TL;DR
This paper presents a computational model of a hippocampal pyramidal neuron that explains the fundamental properties of long-term potentiation (LTP) – cooperativity, input specificity, and associativity – through a unified mechanism. The model highlights the crucial role of a local positive feedback loop involving BDNF protein synthesis at individual synapses, demonstrating how this mechanism can lead to bistable synaptic states essential for learning and memory.
What Problem Does This Paper Address?
Long-term potentiation (LTP) is a cornerstone of learning and memory, representing a persistent strengthening of synaptic connections. While LTP is known to exhibit key properties like cooperativity (requiring multiple inputs), input specificity (affecting only activated synapses), and associativity (linking weak and strong inputs), the underlying molecular mechanisms that give rise to these properties have remained unclear. Existing computational models often focus on specific aspects of LTP or rely on simplified representations of synaptic processes. This paper aims to bridge this gap by developing a more comprehensive computational model that integrates molecular processes, including gene expression and local protein synthesis, to provide a unified explanation for these fundamental LTP properties.
Methods
The authors developed a computational model of a hippocampal pyramidal neuron with multiple excitatory synapses. The model incorporates detailed molecular processes involved in synaptic plasticity, including:
- Synaptic Transmission: Modeling glutamate release from presynaptic terminals and its interaction with postsynaptic AMPA and NMDA receptors.
- Calcium Signaling: Simulating the influx of calcium through NMDA receptors and its downstream effects.
- BDNF Signaling Pathway: Crucially, the model includes a local positive feedback loop where BDNF in the synaptic cleft triggers the synthesis of more BDNF within the dendritic spine via mTOR activation. This synthesized BDNF is then secreted back into the cleft, creating a self-reinforcing cycle.
- Gene Expression: Modeling the transcription of BDNF mRNA, influenced by CREB activation.
- Synaptic Plasticity Dynamics: The model captures changes in AMPA receptor conductance, which directly influence excitatory postsynaptic potentials (EPSPs) and currents (EPSCs), serving as a measure of synaptic strength.
To analyze the model's behavior and understand the mechanisms, the authors employed:
- Numerical Simulations: Using MATLAB to solve the differential equations governing the model's dynamics under various stimulus protocols.
- Bifurcation Analysis: Employing XPPAUT to analyze the stability of the BDNF feedback loop and identify parameter regions leading to bistability, which is hypothesized to be the basis of LTP induction.
Datasets
Not discussed in the paper. The study relies on a computational model and simulated data, not experimental datasets.
Key Insights and Ideas
- Unified Mechanism for LTP Properties: The core insight is that a local positive feedback loop of BDNF protein synthesis at each synapse can simultaneously explain cooperativity, input specificity, and associativity. This loop creates bistability, allowing synapses to switch from a low-strength state to a high-strength state and remain there.
- Bistability as the Basis of LTP: Bifurcation analysis of the BDNF feedback loop reveals that it exhibits bistability, meaning it can exist in two stable states (low and high BDNF levels). This bistability is crucial for LTP induction, as a transient strong stimulus can push the synapse into the high-activity state, which is then maintained.
- Role of Local Protein Synthesis: The model emphasizes that local protein synthesis, specifically of BDNF, is not just a modulator but a fundamental component for achieving long-lasting plasticity.
- Interplay of Pre- and Postsynaptic Activity: The study confirms that both presynaptic neurotransmitter release and postsynaptic depolarization are necessary for LTP induction, primarily through their roles in activating NMDA receptors and initiating downstream signaling.
- Soma-Spine Coupling for Associativity: The authors found that the electrical coupling between the soma and dendritic spines plays a critical role in associativity, allowing the depolarization from strong inputs to influence weakly stimulated synapses and induce LTP there.
Key Results
- Cooperativity: A single weak stimulus to one synapse did not induce LTP. However, simultaneous stimulation of multiple synapses (20 out of 40 in the P1 pathway) led to a significant and persistent increase in EPSP and EPSC amplitudes, demonstrating cooperativity.
- Input Specificity: When strong stimulation was applied to the P1 pathway, only synapses in P1 showed potentiation, while synapses in the unstimulated P2 pathway remained unaffected, illustrating input specificity.
- Associativity: A weak stimulus to the P2 pathway alone did not induce LTP. However, when paired with a strong stimulus to the P1 pathway, both P1 and P2 synapses showed significant potentiation, demonstrating associativity.
- BDNF Dynamics: Simulations showed that strong stimulation led to a sustained increase in postsynaptic calcium and BDNF concentration, which then drove an increase in AMPA channel conductance and, consequently, EPSC/EPSP amplitudes.
- Threshold Effects: The model revealed threshold phenomena for both the number of active synapses required for cooperativity and the coupling conductance between soma and spines for associativity.
Limitations and Future Work
- Simplification of Molecular Processes: While the model incorporates key molecular players, it acknowledges that many complex biochemical reactions within the postsynaptic density are simplified. Future work could involve adding more detailed representations of these processes.
- Return to Low State: The model shows that once LTP is induced, the synaptic efficacy might remain in the high state for a long time, which might not always align with experimental observations of LTP decay. The authors suggest that incorporating processes like neuronal degeneration, random inactivation of gene expression, or epigenetic changes could help model the return to a low state.
- Short-Term Plasticity (STP): The current model focuses on long-term potentiation. Extending the model to include mechanisms for short-term plasticity would allow for a more complete representation of synaptic dynamics.
- Neural Circuits: The authors suggest that the developed model, which captures synaptic plasticity at a single synapse level, could serve as a module for building more complex computational models of neural circuits involved in learning and memory.
- Fear Memory: The study's findings on associative LTP are relevant to fear memory consolidation, suggesting the model could be applied to investigate the mechanisms of fear memory.
Summary updated