Gene Expression by Mouse Inner Ear Hair Cells during Development

Abstract

Hair cells of the inner ear are essential for hearing and balance. As a consequence, pathogenic variants in genes specifically expressed in hair cells often cause hereditary deafness. Hair cells are few in number and not easily isolated from the adjacent supporting cells, so the biochemistry and molecular biology of hair cells can be difficult to study. To study gene expression in hair cells, we developed a protocol for hair cell isolation by FACS. With nearly pure hair cells and surrounding cells, from cochlea and utricle and from E16 to P7, we performed a comprehensive cell type-specific RNA-Seq study of gene expression during mouse inner ear development. Expression profiling revealed new hair cell genes with distinct expression patterns: some are specific for vestibular hair cells, others for cochlear hair cells, and some are expressed just before or after maturation of mechanosensitivity. We found that many of the known hereditary deafness genes are much more highly expressed in hair cells than surrounding cells, suggesting that genes preferentially expressed in hair cells are good candidates for unknown deafness genes.

Keywords: FACS; RNA-Seq; cochlea; development; hair cell; vestibule.

Figure 1.

FACS purification and RNA-Seq results. A, GFP signal in transverse sections of cochlear and utricular HCs from P4 Tg(Pou4f3-Isl1-eGFP) mice. Note the specific GFP expression in cochlear IHCs and OHCs. Supporting cells and other SCs are not labeled. DAPI and fluorescent phalloidin counterstains outlined surrounding areas. B, Dissected tissues used for FACS at P4, shown as combined fluorescence and bright field. Hair cells expressing GFP are green. Cochlear samples included hair cells and supporting cells of the organ of Corti, spiral limbus, basilar membrane, part of the spiral ligament, and spiral ganglion neuronal processes. Utricle sensory epithelium was trimmed away from crista ampullaris, vestibular ganglion, and the roof of the utricle. C, FACS analysis of dissociated sensory epithelia. Gating on GFP⁺ DAPI⁻ cells, coupled to the size of the cells, allowed specific separation of the hair cells from surrounding cells and exclusion of dead cells and doublets. To avoid contamination, only cells exhibiting the highest intensity levels in the GFP channel were harvested as GFP⁺ cells. GFP⁻ cells were those with low GFP fluorescence, as delineated. D, Number of GFP⁺ and GFP⁻ cells harvested by FACS at all ages in cochlea and utricle.

Figure 2.

Tissue and age differences in expression. A, Principal component analysis plot of all 16 samples (light colors for utricle and dark for cochlea) for all 20,207 genes represented in the NCBI build 37/mm9 mouse genome assembly. PC1 correlated with cell type: GFP⁻ (purple) to the left and GFP⁺ (green) to the right, and PC2 correlated approximately with age. Most points are expression from Pou4f3-eGFP mice; those denoted as -T1 and -T2 are from mice expressing tdTomato under control of Gfi1. B, C, Analysis of the HTS results. Differential expression in GFP⁺ and GFP⁻ samples was assessed using the average number of reads at all four ages. Only genes with a minimum of 15 total reads across all samples were considered. There were 5364 enriched in hair cells (GFP⁺/GFP⁻ > 2), 3230 genes were enriched in surrounding cells (GFP⁺/GFP⁻ < 0.5), and 9539 genes were nonspecific (0.5 < GFP⁺/GFP⁻ < 2). D, Differential mRNA expression in postnatal utricular and cochlear HC samples. Genes included in this plot were at least four-fold enriched in HCs compared with SCs in either utricle or cochlea (or both). HC genes preferentially expressed (>5×) are shown in light green (utricle) or dark green (cochlea); not preferentially expressed are black.

Figure 3.

Representative HTS expression profiles for Slc26a5 (prestin), Atoh1, Tmc1, Gfi1, Fscn2, Pcdh15, Lhfpl5, Chrna9, Coch, and Wnt5a. Histograms display the normalized number of reads in hair cells (green) and surrounding cells (purple) in samples from the cochlea (co; dark colors) and utricle (ut; light colors) at ages E16, P0, P4, and P7. Indicated for each gene are the FC representing the GFP⁺/GFP⁻ counts ratio and the multiple test adjusted FDR determined by the Benjamini–Hochberg procedure.

