The distribution of intrahippocampal projections arising from the CA3 region of the rat hippocampus was investigated using in vitro and in vivo methods. In the in vitro hippocampal slice preparation, single CA3 pyramidal cells were intracellularly labeled with horseradish peroxidase (HRP), and the three‐dimensional organization of the axonal plexus was analyzed by using a computer‐aided digitizing system. As many as eight primary collaterals originated from the principal axon of CA3 pyramidal cells and these commonly bifurcated further and innervated stratum oriens and stratum radiatum of CA3 and CA1. Within the 400 μm slice, the summed length of all visible collaterals per neuron ranged from 2.6 mm to approximately 12.5 mm. While the CA3 principal axon tended to be relatively smooth, the axonal collaterals bore numerous varicosities that electron microscopy confirmed to be presynaptic boutons. These varicosities occurred, on average, once every 7 μm of collateral length. The distribution of axonal collaterals differed depending on the location of the parent pyramidal cell. Only rarely could CA3 collaterals be followed in the slice to their terminations within CA1.

To study the topographic organization of CA3 projections both to other levels of CA3 and to CA1, the anterograde tracer, Phaseolus vulgaris leucoagglutinin (PHA‐L) was injected into various transverse and septotemporal levels of CA3. Immunohistochemical visualization of the lectin was conducted in dissected and “extended” hippocampi to facilitate analysis of the topographic distribution of projections along the long or septotemporal axis. Projections from all portions of CA3 reached widespread regions of CA3, CA2, and CA1, but only a few fibers entered the subicular complex and there were no projections to the entorhinal cortex. There were also some CA3 and CA2 projections to the hilus of the dentate gyrus, but these did not enter the granule cell or molecular layers. The CA3 projections to CA1 were organized according to several distinctive and consistent gradients that can generally be summarized as follows. 1. CA3 cells located close to the dentate gyrus (proximal CA3), while projecting both septally and temporally, tended to project more heavily to levels of CA1 located septal to the injection site. CA3 cells located closer to CA1, in contrast, projected more heavily to levels of CA1 located temporally to the injection site. 2. At, or close to, the septotemporal level of the injection, cells located proximally in CA3 gave rise to collaterals that tended to terminate more superficially in stratum radiaum than did those arising from mid and distal levels of CA3. Conversely, cells located more distally in CA3 gave rise to collaterals that terminated deeper in stratum radiatum and in stratum oriens. 3. At, or close to, the septotemporal level of the injection site, CA3 pyramidal cells located near the dentate gyrus tended to project somewhat more heavily to distal portions of CA1 (near the subicular border), whereas CA3 projections arising from cells located distally in CA3 terminated more heavily in portions of CA1 closer to the CA2 border. 4. Regardless of the septotemporal or transverse location of the injection, the highest density of terminal and fiber labeling in CA1 tended to shift to deeper parts of stratum radiatum and stratum oriens at levels septal to the injection and shifted out of stratum oriens and into superficial parts of stratum. radiatum at levels temporal to the injection site.

Moreover, the highest density of fiber and terminal labeling in CA1 shifted proximally (towards CA3) at levels septal to the injection and shifted distally (towards the subiculum) at levels temporal to the injection site. The …