Hippocampal memory consolidation during sleep: a comparison of mammals and birds

Abstract

The transition from wakefulness to sleep is marked by pronounced changes in brain activity. The brain rhythms that characterize the two main types of mammalian sleep, slow-wave sleep (SWS) and rapid eye movement (REM) sleep, are thought to be involved in the functions of sleep. In particular, recent theories suggest that the synchronous slow-oscillation of neocortical neuronal membrane potentials, the defining feature of SWS, is involved in processing information acquired during wakefulness. According to the Standard Model of memory consolidation, during wakefulness the hippocampus receives input from neocortical regions involved in the initial encoding of an experience and binds this information into a coherent memory trace that is then transferred to the neocortex during SWS where it is stored and integrated within preexisting memory traces. Evidence suggests that this process selectively involves direct connections from the hippocampus to the prefrontal cortex (PFC), a multimodal, high-order association region implicated in coordinating the storage and recall of remote memories in the neocortex. The slow-oscillation is thought to orchestrate the transfer of information from the hippocampus by temporally coupling hippocampal sharp-wave/ripples (SWRs) and thalamocortical spindles. SWRs are synchronous bursts of hippocampal activity, during which waking neuronal firing patterns are reactivated in the hippocampus and neocortex in a coordinated manner. Thalamocortical spindles are brief 7-14 Hz oscillations that may facilitate the encoding of information reactivated during SWRs. By temporally coupling the readout of information from the hippocampus with conditions conducive to encoding in the neocortex, the slow-oscillation is thought to mediate the transfer of information from the hippocampus to the neocortex. Although several lines of evidence are consistent with this function for mammalian SWS, it is unclear whether SWS serves a similar function in birds, the only taxonomic group other than mammals to exhibit SWS and REM sleep. Based on our review of research on avian sleep, neuroanatomy, and memory, although involved in some forms of memory consolidation, avian sleep does not appear to be involved in transferring hippocampal memories to other brain regions. Despite exhibiting the slow-oscillation, SWRs and spindles have not been found in birds. Moreover, although birds independently evolved a brain region--the caudolateral nidopallium (NCL)--involved in performing high-order cognitive functions similar to those performed by the PFC, direct connections between the NCL and hippocampus have not been found in birds, and evidence for the transfer of information from the hippocampus to the NCL or other extra-hippocampal regions is lacking. Although based on the absence of evidence for various traits, collectively, these findings suggest that unlike mammalian SWS, avian SWS may not be involved in transferring memories from the hippocampus. Furthermore, it suggests that the slow-oscillation, the defining feature of mammalian and avian SWS, may serve a more general function independent of that related to coordinating the transfer of information from the hippocampus to the PFC in mammals. Given that SWS is homeostatically regulated (a process intimately related to the slow-oscillation) in mammals and birds, functional hypotheses linked to this process may apply to both taxonomic groups.

Fig. 1

(A, B) Histological comparison of the hippocampal formation between the rat and pigeon. (A) Layers of the dentage gyrus (DG) and cornu ammonis (CA) are conspicuous in the rat. (B) The V-shaped layer is the only readily apparent structure in the pigeon hippocampus (Hp). APH: area parahippocampalis. (C, D) Location and extent of the pigeon hippocampal formation (HF, dark gray) and dorsolateral corticoid area (CDL, light grey). (C) Dorsal view. (D) Transverse section. Scale bars in A, B = 1 mm. Reproduced with permission from Atoji & Wild (2007).

Fig. 2

Hypothesized homology of hippocampal formation (HF) subregions between mammals and birds. The medioventral V-shaped layer (shaded light grey) is comparable to the mammalian dentate gyrus, the dorsomedial region (DM) to cornu ammonis (CA) and the subiculum, and the dorsolateral region (DL) to the entorhinal cortex. Other histologically identified regions include the magnocellular region (Ma), the parvocellular region (Pa), and the cell-poor region (Po). CDL: dorsolateral corticoid area. Reproduced with permission from Atoji & Wild (2007).

