Neurogliaform cells mediate feedback inhibition in the medial entorhinal cortex

Abstract

Layer I of the medial entorhinal cortex (MEC) contains converging axons from several brain areas and dendritic tufts originating from principal cells located in multiple layers. Moreover, specific GABAergic interneurons are also located in the area, but their inputs, outputs, and effect on local network events remain elusive. Neurogliaform cells are the most frequent and critically positioned inhibitory neurons in layer I. They are considered to conduct feed-forward inhibition via GABAA and GABAB receptors on pyramidal cells located in several cortical areas. Using optogenetic experiments, we showed that layer I neurogliaform cells receive excitatory inputs from layer II pyramidal cells, thereby playing a critical role in local feedback inhibition in the MEC. We also found that neurogliaform cells are evenly distributed in layer I and do not correlate with the previously described compartmentalization ("cell islands") of layer II. We concluded that the activity of neurogliaform cells in layer I is largely set by layer II pyramidal cells through excitatory synapses, potentially inhibiting the apical dendrites of all types of principal cells in the MEC.

Keywords: GABA; entorhinal cortex; feedback inhibition; microcircuit; neurogliaform cells; optogenetics.

FIGURE 1

Axons of layer II pyramidal cells mainly target layer II. (A) Low-magnification confocal images of the horizontally sectioned MEC of a Calb-Cre mouse showing Cre-dependent local ChR2 expression (yellow). This ChR2 expression is colocalized with the specific LII pyramidal cell markers WFS1 (green) and calbindin (red) (scale bars: 200 μm). (B) High-magnification confocal images of a biocytin-filled layer II stellate cell and the surrounding ChR2-mCherry-positive axons (yellow) in the MEC of a Calb-Cre mouse (scale bars: 50 μm). (C) Schematic representation of optogenetic recording setup (left) and response to hyperpolarizing and depolarizing current steps of the recorded stellate cells (RMP: –65 mV and –100 and +200 pA, above, scale bar: 30 mV and 200 ms). (D) Response of the stellate cell to a suprathreshold (5 mW, above) and a subthreshold (0.25 mW, below) single 5 ms (blue bar) light pulse (RMP: –65 mV and –68 mV, respectively, scale bar: 20 mV and 5 mV, respectively, 20 ms). The black trace represents the average, while the gray traces represent 4 consecutive sweeps. (E) High-magnification confocal image of a parvalbumin-positive (immunoreactivity shown in the inset, scale bar: 10 μm) cell and the surrounding ChR2-mCherry-positive axons (yellow) in the MEC of a Calb-Cre mouse (scale bars: 50 μm). (F) Optogenetic recording schematics (left) and response to hyperpolarizing and depolarizing current steps of the recorded parvalbumin-positive cell (RMP: –67 mV and –100 and +200 pA, scale bar: 30 mV and 200 ms). (G) Response of the parvalbumin-positive fast spiker interneuron to suprathreshold (5 mW, above) and a subthreshold (0.25 mW, below) paired 5 ms light pulses (RMP: –67 mV and –69 mV, respectively, scale bar: 20 mV, 5 mV, and 20 ms). The black trace represents the average, while the gray traces represent 4 consecutive sweeps. The delay between the two pulses was 60 ms (scale bar: 20 mV, 5 mV, and 20 ms).

