Figure 3: Basolateral amygdala gates the activity of pMSNs in the aDLS via glutamatergic mechanisms in nucleus accumbens core.

(a) Schematic representation of the preparation used for the extracellular electrophysiological recordings (adapted from ref. 5). Stimulating electrodes were placed in the motor cortex (M1) and the basolateral amygdala (BLA) while a recording electrode was inserted in the aDLS where action potentials of putative medium spiny neurons (pMSNs) were recorded. The influence of BLA stimulation before M1 stimulation over the spike probability of aDLS pMSNs was measured at various interstimulation intervals (IsI). A cluster analysis (b) performed on the spike probability following the 100–300 ms IsI of ∼50 aDLS pMSNs was recorded, in which all displayed similar M1-stimulation-evoked action potentials, with similar spike latencies and duration (c and see Supplementary Fig. 7), revealed that for IsI ranging from 100 to 300 ms one subset of neurons showed a facilitation of firing (upregulated neurons, n=19), a second subset (downregulated neurons, n=16) showed a robust inhibition while a last subpopulation showed no response to BLA stimulation (non-regulated, n=13) (main effect of neuron class: F_2,45=50.58, P<0.001, partial η²=0.69 and IsI × neuronal class interaction: F_18,405=12.43, P<0.001, partial η²=0.35) (d). BLA stimulation alone triggered no AP in aDLS pMSNs (d) nor did BLA stimulation 1, 5 or 50 ms before M1 stimulation (main effect of stimulation F_3,111=1.58, P>0.19, partial η²=0.04 and stimulation × neuron class interaction: F_6,111=2.06, P>0.05, partial η²=0.1) However, when the BLA was stimulated 100, 200 or 300 ms before M1 stimulation, it induced a marked (that is, beyond the s.d. of the distribution of spike probability of pMSNs measured at the 1 ms IsI, displayed in grey) increase, or decrease, in the spike probability of upregulated or downregulated neurons, respectively (all Ps<0.05 versus M1 stimulation, Dunnet post hoc test) (d). This differential effect was not observed for longer latencies (500 or 1,000 ms) (Ps>0.05) and did not result in recurrent network effects as post-challenge M1 stimulation alone resulted in aDLS pMSNs spike probabilities that were similar to baseline, pre BLA stimulation, levels (all Ps>0.05) (d). The remote control exerted by the BLA over aDLS pMSNs activity was dependent on antecedent glutamatergic transmission because it was abolished by a concomitant infusion of glutamate receptor antagonists into the AcbC (e) (main effect of neuron class × treatment interaction: F_4,18=9.38, P<0.001, partial η²=0.67) (f). Thus while the BLA-modulated firing probability of up-, down-, and non-regulated aDLS pMSNs (72%, 25.7% and 50%, respectively, averaged across three IsI measurements/neuron) differed from each other before AcbC glutamate receptor blockade, they all fell to a spike probability of an ∼50% after the intra-AcbC infusion of a AP5–CNQX mixture (averaged across three IsI measurements per neuron), thereby not differing anymore from M1 stimulation-alone condition (all Ps<0.05, Newman–Keuls post hoc test) (f). *P<0.05 versus baseline (M1 stimulation-alone condition). £: different from up- or downregulated neurons. AcbC, nucleus accumbens core; AcbS, nucleus accumbens shell; BLA, basolateral amygdala; CeN, central nucleus of the amygdala; post-BLA, M1 stimulation following BLA stimulation alone; post-LS, post Latin-square, M1 stimulation following the BLA–M1 co-stimulation protocol performed according to a Latin-square design.