Single cell mass cytometry identifies distinct cell types in rat embryonic spinal cord cultures.

(A) Embryonic day 14 (E14) rat spinal cord are cultured to day in vitro (DIV) 3.5 before a 12hr supplement starvation and subsequent treatment. Samples are collected at indicated time points by rapid dissociation (<90s) and fixation by 2% paraformaldehyde (PFA), then stored at -80°C. Samples are barcoded, pooled, and simultaneously stained with isotopically pure metal-conjugated antibodies before and after methanol (MeOH) permeabilization and fixation. Stained samples are then analyzed by single-cell (suspension) mass cytometry. Created using BioRender. (B) Leiden clustering by cell identity markers is visualized by dimensionality reduction on a 3-Dimensional Uniform Manifold Association Plot (UMAP) colored and labeled by identified cell types. (C) The resulting UMAP is colored by the level of individual cell identity markers to highlight the observed neuronal maturation gradient that shifts from Sox2+, BLBP+ progenitors to more mature neuronal identity with increasing levels of ßIII-Tubulin, MAP2, and NeuN. (D) Non-neuronal cell identity marker levels are plotted to define glial cell populations. Individual cell identity marker expression is normalized from 0 to 1, and gradient is shifted to emphasize locations of marker enrichment, with cells above the intensity threshold colored cyan.

Psuedobulk analysis defines responsive subpopulation of cells responsible for global BDNF-induced signaling.

(A) All cells were included in the pseudobulk analysis of the mean fold change (FC) of responsive cells relative to 0hr, as depicted by heatmap for select markers across time after either continued deprivation (Cont. Depr.) or BDNF treatment. No change is indicated by white (0.67 < FC < 1.5), the level of decrease is visualized by a blue scale (FC ≤ 0.67), and the level of increase is visualized by a red scale (FC ≥ 1.5). Values represent the mean FC (0hr, n = 4; Cont. Depr., n = 2-3; BDNF, n = 3). (B) Co-treatment of BDNF with a Trk inhibitor (K252A) largely abrogated BDNF-induced responses at 1hr of treatment (n = 3). (C) The percent of cells that are defined as responsive plotted for individual signaling markers across time (0hr, n = 4; Cont. Depr., n = 2-3; BDNF, n = 3). (D) Schematic visualization depicting the cells that contribute to a bulk readout, where not all treated cells respond to the stimulus. Created in BioRender. (B, C) Individual points indicate replicate values summarized by mean±SD; p-values are relative to the 0hr samples, colored by treatment, and determined by unpaired Student t test (* p < 0.05; ** p < 0.01; *** p < 0.001 - see Supplemental Table 3 for values).

BDNF-induced signaling dynamics differ between identified cell types.

(A) UMAP highlighting cell identity clusters that are subsequently analyzed. (B) Dot plot depicting the percent of cells that increase activation of the selected signaling molecule within each identified cell type. Size indicates the percent of the given cell type that contribute to the response, while color indicates the mean fold change relative to 0hr. (C) Line graph representation of select data from (B) emphasizes the comparison of responsive cells in the neuronal maturation path. (D) Signaling responses to rescue treatment in cell types that do not exhibit a robust response to BDNF. (E) Line graph representation of select data from (B) to compare responses in non-neuronal cell types. (F) Co-treatment of BDNF with a Trk inhibitor (K252A) largely abrogated cell-type-specific BDNF-induced responses at 1hr of treatment (n = 3). (C-F) Individual points indicate replicate values (0hr, n = 4; Rescue, n = 2-3; BDNF, n = 3) and summarized by mean±SD. p-values are relative to the 0hr samples, colored by cell type, and determined by unpaired Student t test (* p < 0.05; ** p < 0.01; *** p < 0.001 - see Supplemental Table 4 for values). All FC values calculated as described in Supplemental Figure 2 for each cell type.

BDNF induces signaling signature defined by time and by cell type.

