Single-cell RNA-sequencing analysis of the developing mouse inner ear identifies molecular logic of auditory neuron diversification

Abstract

Different types of spiral ganglion neurons (SGNs) are essential for auditory perception by transmitting complex auditory information from hair cells (HCs) to the brain. Here, we use deep, single cell transcriptomics to study the molecular mechanisms that govern their identity and organization in mice. We identify a core set of temporally patterned genes and gene regulatory networks that may contribute to the diversification of SGNs through sequential binary decisions and demonstrate a role for NEUROD1 in driving specification of a I_c-SGN phenotype. We also find that each trajectory of the decision tree is defined by initial co-expression of alternative subtype molecular controls followed by gradual shifts toward cell fate resolution. Finally, analysis of both developing SGN and HC types reveals cell-cell signaling potentially playing a role in the differentiation of SGNs. Our results indicate that SGN identities are drafted prior to birth and reveal molecular principles that shape their differentiation and will facilitate studies of their development, physiology, and dysfunction.

Fig. 1. Developmental diversification of spiral ganglion neuron lineages.

a UMAP visualization of developing SGNs, showing the trajectory from unspecialized SGNs to distinct subtypes from E14.5 to P3 (Cl.1-17). Cl.18 and 19 denote HC and OM cell clusters. b Plots showing expression of marker genes for SGNs, HCs and OM cells. c, d Plots showing the different developmental time points along the trajectory. e Dot-plot of the top 5 most differentially expressed genes (DEGs) for each cluster shown in a. f Plots showing different marker genes delineating different SGN subtype trajectories. g, h In vivo confirmation of marker genes from f using immunostaining and RNAscope on SG sections from E16.5 (g) or E18.5 and P0 (h). At E16.5, Lypd1 and Mgat4c colocalize and label the emerging I_c population, while the I_a/I_b/II lineage is marked by Islr2 expression. At E18.5, RUNX1, Lypd1 and Mgat4c colocalize and label the I_c population; SGNs expressing only CR (calretinin) represent the I_a population, and SGNs expressing Lypd1, CR and CALB1 represent the I_b population. Type II SGNs are marked by Igfbpl1 and by Plk5/PRPH expression at E18.5 and P0. Scale bars: 20 µm. Data in b and f are MAGIC imputed log10(fpm) expression. DE differentially expressed, HC hair cell, OM otic mesenchyme, SGN spiral ganglion neuron.

Fig. 2. Molecular trajectories of developing SGN cell types.

a Dendrogram recapitulating the branched trajectory of the developing SGNs based on the transcriptional similarity of pseudotime-ordered cells. b Normalized fraction of cells corresponding to each time point of collection across pseudotime, showing that pseudotime is aligned with age. c–g Heatmaps representing overview analysis of the different trajectories, together with the top ten upregulated genes along the differentiation of each branch. h, i Genes predicted to be likely involved in cell state and cell type divergence along the diversification tree; the top 4 up- and 4 downregulated genes (h) and transcription factors (i) are shown for each branch/cell state, as defined in a. j Plots showing different marker genes [in MAGIC imputed log10(fpm) expression] delineating different SGN subtype trajectories. Genes referenced through the text are highlighted in red. The color bars at the top indicate cell states as in a. Data in c–g are expressed in minimax normalized fitted log10(fpm). Dev. developmental, Int. intermediate.

Fig. 3. Gene regulatory network analysis identifies regulons essential for SGN diversification.

