Activity of a direct VTA to ventral pallidum GABA pathway encodes unconditioned reward value and sustains motivation for reward

Abstract

Dopamine signaling from the ventral tegmental area (VTA) plays critical roles in reward-related behaviors, but less is known about the functions of neighboring VTA GABAergic neurons. We show here that a primary target of VTA GABA projection neurons is the ventral pallidum (VP). Activity of VTA-to-VP-projecting GABA neurons correlates consistently with size and palatability of the reward and does not change following cue learning, providing a direct measure of reward value. Chemogenetic stimulation of this GABA projection increased activity of a subset of VP neurons that were active while mice were seeking reward. Optogenetic stimulation of this pathway improved performance in a cue-reward task and maintained motivation to work for reward over days. This VTA GABA projection provides information about reward value directly to the VP, likely distinct from the prediction error signal carried by VTA dopamine neurons.

Fig. 1.. VTA GABAergic neurons send long axons to the VP and form synaptic contacts with VP resident neurons.

(A) Injection of AAV2-EF1a-DIO-hChR2-eYFP into the VTA of GAD65-Cre mice. (B) Sagittal view of a mouse brain expressing eYFP in VTA GAD65⁺ neurons. The green fibers show the VTA GABA projection pathways. (C) Brain slices were cut from the mice treated as in (A), and a VP cell was then filled with biocytin and stained with Alexa Fluor 594. Images show that eYFP-expressing axons from VTA GAD65⁺ neurons wrap the cell body and a proximal dendrite of the VP neuron. (D) Diagram for infection of H129ΔTK-TT into the VTA of GAD67-GFP/GAD65-Cre mice. The HSV construct is activated by Cre recombinase and transports viral replicates to connected postsynaptic neurons in the VP. (E) Confocal images taken from the VP, depicting postsynaptic targets of VTA GAD65⁺ neurons infected by H129ΔTK-TT. tdT, GFP, and ChAT immunoreactivities denote HSV-infected, GAD67⁺, and cholinergic neurons, respectively. Merged image shows colocalization of tdT with the molecular tags for identifying postsynaptic GABAergic and cholinergic neurons. (F) Images depicting colocalization of tdT with CaMK2α labeling in the VP. (G) Percentage of different subtypes of VP resident neurons postsynaptic to VTA GABAergic inputs.

Fig. 2.. Optical stimulation of VTA-to-VP GABAergic axon terminals evokes postsynaptic currents (PSCs) in VP neurons.

(A) Example traces from whole-cell patch-clamp recordings in a VP neuron. The brain slice was from a mouse treated as in Fig. 1A. Upper blue line indicates blue light pulses. (B) Same as (A) except that PTX was applied at 500 μM, and 10 μM CNQX was added to fully suppress light-evoked currents. (C) Characterization of light-evoked PSCs, current amplitude versus probability plot. The red dot denotes the average current size and probability (mean ± SE, apply to all figures). Probability = number of detectable evoked synaptic responses/number of stimulus light pulses. (D) Upper panel shows the measure of synaptic delay (SD). The gray trace was recorded in a postsynaptic VP neuron, and the black trace was recorded in a ChR2-expressing VTA neuron. The delay in the black trace (D0) was considered to be a systematic delay. In the lower panel, the black bar corrects the systematic delay recorded in presynaptic ChR2-expressing cells (8.5 ms) to 0; gray bars show corrected synaptic delay in postsynaptic VP neurons. (E) Characterization of action potential half-width versus cell capacitance plot in postsynaptic VP neurons. Red circles denote cells showing a continuous firing pattern (fig. S4B1), and blue circles denote cells exhibiting marked spike accommodation (fig. S4B2). The red and blue dots denote the average of the two groups. (F) Composition of light-evoked PSCs. In group 1, the PSCs were blocked by PTX. Addition of CNQX resulted in no further significant changes. In group 2, the PSCs were blocked by a combination of PTX and CNQX. Student’s t test, *P < 0.05; **P < 0.01.

Fig. 3.. Activity of VTA-to-VP GABA neurons during rewarded behaviors.

(A) Timeline of behavioral procedures and fiber photometry sessions (blue boxes). (B) Fluorescence changes (dF/F) during FR1 responding. Signals were aligned to US retrieval [black dashed lines in (B) to (I) indicate timing of the events to which the signals were aligned]. Black trace represents the averaged result; gray traces in (B), (D), (F), and (G) represent averaged signals in each individual mouse. N = 8 mice. (C) Fluorescence signals when mice visited the magazine when no reward was available. Gray traces in (C), (E), (H), and (I) represent signals to each reward event. (D) Same as (B) except signals were aligned to active nosepokes. (E) Fluorescence signals when mice made inactive nosepokes. (F) Fluorescence changes during CRT training. Signals were aligned to US retrieval. (G) Same as (F) except signals were aligned to the onset of tones. The dashed lines marked “NP” and “US retrieval” show the timing of nosepoke and magazine entry following the corresponding tones. Red arrows indicate the signal following the three behavioral events: tone, nosepoke, and US retrieval. (H) Signals when mice received tones in the beginning of the first CRT session before any reward was presented. (I) Signals during conditioned reinforcement (CR) testing when mice received tones alone.

Fig. 4.. Activity of the VTA-to-VP GABA pathway scales with the value of a primary reward.

