Figure 7. Quantifying value correlates in putative direct pathway SPNs of the dorsomedial striatum.

(A) Schematic of experimental scheme. Control (Nrxn1α^+/+; Nex^Cre/+, n = 7) mice were injected with a retro-AAV2-EF1α−3xFLAG-Cre virus in the substantia nigra, pars reticulata (SNr). Ipsilateral injection of Cre-dependent GCamp6f allowed for enrichment of putative direct pathway SPNs (p-dSPNs). (B, top) Sagittal section of Nex^Cre brain showing GCamp6f expression in dorsal striatal SPNs and placement of 400 µm optic fiber (white arrow). (B, bottom) Magnified view of striatum showing colocalization of nuclear FLAG-Cre and cytoplasmic GCamp6f. (B, bottom left) Location of fiber placements in Nex^Cre/+. (C, top) Trial schematic and relationship of specific task epochs with p-dSPN Ca²⁺ signal (bottom). (D) Peristimulus time histogram (PSTH) of ΔF/F for Nex^Cre/+ aligned to initiation event (all trials). The initiation of the action sequence (green bar) is associated with a rise in p-dSPNs activity. (E) Representative heat map of individual animal trials segregated by reward outcome on (t−1) trial (sorted by the latency to initiate). Trials following a large reward have greater signal suppression than those following small reward. (F) PSTH of ΔF/F for Nex^Cre/+ aligned to initiation event (segregated by outcome on (t−1)). Preinitiation of p-dSPN dynamics exhibits two components – a slow ramping phase (yellow, time_-10→-1) followed by a fast spike phase (green, time_-1→init), both of which are modulated by (t−1) reward outcome. (G) The slow ramping phase is quantified by the integral of GCamp signal −10 s to −1 s before initiation. (H) There is a significant effect of (t−1) reward volume on the preinitiation integral during slow ramping with large rewards showing greater silencing of p-dSPN activity (paired t-test, ***p=0.0002). (I) Preinitiation integral inversely correlates with the comparative action value of the upcoming trial, which is calculated using probability estimates from fitted reinforcement learning models and reflects the disparity in choice value on a trial to trial basis. (J) The dynamics of the fast peak phase are represented by the average slope of GCamp signal from −1 s till initiation. (K) There is a significant effect of (t−1) reward volume on preinitiation slope during the fast peak phase (paired t-test,***p=0.0006) with large rewards showing steeper subsequent preinitiation slopes. (L) Preinitiation slope positively correlates with the comparative action value of the upcoming trial.

Figure 7—figure supplement 1. Additional Photometry Analyses in Wildtype and Mutant Mice.

(A) Mean photometry signals for animals in engaged and disengaged epochs of the task revealed statistically smaller signals during periods of task engagement, without phenotypic differences. Engaged periods were defined as any time point within a window of 5 s before an initiation event (to start a trial) and after a nosepoke exit (to end a trial). (B) Grand average of the putative-dSPN fiber photometry signal in the Nex^Cre control mice from the 405 nm channel aligned to trial initiation, demonstrating no significant waveform deflections (compared to Figure 7D). (C and D) Averaged fiber photometry waveforms aligned by the choice timestamp and selected according to prior trial outcome (red-small prior reward and blue-large prior reward) for control (C) and Nex-deleted Nrxn1α (D). (E and F) Summary data of slopes from linear fits of the Ca²⁺ waveform (t = −0.5 s → 0.2 s) did not reveal statistically significant differences based on prior trial outcome in either control (E, paired t-test, p=0.16) or Nex-deleted Nrxn1α mice (F, paired t-test, p=0.23). A analyzed by two-way RM ANOVA; E and F analyzed by paired t-test. All data represented as mean ± SEM.