In a competition between job candidates, the winner is usually the one who brings a little extra to the table. The same may be true for neurons involved in memory. Although approximately 70% of the 20,000 neurons in the lateral amygdala are wired for associative memory, only roughly 25% show plasticity associated with learning following auditory fear conditioning. What sets these neurons apart? Han et al. report that neurons with elevated levels of the transcription factor CREB are most likely to be involved in associative memory in a recent article in Science.
In auditory fear conditioning, mice freeze with fear in response to a tone that they have learned to associate with a shock. Following auditory fear conditioning, the authors found activated (phosphorylated) CREB in approximately 25% of lateral amygdala neurons, consistent with the proportion of lateral amygdala neurons showing plasticity associated with memory. The authors hypothesized that CREB is important in determining which lateral amygdala neurons participate in the memory.
The activity-regulated cytoskeleton-associated protein (Arc) gene is a marker for neuronal activity. Arc expression localizes to neurons active 5 to 15 minutes before testing. The authors reasoned that if CREB expression determines which neurons participate in a memory trace, then neurons expressing the most CREB will be most likely to express Arc after fear conditioning. They microinjected CREB fused to green fluorescent protein (GFP) into the lateral amygdala and looked for cells coexpressing GFP and Arc after fear conditioning using fluorescent in situ hybridization. In wild-type and CREB-deficient mice, GFP-positive neurons expressed Arc more frequently than did GFP-negative neurons, suggesting that extra CREB conferred an advantage.
Neurons with reduced CREB activity tended to show less Arc expression. The authors injected wild-type mice with dominant-negative CREB fused to GFP. Following auditory fear conditioning, GFP-positive lateral amygdala neurons expressed Arc 12 times less frequently than did GFP-negative neurons.
The authors reasoned that if the selection of lateral amygdala neurons to participate in a memory trace is cell-autonomous, then the percent of lateral amygdala neurons activated by learning and memory would fluctuate in relation to CREB manipulation. However, Arc localized to the same proportion of lateral amygdala cells following auditory fear conditioning regardless of CREB expression. Roughly 20% of lateral amygdala cells expressed Arc following auditory fear conditioning in wild-type mice injected with vehicle, CREB or dominant-negative CREB and in CREB-deficient mice microinjected with CREB, consistent with the proportion of neurons showing plasticity associated with learning. According to the authors, these data indicate that the selection of neurons involved in a memory trace is competitive, with the relative amount of CREB determining participation.
How does CREB impart a competitive advantage? Wild-type mice injected with constitutively active CREB did not show increased Arc expression, suggesting that CREB does not induce the expression of Arc. The authors speculate that neurons expressing more CREB may be more excitable than their neighbors.