Avalilação PREreview de Semaglutide drives weight loss through cAMP-dependent mechanisms in GLP1R-expressing hindbrain neurons

Review of “Semaglutide drives weight loss through cAMP-dependent mechanisms in GLP1R-expressing hindbrain neurons”

Summary

This study by Gao et al. provides a comprehensive mechanistic dissection of how semaglutide drives weight loss through intracellular signaling in hindbrain GLP1R neurons. Using a powerful combination of conditional genetic knockouts, two-photon imaging, and circuit tracing, the authors reveal that semaglutide acts through both Gs- and Gq-mediated pathways in the AP to produce cAMP-dependent neuronal activation. The demonstration that cAMP signaling in Glp1r-expressing neurons in AP rather than neurons in NTS is necessary for semaglutide-induced weight loss, and that this signaling engages downstream parabrachial circuits, represents an important advance in understanding GLP1R agonist action. Conceptually, the study bridges intracellular signaling and systems-level neural control of food intake and weight loss, establishing a framework that may inform design of next-generation obesity therapeutics. Several points could benefit from clarification or additional support to strengthen the conclusions, including the distinct role between AP and NTS and role of Gq signaling in weight loss. I detail these points below in Major Comments and Minor Comments sections.

Major Comments

1.     In Fig. 2q, semaglutide-induced DVC neuronal activity is markedly reduced by Gq inhibition (FR 900359) even in GnasWT neurons, where Gs signaling remains intact. Although the authors conclude that Gs drives the sustained response with only a minor contribution from Gq, this result may be interpreted as an important role for Gq. Including in vivo tests of Gq inhibition during semaglutide treatment, or additional controls directly comparing no treatment, Gq inhibition alone, and combined Gs + Gq inhibition in GnasWT neurons, would clarify whether Gq signaling plays an important role in semaglutide weight-loss action.

2.     Along the same line, the authors report that Gs knockout in vivo abolishes semaglutide-induced weight loss, suggesting that Gq signaling does not play a major role in mediating the behavioral effects of semaglutide. However, this contrasts with the slice imaging data indicating a measurable Gq component in DVC^Glp1r neuronal activation (Fig. 2p, q). Have the authors assessed the behavioral consequences of Gq inhibition in vivo during semaglutide treatment? Demonstrating whether Gq contributes to weight loss by semaglutide would be novel, nausea-related behaviors, or other physiological outputs would help resolve this question.

3.     In Fig 1 and 4, the authors report that Gs signaling in NTS neurons increases body weight on a high-fat diet, implying an important role for NTS Gs signaling in energy balance. However, this challenges the previous finding that the NTS^Glp1r circuit is necessary for aversion-independent satiety and efficacy, since inhibiting the AP aversion pathway does not block anorexia (K. Huang and A. Alhadeff et al., Nature, 2024). Further clarification of how NTS^Glp1r Gs signaling contributes to semaglutide’s behavioral effects, either through additional analyses or expanded discussion, would help reconcile these observations and strengthen the mechanistic framework.

4.     The TRAP-based mapping identifies elPBN neurons as a downstream population mediating conditioned taste aversion, but the molecular identity of these neurons remains unclear. In situ hybridization could determine whether these neurons overlap with Calca and/or other known subtypes. This would provide insightful results on the molecular identity of semaTRAP neurons in the PBN.

5.     In Fig. 2a, the authors describe a genetic cross in which the Glp1r^iresCre line is crossed with two additional mouse lines. Because some Cre-driver lines exhibit germline leakiness, it is important to confirm that Gnas deletion is restricted specifically to Glp1r neurons within the DVC. Providing this validation would strengthen the study and increase confidence in the fidelity of genetic manipulation.

Minor Comments

1.     To enhance the background knowledge of general readers, it would be helpful to include introductory paragraphs about GPCR downstream signaling pathways, such as the differential signaling mechanisms between Gs and Gq that result in Ca²⁺ modulation in neurons.

2.     In Fig. 1g, the animals shown appear to be the same animals included in the DVC^ΔGnas and AP^miss groups in Fig. 1f. To improve clarity, it would be helpful to mark the group assignment of each individual animal. This would provide additional information regarding the relationship between virus expression and changes in body weight.

3.     The authors state, “we observed no significant differences (Extended Data Fig. 4f), suggesting that transient cAMP responses to semaglutide may be regulated by β-arrestin–independent processes.” Have the authors examined whether body weight loss is altered in β-arrestin knockout animals? This information would further support the conclusion.

4.     Fig. 2o nicely quantifies fluorescence changes as AUC. However, the similar AUC levels between semaglutide and KCl are not intuitive, as KCl would be expected to induce maximal neural activity. If the authors calculated mean fluorescence (AUC divided by time) instead of AUC alone for this and subsequent graphs or measure Gs and Gq Ca²⁺ responses as relative to maximum KCL response, the results might be more intuitive.

5.     In Fig. 3, because cADDis is an inverse sensor, increases in fluorescence may confuse readers. Although the authors stated in the legend, adding a label in the figure that explicitly states that the fluorescence traces are inverted would improve clarity. Similarly, scale bar values (e.g., −0.05 ΔF/F0) should be absolute value rather than negative.

6.     In the statement, “Our TRAP-based approach allows for the capture and transfection of a broader subset of neurons than those readily labeled by genetic markers such as Cgrp...,” please correct the gene name: CGRP corresponds to Calca.

7.     It would be helpful to provide additional limitations of the study beyond the need for in vivo cAMP measurements.

8.     In the Methods section, the authors carefully describe histology analysis methods such as axonal density quantification. However, it would be great to see the quantification methods used in Fig. 1d (% Cre expression) and Fig. 1g (% viral expression of region) as well. Clarifying whether these values reflect counts or fluorescence area would help readers interpret the targeting accuracy.

9.     The authors propose that Gq signaling complements Gs-driven cAMP production in AP^Glp1r neurons. However, it remains unclear whether Gq activation serves as a compensatory mechanism or represents a parallel GLP1R-coupled pathway. Temporal and pharmacological dissection of Gq–Gs interactions in vivo would strengthen this conclusion.

10.  Given known sex differences in obesity and Glp1 and Glpr1 neuronal function (L Lopez-Ferreras et al., Front. Neurosci., 2023 and SM Hofmann et al., Diabetes, 2023), it would be informative to comment on sex differences for data analysis.

Editorial suggestions

1.     To ensure clarity, I recommend avoiding the abbreviation of PDE4-cat as PDE+, as PDE+ may be misinterpreted as wild-type PDE4.

2.     Since this manuscript does not address viral Fos, there seems to be no need to distinguish between viral and cellular Fos in this context; hence, use “Fos” alone would be sufficient than cFos and a-cFos to help readers clarity.

3.     In the text using ‘subcutaneous Semaglutide’ would provide greater clarity than ‘SQ Semaglutide’. As SQ is jargon for subcutaneous, SQ in the figures could be defined in legend.

4.     When referring to genes, gene names should be italics, e.g., Gnas^fl/fl should be italics (Gnas^fl/fl) because the fl/fl refers to alleles of the gene.

5.     If Glp1r-iresCre is targeted to the gene locus, it represents allele of that gene. Consequently, the iresCre should be superscript and italic (Glp1r^iresCre) for the consistency of nomenclature.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.