Fig 1.
Single-cell gene expression profiling of adult mouse cochleae during ectopic Atoh1-induced conversion.
(A) tSNE plot of the SCANPY clusters. Cell types were annotated based on previous cell annotations [7]. (B) tSNE mapping of normalised count expression levels of specific markers. Slc17a8 and Ocm were highly expressed in mature IHCs and OHCs, respectively, but not in cHCs (annotated as Atoh1+ in the figure) or SCs (annotated as DC/PCs, IPs/IB+ in the figure). HA (Atoh1-HA) and Fgfr3 were highly expressed in cHCs clusters. (C) Top 25 differential genes when comparing IHCs with OHCs. (D) Ranking for highly differentially expressed genes in the Atoh1-HA+ cHC, DC/PC, IHC and OHC clusters, respectively. RM: Reisner membrane; DCs: Deiters’ cells; PCs: pillar cells; HCs: hair cells; Iphs: inner phalangeal cells; IBs: inner border cells; SGs: spiral ganglia.
Fig 2.
Gene regulatory network (GRN) inference using WGCNA and SCENIC.
(A) WGCNA analysis of cHCs and endogenous HCs, showing the correlation of each cell type and the expression pattern of eigengenes in each module. Each spike represents one cell of the cell types indicated below. (B) Correlation matrix depicting the correlation coefficients of different modules, with all seven modules included and self-comparisons removed from the graph. The size of the box and its color gradient indicate the correlation coefficient. (C) SCENIC identified cell-type-specific GRNs during SC-to-cHC conversion. The binary regulon activity matrix shows correlated regulons (absolute correlation > 0.3) active in at least 1% of cells. Cells are colored according to cluster identity, and genes are colored based on expression level. Regulons exceeding a manually adjusted AUC threshold are shown in black, while inactive regulons are white. Each stage of the conversion is shown at the top of the graph in different colours. (D) Venn diagram displaying the five overlapping key regulators identified by both WGCNA turquoise module (cHC3) and SCENIC in cHC2/3, compared to regulons in OHCs.
Table 1.
Regulators in WGCNA-derived co-expression modules.
Fig 3.
Comparison of expression patterns of converted HC clusters to P1 wild-type OHCs and other naive OHC stages.
(A) PAGA trajectory analysis of wild-type cochlear cells across different developmental stages. L.PsCs, Lateral pro-sensory cells; IPC: Inner pillar cells; DC/OPCs: Deiters’ cells/Outer pillar cells. (B) UMAP analysis of eight different cell types, including cHC1-3 and OHC_P33 from Yamashita et al. [7], and OHC at different stages of natural hair cell development [1]. iHC_E14: immature hair cell at E14; iOHC_E16: immature outer hair cell at E16. (C) The SCT-normalization method effectively reduced the influence of technical factors in the combined dataset. (D) Heatmap showing the top 10 (or all, if less than 10) markers for each cluster. (E) Correlation plot of cell types revealed that cHC3 is more similar to OHC_P1 (r = 0.32) than to other naive OHC stages (iHC_E14:r = 0.1; iOHC_E16: r = 0.2; OHC_P7: r = 0.032; OHC_P33: r = 0.031).
Fig 4.
Transcription factors identified using VIPER in the comparison of the expression profiles of cHC3 versus naive OHCs at P1.
VIPER plot shows the top 10 up-regulated (red) and down-regulated (blue) transcription factors. The strength of protein activity (Act) and RNA expression (Exp) are represented by color intensity, with darker color indicating higher values. Prox1 is shown to be insignificant when using 100 bootstraps. The lines in the middle of the plot are the individual gene expression values for each cell in the dataset and provide additional information about the variability in gene expression within each cell type.