Sleep and memory consolidation: a common mechanism across species?

Abstract

In humans, memory consolidation can be aided by the representation of an odor previously associated with target information during sleep. In an elegant study, Zwaka et al. (Curr Biol 25: 2869–2874, 2015) have demonstrated that the same process occurs in honeybees, suggesting that the relationship between sleep and memory may be similar across different animal species.

human sleep is one of the most complex physiological mechanisms, so much that its prominent function is still unclear (Krueger et al. 2016). In the last few decades a wide range of studies converged toward the idea that sleep may be the optimal state for memory processing (Rasch and Born 2013). Memory, and more broadly cognitive, benefits provided by sleep have been observed not only in mammals but also in phylogenetically different animal species, such as birds (i.e., Zebra finches, European starlings) and insects (i.e., Drosophila melanogaster, Apis mellifera; Vorster and Born 2015). These animals show a sleeping behavior very similar to that detectable in mammals and characterized most notably by behavioral quiescence, increased arousal threshold, and state reversibility with stimulation.

Honeybees, for example, show complex behavioral correlates of sleep that have functional equivalents in mammals, humans included. For instance, in honeybees three different sleep stages can be distinguished by using behavioral criteria (i.e., antennal movements, body posture, sleep bout duration, and response threshold), and the absolute immobility of their antennae is considered a sign of deep sleep, equivalent to the slow-wave sleep stage of human non-rapid eye movement (NREM) sleep (Eban-Rothschild and Bloch 2008).

Capitalizing on the detectability of bees' sleep phases, Zwaka et al. (2015) demonstrated that memory consolidation can be improved in honeybees by re-presenting the learned context odor during deep sleep. Zwaka et al. assessed learning and memory consolidation by measuring the proboscis extension response, as follows: after a conditioning phase—the association between conditioned and unconditioned stimuli (i.e., thermal stimulus-sucrose reward pairing)—bees were exposed only to the conditioned stimulus (i.e., thermal stimulus) so that the unconditioned response (i.e., proboscis response) could be measured.

By pairing thermal-sucrose associations with the presentation of a context odor, Zwaka et al. (2015) showed that reexposing bees to the context odor (learned during the acquisition phase) five times during the subsequent deep-sleep phase selectively enhanced the proboscis extension response (i.e., memory retention). This was not evident in a control group of bees exposed to a noncontextual odor (paraffin oil, which is not perceived as an odor by the bees). This result suggests that the reexposure to a contextual odor during sleep may improve retention of the memory encoded while smelling the same odor during learning.

However, this result alone was not enough to prove that sleep plays a role in bees' memory consolidation. Indeed, the observed performance improvement could have resulted from the mere exposure to any (perceivable) odor during sleep, or from the exposure to the contextual odor irrespective of sleep or specific sleep phase involvement. To rule out these possibilities, Zwaka et al. (2015) addressed these confounding aspects through a series of additional experiments. Overall, they showed that reexposing honeybees during deep sleep to a context odor improves memory of the odor-paired material. They demonstrated that this effect is specific to deep sleep and does not occur during wake phases or other sleep phases. Furthermore, they showed that this effect may extend to the consolidation of “weak” memories (i.e., memory based on single-trial conditioning), which are usually quickly forgotten (Müller 2012).

Notably, these results are consistent with findings in humans. Indeed, it has been shown in humans that matching an olfactory stimulus with target information (e.g., pairs of words, objects, motor sequences) and then re-presenting the odor alone during sleep facilitates memory consolidation (Rasch and Born 2013). Moreover, this process seems to occur only in specific sleep phases (i.e., during N2 and N3 NREM sleep stages) and not in other stages or during wake phases (Rasch et al. 2007). Lastly, even in humans, “weak” encoding may be enhanced after reexposure is cued during sleep (Creery et al. 2015).

These similarities should be interpreted in light of some confounding aspects. For instance, as already mentioned by Zwaka et al. (2015), even though the bees were behaviorally categorized in diverse sleep stages and the successful reexposure was performed during deep sleep, we do not know whether this stage is the electrophysiological equivalent of slow-wave sleep in humans. Considering that in humans odor presentation during deep sleep modifies sleep features associated with memory consolidation such as EEG delta waves and sleep spindle activity (Perl et al. 2016), further studies are needed to assess whether these sleep-related sensory processes are also present in insects such as honeybees.

In addition, different sensory stimuli may affect sleep in specific ways depending on the species-specific importance of sensory modalities. For example, in bees, odors are detected by the antennae, which project to the antennal lobes (i.e., the primary olfactory center) and then to the mushroom body (MB), which is traditionally considered to be an associative memory network functionally similar to the mammalian cerebral cortex, where higher order processing of olfactory stimuli occurs. In bees, a precise elaboration of odors is fundamental for surviving behaviors such as navigating the environment, finding food, interacting with other bees, and finding a mate. Given the key role of odors, and the olfactory system, in honeybees' survival, odors may be particularly effective in modulating memories in this species.

Also, in humans, odors have a great impact on memory encoding and consolidation, and olfactory stimuli represent exceptional sensory stimuli given the lack of a necessary thalamic relay, which guarantees a privileged cortical access and the fact that during slow-wave sleep, the functional connectivity between olfactory, limbic, and cortical areas is enhanced (Barnes and Wilson 2013). Additionally, sleep-related memory enhancement in humans can be triggered by auditory stimuli (Creery et al. 2015; Rasch and Born 2013). However, whether sleep-related enhancement in bees is odor specific or may be triggered via other sensory pathways (e.g., acoustic or tactile stimulation) is still unknown.

