Dense, unspecific connectivity of neocortical parvalbumin-positive interneurons: a canonical microcircuit for inhibition?

Abstract

GABAergic interneurons play a major role in the function of the mammalian neocortex, but their circuit connectivity is still poorly understood. We used two-photon RuBi-Glutamate uncaging to optically map how the largest population of cortical interneurons, the parvalbumin-positive cells (PV+), are connected to pyramidal cells (PCs) in mouse neocortex. We found locally dense connectivity from PV+ interneurons onto PCs across cortical areas and layers. In many experiments, all nearby PV+ cells were connected to every local PC sampled. In agreement with this, we found no evidence for connection specificity, as PV+ interneurons contacted PC pairs similarly regardless of whether they were synaptically connected or not. We conclude that the microcircuit architecture for PV+ interneurons, and probably neocortical inhibition in general, is an unspecific, densely homogenous matrix covering all nearby pyramidal cells.

Figure 1.

Characterization and photostimulation of parvalbumin-positive fast-spiking basket cells. A, Reconstruction of a GFP-positive interneuron in somatosensory cortex layer 2/3 from the G42 transgenic mouse line (axons in blue, dendrites in red). B, Electrophysiological recordings revealed the classic high rheobase and fAHP current (top). A fast-spiking response was elicited after injecting the interneuron with 1 nA of current (bottom). C, This subtype of interneuron is characterized by a rectifying IV curve (Woodruff et al., 2009). D, E, A two-photon laser beam was multiplexed across time by sequentially stimulating multiple targets around a cell body (D) and across space by engaging a diffractive optical element to split the laser beam into five beamlets (E) to uncage RuBi-Glutamate around a cell body (laser spots are drawn at actual size, see H–I). F, Single action potentials were elicited via illumination with 150mW of 800 nm light on the sample using a 20× 0.5 NA objective (black bar shows duration of illumination). G, At a higher power (300 mW) multiple action potentials were elicited with the characteristic fast-spiking waveform. H, The point-spread function of the diffractive optical element beamlets was measured with the 20× 0.5 NA objective we used for mapping and 0.17 μm fluorescent beads. The lateral size of each diffractive optical element beamlet is ∼1 μm (inset, raw image). I, The axial resolution of each beamlet is 6 μm. J, Direct test of two-photon photostimulation spatial resolution. The locations of 180 stimulation targets that were stimulated during the recording of the PC (center, black) are indicated by black circles. Location 1, The target indicated in blue, directly on the soma of the PC, was photostimulated directly to produce bursts of action potentials 10 times (one representative example is shown at right, top). Location 2, This target (upper red circle), which was just slightly offset from the center of the patched PC, caused the cell to fire only once (right, middle). Location 3, Photostimulation of this target (lower red circle) also caused the patched PC to fire single action potentials (right, bottom).

Figure 2.

Mapping inhibitory inputs of pyramidal cells. A, Mapping potential connections from five interneurons (numbered 1 to 5) to one PC (blue arrow) in somatosensory layer 2/3. Gray circles, Unconnected interneurons; red circles, connected. B, Electrophysiological recordings obtained from the PC at holding potentials of +40 mV (top) and −40 mV (bottom) during the photostimulation of the interneurons shown in A. Note the synaptic response from one interneuron (A, red arrow; higher magnification in D) while an interneuron directly nearby (A, gray arrow; higher magnification in F) showed no response. C, E, The red PV+ interneuron neuron was patched and confirmed to be synaptically connected while no response was recorded for the gray neuron (F), which was determined to be unconnected by photostimulation. G, True positive response. A PC patched in a field of view showing many GFP-positive PV+ interneurons nearby, one of which (red circle) was determined to be putatively connected by optical stimulation. H, While holding the PC at +40 mV (top) or −40 mV (bottom), we photostimulated the cell circled in G, eliciting IPSCs. The reversal potential for GABA was set at −80 mV in the internal solution so that EPSCs, but not IPSCs, would change directions between +40 and −40 mV. I, J, The photostimulated cell was patched, confirming electrophysiologically that it was indeed synaptically connected. K, Negative responses. A PV+ interneuron (gray circle) was targeted for photostimulation during whole-cell recording of two nearby PCs. L, The recordings from the PCs in K show no evidence of a synaptic connection during photostimulation of the PV+ interneuron. M, N, The photostimulated cell was patched, confirming electrophysiologically the lack of synaptic connections from the PV+ interneuron onto either PC. O, False positive response. A PV+ interneuron (black circle) directly next to the recorded PC was targeted for photostimulation. P, A slow response, which was outward at +40 mV but flipped at −40 mV, was distinguished from the synaptic event shown in H as a direct stimulation of the patched neuron. Q, A PV+ interneuron (black circle), located distal to the recorded PC (blue arrow), was targeted for photostimulation. R, An excitatory cell connected to the recorded PC was stimulated, resulting in a false positive distinguished by the presence of EPSCs.

