Inhibition of Slc17a7 expressing neurons in the basolateral amygdala which project to the nucleus accumbens shapes the fidelity of motivated behavior

Reviewer #1 (Public review):

Summary:

The authors aimed to determine whether reward conditioning increases inhibitory regulation of Vglut1-expressing BLA→NAc neurons and whether this inhibition shapes motivated behaviors. They used whole-cell electrophysiology to measure conditioning-induced changes in synaptic inhibition and intrinsic excitability. Subsequently, they employed dual-recombinase chemogenetics to selectively inhibit this projection during behavioral tasks. The goal was to test whether suppressing the activity of Vglut1-expressing neurons would alter reward learning, valuation, and fear discrimination.

Strengths:

(1) The combination of electrophysical and behavioral assessments to dissect the function of Vglut1-expressing BLA→NAc neurons.

(2) The various behavioral assessments employed to determine the effect of silencing Vglut1-expressing BLA→NAc neurons.

Weaknesses:

(1) The introduction underscores the importance of molecular identity and population dynamics when studying the function of BLA→NAc neurons. Yet, the experiments and manuscript provide little to no information about the Slc17a7-expressing population under study. In fact, there is no evidence that the viral manipulations targeted this neuronal population (e.g., extent and specificity of viral transduction). Regarding population dynamics, evidence is meant to be provided by Experiment 1, but the results are difficult to interpret. The control mice were not exposed to the conditioning chambers, stimuli, or food rewards. These exposures may have been sufficient to produce the changes observed in the experimental mice (i.e., they may have had nothing to do with cue-reward learning). Further, the experiments provide no evidence that the observed effects result from prolonged conditioning, since there is no group receiving a single conditioning session.

(2) The dual-recombinase approach employed does not permit conclusions about the BLA→NAc pathway specifically, because the effects of silencing NAc-projecting BLA neurons could be driven by modulation of activity in other brain regions innervated by these same neurons through collateral projections. This limitation must be clearly acknowledged by the authors, and the manuscript should refrain from making definitive claims about the BLA→NAc pathway per se.

(3) The experimental parameters and measures used for cued-reward conditioning complicate any firm conclusions about the observed effects. The use of a 2-second cue provides a minimal temporal window to monitor cue-related behavior. This issue is masked in the data presented because what is labeled as "cued responses" includes responses that occur after the cue has terminated and overlap with those triggered by sucrose delivery itself. These post-cue responses cannot be classified as cue-reward responses since the cue is no longer present; they are reward-related responses. Perhaps the z-score calculation addresses this issue, but this is difficult to assess since the authors do not explain how this calculation was performed or what baseline period was used.

(4) Throughout the manuscript, there is conceptual confusion regarding the fundamental distinction between Pavlovian (cue-outcome) and instrumental (action-outcome) responses. It is unclear why the authors aimed to study both types of conditioning, but greater caution is necessary when interpreting the findings labeled as "instrumental conditioning." First, no evidence is provided that initiation port entries constitute an instrumental or goal-directed response rather than a Pavlovian approach behavior. Second, many of the conclusions are based on analyzing reward port entries-a Pavlovian conditioned response identical to that measured in the cued-reward conditioning task. This conflation undermines claims about instrumental learning.

(5) The data from the reward valuation and reversal learning experiments are difficult to interpret. The animals are not tested under extinction conditions (with the flavors present but without reward delivery), making it impossible to establish whether their behavior relies on learned associations or ongoing reinforcement. Further, the behavior generated by these procedures appears unreliable, with substantial inconsistencies across figures (compare Figure 4A with Figures 5B, C, G, H).

(6) The results from the auditory fear discrimination procedure are also difficult to interpret. No conditioning data are presented, and the "enhanced discrimination" could simply reflect reduced overall responding to the CS-. It is not clear how this selective impact on the CS- fits with the authors' conclusions about enhanced associative salience (noting that the meaning of the latter remains obscure).

(7) The manuscript contains several statements about behavioral outcomes that are not supported by statistical evidence. The list provided here is non-exhaustive, and the authors should carefully correct any conclusions that lack statistical support.
a) Line 294 (Figure 2F): the control mice gradually reached a similar performance to the experimental mice.
b) Lines 301-303 (Figures 3D-F): inhibition strengthened the temporal association between initiation and reward consumption.
c) Lines 337-339 (Figure 4A): both groups increased their preference for 10% sucrose.

(8) The manuscript suffers from a lack of clarity and/or transparency about experimental parameters and data. Clarifications about the following would be necessary for the reader to confidently interpret the findings.
a) Number of animals of each sex in each group.
b) Number of animals excluded and justification.
c) Analysis of sex differences.
d) A clarification on the control group used in the electrophysiological experiment.
e) Whether the same animals progress through multiple behavioral paradigms or if separate cohorts are used.
f) All protocols should be described in the methods section.

Without clarifying the points made above, a reliable and fair assessment of the discussion is impossible.