(A) Depiction of re-clustering of cell ID clusters (excluding Unassigned) using signaling markers to identify signaling signatures that are agnostic to the initial cell identity. (B) UMAP colored by time point within continued deprivation or BDNF treatment to show the temporal shift in cell distribution across the signaling signature landscape. (C) UMAP of signaling clusters that dynamic shift in cell distribution upon BDNF induction. Signaling clusters are categorized into Non-signaling Signature, Basal Signature, Signaling Signature A, Signaling Signature A, and Signaling Signature C based on the BDNF-induced temporal shifts in distribution. (D) Expression heatmap for indicated signaling clusters to molecularly define the identified signaling signatures. (E) Select cell ID clusters are overlayed on the signaling signature UMAP to identify signaling-driven grouping of cells. (F) UMAP colored by time point across cell types treated with BDNF to depict temporal shift in cell distribution. (G) Signaling dynamics are summarized for selected cell types. Created in BioRender.

TrkB internalization defines subpopulations exhibiting distinct signaling patterns within the same cell type.

(A) Distribution and categorization of TrkB (total) expression level in psuedobulk data. Inset depicts percent of all cells within each expression category. (B) Pseudobulk signaling response within each expression category to define general trends correlating to TrkB expression. (C) Distribution of signaling signatures by cell ID and colored by TrkB (total) expression levels. Dotted lines indicate Signaling Signature A (rapid pAkt/pERK; magenta), B (sustained pS6; yellow), and C (complex multi-marker activation; green). Insets highlight the BDNF-induced signatures. (D) TrkB (total) expression distribution across selected cell types, indicating each cell type expresses some degree of TrkB. Dashed lines represent to the same values set on pseudobulk data in (A). (E) Diagram of TrkB surface staining used as a proxy to measure TrkB internalization. (F) Temporal trend of TrkB (surface) in response to continued deprivation (Cont. Depr.), Rescue, or BDNF within select cell types. Individual points represent the fold change of the mean of each replicate (0hr, n = 4; Cont. Depr., n = 2-3; Rescue, n = 2-3; BDNF, n = 3) and summarized by mean±SD. p-values are relative to the 0hr samples, colored by cell type, and determined by unpaired Student t test (* p < 0.05; ** p < 0.01; *** p < 0.001). (G) Results suggest that surface dynamics of TrkB is a better predict of BDNF responsiveness than static TrkB abundance. Created in BioRender.

Identical BDNF receptor profiles do not confer identical responses across cell types.

(A) Distribution and categorization of TrkB and p75NTR expression level in psuedobulk data. Inset depicts percent of all cells within select expression categories defining TrkB– /p75–, TrkB–/p75+. TrkB+/p75–, TrkB+/p75+ subpopulations. (B) Pseudobulk signaling response within each expression category highlighted in (A) to define general trends correlating to TrkB, p75NTR co-/expression. (C) Distribution and categorization previously characterized cell types based on TrkB, p75NTR co-/expression. (D, E) Comparison of signaling dynamics across cell types with a (D) TrkB-Med/p75+ or (E) TrkB-Med/p75+ expression pattern. (F) Schematic depicting cells with the same BDNF receptor profile exhibiting different signaling responses. Created in BioRender. (D-E) Individual points indicate replicate values (0hr, n = 4; BDNF, n = 3) and summarized by mean±SD; p-values are relative to the 0hr samples, colored by cell type, and determined by unpaired Student t test (* p < 0.05; ** p < 0.01; *** p < 0.001).

Factors contributing to the heterogeneity of BNDF signaling responses.

Stimulation with BDNF induces different cell signaling responses across different cell types. The pattern of BDNF receptor abundance is necessary to regulate how cells respond; however, stoichiometry alone is not sufficient to explain how some cell types are non-responsive, despite high levels of TrkB. A better predictor of BDNF sensitivity is the prolonged reduction of surface TrkB receptor (i.e. internalization), where the ratio of TrkB isoforms may add a level of developmental competency to the overall regulation network. Ultimately, the cell-intrinsic environment acts as the final filter through which a cell interprets the BDNF message. Created in BioRender.