Figure 3.

Simultaneous mapping of PV+ inputs to four PCs. Four PCs in somatosensory cortex layer 2/3 were patched and the surrounding PV+ interneurons in four different focal planes were optically stimulated. Each column shows the result of mapping the GFP-labeled PV+ interneurons for each of the four PCs. Each row shows the result from a different focal plane, with the projection of all the focal planes in the top row. Connected PV+ interneurons are circled in red, unconnected PV+ interneurons are circled in gray, and PV+ interneurons at which a direct stimulation of the patched PC occurred are circled in black.

Figure 4.

Dense PV+ interneuron inputs to PCs. A, Representative map from S2/3 showing inputs from PV+ interneurons to PCs (red, connected; gray, unconnected; black, false positive). B, Number of PV+ interneurons tested optically per patched PC (no significant differences) and how many were connected in each cortical area and layer sampled. C, Schematic of the cube of tissue we sampled in our maps. D, Histogram showing all the intersomatic distances of the PV+ interneurons from the PCs whose connectivity was tested (green bars) in the different cortical layers and areas. False positive responses (black bars) were few in number. E, Histogram of the connection probability depending on the intersomatic distance between interneurons and PCs. The connection probability of PV+ interneurons to PCs in all areas and layers was very high when the PV+ interneurons were close to the PC, and fell off with distance. F, The probability of observing a connection from a PV+ interneuron to a PC within 200 μm in S2/3, S5, and F2/3. G, Connection probability for all optically stimulated interneurons across different cortical areas and layers. The probability was lower in somatosensory layer 2/3 than in somatosensory layer 5 or frontal layer 2/3. H, The probability of observing a connection from a PV+ interneuron to a PC within 200 μm in older animals (no statistical differences). I, Connection probability for all optically stimulated interneurons in older animals (no statistical differences). n.s., Not significant. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 5.

Confirmation of dense PV+ innervation of PCs. A, Anatomical reconstruction of a connected pair with a PV+ interneuron connected to a PC. Blue, PV+ axon; red, PV+ dendrites; green, PC axon; black, PC dendrite. B, Electrophysiological recording of this PV+ interneuron to PC connection showing IPSCs in the PC in response to a train of eight action potentials evoked at 50 Hz in the PV+ interneuron. The IPSCs depress over the course of the action potential train. C, Intersomatic distances of the tested pairs from dual whole-cell recording experiments (red, connected; gray, unconnected). D, Probability of connection observed at different intersomatic distances. E, Connection probability within 100 μm did not differ whether observed with mapping or patching in all tested cortical areas and layers. Within this nearby range, there was no difference in connection probability between the layers and areas.

Figure 6.

Lack of specificity in PV+ interneuron connectivity. A, The common connection probability, calculated as the number of PV+ interneurons that connect to two PCs out of the number of PV+ interneurons that connect to either PC, does not differ depending on whether the postsynaptic PCs are connected to each other. B, The common connection probability is inversely proportional to the distance between the PCs (r = −0.3898).

Figure 7.

Spatial pattern of connected PV+ interneurons varies between layers. A, Position of the connected (red) and unconnected (gray) PV+ interneurons plotted relative to the recorded PC (center of each plot). B, Polar plots show the probability of connection from a given angular region with the pial surface at the top of the circle. The probability of connection is 1 at the outer edge of a circle and 0.5 at the middle dotted circle. Polar plots in each column represent different cortical areas and layers. Rows show the angular distribution of connection counting only interneurons at certain distances (within 200 μm, from 200 to 600 μm, or all distances).

Figure 8.

Model showing convergence of PV+ interneurons onto one PC. A, A cube of neocortex 500 μm on each side from somatosensory layer 2/3 with 5000 neurons. The cube was constructed with PV+ interneurons and PCs using the connectivity profile we observed experimentally in S2/3, with one representative PC in black at the center. Other PCs are in blue, connected PV+ interneurons in red, and unconnected PV+ interneurons in gray. The density of connected PV+ interneurons converging on this one PC can be visualized in three dimensions. B, Schematic of the densely connected circuit architecture from PV+ interneurons onto local PCs.