Local Connections of Pyramidal Neurons to Parvalbumin-Producing Interneurons in Motor-Associated Cortical Areas of Mice

Light and electron microscopic findings of close appositions formed between the axon varicosities of pyramidal neurons and somas/dendrites of PV neurons. A biocytin-labeled pyramidal neuron was developed in blue-black with the DAB/nickel reaction, whereas almost all the somas and dendrites of PV neurons were visualized in pink by immunostaining for GFP with the TAPM/p-cresol reaction (A). Arrows and arrowheads in B–E indicate the axon varicosities, which were closely apposed to dendrites (B–D) and the cell body (E) of PV neurons. Some apposed varicosities formed multiple appositions to a dendritic branch or a soma; for example, five, two, and three varicosities closely apposed to a dendritic branch and a soma (arrows in C, D, E, respectively). White arrowheads in F indicate a typical asymmetric synaptic contact that was made between the biocytin-labeled axon terminal (AT) and the dendrite (Den) with GFP immunoreactivity. Biocytin and GFP were visualized with the DAB/nickel and DAB reactions, respectively. Extended Data Figure 2-1 shows other examples of asymmetric synaptic contact. Scale bars: 100 μm (A), 10 μm (in E; applies to B–E), and 500 nm (F).

Distribution of axon varicosities of layer 2/3 pyramidal neurons in close appositions to PV neurons. Two representative layer 2/3 pyramidal neurons are shown (A, B). Axons of pyramidal neurons were two-dimensionally reconstructed and projected onto the frontal plane. Black lines and filled circles represent axons and cell bodies of pyramidal neurons, respectively; red and blue circles indicate the axodendritic and axosomatic appositions, respectively. One each apposed varicosity is represented with one red or blue circle. Dark green and gray lines indicate reconstructed dendrites and their actual positions in the cortical layers, respectively. Recently, layer 2/3 pyramidal neurons are suggested to be divided into layer 2 and layer 3 neurons based on the morphology of their dendrites. In layer 2 neurons, the horizontal span of the apical dendrites is larger than that of the basal dendrites, whereas, in layer 3 neurons, the horizontal span of the basal dendrites is larger. According to this criterion, neuron 4 (B) shown here is presumed to be layer 2 neurons because the horizontal span of their apical dendrites is larger than that of basal dendrites. On the other hand, neuron 3 (A) shown here is assumed to be layer 3 neurons because the horizontal span of their basal dendrites is larger than that of apical dendrites. The other layer 2/3 pyramidal neurons are shown in Extended Data Figure 3-1. Scale bar: 500 μm.

Distribution of axon varicosities of layer 4 pyramidal neurons in close appositions to PV neurons. Two representative layer 4 pyramidal neurons are shown (A, B). Axons of pyramidal neurons were two-dimensionally reconstructed and projected onto the frontal plane. Black lines and filled circles represent axons and cell bodies of pyramidal neurons, respectively; red and blue circles indicate the axodendritic and axosomatic appositions, respectively. One each apposed varicosity is represented with a red or blue circle. Dark green and gray lines indicate reconstructed dendrites and their actual positions in the cortical layers, respectively. The other layer 4 pyramidal neurons are shown in Extended Data Figure 4-1. Scale bar: 500 μm.

Distribution of axon varicosities of layer 5 pyramidal neurons in close appositions to PV neurons. Representative layer 5a (A) and layer 5b (B) pyramidal neurons are shown. Axons of pyramidal neurons were two-dimensionally reconstructed and projected onto the frontal plane. Black lines and filled circles represent axons and cell bodies of pyramidal neurons, respectively; red and blue circles indicate the axodendritic and axosomatic appositions, respectively. One each apposed varicosity is represented with a red or blue circle. Dark green and gray lines indicate reconstructed dendrites and their actual positions in the cortical layers, respectively. The layer 5b pyramidal neuron (B) had well developed apical dendrites than the layer 5a neuron (A). The other layer 5a and 5b pyramidal neurons are shown in Extended Data Figures 5-1, 5-2, respectively. Scale bar: 500 μm.

