Tracing the Synaptic Plasticity of Associative Learning on a Behavioral Timescale

The core component of Hebbian plasticity is contiguity, that is the activation of pre- and postsynaptic neurons should temporally coincide. To form an association, Hebbian plasticity requires two events to occur no more than a couple of milliseconds apart. This is in direct contrast to most of our day-to-day learning where we effortlessly associate events that are many seconds to minutes apart. If the Hebbian model, at least as it is understood, is insufficient in explaining the cellular mechanism of learning, what are the alternatives? Currently, there are no consensual theoretical models, let alone experimental paradigms, that have been able to capture this problem of associative learning without leading to inconsistent conclusions and biologically implausible predictions.

For this purpose, we will use  an associative learning in which  two events are separated by many seconds. 

 

Associative learning

Synaptic Mechanisms Underlying Memory Decay

The field of memory formation and learning has been a centerpiece of research in neuroscience community for many years. Its counterpart, memory decay, however, has been largely left behind. This lack of progress is not the result of a lack of interest. In contrary, there is a tremendous interest in public as well as scientific community on the cause of forgetting (i.e. memory decay). The cause for this lack of progress boils down to the slow temporal scale of memory decay. This has made the phenomenon a challenge to scientific investigation. In his classic essay, Charles Steven set four criteria to test the cause of memory formation: detectability, blocking, reversibility, and mimicry. These criteria, particularly the first two are of a less challenge to test for memory formation as learning occurs within seconds to minutes. Electrophysiological studies, chemical antagonists and genetic manipulations have been instrumental to link LTP to memory formation. The process of memory decay, on the other hand, happens in a much slower time scales (from days to months in rodents). 

As such, conventional approach for monitoring synaptic strength (detectability criteria) or blocking different proteins and receptors (blocking criteria) for days and longer is not ideal. We propose to tackle this problem on two grounds: 1) Establishing a weak associative learning; 2) Using a circuit in which the memory trace can be monitored at the synaptic scale.

 

Memory decay

Mapping the Neural Circuit for an Innate Fear Behavior 

Fears are either innate or learned. Examples of innate fears include fear of snakes or height. The fear of a gun pointing at us is learned, as when a man of the 12th century would not have been frightened. In rodents a learned fear is induced by associating a neutral stimulus such as a tone with an aversive event, such as an electric shock. Innate fear is induced simply by exposing animals to a fake predator such as a toy snake or a looming object.

At the cellular level, the circuit for innate fear is hardwired in our brain. Learned fear, on the other hand, is formed with experiences by synaptic plasticity. At behavioral level, however, an animal’s reaction to both types of fear is virtually the same, expressed by a freezing response. Recent findings indicates that both types of fears engage overlapping regions within the brain. This raises a dilemma: how do two pieces of qualitatively different information, one signaling a neutral object, the other a predator, activate the same regions, and yet cause the animal to response appropriately. In this project we test the hypothesis that neutral stimuli (such as tone) and innately aversive stimuli (such a looming object) target different cell types within the regions for defensive behaviors.

Fear memory

During the formation of a memory or induction of an L-LTP, a group of proteins (plasticity-related proteins or PRPs) are synthesized. These proteins, it has been proposed, are essential for stabilization of memories and L-LTP. According to the current dogma, the PRPs share two features: 1) They are synthesized during memory formation/LTP induction; 2) They are transported to the synapses within 30-60 minutes after their synthesis. Consequently, to identify these proteins we need an approach with high temporal and spatial resolution. The aim of this project is to develop the tools that provide such resolutions. This project is funded and directed by the Center for Proteins in Memory (PROMEMO).

Plasticity-related proteins

From Synapses to Behavior 

It is shown that long-term potentiation (LTP) is the cellular basis of memory formation. However, since all but small fraction of memories are forgotten, LTP has been further divided into early LTP (e-LTP), the mechanism by which short-term memories are formed, and a more stable late LTP (L-LTP), by which long-term memories are formed. Remarkably, it has been shown that an e-LTP can be stabilized if it is preceded or followed by heterosynaptic L-LTP.

According to Synaptic Tagging and Capture (STC) hypothesis, e-LTP is stabilized by capturing proteins that are made by L-LTP induction (please see section II for our approach to characterize these proteins). The model proposes that this mechanism underlies the formation of late associative memory, where the stability of a memory is not only defined by the stimuli that induce the change but also by events happening before and after these stimuli. As such, the model explicitly predicts that a short-term memory can be stabilized by inducing heterosynaptic L-LTP.

In this project, we put this hypothesis into test. Specifically, we will test two explicit predictions of STC model: 1) A naturally formed short-term memory can be stabilized by induction of heterosynaptic L-LTP. 2) This stabilization is caused by the protein synthesis feature of L-LTP. 

Synaptic tagging and capture

Synaptic plasticity remains an almost indisputable candidate model for learning and memory formation. Although the majority of studies devoted to the mechanisms underlying synaptic plasticity rely on in vitro recordings, many behavioral phenomena are inconsistent with the findings obtained with such approach. Therefore, the goal of our research is to understand the rules that govern synaptic plasticity in vivo by performing recordings simultaneously to associative learning. Our projects emphasize types of associative learning and forgetting confined to synaptic and circuit levels of analysis that cannot be reconciled with the models advanced by in vitro recordings studies. The behavioral models we use consist of instances of associative learning and forgetting that can be monitored and manipulated at the cellular and synaptic scales. Importantly, we do not confine ourselves to a particular set of invasive techniques, with our projects involving not only electrophysiology and imaging, both in vivo and in vitro, but also optogenetics and  brain-circuit mapping. Finally, we are also interested in investigating the neural circuits underlying innate fear behavior with the aim of clarifying how the neuronal populations involved in this behaviour differ from those involved in learned fear.

 

nabavi lab

Funding agencies and affiliation: