Patterns of inter- and intralaminar GABAergic connections distinguish striate (V1) and extrastriate (V2, V4) visual cortices and their functionally specialized subdivisions in the rhesus monkey

Local GABAergic connections are undoubtedly important for the operation of cerebral cortex, including the tuning of receptive field properties of visual cortical neurons. In order to begin to correlate specific configurations of GABAergic networks with particular receptive field properties, we examined the arrangement of GABAergic neurons projecting to foci in compartments of known functional specialization in striate (area V1) and extrastriate (areas V2, V4) cortices of rhesus monkeys. GABAergic cells were detected autoradiographically following microinjections into supragranular, granular, or infragranular layers of 5, 10, or 50 nl of ³H-nipecotic acid, which selectively exploits the GABA reuptake mechanism. These injections produced complex inter- and intralaminar distributions of retrograde perikaryal labeling that was selective for GABA-immunopositive neurons and glia. The pattern of retrograde labeling depended on both the laminar and cytoarchitectonic location of injection sites. In all cases, a high density of labeled neurons was present in the immediate vicinity of injection sites, with the density of labeled neurons decreasing for the most part uniformly with horizontal distance. Injections in supragranular layers produced relatively widespread labeling (up to 1.5-1.7 mm from the center of injections) in upper layers, whereas in granular and infragranular layers, labeling was confined to a radius of 0.25-0.5 mm. Conversely, injections in infragranular layers produced labeling that was widest (up to 1 mm) in lower layers, but more laterally restricted in supragranular layers. Injections in granular layers, on the other hand, produced an even distribution of labeling, 0.6-1.0 mm in diameter, throughout all layers. Comparably placed injections in V1, V2, and V4 resulted in patterns of labeling that were distinguished by features including stepwise increases in the lateral extent of labeling from striate to extrastriate areas, and the circular versus markedly elongated intralaminar distribution of labeled neurons in V1 and V4 versus V2. Further, for superficial injections, labeling was present in all layers in V1 and V2, but did not extend below the top of layer V in area V4. These findings offer clear examples of organizational differences in the intrinsic inhibitory connections of visual cortices. The results also demonstrate that the number of GABAergic neurons projecting to any spot in cortex decreases systematically with horizontal distance from the spot, and that radiolabeled cells do not coalesce to form slabs, columns, or clusters. This relatively even distribution of retrogradely labeled cells in the tangential plane is consistent with recent computer simulations (Worgotter and Koch, 1991) that suggest that inhibitory neurons broadly tuned as a population can produce the specific response properties of cortical neurons. Injections fortuitously confined to cytochrome oxidase compartments, for example, blobs or stripes in V1 and V2, produced most labeling within like compartments. However, some labeling always traversed these histochemical boundaries. Thus, inhibitory connections within V1 and V2 may provide a means for cross talk between discrete channels of visual information, whose anatomical segregation has been a hallmark of V1 and V2 organization.