Distribution of axon varicosities of layer 6 pyramidal neurons in close appositions to PV neurons. Representative layer 6 CT-like (A) and CC-like (B) pyramidal neurons are shown. Axons of pyramidal neurons were two-dimensionally reconstructed and projected to the frontal plane. Black lines and filled circles represent axons and cell bodies of pyramidal neurons, respectively; red and blue circles indicate the axodendritic and axosomatic appositions, respectively. One each apposed varicosity is represented with a red or blue circle. Dark green and gray lines indicate reconstructed dendrites and their actual positions in the cortical layers, respectively. The other layer 6 CT-like and CC-like pyramidal neurons are shown in Extended Data Figures 6-1, 6-2, respectively. Scale bar: 500 μm.

Comparisons of morphological and physiological properties between layer 6 CT-like and CC-like pyramidal neurons. A, A scattergram showing the relationship between horizontal spans of axon collaterals of layer 6 pyramidal neurons and fast afterhyperpotentials. Layer 6 pyramidal neurons were subdivided into two groups based on the horizontal spans of their axon collaterals; layer 6 CT-like pyramidal neurons, which had axon collaterals restricted within 500-μm width, and layer 6 CT-like pyramidal neurons, which possessed wider axon collaterals extended >1200-μm width. All the five layer 6 CT-like pyramidal neurons showed deeper fast hyperpolarizing afterpotentials than those of the layer 6 CC-like pyramidal neurons. B–H, Comparison of morphological properties between layer 6 CT-like and CC-like pyramidal neurons. The number of all the labeled varicosities of layer 6 CT-like neurons was smaller than that of layer 6 CC-like neurons (C). In contrast, the percentage of apposed varicosities to all the labeled varicosities of layer 6 CT-like pyramidal neurons was significantly larger than that of layer 6 CC-like pyramidal neurons (G). Marks, bars, and error bars in B–H indicate individual values, means, and SDs, respectively.

Quantitative comparisons of varicosities and appositions of pyramidal neurons in each layer. A, The number of varicosities per single pyramidal neurons in each layer. Varicosities of pyramidal neurons in layers 2–5a were mainly distributed in layers 1–5a, whereas those of pyramidal neurons in layers 5b–6 were mostly located in layers 5b–6. B, The number of appositions per single pyramidal neuron in each layer. In layers 1–5a, the largest average number of appositions from each pyramidal neuron to PV-neuron dendrites was observed for layer 4 pyramidal neurons, followed by layer 2/3 pyramidal neurons. Layer 6 CC-like pyramidal neurons formed the largest number of appositions to PV-neuron dendrites in layers 5b–6. C, Percentage of apposed varicosities to all the labeled varicosities in each layer. Pyramidal neurons in layers 2–5a most frequently formed appositions to PV-neuron dendrites in layer 5a, whereas pyramidal neurons in layers 5b–6 most frequently apposed to PV-neuron dendrites in layers 5b and 6a. Marks and error bars in A–C indicate means and SDs, respectively.

Area proportion of PV-neuron dendrites in each layer, and κ. The area proportions of PV-neuron dendrites were measured in the primary motor (M1; A), secondary motor (M2), forelimb (FL), and hindlimb (HL) areas. The frontal sections from PV/myrGFP-LDLRct mice were stained with DAPI and immunostained for VGluT2 (Alexa Fluor 594) to determine cortical layer structure, and observed and captured images under a fluorescent microscope or under a confocal laser-scanning microscope (A). In the captured images, GFP-signal was binarized and the GFP-positive area (except cell bodies) was measured. The area proportion of PV-neuron dendrites was plotted in B. Among the motor-associated areas, the area proportion of PV-neuron dendrites showed a similar distribution pattern; PV-neuron dendrites existed most densely in layer 4, followed by layer 5a and layer 5b. To quantify input maps from a pyramidal neuron to PV neurons, we introduced κ (C). The κ of pyramidal neurons in layers 2–5 (yellow, green, dark-blue, and peal-blue lines, respectively, in C) were almost constant and around 3 in all layers, indicating that these neurons uniformly form appositions to PV-neuron dendrites in all layers. In contrast, κ for layer 6 CT-like pyramidal neurons (purple line in C) was dynamically changed: κ was ∼3 in layers 1–5a, whereas κ was much larger than 3 in layers 5b–6b. The average of κ was compared between layers 1–5a and layers 5b–6b (D; upper graph) or layers 5b–6a (D; lower graph). Statistical significance was assessed with Tukey’s multiple comparison test after a two-way analysis of variance (****p < 0.0001). Marks/bars and error bars in B–D indicate means and SDs, respectively. Scale bar: 100 μm (A).