CAl, and the Entorhinal Cortex in Rats

In order to examine whether the entorhinal-hippocampal-entorhinal circuit is reciprocal and topographic, the connections between the subiculum, the CA1 field, and the entorhinal cortex were studied with the carbocyanine dye (Dil), which moves in both retrograde and anterograde directions. We investigated the organization of reciprocal connections revealed by injections of Dil in the entorhinal cortex along the rhinal sulcus. Anterograde fluorescent labeling showed the same pattern reported in previous studies of the dorsal hippocampus. When the injection site of DiI extended into the deep layers (IV-VI) of the same cortical column, the anterograde labeling of the perforant path was accompanied by retrograde labeling of the subicular neurons and the CA1 neurons. The distribution of labeled cells overlapped the distribution of labeled fibers, and the distribution of labeled cells paralleled that of the labeled fibers in the CA1 field. DiI injection into the medial entorhinal cortex revealed fewer retrogradely labeled subicular neurons than injection into the lateral entorhinal cortex, whereas the number of labeled CAI neurons was not dependent on the injection site. The number of labeled CAI neurons was always several times greater than the number of subicular neurons. Thus, the amount of information conveyed by the CA1 projection might be higher than that conveyed by the subicular projection. These results indicate that the entorhinal cortex, CA1, and the subiculum are connected reciprocally and topographically. We believe that the framework of the major hippocampal circuit proposed in previous studies should be reconsidered. We propose that the CA1 projection, rather than the subicular projection, is the main projection that feeds back information from the hippocampus to the entorhinal cortex. c 1995 Wiley-Liss, Inc. Indexing terms: hippocampus, perforant path, CA2, reciprocal connection, DiI The formation of memory is proposed to occur via a neocortex-hippocampus-neocortex circuit (Mishkin, 1982; Van Hoesen, 1982). In support of this idea, neocortical events initiated by multimodal sensory input are stored in the neocortex as long-term memory (Sakai and Miyashita, 1991), and hippocampal dysfunction contributes primarily to an anterograde memory impairment (Scoville and Milner, 1957; Zola-Morgan et al., 1986). Therefore, the hippocampal formation has been expected to have access to cortically processed sensory information and to the memory storage sites in the neocortex through afferent and efferent connections. Accordingly, previous studies have reported that the connections between the parahippocampal gyrus and the neocortex are reciprocal (Van Hoesen, 1982; Amaral, 1987). However, demonstration of reciprocal connections between the parahippocampal gyrus and the hippocampal formation is lacking. In fact, the main connections between the parahippocampal gyrus and the hippocampus have been regarded as a one-way loop. On the other hand, recent physiological observations in rat behavior in vivo (Buzsaki, 1989) as well as in the isolated guinea pig brain in vitro (deCurtis et al., 1991) have shown that the entorhinalhippocampal loop can give rise to a few cycles of reverberant activity at a frequency of approximately 20-40 Hz following high-intensity entorhinal stimulation. This predicts that the output from one cortical column in the entorhinal cortex projects to a region of the hippocampus that, in turn, projects to the same cortical column in the entorhinal cortex. Such an organization would provide an anatomical basis for the observed reverberant activity as well as providing the reciprocal connections necessary for the formation of memories. To test whether this predicted organization exists, we investigated the reciprocal connecAccepted August 24,1994 Address reprint requests to Nobuaki Tamamaki, Department of Anatomy, Fukui Medical School, Matsuoka, Fukui 910-11, Japan. o 1995 WILEY-LISS, INC.