CREB and the enhancement of long-term memory

Abstract

Studies performed on invertebrates and mammals indicate that temporally spaced behavioural training is more effective for eliciting long-term memory (LTM) than massed training. At the molecular level, LTM, but not short-term memory (STM), requires new protein synthesis for its formation. How is the formation of LTM tuned to specific temporal patterns of training?A study by Josselyn et al. 1xLong-term memory is facilitated by cAMP response element binding protein overexpression in the amygdala. Josselyn, S.A. et al. J. Neurosci. 2001; 21: 2404–2412PubMedSee all References now provides the first direct evidence that overexpression of CREB (cAMP response element-binding protein) in the mammalian brain improves the formation of LTM after massed training. CREB is a transcription factor that is known to regulate the synthesis of proteins important for the formation of LTM. In their study, Josselyn et al. hypothesized that overexpression of CREB in one region of the rat brain, the basolateral amygdala (BLA), should enhance LTM for fear conditioning following massed training trials, which normally do not elicit robust LTM. In fear conditioning, rats learn to associate a conditioned stimulus (e.g. a light) with a harmful stimulus, such as footshock. Successful retention of this association is evident from enhanced startle responses displayed by rats when the startle reflex is subsequently elicited in the presence of the conditioned stimulus. Expression of fear-potentiated startle, and acquisition of fear conditioning, are dependent upon the BLA. Josselyn et al. showed that temporally spaced training trials elicited robust LTM for fear-potentiated startle in control rats, whereas temporally massed training produced weaker LTM in the same animals. They then used herpes simplex virus (HSV) vector-mediated gene transfer to overexpress CREB (‘HSV-CREB’) in the BLA of rats before behavioural training. Josselyn et al. confirmed that CREB overexpression was specific to the lateral (but absent from the central) nucleus of the amygdala and that injection of HSV vectors per se had no harmful effects on their behavioural measurements. In rats that overexpressed CREB in the BLA during massed fear conditioning, LTM was significantly enhanced to levels seen following spaced training. Rats that received HSV-CREB injections in adjacent brain regions that are not crucial for fear conditioning did not show memory improvement. Injection of mutant CREB into the BLA also did not enhance LTM compared with saline-injected animals. More importantly, HSV-CREB injection into the BLA did not enhance STM and it did not improve LTM following spaced training using both weak and strong footshock intensities. The latter finding shows that enhancement of LTM is specific to massed training.These data support the notion that CREB exists as activator and repressor isoforms. Activator isoforms are hypothesized to inactivate more slowly than repressor isoforms. Spaced training might elicit more effective LTM because activator isoforms have time to accumulate during the rest intervals between training trials. Massed training elicits weaker LTM presumably because the lack of rest intervals does not permit activator isoforms to accumulate (i.e. repressor isoforms outweigh the activators). The finding by Josselyn et al. that CREB overexpression enhanced LTM after massed training (but not after spaced training using weak or strong shock intensities) supports this model. Spaced training might allow sufficient accumulation of activators (even at weak shock intensities) such that artificial overexpression of CREB might not produce further improvement of LTM. Some questions that are ripe for future exploration include: which downstream genes regulated by CREB are crucial for LTM? Can CREB over-expression in other specific brain regions enhance other distinct types of behavioural memory?