Two sets of experiments were carried out to examine the organization of associational connections within the rat entorhinal cortex. First, a comprehensive analysis of the areal and laminar distribution of intrinsic projections was performed by using the anterograde tracers Phaseolus vulgaris-leuocoagglutinin (PHA-L) and biotinylated dextran amine (BDA). Second, retrograde tracers were injected into the dentate gyrus and PHA-L and BDA were injected into the entorhinal cortex to determine the extent to which entorhinal neurons that project to different septotemporal levels of the dentate gyrus are linked by intrinsic connections. The regional distribution of intrinsic projections within the entorhinal cortex was related to the location of the cells of origin along the mediolateral axis of the entorhinal cortex. Cells located in the lateral regions of the entorhinal cortex gave rise to intrinsic connections that largely remained within the lateral reaches of the entorhinal cortex, i.e., within the rostrocaudally situated entorhinal band of cells that projected to septal levels of the dentate gyrus. Cells located in the medial regions of the entorhinal cortex gave rise to intrinsic projections confined to the medial portion of the entorhinal cortex. Injections made into mid-mediolateral regions of the entorhinal cortex mainly gave rise to projections to mid-mediolateral levels, although some fibers did enter either lateral or medial portions of the entorhinal cortex. These patterns were the same regardless of whether the projections originated from the superficial (II-III) or deep (V-VI) layers of the entorhinal cortex. This organizational scheme indicates, and our combined retrograde/anterograde labeling studies confirmed, that laterally situated entorhinal neurons that project to septal levels of the dentate gyrus are not in direct communication with neurons projecting to the temporal portions of the dentate gyrus. These results suggest that entorhinal intrinsic connections allow for both integration (within a band) and segregation (across bands) of entorhinal cortical information processing.

The organization of CA1 projections to the rat subiculum was investigated with the anterograde tracer, Phaseolus vulgaris leucoagglutinin (PHA-L). Discrete iontophoretic injections of PHA-L were placed into various transverse positions of the CA1 field at different septotemporal levels of the hippocampus. The distribution of CA1 projections was observed in dissected and extended hippocampal preparations. CA1 cells located proximally in the field, i.e., close to the CA2 field, gave rise to projections that terminated in the distal third of the subiculum, i.e., close to the presubiculum. CA1 cells located distally in the field, i.e., close to the subiculum, gave rise to projections that terminated proximally in the subiculum, i.e., just across the CA1/subiculum border. CA1 cells in the middle of the field projected to a midtransverse portion of the subiculum. The same general pattern of projections was observed at all septotemporal levels of the hippocampus. Varicose fibers from the CA1 neurons terminated among the basal dendrites of the subicular pyramidal cells, within the pyramidal cell layer, and in the deep portion of the molecular layer. In addition to the CA1 to subiculum projections, the discrete PHA-L injections provided the opportunity of examining the extent of local and associational connections within CA1. In general, associational connections in CA1 are far less extensive than in CA3. CA1 is not entirely without local connections, however. CA1 cells located close to the subicular border, for example, originated axons that first innervated the proximal subiculum and then reentered the CA1 field at the interface between stratum radiatum and stratum lacunosum-moleculare. In most of the experimental cases, there were collaterals located in stratum oriens of CA1 that branched from the fibers directed toward the subiculum. Thus, the basal dendrites of CA1 cells may receive associational inputs. The organization of the CA1 projections to the subiculum is discussed in relation to the organization of CA3 projections to CA1 and the differential output of transverse regions of the subiculum. The possibility is raised that information may be "channeled" through the hippocampal formation via the transverse organization of these connections and ultimately distributed to different recipients of hippocampal efferent projections.

By using three-dimensional computer reconstruction techniques and the production of two-dimensional unfolded maps, we analyzed the topographic organization of projections from the entorhinal cortex of the rat to the dentate gyrus. The retrograde tracers, Fast blue and Diamidino yellow, were injected at all septotemporal levels of the dentate gyrus, and the distribution of retrogradely labeled layer II cells in the entorhinal cortex was plotted by using computer-aided microscopy systems. Discrete injections of fluorescent dyes into the dentate gyrus labeled bands of layer II neurons in the entorhinal cortex that covered approximately 45% of its surface area. Injections confined to the septal half of the dentate gyrus resulted in a band that occupied the most lateral and caudomedial portions of the entorhinal cortex. Although there were subtle changes in the density of labeled cells in this region, essentially the same region of cells was labeled after any injection into the septal half of the dentate gyrus. Injections into mid-septotemporal levels of the dentate gyrus (50-75% of the distance from the septal pole) led to a distinctly different pattern of retrograde labeling. A more medial portion of the lateral entorhinal cortex and a more rostral portion of the medial entorhinal area were labeled in these cases. Another change in entorhinal labeling occurred when the injection involved the most temporal quarter of the dentate gyrus. Injections into this area led to a constrained region of entorhinal labeling that included the most medial portion of the lateral entorhinal area and the most rostral portion of the medial entorhinal area. Although the domains of cells projecting to septal, mid-septotemporal, and temporal levels of the dentate gyrus were not entirely segregated, there was relatively little overlap of the three populations of neurons. These data raise the possibility that different portions of the entorhinal-hippocampal circuit are capable of semiautonomous information processing, at least at the stage of input to the dentate gyrus.

By using three-dimensional computer reconstruction techniques and the production of two-dimensional unfolded maps, we analyzed the topographic organization of projections from the entorhinal cortex of the rat to the dentate gyrus. The retrograde tracers, Fast blue and Diamidino yellow, were injected at all septotemporal levels of the dentate gyrus, and the distribution of retrogradely labeled layer II cells in the entorhinal cortex was plotted by using computer-aided microscopy systems. Discrete injections of fluorescent dyes into the dentate gyrus labeled bands of layer II neurons in the entorhinal cortex that covered approximately 45% of its surface area. Injections confined to the septal half of the dentate gyrus resulted in a band that occupied the most lateral and caudomedial portions of the entorhinal cortex. Although there were subtle changes in the density of labeled cells in this region, essentially the same region of cells was labeled after any injection into the septal half of the dentate gyrus. Injections into mid-septotemporal levels of the dentate gyrus (50-75% of the distance from the septal pole) led to a distinctly different pattern of retrograde labeling. A more medial portion of the lateral entorhinal cortex and a more rostral portion of the medial entorhinal area were labeled in these cases. Another change in entorhinal labeling occurred when the injection involved the most temporal quarter of the dentate gyrus. Injections into this area led to a constrained region of entorhinal labeling that included the most medial portion of the lateral entorhinal area and the most rostral portion of the medial entorhinal area. Although the domains of cells projecting to septal, mid-septotemporal, and temporal levels of the dentate gyrus were not entirely segregated, there was relatively little overlap of the three populations of neurons. These data raise the possibility that different portions of the entorhinal-hippocampal circuit are capable of semiautonomous information processing, at least at the stage of input to the dentate gyrus.