PREreview del Ultrasensitive proteomics uncovers nociceptor diversity and novel pain targets

Peer Review Report

Ultrasensitive proteomics uncovers nociceptor diversity and novel pain targets

The manuscript by Chakrabarti et al describes an innovative application of single-cell proteomics to dissect the molecular diversity of nociceptors. The authors combine electrophysiology with deep visual proteomics to profile thousands of proteins from functionally characterized subtypes of mouse sensory neurons. They focus on peptidergic vs. non-peptidergic nociceptors, identifying subtype-specific proteomic signatures and novel protein markers that were not apparent in prior snRNAseq studies. Notably, the proteome profiles show broad agreement with known transcriptomic differences. For example, TrkA, CGRP, Mrgprd, and P2X3 are appropriately enriched and reveal meaningful discrepancies, suggesting post-transcriptional regulation in these neurons. The authors demonstrate the sensitivity of their approach by detecting ~3000 proteins from thin DRG slices (~ a quarter of neurons). Functionally, they show that treating cultured nociceptors with an inflammatory stimulus such as NGF + a PKC activator induces acute mechanical hypersensitivity in peptidergic neurons, accompanied by widespread proteome changes. Through comparative proteomic analysis and targeted knockdown experiments, they identify glycosyltransferase B3GNT2 as an upregulated protein that appears to mediate NGF-induced sensitization. Overall, the work provides a high-resolution proteomic atlas of sensory neuron subtypes and uncovers a new potential molecular target, B3GNT2, for pain modulation. These contributions significantly advance our understanding of how protein expression underlies sensory neuron diversity and plasticity in pain. The comments below focus on specific points that could be clarified to further strengthen the manuscript.

Major Comments

1.     The manuscript would benefit from additional clarity regarding how protein intensities were normalized across samples and neuronal subtypes. It is not immediately clear whether normalization was performed globally, adjusted to total protein content, or scaled in some other manner. Given the substantial size differences between mechanoreceptors and nociceptors, the chosen normalization approach could influence the interpretation of relative protein abundance. Additional detail on the normalization strategy used would help clarify quantitative comparability across cell types.

2.     The siRNA transfection protocol is described in detail in the Methods section (page 18). However, the manuscript does not report validation of knockdown efficiency for B3gnt2 or Rrad at the transcript or RNA level. Given the role of these knockdown experiments in supporting the functional conclusions, inclusion or discussion of knockdown efficiency measurements, for example, by qPCR and protein-level assessment would help strengthen confidence in the specificity of the observed effects.

3.     The authors note that approximately 90% of proteins identified in cultured neurons overlap with those detected in intact tissue, indicating substantial correspondence between conditions. The remaining non-overlapping fraction may nonetheless reflect biologically meaningful differences, such as culture-associated adaptations or proteins preferentially present in the in vivo environment. While gene ontology enrichment is mentioned, further discussion of the nature of these distinct protein sets could provide additional insight into how culture conditions shape the neuronal proteome.

4.     In Fig. 1h, PCA reveals clear clustering of peptidergic and nonpeptidergic nociceptors and mechanoreceptors across biological replicates. Could a UMAP visualization add complementary insight into the relationships between individual neuronal populations?

5.     A correlation of r = 0.73 between tissue-derived and culture-derived proteomic profiles indicates strong similarity overall, but also reflects persistent differences between conditions. Because roughly half of the variance is not accounted for, slightly more cautious language, such as “largely conserved” or “strongly correlated,” may better align with the data in Fig. 2C.

6.     The detection of glial and immune-associated proteins in the neuronal dataset (Figure 2E) is potentially informative. The authors note that these signals could represent either contamination or physiological expression, but the discussion would benefit from considering the influence of laser microdissection. Given the close physical association between DRG neurons and satellite glial cells, it may be useful to consider in your discussion whether partial co-collection during laser microdissection could contribute to these observations, and how this possibility might be distinguished from true neuronal expression.

7.     In Fig 3A on Cellpose-derived segmentation masks shows that there are some neurons that overlap with each other, but it is not immediately clear how the authors controlled for potential overlap of adjacent neuronal somata within a single tissue section or culture area. This information could help interpret the extent to which the reported datasets represent true single-cell versus pooled or micro-bulk proteomes.

8.     Approximately 100 CGRP⁺ and P2X3⁺ nociceptors were initially collected by laser microdissection from three mice, but only 20 single neurons remained after quality control, as shown in Fig. 3B. Could further details be provided on the criteria used for cell exclusion during QC, and on the most common reasons for removal? Additional context on whether exclusions were driven by low protein yield, poor spectral quality, segmentation ambiguity, or other factors would clarify how the final sample size was determined.

Minor Comments

1.     The Methods describe an MM3A micromanipulator and indentation range (1–9 μm). It might be helpful to specify the time period over which these indentations were applied, for example, increments of 1 μm at 1 Hz, or some other rate as they can influence current adaptation properties.

2.     The Methods and Extended Data Fig. 2 and 3 state “numbers of precursors and proteins identified”. Please clarify the identification criteria used, including the peptide- and protein-level FDR thresholds, the protein inference or grouping approach, and whether protein counts correspond to protein groups or individual accessions, as this would help contextualize the values presented in these figures.

3.     The Methods section would benefit from specifying the duration of neuronal culture before downstream electrophysiological, proteomic, and immunostaining analyses, and from clarifying whether this timing was consistent across assays.

4.     Fig 1D and E.  It is not clear whether IB4^- populations are intended to include the TrkA+ nociceptors and mechanoreceptors. Please clarify how these populations are defined.

5.     Fig 1 title wording: The phrase “electrophysiology-guided” could potentially be interpreted as implying that the same neurons underwent both electrophysiology and proteomics. An alternative phrasing such as “electrophysiology-informed” might better reflect the experimental design.

6.     The Euclidian distance scaling in Fig. 3F and Extended Data Fig. 3B (high and low similarity) could benefit from further clarification.

7.     Extended Data Fig. 4 reports a relatively small sample size (N = 2). Additional context regarding the rationale or intended interpretation of this dataset may be helpful.

8.     Sample sizes are not consistently indicated across figures. Consider including this information throughout the manuscript to aid interpretation.

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they used generative AI to come up with new ideas for their review.