Understanding Memory at the Molecular Level

Probably the most extensively researched facet of cognition, memory has proven to be a process that is as difficult to unravel as it is essential to the human experience. Monumental developments in memory research are occurring regularly, however, although they are often under the public radar as they are only pieces of a puzzle we are still incapable of fully assembling. Regardless, the work being done on these pieces will one day allow for an understanding of memory so extensive it will seem to have little in common with our traditional conceptions of what memory is.

An important share of that work is being done at The Scripps Research Institute. A group of researchers there have been focusing their studies on the molecular mechanisms of memory. Last year, they developed a transgenic mouse with genes that cause neurons activated within a particular timeframe to be tagged with a fluorescent marker. Using this mouse, the group demonstrated that the same neurons used during the learning of a fear response are activated during retrieval of the memory. The number of neurons activated also correlated with the intensity of the response, indicating a steady relationship between experience and neural representation. This work helped to elucidate the structure of neural networks involved in memory consolidation and retrieval.

The group then turned their attention to the mechanism involved in long-term memory formation. Receptors for glutamate, the primary excitatory neurotransmitter in the central nervous system, are essential for long-term potentiation (LTP), which is the enhancement of synaptic communication thought to underlie long-term memory formation. It has been suggested that LTP may be the result of the integration of AMPA glutamate receptors (AMPARs) into the synapse that is strengthened. The additional receptors would make the neuron more sensitive to glutamate, leading to LTP.

It has been shown that protein synthesis in neuronal cell bodies is necessary for consolidating memories as well. Thus, it was hypothesized that proteins made in the soma (cell body) are involved in the insertion of AMPARs into synapses associated with memory, resulting in quicker transmission at these synapses and the capacity for memory recall. What has been unclear, however, is exactly how the proteins synthesized in the soma cause plasticity to occur only at the specific synapses associated with a memory.

A popular explanation for this process involves something called synaptic tagging. In this scenario, neuronal stimulation causes the creation of a synaptic tag, a kind of signpost on the neuron that is used to attract proteins necessary for plasticity (such as proteins involved in forming AMPARs). This tagging would occur only at synapses involved in processing the LTP-inducing stimuli, and thus could account for the localization of memory to specific groups of neurons.

The researchers at The Scripps Institute, Naoki Matsuo, Leon Reijmers, and Mark Mayford, again used transgenic mice, this time to investigate the concept of synaptic tagging. They engineered mice to express a subunit of the AMPAR, referred to as GluR1, fused to a green fluorescent protein. They suppressed expression of this gene through the use of doxycycline until the experiment. They then exposed some of the mice to a fear-conditioning program, where they learned to associate foot shock with a particular environment. GluR1 is necessary for the formation of AMPARs. Thus, new AMPAR formation was measurable after the fear conditioning by examining dendritic spines of hippocampal brain slices for fluorescence.

Dendritic spines are regions that protrude from dendrites, where synapses are located and input from other neurons received. There are three morphological types: thin, stubby, and mushroom-shaped.

As they expected, the researchers found an increased proportion of fluorescent GluR1 subunits on the dendrites of hippocampal neurons in those mice that underwent the fear conditioning. Specifically, the fluorescent GluR1 was found on the mushroom-shaped spines, and not in significant amounts on the other spines. It seems a mechanism that resembles synaptic tagging plays a role in directing GluR1-containing AMPARs to mushroom spines.

This study is important for a number of reasons. Understanding the morphological changes that result in LTP is necessary in developing a working physiological model of memory. This study provides insight into the mechanism of these changes and makes our understanding of the memory process a little more concrete. In addition, synaptic modifications that lead to behavioral changes are thought to underlie a number of human tendencies. The successful use of transgenes in this memory study could thus provide a basis for their use in studying these other behaviors affected by neuronal enhancement, e.g. addiction.


Matsuo, N., Reijmers, L., Mayford, M. (2008). Spine-Type-Specific Recruitment of Newly Synthesized AMPA Receptors with Learning. Science, 319, 1104-1107. DOI:10.1126/science.1149967