Figure 4.

Validation of the HTS results. A, HTS expression profiles at all ages of key proteins of the Notch pathway. The Notch receptor genes, Notch1, 2, and 3, as well as the downstream effector, Hes1, were mostly expressed in surrounding cells (purple). Hes5 and Jag1 had nonspecific expression (black). The ligand-encoding genes Jag2, Dll1, Dll3, and Dll4, were enriched in hair cells (green). B, HTS results (blue) compared with qPCR (gray). The average counts and qPCR results in P0, P4, and P7 cochlea samples relative to the E16 sample were plotted as fold change.

Figure 5.

Deafness gene distribution. A, Expressed genes (total reads > 15) ranked in order of the GFP⁺/GFP⁻ read ratio at all ages. Green indicates the HC-enriched genes (fold change > 2), purple the genes enriched in SCs (fold change < 0.5), and black the nonspecific genes (NS). Known deafness genes are indicated in each category. B, Histogram showing the distribution of deafness genes relative to their GFP⁺/GFP⁻ rank at all ages. Each bin represents 1000 genes. The majority of deafness genes previously shown to be expressed in hair cells was found in the top 2000 genes, and deafness genes expressed in surrounding cells were found within the last 3000 genes.

Figure 6.

Spatial and temporal expression patterns of genes by hierarchical clustering. Heat maps are shown for selected tree branches clustered by spatial and temporal expression patterns; red represents above-average expression levels and blue below-average levels. Each row represents a gene, and each column a sample. Plotted below each heat map are the average standardized gene expression values across samples for that cluster. Examples of genes in the cluster are listed. The GFP⁺ samples are shown in green and the GFP⁻ samples in purple. Clusters represent embryonic and postnatal hair cell-enriched genes (A), postnatal hair cell-enriched genes (B), cochlear hair cell-enriched genes (C), utricular hair cell-enriched genes (D), and surrounding cell-enriched genes (E). See also Tables 1 and 2. Coch, cochlea; Ut, utricle.

Figure 7.

Validation by in situ hybridization of cell specificity predicted from hierarchical clustering. Shown for each gene are the normalized HTS read counts for each condition (left), a schematic of the predicted cell specificity based on these counts (middle), and in situ hybridization in cochlear and vestibular sensory epithelia (right). Scale bars: 20 μm. A, Grp was expressed in both cochlear and saccular hair cells at E18. B, Grxcr2 was expressed in both cochlear and utricular hair cells at P6. C, D, Strip2 and Lmod3 were detected only in P6 OHCs. E, Dnah5 expression was restricted to utricular hair cells at P6. F, Emcn, predicted to be in cochlear surrounding cells, was found mostly in basilar membrane at P6. G, Ptgds, predicted to be mainly in vestibular surrounding cells, was found adjacent to the sensory epithelium of the ampulla but also in the stria vascularis of the cochlea at E18. Co, cochlea; ut, utricle.

Figure 8.

Validation by immunocytochemistry of predicted cell specificity at P6. Normalized HTS read counts for each condition, predicted cell specificity, and antibody labeling in cochlear and vestibular sensory epithelia. Scale bars: 20 μm. A, Lmod3 was found mostly in cochlear but not utricular hair cells; within the cochlea, it was expressed mainly in OHCs. B, Lmod1 was found in all cochlear and utricular hair cells and was located throughout the cell bodies. C, Ldb3 was found in cochlear and utricular hair cells; in the cochlea, it was expressed mainly in IHCs. LDB3 immunoreactivity was localized to the zonula adherens region. Co, cochlea; ut, utricle.

Figure 9.

Enrichment of gene ontology categories expressed by HCs at E16 (gray), P0 (purple), P4 (blue), and P7 (red). The y-axis represents the enrichment (observed number of genes/expected).