A Single-Cell Signaling Atlas of Spinal Cord BDNF Responses Reveals Determinants Beyond Receptor Expression

Reviewer #1 (Public review):

The manuscript by the Deppmann group is an important contribution to understanding how growth factor signaling is controlled at a per-cell basis, in contrast to bulk biochemistry results. Their system uses cell culture and single-cell signalling proteomics methods to measure responses of cells of different developmental stages (from E14 rat) with complex but relatively clear-cut phenotypes, allowing the effects of BDNF to be compared. This work validates the method for the discovery of future insights from less well-studied ligand-receptor investigations.

Strengths include:

(1) The methods are cutting-edge and powerful.

(2) Clearly written. It leads the reader through the rationale of methodological steps.

(3) Step-by-step data interrogation rather than leaping into complex models of analysis.

(4) "sanity check" controls e.g., mimicking bulk culture expected signaling /expression changes.

(5) Testing biologically of certain findings within the presentation of the results ( e.g., progenitors not responding to BDNF also not internalising TrkB).

(6) Effort to make complex figures/data as understandable as possible.

(7) Not overstating conclusions.

(8) Important conclusion of receptor stoichiometry sets the potential for BDNF sensitivity, and that the intrinsic environment allows for a cell to engage that potential, something possibly thought but not demonstrated previously.

Major points:

(1) Apply appropriate statistics: Student's t-tests are used throughout. It would be more appropriate to utilise ANOVA, at least one-way, to compare across timepoints for a given phospho-protein within one treatment condition (e.g., pERK following BDNF stim), or even multiple t-tests. Also, multiple testing adjustments. are likely needed (not my expertise).

(2) Some data points are n=2; for statistical rigour n=>3 would be appropriate.

(3) They measured pTrkB with antibody targeting site Y816, which couples to PLCy/PKC/Ca2+, but not Shc (for PI3K/MEK pathways), why? Did they get any measurements using an antibody targeting the phosphorylation sites in the activation loop of the kinase? Could this explain the relatively low abundance of active TrkB, compared to the measured TrkB-dependent signalling outcomes? Especially considering the "unresponsive" cells. E.g. https://doi.org/10.1016/S0896-6273(00)00035-0.

(4) Was TrkC ( or A) expressed in any TrkB population that could potentially mediate BDNF signaling?

Reviewer #2 (Public review):

In this study, Sewell et al. use a novel approach to understand cell-specific BDNF signaling in the developing spinal cord. Using cultured E14 spinal cord, the authors used a mass cytometry approach to identify the levels of TrkB and p75NTR receptor expression, as well as 19 signaling markers and cell identification markers, to delineate activation of BDNF signaling in different cell types within a complex population. They identified that the level of receptor expression, while necessary, is not sufficient to determine the activation of signaling cascades. It has been known for some time that TrkB, indeed all RTKs, have the capacity to activate certain canonical signaling pathways; however, not all these pathways are always activated upon ligand treatment. This study begins to identify the conditions under which specific signaling pathways are activated by ligand. Specifically, the type of cell and maturation state are critical for determining signaling. The cytometry approach allows the clustering of cell types according to expression of specific markers, and overlaying those clusters onto the expression status of TrkB and p75 receptors, as well as specific activated signaling proteins. This study provides greater insight into when specific signaling events can be activated by BDNF than was previously known.

The comparison of levels of expression of TrkB and p75NTR is interesting to demonstrate which pathways may require one or both receptors for specific signaling responses.

It is very interesting that progenitors do not respond to BDNF despite abundant expression of TrkB, although they responded to the rescue treatment with phosphorylation of Erk and Akt. The development of competence to respond to BDNF is an interesting question for future analysis, and the authors suggest some possibilities in their Discussion.

The responses of glial cells in their culture preparation are also interesting. They see signaling responses to BDNF in astrocytes and "laden" microglia (presumably phagocytic). E14 spinal would not be expected to have a large population of glia at this stage of development, although the serum in their plating media would allow for the proliferation of the progenitors. Astrocytes are generally considered to have the truncated TrkB receptor, yet they see P-Erk, P-Akt, etc. in these cells in response to BDNF. This raises the question of which receptors are expressed in the glial populations and whether the responses in these cells are also maturation dependent, since the glia in their culture conditions are also likely to be immature.

