Very few diencephalic structures project directly (monosynaptically) to the hippocampal formation (HF). Two main diencephalic nuclei with direct projections to the HF are the nucleus reuniens (RE) of the midline thalamus and the supramammillary nucleus (SuM) of the caudal hypothalamus 1, 2, 3. The RE distributes to the stratum lacunosum moleculare (slm) of CA1 and to the ventral subiculum but lacks projections to the dentate gyrus (DG) and to CA2/CA3 of the HF 2, 4. By contrast, the SuM projects selectively to the DG and to CA2 of the hippocampus 1, 5, 6. While RE and SuM share some cortical and subcortical targets, RE and SuM distribute to separate or nonoverlapping sites in the HF. As discussed herein, this suggests that RE and SuM may participate in separate, or in some cases, complementary hippocampal-associated functions.

RE is reciprocally connected with the HF and the medial prefrontal cortex (mPFC) and in the absence of direct projections from the mPFC to the HF 7, 8, 9, RE is the major (indirect) route from the mPFC to the HF 10, 11, 12 (Figure 1). Accordingly, the functional properties of RE are closely aligned with those of the HF, the mPFC, and their interactions [13].

RE has been most extensively examined with respect to working memory (WM) — or more specifically to spatial working memory (SWM) 3, 14, 15. The hippocampus and the mPFC serve well-recognized roles in SWM (for review 13, 15), and as a bridge between these structures, RE has been firmly linked to SWM. For instance, Viena et al. [16] reported that the inactivation of RE produced deficits on a delayed nonmatch to sample T-maze task at various delays. Griffin and colleagues [17] similarly demonstrated that inactivating RE disrupted performance on a SWM version of a tactile/visual T-maze task. This same group [18] subsequently reported that theta/gamma oscillations of the mPFC became entrained to hippocampal theta during successful performance on the T-maze task and that the inactivation of RE disrupted HF–mPFC synchrony and task performance. Whereas RE serves a well-established role in SWS, Cassel and colleagues [19] reported that the inactivation of RE did not alter performance on a double-H water maze in rats. While the authors stated that there was no ‘convincing answer’ to the discrepancy between their results and prior ones, they speculated that the ‘higher spatial load’ of the double-H maze may have placed greater burden on the hippocampus, rather than on HF–mPFC interactions, rending this task less sensitive to disruptions of RE.

Fewer studies have examined the role of RE in executive functions — such as attention, behavioral flexibility, or goal-directed behaviors. Early reports, however, showed that lesions/inactivation of RE produced the types of impairments generally recognized as ‘executive’ dysfunctions. In particular, the inactivation of RE resulted in maladaptive search strategies on the water maze [20], an inability to switch strategies on ‘double H’ water maze task [21], and premature (impulsive) responding on a 5-choice serial reaction time task [22]. More recently, Linley et al. [23] reported impairments in attentional set formation and reversal learning in RE-lesioned rats using an odor/tactile set shifting task [24] — indicating an executive dysfunction.

The RE has also been shown to serve a role in affective behaviors, mainly fear conditioning 25, 26. For instance, Xu and Sudhof [27], using a combination of techniques, showed that the selective inactivation of mPFC terminals to RE produced an overgeneralization of contextual fear memories, and further that the direct suppression of RE with tetanus toxin or the activation of RE with neuroligin-2, produced, respectively (1) increases or decreases in fear memory generalization; and (2) inhibition or activation of CA1 and mPFC neurons measured with c-fos. In line with this, Quet et al. [28] reported that RE/rhomboid (RH) lesions (RH is often affected together with RE) disrupted remote but not recent contextual fear memory, indicating a critical role for RE in the systems consolidation of memories.

Maren and colleagues 29, 30, 31 described a critical role for RE in the acquisition, retrieval, and extinction of fear memory. Specifically, they showed that the inactivation of RE disrupted the acquisition of conditioned fear [29], and further resulted in a failure to extinguish fear — both cued and contextual fear. [31]. Regarding the latter, they proposed [31] that RE actively suppresses the reemergence of extinguished fear memories, stating “RE serves as a hub by which the mPFC regulates hippocampal activity to suppress the retrieval of fear memories.” In summary, the foregoing brief overview supports a critical role for RE in SWM, executive functions, and affective behavior, mainly fear.

The SuM has both direct and indirect projections to the HF and exerts effects on the HF via both routes. The major indirect SuM projection to the HF is through the medial septum and this projection has been extensively examined with respect to the hippocampal theta rhythm, that is, SuM is a major link in an ‘ascending synchronizing system’ responsible for the generation of theta 32, 33, 34. The present review, however, focuses on the direct SuM projections to the HF, that is, to the DG and to CA2 of the HF (Figure 1).

Whereas studies devoted to RE are ever increasing, fewer reports have focused on the SuM. Nonetheless, there has been a significant renewed interest in the properties of SuM mainly with respect to SuM effects on the DG of the HF. Based largely on SuM connections with the hippocampus, early studies primarily examined SuM effects on learning and memory (for review, [35]). Specifically, Shahidi et al. [36] initially showed that the inactivation of SuM produced deficits on a passive avoidance task, and subsequently [37] that inhibiting SuM disrupted spatial reference and WM on the water maze. Consistent with this, it was later shown that lesions [38] or pharmacological inactivation [39] of SuM severely impaired SWM on a delayed-matching-to-position task. Finally, Gutiérrez-Guzmán et al. [40] reported that the selective elimination of serotonergic fibers to SuM altered performance on a water maze task and disrupted the hippocampal theta rhythm.

The SuM has also been linked to various ‘affective’ processes, including novelty detection, response to stress, and reductions in anxiety. For instance, it was shown that (1) levels of c-fos expression in SuM were substantially elevated following exposure to novel environments, indicating an enhanced SuM response to novelty 41, 42; (2) select populations of SuM cells projecting to the HF were highly activated by immobilization stress [42]; and (3) SuM lesions produced anxiolytic effects on the elevated maze [43]. In summary, the foregoing supports a direct involvement of SuM in learning/memory and emotional behaviors — as alterations of SuM were reported to produce significant deficits in various mnemonic and affective tasks.