a Dotplot showing the developmental progression of regulons associated with the different cell states and transitions along the diversification tree shown in Fig. 2a; the color bars on the left indicate cell states as in Fig. 2a. b Plot showing developmental progression of Neurod1(+) after reducing the threshold score. c Plots showing developmental (pseudotime) progression of two regimes of activity of Neurod1(+) along the differentiation tree. A first regime is active at the beginning of the unspecialized population state, is associated with neurogenesis targets and decreases progressively towards the first bifurcation (representing fate choice between I_c and I_a/I_b/II); a second regime is progressively active in cells as they reach the first bifurcation and continues to be active in the I_c trajectory. d Gene regulatory network representation of Neurod1(+) showing its two regimes of activity associated with either neurogenesis (left) or I_c targets (right). e Genetic strategy for conditional deletion of Neurod1 from postmitotic SGNs. f–j Neuronal diversification phenotype in cochlea of Neurod1^cKO mice at E16.5, P0 and P3. At P3 (f, g) and P0 (h), I_a-, I_b- and I_c-SGNs are CR⁺, CR⁺/Lypd1⁺ and Lypd1⁺/CR^- (arrowheads in f), respectively. At E16.5 (i, j), SGNs are immunostained for βIII-tubulin and only emerging I_c-SGNs express Lypd1 (see Fig. 1). Quantifications of labelling are shown in g, h and j. Data in a represent max normalized fitted AUC scores. Data in c represent mean of minmax normalized gene expression from I_c targets and neurogenesis targets, shown as single cell data points as well as gam fit (line) over pseudotime trajectory. Data in g, h and j are presented as mean ± SEM; circles represent values from individual animals (1 cochlea per animal analyzed, minimum of 5 sections per cochlea, basal and mid-basal regions; n = 2–6 animals per genotype, per stage); t-test, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file. Genes referenced through the text are highlighted in red in a and d. Scale bars: 20 µm. Ctrl. control.

Fig. 4. Gene modules defining cell fate choice and commitment along the differentiation trajectories.

a–i Analysis of the bifurcations of the differentiation tree representing cell fate selection between distinct neuronal cell lineages: I_c versus I_a/I_b/II (a–c), II versus I_a/I_b (d–f) and I_a versus I_b (g–i). a, d, g Scatter plots show average expression of lineage-specific modules in each cell along the trajectory. Early competing modules show gradual co-activation, followed by selective upregulation of one fate-specific module and downregulation of the alternative fate-specific module. Late modules show almost mutually exclusive expression within the two lineages after bifurcation reflecting commitment to either fate. Colors encode tree branches as in Fig. 2a. b, e, h Representative genes in each module; asterisks highlight TFs (see also Supplementary Data 7). c, f, i Average local correlations of early gene modules with branch-specific correlations, in cells with similar developmental pseudotime (in black on the trajectories); the difference between intra- and inter-module correlations is shown in the upper right corner of the correlation plots and would reflect the repulsion between modules. j Summary scheme of the cellular diversification of developing SGNs via distinct trajectories. The transcriptional regulators predicted to be involved in the unfolding of the different lineages are represented on the different trajectories and are derived from the analysis of Figs. 2–4. Genes in colored boxes indicate TFs of early modules from the bifurcation analysis and potentially driving specific cell fate choice (arrows). The arrow symbols after the genes indicate an upregulation or downregulation within a trajectory, while they indicate a high or low expression, respectively, in the final states (I_a, I_b, I_c and II states at the far-right end of the differentiation tree). Int., intermediate.

Fig. 5. Cell–cell communication signatures defining SGN differentiation.

a Expression of genes linked to morphogen signaling, axon guidance and cell adhesion in each neuronal trajectory (see also Supplementary Fig. 4 and 5). The color bars at the top indicate cell states as in Fig. 2a. b tSNE plot of the IHC and OHC clusters at E18.5. c Representation of selected genes specific to IHC (Fgf8 and Trh) and OHC (Bcl11b and Scn11a) on the tSNE plot. d In vivo confirmation of Scn11a and Trh expression in HCs at E16.5 and E18.5 using immunohistochemistry and in situ hybridization (RNAscope). Note that Scn11a is a general marker for both HC types at E16.5. e Dot-plot of the top 25 DEGs between IHC and OHC at E18.5. f CellPhoneDB analysis of the top selected potential outgoing cell-to-cell signalling from HCs to SGNs (see also Supplementary Fig. 7). Data in a, c and f represent max normalised fitted gene expression, Knn smoothed log10(fpm) expression and log2 mean of (interacting molecule 1, interacting molecule 2), respectively. Genes referenced through the text are highlighted in red in a and f. Scale bar: 20 µm. OC organ of Corti.

Fig. 6. Expression of deafness genes in developing SGNs and HCs.

a, b Heatmap of deafness genes related to hereditary hearing loss present in our dataset within SGN lineage trajectory (a) as in Fig. 2a and in HC clusters (b) from Fig. 5b. Genes referenced through the text are highlighted in red. c In situ confirmation of Gjb2 and Gjb6 expression in both supporting cells and IHCs (MYO7a⁺) at E18.5, using RNAscope. Data in a represent max normalized fitted gene expression. Scale bar: 10 µm.