Fig. 3

Afferent and efferent connections of the caudolateral nidopallium (NCL). Primary sensory areas are depicted in red, secondary and tertiary areas in pink. The primary sensory areas project to secondary and tertiary structures (small black arrows), which have reciprocal connections with the NCL (red arrows). The visual thalamofugal and tectofugal systems correspond to the geniculocortical and colliculo-pulvino-extrastriate systems of mammals, respectively. The area labeled “motor” is the arcopallium, which has descending projections to various motor and premotor structures. Thalamic afferents arise from the nucleus dorsolateralis posterior thalami (DLP). Dopaminergic afferents stem from the area ventralis tegmentalis (AVT) and the substantia nigra (SN). GP: globus pallidus. Reproduced with permission from Springer Science+Business Media: Encyclopedia of Neuroscience, Evolution of association pallial areas: in birds, 2009, pp. 1215–1219, Figure 2, Rose, Güntürkün & Kirsch.

Fig. 4

Mammalian-like slow-wave sleep (SWS) homeostasis in the pigeon (Columba livia). Spectral power density during SWS for the four 3-h quarters (black, first; red, second; blue, third; green, fourth) of the baseline night and the recovery night following 8 h of daytime sleep deprivation. The data presented is from bipolar electroencephalograms of the right anterior pallium. Black squares at the bottom of each plot reflect statistical significance (P < 0.05) for the first quarter (top row) to the fourth quarter (bottom row). Significance in the baseline plot reflects the comparison of power density at a specific frequency bin for that quarter of the baseline night to the all-night baseline SWS mean. In the recovery plot, significance reflects the comparison of power density at a specific frequency bin for that quarter of the recovery night to the corresponding frequency bin and quarter of the baseline night. Power increased most markedly in the low-frequency range (< 5.0 Hz) during the first quarter of the recovery night and progressively declined thereafter, a response indicative of mammalian-like SWS homeostasis. Modified from Martinez-Gonzalez et al. (2008). Pigeon image: Niels Rattenborg.

Fig. 5

Mammalian hippocampal formation showing sharp-wave/ripple complexes (SWRs). Self-organized bursts of activity in the hippocampal CA3 region produces a sharp-wave field potential in the dendritic layer of CA1 and a short-lived fast-frequency field oscillation (200 Hz ripple) within the stratum pyramidale, as well as a phase-related discharge of the neurons. Hippocampal output, in turn, produces similar SWRs in the subiculum (Sub), parasubiculum (Para) and deep layers of the entorhinal cortex (EC). Although SWRs occur spontaneously during quiet wakefulness, neocortical input biases their timing to the up-state of the neocortical slow-oscillation during slow-wave sleep. Reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Buzsáki & Chrobak, 2005.

Fig. 6

Proposed roles of the prefrontal cortex (PFC) in the formation and recall of remote memories. Initially, memories are encoded in hippocampal–neocortical networks (A, thick lines). At this early time point, the hippocampus is crucial for integrating information from distributed cortical modules, each representing individual components of a memory. However, over time direct projections from the hippocampus are thought to transfer a high-order representation of the memory to the PFC (B), which then uses this information to facilitate the transfer of information from the hippocampus to the neocortex, via the entorhinal and perirhinal cortices. As initially proposed for the hippocampus, the PFC may also use this version of the memory to strengthen the connections between the distributed cortical modules involved in the memory (thick lines), and to integrate the memory within related preexisting memories. Later, the PFC may also use this memory to identify and recall context-relevant information from remote memory stores. Finally, during recall of remote memories, the PFC appears to inhibit hippocampal activity (blue line), thereby preventing the encoding of redundant information. Reprinted by permission from Macmillan Publishers Ltd: Nature Reviews Neuroscience, Frankland & Bontempi, 2005.

Fig. 7

Functional implications of the similarities and differences between mammalian and avian sleep-related brain rhythms. REM: rapid eye movement. Rhythms shared by mammals and birds (grey overlap between circles) are likely to be involved in a fundamental function of each sleep state, whereas rhythms occurring only in mammals are most likely involved in mammal-specific functions. Images: pigeon (Niels Rattenborg), rat reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Mehta, 2007.