Emotional Enhancement of Memory Via Amygdala- Driven Facilitation of Rhinal Interactions
Total Page:16
File Type:pdf, Size:1020Kb
ARTICLES Emotional enhancement of memory via amygdala- driven facilitation of rhinal interactions Rony Paz, Joe Guillaume Pelletier, Elizabeth P Bauer & Denis Pare´ Emotions generally facilitate memory, an effect mediated by the basolateral amygdala (BLA). To study the underlying mechanisms, we recorded BLA, perirhinal and entorhinal neurons during an appetitive trace-conditioning task. We focused on the rhinal cortices because they constitute the interface between the hippocampus, a mediator of memory consolidation, and the neocortex, the storage site of declarative memories. We found that, after unexpected rewards, BLA activity increased impulse transmission from perirhinal to entorhinal neurons and that this effect decayed as the association between conditioned stimuli and rewards was learned. At this late phase of learning, the BLA effect occurred when the animals were anticipating the reward. By enhancing the processing of sensory cues, the BLA-mediated facilitation of rhinal interactions may explain how the amygdala promotes memory formation in emotional conditions. http://www.nature.com/natureneuroscience Humans generally form more vivid memories of emotionally charged recording neurons of the BLA, perirhinal areas 35 and 36, and events than of mundane experiences1. How do emotions facilitate entorhinal cortex in cats performing a trace-conditioning task memory? It is known that the facilitation of memory by emotions known to be dependent on the amygdala, rhinal cortices and requires an intact BLA in humans2–4 and animals5,6. Moreover, there is hippocampus19–23. Indeed, the rhinal cortices receive massive BLA evidence that the amygdala facilitates memory by enhancing attention inputs24,25 and form the interface26,27 between the hippocampus, a during encoding and by modulating consolidation and storage critical mediator of memory consolidation, and the neocortex, thought after learning5,7. to be long-term repository of declarative memories28–30.Wefoundthat In humans, attention is enhanced by emotional stimuli and this BLA activity enhances the processing of sensory cues during behavio- Nature Publishing Group Group Nature Publishing effect is absent in a subject with amygdala lesions8. Consistent with rally salient events by facilitating impulse transmission from the 6 these results, functional imaging studies indicate that the amygdala is perirhinal to the entorhinal cortex. Furthermore, this effect was tightly 200 activated by emotional stimuli9. In addition, there is a strong positive linked to learning. © correlation between amygdala activity at encoding and the long-term recall of emotional material10,11. RESULTS In animals, evidence shows that the amygdala is a critical site of To test the possibility that amygdala projections to the rhinal cortices plasticity when using classical fear-conditioning protocols12. However, facilitate memory by promoting impulse transfer between the neo- for many other types of emotional memories, including striatal- and cortex and hippocampus, we simultaneously recorded from BLA hippocampal-dependent ones, the amygdala does not seem to be a (n ¼ 390), perirhinal (n ¼ 282) and entorhinal (n ¼ 232) neurons storage site but rather modulates memory consolidation in its targets. in cats, by means of 24 microelectrodes (Fig. 1a,b). We analyzed BLA- Indeed, post-training injections of drugs that presumably enhance or related changes in rhinal activity in four conditions: under anesthesia, reduce BLA activity respectively facilitate or impair retention, even during waking periods where no rewards were administered, after when memory was tested long after the effects of these drugs has unexpected rewards and during a trace-conditioning task. The following dissipated13. In contrast, injections of lidocaine into the BLA days after analyses include only neurons whose location was confirmed by post-hoc training, just before testing retention, do not affect performance on histological reconstructions of microelectrode tracks (Fig. 1c–f). striatal- and hippocampal-dependent memory tasks14. Although emotional arousal recruits the BLA by means of stress Impact of BLA activity on spontaneous rhinal interactions hormones15–17, causing a long-lasting increase in BLA firing rates18,the To study how BLA activity affects impulse traffic in the rhinal cortices, impact of this increased activity on target structures has received little we first computed cross-correlations of spontaneous firing generated attention so far. Here, we investigated how the presentation of by perirhinal and entorhinal neurons located in the same coronal plane biologically significant and arousing stimuli affects neuronal inter- (Fig. 1g–i and Supplementary Fig. 1 online). Despite the strong actions between the BLA and rhinal cortices, by simultaneously reciprocal connections between the perirhinal and entorhinal cortices, Center for Molecular and Behavioral Neuroscience, Rutgers, The State University of New Jersey, 197 University Avenue, Newark, New Jersey 07102, USA. Correspondence should be