Fear-context association during memory retrieval requires input from granular to dysgranular retrosplenial cortex

The contribution of the granular (area 29, A29) and dysgranular (area 30, A30) subdivisions of the retrosplenial cortex (RSC) to contextual fear memory (CFM) retrieval remains elusive. Here, intact and orchiectomized (ORC) male rats received an intraperitoneal (I.P.) injection of saline (control) or 5 mg/Kg MK801 after training and memory formation. In ORC, but not in intact males, this MK801 treatment selectively induces overt loss of neurons in layers IV-Va of A29 (A29^MK801 neurons) (Sigwald et al., 2016). Compared to ORC-saline, ORC-MK801 rats showed impaired CFM retrieval in an A-B-A design for contextual fear conditioning (CFC), however context recognition was not affected. In ORC-MK801 rats, neither novel object recognition nor object-in-context discrimination were impaired, further indicating that A29^MK801 neurons are not required for contextual recognition. Elevated plus maze test showed that anxiety-like behavior was not affected in ORC-MK801 animals, suggesting that loss of A29^MK801 neurons does not affect the emotional state that could impair freezing during test. Importantly, in a sensory preconditioning test, higher order CFM retrieval was abolished in ORC-MK801, but not in male-MK801. Collectively, these observations indicate that A29^MK801 neurons are critically required for retrieving fear-context association. For dissecting the anatomofunctional contribution of A29^MK801 neurons to CFM retrieval, expression of c-Fos and Egr-1 was used to map brain-wide neuronal activity. In control male rats CFC and CFM retrieval was associated with significant enhancement of both proteins in limbic structures and A30, but not in A29, suggesting that neurons in A30 and limbic structures encode and store the associative experience. Notably, in ORC but not in intact males, MK801 impairs CFM retrieval and expression of c-Fos and Egr-1 proteins in A30, without affecting their expression in limbic structures. Thus, the loss of A29^MK801 neurons after CFM formation precludes activation of associative neurons in A30, impairing CFM recall. FluoroGold retrograde track-tracing confirmed that A29^MK801 neurons project to A30. Silver staining provide evidence that MK801 in ORC rats induces axonal deafferentation of A29^MK801 neuron in A30. Collectively, our experiments provide the first evidence that A30 neurons participate in encoding and storing CFM while A29 is required for their activation during recall.