Furthermore, different types of learning may be affected in different ways by sleep across species. For example, in humans, extinction of aversive conditioning is facilitated when individuals are reexposed to the contextual odor during N3 sleep, whereas in mice and rats, odor stimulation during sleep enhances fear responses (Diekelmann and Born 2015). Since Zwaka et al. (2015) used appetitive conditioning, it would be interesting to test whether and how contextual stimulation during sleep can modify aversive learning.

Last is a question that has been neglected in both animal and humans studies: the role of individual differences in memory consolidations. For example, in Zwaka et al.'s (2015) study, bees' learning abilities (during the acquisition phase) showed a high variability through the experiments reported in the study. Assuming that the general procedure in the experiment was kept fixed, it is parsimonious to consider that individual differences may play an important role in acquiring new information. How these individual differences affect the following sleep-related memory improvement remains unknown and needs further exploration, in both humans and animals.

Notwithstanding these confounding issues, overall the findings from both species can be interpreted within the theoretical framework of the active system consolidation (ASC) model (Rasch and Born 2013). The ASC proposes that memory consolidation is the result of an active mechanism that allows the reactivations of newly encoded memory representations during NREM sleep. These reactivations would mediate the redistribution of the temporally stored memory representation to long-term storage sites where they are integrated into preexisting memory traces via a parallel synaptic consolidation process that can be mediated by sleep spindles or take place during REM sleep. The data provided by Zwaka et al. (2015) are consistent the ASC theoretical model and suggest that this reactivation mechanism, aimed to transform newly encoded labile memory representations into stabile long-term information, also known as neural replay, may also be present in honeybees. Therefore, the findings of Zwaka et al. intriguingly indicate that the neural replay may be a common mechanism across several species, at least mammals, birds, and invertebrates such as bees and fruit flies.

However, although these behavioral results in bees overlap observations in humans, the neural circuits underlying memory processing during sleep remain unknown in bees. Nevertheless, some speculations have been advanced on the basis of findings of a very similar organism, Drosophila melanogaster (Vorster and Born 2015).

In Drosophila, the MB has been associated with sleep, but the subtle contribution of specific pathways involved in sleep in this structure have not been yet identified. Sitaraman et al. (2015) have revealed the functional connections between subsets of cells in a region of MB (i.e., Kenyon cells in MB calyx) and particular MB output neuron classes, identifying two synaptic microcircuits underlying the transmission of homeostatic sleep signals in MB. One circuit originates in γd Kenyon cells and converges onto cholinergic γ2α′1 MB output neurons, and it is supposed to promote sleep. The other circuit originates in α′/β′ and γm Kenyon cells and converges onto glutamatergic γ5β′2a/β′2mp/β′2mp_bilateral MB output neurons, and it is thought to promote wakefulness. These two microcircuits are relatively inactive under normal physiological conditions, but they are endogenously activated under conditions that alter sleep pressure (Sitaraman et al. 2015). For example, sleep deprivation increased the spontaneous activity of sleep-promoting γd Kenyon cells and concurrently decreased the activity of wake-promoting cells. However, if this sleep-promoting circuit is temporally thermogenically silenced following a sleep deprivation period, the expected homeostatic sleep rebound does not occur. Moreover, when this circuit is restored, no homeostatic sleep recovery is observed (Sitaraman et al. 2015). Thus we could speculate that this microcircuit would activate sleep after detection of unbalanced homeostasis for the recovery or the activation of specific neuron-dependent functions (e.g., need to consolidate information or to reduce synaptic weight) in a timely way, without accumulating the homeostatic sleep debt over time.

What is intriguing is that not only are these MB microcircuits involved in homeostatic sleep regulation, but they also contribute to various forms of associative memory formation. Indeed, as shown by Aso et al. (2014) in a series of experiments using optogenetic activations, the same wake-promoting circuit (i.e., γ5β′2a/β′2mp/β′2mp_bilateral MB output neurons) mediates innate avoidance, whereas the sleep-promoting circuit (i.e., γ2α′1 MB output neurons) modulates attraction behavior. Therefore, it is possible that these circuits mediate sleep homeostatic pressure as a function of increased need to consolidate memories and/or to reduce synaptic weight, perhaps prioritizing specific information (e.g., appetite memory). It is possible that MB circuits similar to the ones highlighted by Sitaraman et al. (2015) in Drosophila can be present in honeybees and that they can mediate the process underlying the sleep-dependent memory consolidation observed in the work by Zwaka et al. (2015). However, why bees' performance is enhanced only when odor cues are released during deep sleep remains unclear. Is it possible that bees while asleep experience several physiological states similar to the different sleep stages in humans? This would imply a phylogenetically complex sleep architecture where different sleep stages would perform different functions (e.g., memory reactivation, synaptic consolidation, synaptic downregulation). Finally, an interesting suggestion might come from the link between the flies' cholinergic sleep-promoting microcircuit and the high cholinergic tone during REM sleep detectable in humans. Perhaps the activity of this microcircuit in flies (but not only) could be modulated during night similarly to mammalians. Further studies integrating behavioral data, including memory and sleep, with genetic and neurophysiological tools suitable to dissect and record the neural networks might provide new insights about the sleep-memory relationship in invertebrates and shed some light on the evolutionary importance of memory consolidation during sleep across species.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

G.F. and N.C. conception and design of research; G.F. and N.C. drafted manuscript; G.F. and N.C. edited and revised manuscript; G.F. and N.C. approved final version of manuscript.