Memory of an event represented within a neuronal ensemble is called a memory trace or engram (Devan and McDonald, 2022, Josselyn et al., 2015, Tonegawa et al., 2015). Investigating how different memory traces are stored and interact is fundamental for understanding adaptive behaviors in humans and animals. Studies have demonstrated that different memory traces may involve non-overlapping neuronal populations, thereby minimizing interference between individual memory representations (Choucry et al., 2024, Rashid et al., 2016). On the other hand, it has been found that different memories can be encoded in overlapping neuronal populations (Cai et al., 2016, Crestani et al., 2019, Rashid et al., 2016, de Sousa et al., 2021, Josselyn and Frankland, 2018, Madar et al., 2019, Yokose et al., 2017). In this case, different memories can either interact or remain independent. For instance, it has been shown that after forming aversive memories for two distinct auditory stimuli, disrupting memory reconsolidation for one stimulus selectively impair the response to that stimulus while preserving the fear response elicited by the other stimulus (Abdou et al., 2018). These findings suggest that different synapses within the same neuron may participate in encoding distinct memory traces (Abdou et al., 2018, Bailey et al., 2015, Choucry et al., 2024, Doyère et al., 2007, Rashid et al., 2016).

Conversely, modifying one memory due to new experiences may affect another memory (Cai et al., 2016, Crossley et al., 2023, de Sousa et al., 2021, Gostolupce et al., 2022, Holland, 2005, Lee et al., 2017, Rashid et al., 2016, Yetton et al., 2019). When humans or animals encounter a new aversive event similar to a previously known one, the old aversive memory can serve as a basis for forming or enhancing the new aversive memory (Cai et al., 2012, Crossley et al., 2023, Rivi et al., 2023). For example, if an animal has been punished for consuming certain food and subsequently encounters a new food, it may avoid the new food by utilizing the old aversive memory to analyze the new experience (Lee et al., 2017). On the other hand, new aversive training can activate an old aversive memory and strengthen its association with certain signals or contexts (Morton et al., 2017).

A significant factor contributing to memory trace interaction is the similarity of memorized stimuli (related stimuli). For instance, second-order conditioning occurred more rapidly when two conditioned stimuli belonged to the same stimulus class rather than different classes (e.g., when both conditioned stimuli were food stimuli) (Gostolupce et al., 2022).

However, in certain cases, new aversive training can lead to extinction or forgetting of the old aversive memory, especially if the new conditions differ from those under which the old memory was initially formed (Clem and Schiller, 2016). Additionally, the formation of a new memory may be independent of an older related memory (Bisaz et al., 2014).

It should be noted that the molecular and cellular mechanisms underlying the interaction of memory traces remain poorly understood. Memory is thought to form through long-term changes in synaptic efficiency, a process known as synaptic plasticity (Bailey et al., 2015, Chen et al., 2014, Gold and Glanzman, 2021, Sossin, 2018). It has been proposed that two different memories may be connected through inter-synaptic interactions via a mechanism known as “synaptic tagging” (Frey and Morris, 1997, Kennedy, 2013, Martin et al., 1997). According to this hypothesis, synaptic efficiency can be altered by the formation of tags at synapses, which capture mRNA and proteins produced during strong activation of other synapses within a limited timeframe. This hypothesis has also been supported by behavioral experiments demonstrating that weak memory can be enhanced if a second “strong” memory is formed during its consolidation period. This phenomenon has been termed “behavioral tagging” (Moncada et al., 2015, Viola et al., 2014). This hypothesis proposes a model for interaction between two active tasks. However, its applicability is less clear when only one memory is active and affects an inactive memory.

According to another hypothesis, synaptic “tagging” and “capture” processes are critical during memory formation, whereas nuclear processes, including epigenetic mechanisms, play a leading role in maintaining long-term synaptic plasticity and memory over subsequent periods (Bédécarrats et al., 2018, Chen et al., 2014, Gold and Glanzman, 2021, Haidar et al., 2024, Jarome and Lubin, 2014, Oliveira, 2016). Besides hypotheses postulating that memory involves various interconnected neuronal mechanisms – from synapse to genome were suggested (Langille and Brown, 2018, Sossin, 2018). It should be noted that mechanisms underlying interactions of different memory traces at the genomic level remain largely unknown.

Overall, studies indicate that different memories can either interact, affecting their storage and expression, or function independently. However, several critical questions remain unresolved. For instance, what molecular and cellular mechanisms ensure the stable long-term storage of information? How do specific memories maintain their identity when interacting with other memory traces stored within the same neuronal ensembles? What molecular changes explain the specificity of information storage and retrieval? What molecular mechanisms enable the association of new information with pre-existing memory associations? Which neurological pathologies may result from disrupted interactions between memory traces?

So, to address these questions, we used a model of conditioned food aversion in snails, which allows us to uncover the molecular mechanisms underlying the interaction of multiple memory traces. We found that these animals can readily develop conditioned aversion responses to either one or two food stimuli (Kozyrev et al., 1992, Kozyrev and Nikitin, 2010, Nikitin et al., 2017). Disruption of memory reconsolidation by NMDA receptor antagonists,or protein synthesis inhibitors induced amnesia. At later developmental stages (≥10 days), this amnesia acquired characteristics of anterograde amnesia – lost skills were not restored by repeated training (Solntseva et al., 2007, Solntseva and Nikitin, 2010). It was further found that a critical mechanism of anterograde amnesia involves DNA methylation-dependent (Solntseva et al., 2023) and histone deacetylation dependent (Kozyrev et al., 2022) repression of genes necessary for long-term memory formation. Retraining amnesic animals during inhibition of DNA methyltransferases (DNMT) or histone deacetylases resulted in accelerated memory formation. The phenomenon of accelerated memory formation is known as the “Ebbinghaus savings effect” or “memory savings,” defined by fewer trials needed to achieve the original criterion level of conditioning compared to original training (Chen et al., 2014, Finnie and Nader, 2012, Pearce et al., 2017). This phenomenon suggests that during anterograde amnesia, memory is preserved in a latent state and can be converted into an active form through removal of epigenetic blockade and retraining (Solntseva et al., 2023).

Furthermore, we observed that administering protein synthesis inhibitors before retraining or RNA synthesis inhibitors after – but not before – retraining prevented memory formation during DNMT inhibition (Solntseva et al., 2023). The earlier dependence of memory formation on protein synthesis compared to RNA synthesis inhibitors suggested that memory-related proteins are translated from previously existing, translationally repressed mRNAs. Thus, we identified two crucial memory and amnesia mechanisms for the first time. First, DNA methylation-dependent repression of gene transcription essential for memory processes is a critical mechanism of anterograde amnesia. Second, translationally repressed mRNAs are the primary mechanism for maintaining latent memory during anterograde amnesia. DNA demethylation grants access to genes involved in memory processes, and transcription factors synthesized during retraining are derived from pre-existing mRNAs (Solntseva et al., 2023).

The model we used as well as our findings became the basis for studying the molecular and genetic mechanisms of memory storage, interactions among related memory types, and interactions between latent old memory and newly acquired memory in grape snails. Conditioned aversion responses to two distinct food stimuli were simultaneously formed in the same animals. Subsequently, the specificity of memory storage for these food stimuli was studied by selectively disrupting memory reconsolidation for only one of the two stimuli. We then induced anterograde amnesia for both conditioned stimuli and investigated the possibility of restoring memory for one stimulus through retraining under altered DNA methylation conditions. Finally, we explored whether aversive memory formation to a new food stimulus could interact with an older latent memory under altered DNA methylation conditions in amnesic animals.