A Model of Late Long-Term Potentiation Simulates Aspects of Memory Maintenance

Late long-term potentiation (L-LTP) appears essential for the formation of long-term memory, with memories at least partly encoded by patterns of strengthened synapses. How memories are preserved for months or years, despite molecular turnover, is not well understood. Ongoing recurrent neuronal activity, during memory recall or during sleep, has been hypothesized to preferentially potentiate strong synapses, preserving memories. This hypothesis has not been evaluated in the context of a mathematical model representing biochemical pathways important for L-LTP. I incorporated ongoing activity into two such models: a reduced model that represents some of the essential biochemical processes, and a more detailed published model. The reduced model represents synaptic tagging and gene induction intuitively, and the detailed model adds activation of essential kinases by Ca. Ongoing activity was modeled as continual brief elevations of [Ca]. In each model, two stable states of synaptic weight resulted. Positive feedback between synaptic weight and the amplitude of ongoing Ca transients underlies this bistability. A tetanic or theta-burst stimulus switches a model synapse from a low weight to a high weight stabilized by ongoing activity. Bistability was robust to parameter variations. Simulations illustrated that prolonged decreased activity reset synapses to low weights, suggesting a plausible forgetting mechanism. However, episodic activity with shorter inactive intervals maintained strong synapses. Both models support experimental predictions. Tests of these predictions are expected to further understanding of how neuronal activity is coupled to maintenance of synaptic strength.

💡 Research Summary

The paper tackles the long‑standing question of how memories, encoded as patterns of strengthened synapses, can persist for months or years despite continual molecular turnover. Building on the hypothesis that ongoing recurrent neuronal activity preferentially reinforces strong synapses, the author incorporates such activity into two computational frameworks of late long‑term potentiation (L‑LTP). The first is a reduced model that abstracts the essential steps of synaptic tagging, gene induction, and brief calcium (Ca²⁺) transients. The second is a more detailed, previously published model that adds explicit activation of key kinases (CaMKII, PKA, MAPK) by Ca²⁺. In both models, ongoing activity is represented as a series of brief Ca²⁺ spikes whose amplitude is positively linked to the current synaptic weight. This creates a positive feedback loop: a higher weight yields larger Ca²⁺ transients, which in turn further increase the weight. As a result, each model exhibits bistability—two stable steady‑states corresponding to a low‑weight (weak) and a high‑weight (strong) synapse.

A tetanic or theta‑burst stimulation can push a synapse from the low‑weight basin into the high‑weight basin. Once there, the continued low‑level Ca²⁺ activity maintains the strong state. Parameter sweeps demonstrate that this bistability is robust: variations in Ca²⁺ pulse frequency, duration, kinase activation rates, and feedback strength do not abolish the two‑state behavior unless the feedback is severely weakened. Importantly, the simulations reveal a plausible forgetting mechanism: prolonged reduction of ongoing activity (e.g., extended periods of neuronal silencing or sleep deprivation) causes the high‑weight state to decay back to the low‑weight state. Conversely, intermittent activity with brief inactive intervals is sufficient to preserve the strong synapse, suggesting that occasional re‑activation during recall or sleep can sustain memory traces.

The models generate experimentally testable predictions. Pharmacological inhibition of CaMKII or MAPK during the maintenance phase should accelerate the decay of the high‑weight state, leading to rapid forgetting. Likewise, manipulations that enhance spontaneous Ca²⁺ transients (e.g., promoting spindle activity during sleep) are predicted to stabilize strong synapses. The paper argues that such activity‑dependent reinforcement provides a unifying principle for memory maintenance across brain regions and time scales.

In summary, by embedding ongoing Ca²⁺‑driven activity into both simplified and detailed L‑LTP models, the study demonstrates that a self‑reinforcing feedback loop can produce durable bistable synaptic states. This framework accounts for both the persistence of long‑term memories and their eventual loss when activity falls below a critical threshold, offering a concrete mechanistic bridge between synaptic biochemistry and behavioral phenomena of memory retention and forgetting.