Figure 7.

dCA1-Connected NAc PV⁺FSIs Are Required for the Retrieval of CPP Memory and Associated MSN Assemblies

(A and B) Intersectional strategy to restrict the expression of ArchT-GFP to Cre-expressing cells (A). The translation of fDIO-ArchT-GFP construct was conditional to the sequential activity of Cre and FlpO recombinases, and validated in cultured HEK293T cells co-transfected with Cre and the cDIO-FlpO construct (B, top) or not (B, bottom).

(C and D) This strategy was used in vivo for the trans-synaptic anterograde targeting of dCA1-connected NAc PV⁺ cells (C). PV⁺FSI^dCA1→NAc indicates a NAc PV⁺ interneuron that receives direct dCA1 inputs and that expresses ArchT (see red and white cell in confocal image of D), whereas PV⁺FSI^NAc indicates a neighboring interneuron that does not (see red-only cell in confocal image of D). PV^dCA1→NAc::ArchT-GFP mice were implanted with tetrodes for dual-site monitoring of dCA1 and NAc ensembles, and optic fibers for light ON^NAc delivery. This strategy was also used for the trans-synaptic targeting of dSub-connected NAc PV⁺ cells (see Figures S7D–S7G).

(E) Mice performed each day the 1-day CPP protocol and explored another enclosure (other), as before (Figures 2 and 3). Yellow light (20 × 30-s pulses over 15 min) was delivered during CPP test ON. Reward seeking was also measured during CPP tests (Figure S7B) and hedonic motivation further tested in another enclosure containing drops of sucrose in light ON^NAc (n = 11 mice tested; 10/10 collected drops of sucrose for all mice). Novel place preference was also tested in these mice (Figure S7C).

(F) Examples of two NAc FSIs recorded in a behaving PV^dCA1→NAc::ArchT-GFP mouse. Shown are average spike waveforms (left) and raster plots of spike discharge in relation to light ON^NAc (right). Note that the firing of the PV⁺FSI^dCA1→NAc cell (top) was suppressed during light delivery whereas the bottom FSI^NAc cell was not.

(G and H) CPP score (mean ± SEM; n = 64 days from 11 mice) during pre-test and test (G: n = 30 light OFF; H: 34 ON^NAc days).

(I) Examples of weight vectors showing NAc assembly patterns formed by short (25 ms) timescale co-activation of MSNs in the CPP apparatus. Each assembly is depicted in the form of a lollipop plot with MSNs ordered and color-coded to highlight those with coincidental spiking.

(J) Activation strength of NAc assembly patterns detected from MSN co-activation in the CPP apparatus prior to test. Each pair of connected data points shows the average strength of an assembly pattern tracked during the Other enclosure and the CPP test (mean ± SEM; n = 16 light OFF, 10 ON^NAc patterns; 2.36 ± 0.43 patterns per mouse recording day; 13.91 ± 3.05 MSNs per mouse recording day).

(K–M) Prediction of MSN spike trains from dCA1 PYR spike trains using generalized linear models (GLMs; K). In light OFF, GLMs trained on spike trains recorded prior to the test predicted MSNs during CPP test significantly above their prediction accuracy when applied in the Other enclosure (L; n = 58 MSNs from 30.25 ± 3.98 PYRs per GLM). Photo-silencing of PV⁺FSIs^dCA1→NAc dropped PYR-to-MSN prediction to the levels seen in the CPP-unrelated enclosure (M; n = 48 MSNs from 26.85 ± 4.12 PYRs per GLM). ^∗∗∗p < 0.001, paired t tests.