FIGURE 2

Layer I neurogliaform cells receive excitatory monosynaptic inputs from layer II pyramidal cells. (A) Confocal image of a biocytin-filled layer I neurogliaform cell surrounded by ChR2-mCherry-positive axons (yellow) (left). The recorded cell shows GABAARα1 (middle) and Reelin (right) immunopositivity (scale bars: 15 μm). (B) Representative reconstruction of somatic locations in a 40 μm thick section of the mouse medial entorhinal cortex showing layer I GABAARα1 (red circle) positive cells together with GABAARα1 and Reelin positive cells (green triangle) (scale bar: 500 μm). (C) The postsynaptic effects seen on neurogliaform cells (black, firing pattern in inset) were reduced by 1 μM TTX (green). Of note, 1 mM 4-AP not only recovered but also increased the amplitude of the postsynaptic effect, indicating monosynaptic input [RMP: –65 mV, excitatory postsynaptic potential (EPSP) amplitude: before TTX application 3.1 ± 0.6 mV, after TTX application 0.3 ± 0.2 mV and after 4-AP application 9.8 ± 2.9 mV, n = 4]. (D) The voltage response of the neurogliaform cell to a 5 ms photo-stimulation (black, single sweep EPSP, n = 34) and the disappearance of the effect after 10 μM CNQX + 10 μM NBQX wash in (red) (RMP: –65 mV, scale bars: 3 mV and 20 ms). Inset: response of the recorded cell to 1 s current injection (RMP: –65 mV and –100 and +150 pA). Note that the blue bars on panel (C) and (D) represent the 5 ms long light-pulses.

FIGURE 3

Neurogliaform and non-neurogliaform cells react differently to layer II excitatory inputs. (A) Up left: schematic representation of the recording setup. Bottom left: confocal image of biocytin-filled parvalbumin-positive fast-spiking interneuron soma (scale bar: 10 μm). Right: response to a 1 s current injection (RMP: –70 mV and –100 and +200 pA, scale bar: 20 mV and 200 ms). (B) 3D waterfall plot of the responses to five 5 ms long and 5 mW (blue bars) light pulses at 17 Hz. Sweeps were run 3 s apart. The last sweep in front (RMP: –70 mV, scale bars: 15 mV and 50 ms). (C) Up left: schematic representation of the recording setup. Bottom left: biocytin-filled GABAARα1 immunopositive neurogliaform cell body (scale bar: 10 μm). Right: response to a 1 s current injection (RMP: –68 mV and –100 and +200 pA). (D) 3D waterfall plot of the responses to five 5 ms long, 5 mW (blue bars) light pulses at 17 Hz at resting membrane potential (–68 mV, bottom), and at a depolarized state (–50 mV, top) (scale bars: top: 10 mV and 50 ms, bottom 5 mV and 50 ms). (E) Response of the parvalbumin-positive fast-spiking interneuron (left) and the neurogliaform cell (right) to paired 5 mW (above) and 0.5 mW (below) intense 5 ms light-pulses (RMP: –68 and –66 mV, respectively, scale bars: 5 mV and 50 ms). The black trace represents the average, while the gray traces represent 4 consecutive sweeps. The delay between the two pulses was 60 ms. The response of the non-neurogliaform interneuron to paired 5 ms long light pulses with 0.5 mW light intensity was 11.9 mV at RMP: –68 mV, while the response of the neurogliaform cell to a single 5 ms long light pulse with the same intensity was 0.7 mV at RMP: –66 mV. It is noted that the response of neurogliaform cell to a 0.5 mW light pulse was tested only with a single 5 ms long light pulse (right, below).

FIGURE 4

The distribution of neurogliaform cells does not correlate with dendritic clustering of layer II pyramidal cells. Neurolucida partial reconstruction of two paired recordings. (A) Neurogliaform cell (soma and dendrites blue, axons red) together with layer II pyramidal cell (soma and dendrites, black). (B) Neurogliaform cell together with layer II stellate cell. Neurogliaform cell dendrites and axons are mostly restricted to layer I (scale bar: 50 μm). (C) Tangential section of the mouse temporal cortex showing the dorsal MEC (dMEC) and the parasubiculum (ParS) (left, scale bar: 200 μm). In layer I, dendrites of layer II pyramidal cells form patches, and Reelin-positive interneurons are located within and in between patches as well (right) (z-stack of 7 μm section, scale bar: 50 μm). (D) Left: drawings of the patch borders (black lines) and the positions of α-actinin+ putative neurogliaform cells (green dots within patches and red dots outside patch structures) in the rat MEC layer I. Right: representative low-magnification confocal image of the dorsal MEC layer I, based on which the drawing was made (z-stack of 30 μm section, scale bar: 50 μm).