(A) Heatmap of calcium-dependent fluorescence signals in VTA-to-VP GABA neurons across five CRT sessions on consecutive days. Signals were aligned to the time of magazine entry (0 s). (B) Area under the curve (AUC) of fluorescence signals measured from (A). Upper panel shows the fluorescence change across five sessions. Lower panel shows the AUCs calculated from the upper panel curves from 0 to 5 s (gray area). Each dot represents the averaged AUC of the fluorescence signals measured in one mouse in one session. One-way analysis of variance (ANOVA) of the signal AUC across five sessions: F_(4,30) = 0.034; P = 0.998. N = 7 mice. (C) Heatmap of fluorescence signals in terminals of VTA-to-VP GABA neurons during FR1 responding to different sizes of reward (reward delivery for 0, 1, 2, 3, 4, and 5 s). (D) AUC of fluorescence signals measured from (C). Upper panel shows the fluorescence change in response to different sizes of reward. Lower panel shows the AUCs calculated from the upper panel curves. One-way ANOVA of the signal AUC across reward sizes: F_(5,30) = 18.6; P = 2.2 × 10⁻⁸. N = 6 mice. (E) Heatmap of fluorescence signals in VTA-to-VP GABA neurons during FR1 responding to different dilutions of reward with the same reward delivery for 2 s. (F) AUC of fluorescence signals measured from (E). Upper panel shows the fluorescence change in response to different dilutions of reward. Lower panel shows the AUCs calculated from the upper panel curves. One-way ANOVA of the signal AUC across different dilutions: F_(4,20) = 14.7; P = 9.4 × 10⁻⁶. N = 5 mice.

Fig. 5.. Stimulation of VTA-to-VP GABA pathway modulates VP activity during reward seeking.

(A) Intersectional strategy targeting Gq-DREADD to VTA-to-VP GABA projections and recording in VP. Gray line, placement of lens. Syn1-GCaMP6s (blue) was expressed in VP for recording from all neuronal subtypes. Retrograde AAV carrying Cre-dependent Flp (rHSV-DIO-flp) was injected into VP of the same GAD65-Cre mice to express Flp recombinase in Cre-expressing neurons with terminals in VP. Last, Flp-dependent Gq-DREADD (AAV-Frt-DREADD) was injected into the VTA, allowing expression of the DREADD only in GAD⁺ VTA neurons with terminals in VP that had taken up the Flp construct. (B) VTA tyrosine hydroxylase (TH) and mCherry staining in a GAD65-Cre mouse with AAV8-Ef1a-fDIO-DREADD Gq-mCherry and VP:rHSV-hEF1α-LS1L-flpo. The DREADD is not expressed in TH⁺ cell bodies. (C) Frame taken through the microscope in VP showing neurons expressing GCaMP6s. (D) Whole-cell patch-clamp recording of Gq-DREADD-mCherry+ VTA neuron. Red arrowhead indicates the time when CNO (10 μM) was added. (E) Scheme for head-fixed sucrose seeking. Rewards earned on an FR10-FI5 schedule (right). (F) Fluorescence changes (dF/F) aligned to onset of each lick bout. Black lines indicate averages of the transients; gray shading indicates 20 to 80% of individual traces. N = 3 mice. (G) Fluorescence changes aligned to time of reward delivery. (H and I) Distribution of event-associated fluorescence changes. (H) Transients aligned to onset of “seeking” lick bout (vehicle versus CNO, P = 0.0008, χ² test). (I) Transients aligned to liquid delivery (vehicle versus CNO, P = 0.0524, χ² test). All cells had been examined using the Wilcoxon signed-rank test, pre-event dF/F versus post-event dF/F. Cells with no significant changes (P ≥ 0.05, gray bars) and cells with significant changes (P < 0.05, red bars) were grouped. (J and K) Scatterplot of individual neuron responses at time of reward seeking versus consumption after (J) vehicle or (K) CNO (Wilcoxon signed-rank test, pre-event versus post-event).

Fig. 6.. Stimulation of VTA-to-VP GABA pathway improves performance in CRT.

(A) Timeline of behavioral procedures and optical stimulation (blue boxes). (B) Settings of behavioral boxes used for CRT. (C to G) During CRT sessions, task duration [(C); F_(1,24) = 7.65, P = 0.011], nosepoke frequency [(D); F_(1,24) = 0.073, P = 0.79], magazine entry frequency [(E); F_(1,24) = 6.20, P = 0.020], effort conversion rates [number of reinforcers/total NPs; (F); F_(1,24) = 17.79, P = 0.0003], and reward acquisition rate [number of rewards/number of cues; (G); F _(1,24) = 9.95, P = 0.0043] were compared between control (n = 13) and “stimulation” (n = 13) groups. Two-way ANOVA was used to compare the two curves over 5 days. Uncorrected Fisher’s least significant difference (LSD) was used for post hoc tests. *P < 0.05; **P < 0.01, ***P < 0.001. (H) Settings of boxes for CR tests. (I) Number of nosepokes in each aperture during the first CR test was compared between control (n = 9) and stimulation (n = 8) groups [two-way ANOVA: F_(1,15) = 0.071, P = 0.794].

Fig. 7.. Stimulation of VTA-to-VP GABA pathway increases motivation to obtain rewards in PR testing.

Following CRT training and CR test as depicted in Fig. 6A, the mice were trained as in Fig. 6B on a PR schedule of reinforcement. Breakpoints [(A); F_(1,17) = 9.23, P = 0.0074], number of earned rewards [(B); F_(1,17) = 11.83, P = 0.0031], task duration [(C); F_(1,17) = 13.5, P = 0.0019], and nosepoke frequency [(D); F_(1,17) = 2.36, P = 0.14] were compared between control (n = 10) and stimulation (n = 9) groups. Two-way ANOVA compared the two curves over 5 days. Uncorrected Fisher’s LSD was used for post hoc tests. *P < 0.05; **P < 0.01; ***P < 0.001.