Some specific comments:

(1) The authors should specify what is meant by "rescue" in the text. What is rescuing the cells from trophic deprivation when no BDNF is added? Is it the B27 and GlutaMax in the Maintenance media, and does this actually rescue the cells?

(2) Figure 3 - K252a blocked activation in most, but not all, lineages, especially in mature neurons. Is some component of the P-Erk activation in these cells TrkB independent?

(3) Figure 5 E, F - The correlation between receptor surface depletion and signaling is based on "surface-specific staining". Does the staining allow you to see internalized receptors to confirm that the receptors are internalized?

(4) The drawbacks to the study - particularly capturing snapshots in time to represent signaling cascades, are fully acknowledged in the Discussion. The interplay between TrkB-T1, TrkB-FL, and p75NTR cannot be elucidated from this study, but again, that is acknowledged and will require a different approach.

Reviewer #3 (Public review):

This study addresses a fundamental and long-standing question in neurotrophin biology, how cellular context shapes the interpretation of a single trophic message, and tackles it with a technically demanding and well-executed single-cell mass cytometry approach. By simultaneously measuring 19 signaling effectors and 18 identity markers across a developmental gradient of spinal cord cell types, the authors substantially expand our understanding of BDNF signaling and provide a compelling demonstration of the limitations inherent to bulk biochemical readouts, which average across heterogeneous populations and obscure the discrete subpopulation behavior that the present data reveal.

The finding that only 47-75% of cells respond at peak activation, that maturation state dictates both the magnitude and the qualitative "signature" of the response, and that identical receptor stoichiometries can yield divergent outcomes across cell types collectively constitute an important conceptual advance. The proposed framework of "prepared competence" is thought-provoking and likely to stimulate follow-up work.

That said, several aspects of the data interpretation deserve more critical discussion. My specific comments are detailed below.

(1) Interpretation of TrkB-independent ERK activation (lines 194-196).

The authors state that the residual pERK induction observed in TrkB-negative ("None") cells and the incomplete suppression of pERK by K252a support the established notion that BDNF signaling is not mediated solely through TrkB. This interpretation is presented without sufficient mechanistic detail and, in its current form, is difficult to follow. If BDNF-induced ERK activation is not mediated by TrkB, which alternative receptors could account for it? Does this reflect signaling through p75NTR, transactivation of other receptor tyrosine kinases, or another mechanism altogether? Likewise, the partial resistance of pERK to K252a is interpreted as evidence of an additional regulatory layer, but the underlying activity is not specified. Is the authors' hypothesis that a distinct pool of ERK is engaged independently of Trk activity? If so, what kinase activity is proposed to drive it? These results are intriguing yet puzzling and merit a more critical and explicit discussion of the candidate mechanisms.

(2) The "progenitor paradox" in light of prior work on PC12 cells (lines 207-208).

The observation that TrkB-expressing progenitors remain insensitive to BDNF is presented as a paradox and interpreted through the lens of impaired internalization. This interpretation would benefit from explicit discussion in the context of the classical work on PC12 cells (Segal and colleagues, among others), which established that plasma membrane-restricted Trk receptors engage the Ras-MAPK pathway with rapid, short-duration kinetics that drive proliferation rather than differentiation, whereas internalized Trk receptors sustain MAPK signaling and promote differentiation. Under this framework, the apparent signaling silence of progenitors could, in fact, reflect transient plasma membrane signaling that the time points sampled in the present study (5 min onward) may not capture. The single-cell mass cytometry approach used here is, in principle, well-suited to resolving such rapid kinetics, and the authors are encouraged to address this possibility, both as an alternative interpretation of their data and as a potential extension of the study.

(3) Astrocyte responsiveness and the TrkB isoform issue.