addressed to D.P. ([email protected]) or R.P. ([email protected]). Received 16 June; accepted 22 August; published online 10 September 2006; doi:10.1038/nn1771 NATURE NEUROSCIENCE VOLUME 9 [ NUMBER 10 [ OCTOBER 2006 1321 ARTICLES Figure 1 Simultaneous recordings of amygdala, R Amygdala abOB g perirhinal, and entorhinal neurons. (a) Ventral view ML 2 of cat brain showing position of microelectrodes C 1 (dots). Cross indicates orientation. EC, entorhinal Entorhinal spikes per s spikes 0 –100 0 100 cortex; OB, olfactory bulb; rh, rhinal sulcus. Time from PRH spike (ms) Perirhinal (b) Spontaneous activity. (c–f) Histological BLA verification of recording sites. Coronal brain rh 800 sections. Arrows, electrolytic lesions performed EC 600 at the end of the experiments to mark the tip Entorhinal of electrodes that coursed through the lateral 400 amygdala (LA; c), basal amygdaloid nuclei Spike number Spike 200 (BL, BM; d), area 35 (e), and area 36 and the entorhinal cortex (f). AHA, amygdalohippocampal 0 area; CE, central nucleus; ec, external capsule; PRH c d h H, hippocampus; ME, medial nucleus; OT, optic 2 tract; V, ventricle. (g–i) Activity of an entorhinal 1 neuron around perirhinal (g), BLA (h) or perirhinal CE Entorhinal spikes per s spikes 0 and BLA (± 30ms, i) firing as illustrated in cross- ec ec OT –100 0 100 correlations (top) or raster plots (bottom). PRH, Time from BLA spike (ms) perirhinal. Scale bars: 3 mm in a,1sinb, 2mminc. LA LA BL ME 800 600 BM AHA 400 there was a spike from the perirhinal cell at Spike number Spike 200 time x and one from the entorhinal cell at time 0 y, one count is added to the matrix bin (x,y). Repeating this process for each BLA spike http://www.nature.com/natureneuroscience BLA e f gradually produces the raw STJH (Fig. 2b). i 2 V V Because we were interested in BLA-related 1 Entorhinal spikes per s spikes 0 rhinal correlation, we tested the raw STJHs H –100 0 100 Time from PRH spike (ms) against two null hypotheses (Methods): (i) that the observed correlation is similar to 500 that expected independently of BLA activity, 400 rh and (ii) that the correlation merely reflects 300 rh independent responses of rhinal neurons to 200 BLA activity. To test the first possibility, we Spike number Spike 100 compared the raw STJH to the average of 50 Nature Publishing Group Group Nature Publishing 0 6 STJHs computed after shuffling the BLA spike PRH (BLA) train (Fig. 2c). This is equivalent to comput- 200 ing the STJH around random times. To test © impulse propagation occurs with a low probability in the transverse the possibility that the correlations evidenced in the STJHs reflect axis of the rhinal cortices31. In keeping with this, we observed a low independent responses of rhinal neurons to BLA activity, we computed proportion of significant perirhinal-entorhinal cross-correlations STJHs by shuffling the BLA spike train of one of the two rhinal cells (Fig. 1g; 20% of 445 cell couples, P o 0.05, Methods) under anesthesia with respect to the other 50 times and averaging the result (Fig. 2d). and during quiet waking. Similarly, entorhinal cells were generally not This technique corresponds to the shift predictor33.Wethenperformed coactive with BLA cells (Fig. 1h; 26% of 245). However, when this bin-to-bin comparisons of significance between the raw and the two analysis was restricted to perirhinal spikes that occurred within 30 ms randomized sets of STJHs, using a Poisson distribution with a thresh- of a BLA spike, an interval corresponding to the average latency of old P-value corrected for multiple comparisons (0.05 divided by the synaptically evoked discharges in the BLA-rhinal network32,thepro- number of bins: 900; Fig. 2e). Finally, we verified that most significant portion of significant perirhinal-entorhinal cross-correlations nearly bins clustered around one dominant peak (Methods and Supplemen- doubled (Fig. 1i; 38% of 445; w2 test, P o 0.001). tary Fig. 2 online). However, cross-correlations have severe limitations because they Another important property disclosed by the STJH is the relative relate the activity of only two cells, forcing one to choose arbitrary timing of perirhinal and entorhinal firing in relation to BLA spikes. intervals when studying how a third neuron affects the relation between Bins above or below the main diagonal of the STJH represent correlated them. To overcome this difficulty, we studied triplets of perirhinal, activity where entorhinal firing respectively follows or precedes entorhinal and BLA cells (n ¼ 445) using spike-triggered joint histo- perirhinal spikes (Fig. 2f). We used this information to compute a grams (STJH), an adaptation of the joint peristimulus time histogram directionality index (Methods). We will return to the direction- (JPSTH) method33, in which BLA spikes were used as a temporal ality index in the section describing the impact of rewards on reference34 to study correlated perirhinal and entorhinal firing. BLA-rhinal interactions. STJHs (Fig. 2a) are computed by taking segments (± 150 ms) of We computed cross-correlations of perirhinal-entorhinal cell cou- rhinal activity around BLA spikes (time 0 in x and y). The spikes (red ples (Fig.