My laboratory is interested in understanding the cellular and molecular mechanisms that underlies synaptic changes during associative and non-associative learning. One major goal is to elucidate the mechanisms responsible for associative changes in synaptic connections during classical conditioning. We are particularly interested in a molecular mechanisms of associative integration. We study this issue in a simple model system: conditioning of the defensive withdrawal reflex of the marine snail Aplysia.
One mechanism that contributes to associative increases in synaptic strength is is associative activation of a dually-regulated enzyme: the Ca2+/calmodulin-sensitive adenylyl cyclase. Working with Eric Kandel, I found that during conditioning, this Ca2+/calmodulin-sensitive adenylyl cyclase provides a site of associative stimulus convergence for two cellular representations of behavioral events: Ca2+ influx and modulatory neurotransmitter. In biochemical studies, we have asked whether the activation properties of this enzyme can account for some of the key features of conditioning. For example in most forms of conditioning, animals learn the predictive relationship between the conditioned stimulus and unconditioned stimulus only if, during training, the conditioned stimulus is presented shortly before the unconditioned stimulus; pairing in the backward direction is relatively ineffective. My laboratory earlier demonstrated that neural adenylyl cyclase displays a sequence preference that parallels that of the behavior: activation is more effective if Ca2+ (the signal from the conditioned stimulus) precedes modulatory neurotransmitter (the signal from the unconditioned stimulus). We recently found that this mechanism involves a temporal shift in the response to a transient Ca2+ stimulus, so that the enzyme is activated with a delay, with most of activation occurring after the stimulus has ended; this delay causes the Ca2+ response to coincide with the modulatory transmitter signal initiated by the unconditioned stimulus. We are now using recombinant mammalian adenylyl cyclase to investigate the kinetic basis for this delayed activation by Ca2+/CaM.
The resulting increase in cAMP levels, which results from convergent activation of adenylyl cyclase initiates short-term and long-term facilitation of synapses from the sensory neurons. We are also studying the associative induction of transcription factors that are involved in the initiation of long-term synaptic facilitation. In electrophysiological studies, we are investigating whether additional sites of associative convergence contribute to enhanced presynaptic facilitation when presynaptic activity is paired with modulatory transmitter. In the past, we have analyzed mechanisms of facilitation that involve the modulatory of voltage-dependent potassium currents. Currently we are investigating an independent mechanism of facilitation involving modulation of the exocytosis process itself.
When these sensory neuron synapses are activated with single action potentials at low frequencies, they rapidly undergo in profound synaptic depression. We have found through computer modeling studies and parallel cellular experiments that this synaptic depression involves the abrupt switching off or silencing of release sites. We recently identified a novel mechanism of synaptic plasticity in which the pattern of presynaptic firing determines whether release sites are switched on or off. Whereas single sensory neuron spikes result in rapid synaptic depression, activation of these synapses with brief bursts of spikes protects the synapses from depression. This burst-dependent protection against synaptic depression is mediated by protein kinase C, which is activated by the additional Ca2+ influx during the brief burst of spikes. Burst-dependent protection may function to maintain responsiveness when sensory afferents are repeatedly activated by moderate, behaviorally significant, stimuli that might otherwise produce habituation.