The authors report that astrocytes are highly responsive to BDNF and exhibit robust ligand-induced depletion of surface TrkB, which they interpret as evidence of signaling-competent full-length TrkB (TrkB-FL) on these cells. However, it is well established that astrocytes predominantly express the truncated isoform TrkB-T1, which lacks the intracellular kinase domain and is thought to function in BDNF capture, clearance, and recycling at synapses rather than in canonical downstream signaling. The robust phosphorylation events observed in astrocytes are therefore difficult to reconcile with TrkB-T1-mediated signaling alone. Could these responses instead reflect transactivation of other receptors through neuron-astrocyte crosstalk, for instance, via ligands released by neurons in response to BDNF? Because the authors explicitly state that their antibody cannot distinguish TrkB-FL from TrkB-T1, this limitation directly impacts the interpretation of the astrocyte data and of the proposed isoform-switch hypothesis for progenitors. This caveat is briefly acknowledged but deserves more thorough discussion, ideally with explicit consideration of the alternative interpretations outlined above.

(4) Pathways resistant to K252a inhibition.

The authors note that K252a fails to fully abolish pERK induction in several lineages, but the specific pathways, differentiation states, and receptor stoichiometries that remain K252a-resistant are currently insufficiently described. A more systematic description would strengthen this section. In addition, it would be helpful to discuss whether the residual signal could reflect the proximity of the response to the detection threshold rather than a genuinely K252a-insensitive pool of activity. More broadly, K252a is a broad-spectrum tyrosine kinase inhibitor with well-documented off-target effects, and the present study relies on this single pharmacological tool to define Trk-dependence. The limitations of this approach, and the desirability of complementary inhibitors or genetic perturbations in future studies, should be acknowledged in the Discussion.

(5) The 12-hour trophic deprivation paradigm as a potential confounder.

All cells in the present study are trophically deprived for 12 hours prior to stimulation. This is a methodologically convenient choice, but sustained deprivation is not a neutral starting point: it activates stress-responsive pathways (JNK, p38, autophagy), alters receptor surface trafficking, and can sensitize cells to subsequent stimulation. Several of the reported observations - including the apparent synergy of p75NTR with TrkB on stress markers (p-c-Jun, p38) and the strong induction of trophic effectors immediately upon BDNF addition - could be amplified, or qualitatively altered, by the prior deprivation state, which does not reflect baseline in vivo physiology. The Rescue control, with complete medium, partially addresses this concern but is non-specific. The authors should explicitly acknowledge this limitation and, ideally, discuss the extent to which their conclusions about cell-type-specific signaling competence depend on the deprivation paradigm.

(6) Direct comparison of pseudobulk data with conventional bulk biochemistry.

The pseudobulk reconstruction of the single-cell data is presented as recapitulating canonical BDNF responses, but this comparison relies on general agreement with the published literature rather than on a direct, parallel measurement in the same cultures. Given that the central conceptual contribution of the manuscript rests precisely on departures from the bulk biochemical view of BDNF signaling, an explicit side-by-side comparison of the pseudobulk profile against a parallel bulk Western blot from sister cultures - for at least a subset of key markers such as pERK, pAkt, and pCREB - would substantially strengthen the validation of the platform. Such a comparison would reassure the reader that the discrete subpopulation behavior reported here is genuinely biological, and not in part a consequence of methodological differences between mass cytometry and conventional biochemistry (e.g., differences in fixation kinetics, epitope accessibility, or sensitivity to low-abundance phosphoproteins).

(7) Manuscript organization and balance between main and supplementary figures.

The manuscript presents an exceptionally rich dataset, but the current organization - seven main figures supported by thirteen supplementary figures, several of which are explicitly labeled as extensions of main-text figures - makes it difficult to follow the argument without continuous cross-referencing between documents. I would encourage the authors to consider a substantive reorganization with the following suggestions: (i) Figure S2 and Figure S3, which respectively define the threshold-based "responsiveness" criterion and assess its robustness, are foundational to the central 47-75% responsiveness claim and would be better integrated into the main text, for example as additional panels of Figure 2; (ii) the methodological and quality-control components of Figure S1 and Figure S2 would be more naturally placed within the Methods section; and (iii) the four "Extension" figures (S4, S7, S12, S13) contain considerable redundancy with the corresponding main figures and could be consolidated, with only the most diagnostic panels retained. Concurrent trimming of the denser main figures (Fig. 4, 5, and 6 each carry six or seven panels) would further improve readability.