Reviewer #2 (Public review):

Summary:

This study by Mercer et al. focused on Vglut1 neurons in the BLA that project to the NAc. They characterized reward conditioning-induced electrophysiological changes in these neurons, including a decrease in membrane excitability and an increase in inhibitory synaptic inputs onto them, and showed the consequences of reducing their activity in enhancing reward-seeking behaviors. Considering that Vglut1 neurons represent the majority of the BLA→NAc projecting neurons, the findings are important for potentially correcting some of the previous biases in understanding the role of BLA-to-NAc projection in reward processing, for example, the notion that this projection generally promotes reward seeking by conveying reward-associated cue information.

Strengths:

The paper is clearly written, with results strongly supporting the main conclusions for the most part.

There are a few weaknesses noted. For example:

(1) They used a retrograde recombinase strategy to drive DREADD expression in these cells; however, it is not known if they project exclusively to NAc or to other brain regions as well, and whether those other potential regions may mediate the DREADDs (Gi) effects on reward seeking. They also did not show which subregions of the NAc were innervated by these neurons.

(2) They did not assess potential changes in excitatory synaptic transmission onto these cells after reward conditioning, which leaves a gap in concluding a shift toward inhibition.

(3) They also did not report on whether the inhibition was specific to Vglut1 neurons.

(4) Some statistics appear missing (Figure 3D-F), not optimal (Figure 5CEF and HJK using separate t-tests rather than repeated measure ANOVA), not clear (Figure 2I on peak timing or port entry), or has low n number (Figure 1 Ephys, animal-based manipulations).

(5) They did not clarify why they used two different doses of the DREADDs ligand Compound 21 at 0.1 or 0.3 mg/kg for different experiments.

Reviewer #3 (Public review):

Summary:

This study by Mercer et al. investigates how inhibitory modulation of basolateral amygdala neurons expressing Vglut1 and projecting to the nucleus accumbens (Vglut1BLA→NAc) influences motivated behavior in both appetitive and aversive tasks. Using a combination of whole-cell electrophysiology, chemogenetic inhibition and behavioral tests, the authors demonstrate that (1) reward conditioning increases inhibitory synaptic input and reduces intrinsic excitability of Vglut1BLA→NAc neurons, (2) chemogenetic inhibition of these neurons enhances the number of conditioned approaches in a Pavlovian task and the number of nosepoke responses in an instrumental task, elevates reward valuation, and increases fear discrimination and (3) these effects are linked to salience assignment and associative strength, rather than altered learning or reversal flexibility. The work challenges the classical excitatory function usually reported about the BLA projection to the NAc and highlights an interesting and thought-provoking result. Nevertheless, the study does not address the potential effect of their manipulation on motoric impulsivity, nor did they provide a theoretical framework explaining this unorthodox yet interesting effect.

Strengths:

The study establishes the initial finding with a correlational approach that informs a causal study. They find convincingly that Pavlovian conditioning induces an increase in inhibitory inputs onto Vglut1BLA→NAc neurons that leads to reduced excitability. Causality is studied using a powerful dual recombinase chemogenetic strategy to selectively inhibit this population of Vglut1BLA→NAc neurons and determine the effect on different behavioral tasks. The use of different tasks provides convergence on their effect. This surprising finding provokes interest and will stimulate further investigation into the mechanisms underlying these effects.

Weaknesses:

Several important aspects of the evidence remain incomplete.

(1) First, an important aspect of the underlying processes at play remains to be investigated. In all behavioral tasks, the authors find that their manipulation increases responding that they interpret as a facilitation of learning. However, none of the appetitive tasks include a control stimulus that could address the specificity of their effect. Given that on the Pavlovian task, responding to the CS is almost 100%, I suspect that their manipulation may induce motoric impulsivity. This aspect would clearly benefit from additional controls.

(2) Second, I have several interrogations about the time-resolved probability of port entries (PSTHs).

a) There is a mismatch between the results presented in Figure 1. Panel D shows a peak of responses on the PSTH at ~2s on day 5 (my remark applies to all days), suggesting that the average should lie around this value. However, panel C reports a latency to respond at ~4sec. Could the authors double-check their PSTHs?

b) More generally, the fact that in the Pavlovian task all PSTHs show a peak at almost exactly 2 sec is quite surprising and raises questions about how they are constructed. Sure, the most salient event is the water drop occurring 2s after cue onset. Yet, if mice responded only to these drops, the peak response should occur at 2s+reaction time, which is not the case. Figure 2 shows that on the first acquisition day, responding is already centered around 2s and does not decrease with learning, except for treated animals.

(3) Several methodological flaws are present.

a) The authors need to report clearly the statistics. In most cases, the statistical test used is mentioned in the figure caption with a single P-value. Thus, on two-way ANOVAs, I do not know whether the P-value relates to the interaction, the main effects, or the post-hoc tests.

b) Another important issue is related to the average time-resolved z-score probability of port entries. The bin size used, the smoothing (that is much too strong), and the baseline period used to calculate the z-score are absent from the methods.

(4) This study reports that manipulating 70% of the glutamatergic projection to the NAc induces an effect opposed to what has been previously reported in many different studies. Such a surprising finding deserves a more elaborate discussion about the mechanism that could be at play.