynamics of forward and reflected synchronized waves. A, Confocal image of an OGB1-loaded slice (left). The dotted line was line-scanned in waves without reflection (middle) and with reflection (right). HF, hippocampal formation; PrC, perirhinal cortex. F, forward waves; R, reflected waves. B, Propagation speeds of waves in the PrC (forward from posterior to anterior, P → A; reflection from posterior to anterior, P → A) and the MEC (forward from anterior to posterior, A → P; reflection from posterior to anterior, P → A). The colors of the columns correspond to those of the arrows shown in A. Error bars represent SDs of seven slices. *p < 0.05, **p < 0.01; Mann–Whitney U test. C, The dotted line of an OGB1-loaded slice (left) was line-scanned during forward and reflected waves. In both waves, layer III was activated earlier than the other layers. F, forward waves; R, reflected waves.

Figure 9.

Wave termination in the MEC. A, High-magnification imaging of a wave that stopped at the border between the MEC and the parasubiculum (PS). The MEC layer II neurons were more active than the layer III neurons. B, Simultaneous voltage-clamp recording from two pyramidal cells in the MEC and parasubiculum (PS) during wave termination. Glutamatergic PSCs (top) and GABAergic PSCs (bottom) were dominated at clamped voltages of −73 and 0 mV. The shaded areas indicate the periods of wave-relevant activation. C, The peak amplitudes of wave-relevant glutamatergic PSCs (left) and GABAergic PSCs (right) in the MEC were larger than those in the parasubiculum. Data are shown as box plots. *p < 0.05, Mann–Whitney U test, n = 5 and 6 cells from a total of six slices for glutamatergic PSCs and GABAergic PSCs, respectively.

Figure 10.

Persistent synaptic inputs into the MEC layer III neurons during wave reflection. A, High-magnification imaging of a wave that reflected at the border between the MEC and the parasubiculum (PS). When compared with the case of wave termination (Fig. 9A), cells in the MEC layer III were more strongly activated; note that during and after a transient decrease in the entire activity of ∼2 s, the layer III activity persisted from 2 to 4 s. Following this persistent activation of layer III cells, the second wave was generated at 4 s, and the network became silent by 6 s. B, Simultaneous recording of the membrane potential (top) and the calcium fluorescence (bottom) of an MEC layer III neuron. F, forward waves; R, reflected waves. C, Representative glutamatergic PSC traces recorded simultaneously from an MEC layer II cell (top) and a layer III cell (bottom) during wave termination (left) and wave reflection (right). The layer III neuron received prolonged synaptic inputs during wave reflection. D, The PSC durations in layer III neurons were longer than those in layer II neurons and longer during wave reflection (reflected) than during wave termination (stop). The duration was calculated as a 25%-peak width of synaptic barrages. Error bars represent SDs of five, five, six, and five neurons from six slices. *p < 0.05, **p < 0.01, Mann–Whitney U test.

Figure 11.

Effect of wave intensity on the reflection probability. A, The peak ΔF/F amplitudes of the forward waves with reflection were larger than in the nonreflected forward waves. Left, The calcium signal was measured from the border area of the MEC and the LEC where the forward wave propagated through. F, forward waves; R, reflected waves. *p < 0.05, Wilcoxon signed-rank test, n = 6 slices. B, The maximal firing frequencies and the total spike numbers of layer III neurons near the border area of the MEC and the LEC during the forward waves were higher in the reflected waves than in the nonreflected waves. *p < 0.05, Wilcoxon signed-rank test, n = 5 neurons of five slices. C, Representative population calcium signal of waves initiated by electrical stimulation of the LEC at 5–80 μA (2 50 μs pulses at 40 Hz) at time 0 s. D, Probability of the stimulation-induced initiation of waves (black) and reflected waves (red) for the 30 stimulation trials. Error bars represent SDs of four slices. *p < 0.05, Mann–Whitney U test.