A population of gap junction-coupled neurons drives recurrent network activity in a developing visual circuit

Abstract

In many regions of the vertebrate brain, microcircuits generate local recurrent activity that aids in the processing and encoding of incoming afferent inputs. Local recurrent activity can amplify, filter, and temporally and spatially parse out incoming input. Determining how these microcircuits function is of great interest because it provides glimpses into fundamental processes underlying brain computation. Within the Xenopus tadpole optic tectum, deep layer neurons display robust recurrent activity. Although the development and plasticity of this local recurrent activity has been well described, the underlying microcircuitry is not well understood. Here, using a whole brain preparation that allows for whole cell recording from neurons of the superficial tectal layers, we identified a physiologically distinct population of excitatory neurons that are gap junctionally coupled and through this coupling gate local recurrent network activity. Our findings provide a novel role for neuronal coupling among excitatory interneurons in the temporal processing of visual stimuli.

Keywords: Xenopus tadpole; gap junctions; microcircuit; optic tectum; recurrent activity.

Fig. 1.

A subpopulation of neurons in the superficial layer displays a distinct pattern of synaptic connectivity. A: a brightfield image of a stage 48 optic tectum as a superficial layer type 1 (SL1) neuron is being recorded in whole cell configuration. Notice that the cell body layer is many somata deep. B: examples of three deep layer (DL) neurons (gray traces) showing the range of retinal ganglion cell (RGC)-evoked responses. Notice that all responses consist of both a monosynaptic and polysynaptic component; examples of three SL1 neurons (black traces) showing the range of RGC-evoked responses. Notice that these neurons only display the monosynaptic component of the response. C: examples of spontaneous excitatory postsynaptic currents (sEPSCs) recorded from a DL (gray) and SL1 (black) neuron (left); averaged sEPSCs for DL and SL1 neurons with amplitudes scaled to compare kinetics (right). Notice that the average SL1 sEPSC has a faster rate of rise and decay compared with the average DL sEPSC. D: dot plots showing sEPSC amplitude (left) and average frequency (right). Each dot represents a single neuron. SL1 neurons display a statistically significantly lower sEPSC frequency compared with DL neurons (P = 0.0024), and greater amplitude (P = 0.00021). The black diamond represents averaged values. **P < 0.01, Mann-Whitney test. ***P < 0.001, unpaired 2-tailed t-test.

Fig. 2.

SL1 neurons display dye coupling and are excitatory. Images of a tectum (whole mounts) in which a single SL1 neuron was filled with biocytin (A), neurobiotin (B), or biocytin in the presence of the gap junction blocker 18β-glycyrrhetinic acid (18β-GA, C). That 18β-GA prevents dye coupling suggests the presence of gap junctions. The scale bar also applies to A and C. D: a merged image showing dye-coupled neurons (red) and γ-aminobutyric acid (GABA, green), visualized via immunocytochemistry methods. Notice that none of the dye-coupled neurons appears to be expressing GABA. Unmerged images showing GABA expression alone (E) and biocytin alone (F). R, rostral; C, caudal.

Fig. 3.

Depolarizing currents observed in SL1 neurons while blocking all synaptic transmission confirm electrical coupling. A: example of maximum RGC-evoked response of an SL1 neuron when clamped at −60 mV (left) and at 5 mV (right). B: example of maximum RGC-evoked response of an SL1 neuron when clamped at −60 mV (left) and at 5 mV (right) in the presence of 18β-GA (20 μm). C: example of maximum RGC-evoked response of a DL neuron when clamped at −60 mV (left) and at 5 mV (right). Notice that only the SL1 displays an RGC-evoked depolarizing current at 5 mV. This depolarizing current is no longer observed when gap junctions are blocked. D: dot plots showing the peak monosynaptic amplitude of RGC-evoked currents with the membrane potential voltage clamped at −60 mV (axis on the left) and then at 5 mV (axis on the right) for SL1 (left), SL1 + 18β-GA (middle), and DL (right). The black diamond represents averaged values.

Fig. 4.

SL1 neurons express different intrinsic properties compared with DL. A: an example of a single DL neuron recorded in voltage clamp as it is stepped to increasingly depolarized potentials. The downward inflections represent inward Na⁺ current, and the upward deflections represent the outward K⁺ current. Data from these recordings are used to generate current (I)-voltage (V) plots shown in B. B: I–V plots showing peak Na⁺ and K⁺ currents as a function of voltage step. Notice the average peak Na⁺ currents in response to essentially every step are greater than the corresponding DL current. C: cumulative probability plots of the maximum peak Na⁺ (left) and K⁺ (right) currents displayed by individual SL1 (black dots) and DL (gray dots) neurons. Each dot represents a single neuron. D: plot showing the no. of action potentials fired as a function of the amount of current injected into the soma. Notice that SL1 neurons fire more action potentials in response to every different amount of current injected, but the variability displayed between SL1 neurons is greater than that displayed by DL neurons. E: sample traces showing the maximal no. of action potentials fired by 2 different DL neurons (left) and 2 different SL1 neurons (right). *P < 0.05, unpaired 2-tailed t-test.

Fig. 5.

Blocking gap junctions with 18β-GA or shutting down the SL1 neuron network with BAPTA inhibits the polysynaptic portion of the RGC-evoked response of DL neurons. A: average traces of maximum RGC-evoked responses of DL neurons and DL neurons in the presence of 20, 30, and 40 μM 18β-GA added to the external recording solution. B: plots showing monosynaptic amplitude of RGC-evoked responses of DL neurons and DL neurons in the presence of 18β-GA. C: plots showing the amount of charge of the polysynaptic portion (30 and 400 ms after stimulation) of the RGC-evoked response of DL neurons and DL neurons in the presence of 18β-GA. For all plots in B and C, each dot represents one neuron, and the black diamond represents average values. Averaged evoked traces shown in A are generated from this same set of neurons. D: average traces of maximum RGC-evoked responses of DL neurons recorded from preparations in which at least 1 SL1 neuron had been filled with BAPTA and the same protocol except with only non-SL1 neurons filled with BAPTA, referred to as control (ctrl) in D, E, and F. E: plots showing monosynaptic amplitudes of RGC-evoked responses of DL neurons from preparations in which only non-SL1 neurons have been filled with BAPTA (Ctrl), and from preparations in which at least 1 SL1 neuron had been filled with BAPTA. F: plots showing the amount of charge of the polysynaptic portion (between 30 and 400 ms after stimulation) of RGC-evoked responses of DL neurons from preparations in which only non-SL1 neurons had been filled with BAPTA (Ctrl), and from preparations in which at least 1 SL1 neuron had been filled with BAPTA. For all plots in E and F, each dot represents one neuron, and the black diamond represents average values. Averaged evoked traces shown in D are generated from this same set of neurons. **P < 0.01 and ***P < 0.001, unpaired 2-tailed t-test.

Fig. 6.

A schematic diagram of the tectum suggested by combined data. Visually driven input enters the tectum via RGC axons, which directly synapse onto both SL1 and DL neurons. Next, activation of the SL1 neurons triggers feedforward recurrent activity among deep layer tectal neurons. The recurrent activity may also include local connections between DL neurons; however, it is not known for sure because synaptically coupled tectal neurons have not been observed.