Dynamic gene expression by putative hair-cell progenitors during regeneration in the zebrafish lateral line

Abstract

Hearing loss is most commonly caused by the destruction of mechanosensory hair cells in the ear. This condition is usually permanent: Despite the presence of putative hair-cell progenitors in the cochlea, hair cells are not naturally replenished in adult mammals. Unlike those of the mammalian ear, the progenitor cells of nonmammalian vertebrates can regenerate hair cells throughout life. The basis of this difference remains largely unexplored but may lie in molecular dissimilarities that affect how progenitors respond to hair-cell death. To approach this issue, we analyzed gene expression in hair-cell progenitors of the lateral-line system. We developed a transgenic line of zebrafish that expresses a red fluorescent protein in the presumptive hair-cell progenitors known as mantle cells. Fluorescence-activated cell sorting from the skins of transgenic larvae, followed by microarray-based expression analysis, revealed a constellation of transcripts that are specifically enriched in these cells. Gene expression analysis after hair-cell ablation uncovered a cohort of genes that are differentially regulated early in regeneration, suggesting possible roles in the response of progenitors to hair-cell death. These results provide a resource for studying hair-cell regeneration and the biology of sensory progenitor cells.

Keywords: alkaline phosphatase; auditory; neuromast; supporting cell.

Fig. 1.

Expression of fluorescent proteins in mantle cells of the posterior lateral-line system. (A) Confocal mosaic images of a living alpl:mCherry;Et20 larva at 4 days postfertilization (dpf) demonstrate expression of mCherry (magenta) overlapping with that of GFP (green) in neuromasts and interneuromast cells. (Scale bar: 500 μm.) The same color code applies in C–E. (B) A neuromast comprises at least three cell types: hair cells (green), supporting cells (aqua), and mantle cells (magenta). (C) mCherry expression in alpl:mCherry larvae is limited to a subset of cells at the periphery of the neuromast. The image represents a confocal slice through a living alpl:mCherry; Tg(−8.0cldnb:lynEGFP)zf106 larva, in which all cells of the neuromast express membrane-tethered GFP. (D) An alpl:mCherry;pou4f3:GFP animal expresses mCherry in peripheral cells, but that marker is excluded from hair cells that express membrane-tethered GFP instead. (E) mCherry and GFP have extensively overlapping but not identical expression patterns in mantle cells of alpl:mCherry;Et20 larvae. The arrowheads indicate two mCh⁺, GFP⁻ cells. (Scale bar: C–E, 10 μm.)

Fig. 2.

Isolation of mCherry-expressing mantle cells and GFP-expressing hair cells. (A) alpl:mCherry;pou4f3:GFP larvae at 4 dpf were terminally anesthetized and their skins were removed with fine forceps. The skins were then dissociated into the component cells, which were sorted by flow cytometry. Red cells represent mCh⁺ mantle cells; green cells denote GFP⁺ hair cells; and gray and black cells represent epidermal, fin, and pigment cells. (B) Representative results from FACS demonstrate the efficient discrimination of GFP⁺ (green) and mCh⁺ (red) cells from alpl:mCherry;pou4f3:GFP larval skins. Each sample included ∼50 complete skins. Singly transgenic and nontransgenic controls confirm the selectivity of the gates. (C) In plots of the cells sorted from alpl:mCherry;Et20 larvae, about 20% of mCh⁺ cells shifted along the abscissa owing to the dual expression of mCherry and GFP. mCh⁺/GFP⁻ cells are represented here in magenta, whereas red cells are mCh⁺/GFP⁺. Numbers in plots represent the percentage of total singlet particles within each associated gate.

Fig. 3.

Confirmation by in situ hybridization of molecular markers for mantle cells. (A) As shown in low-magnification micrographs (Left; neuromasts indicated by black arrowheads) and confocal differential-interference-contrast images (Right), fat1a, fat1b, fgfr1a, fndc7, robo3, and tspan1 are expressed in neuromasts of the posterior lateral line. Note that each transcript is most prominent in a subset of cells at the periphery of the neuromast. (Scale bar: Left, 500 μm; Right, 10 μm.) (B) fat1a, fat1b, and robo3 each display a pattern of expression distinct from that of the Et20 transgene. FISH (Left; Fast Red) followed by immunofluorescent labeling of GFP (Middle) in Et20 larvae permits the comparison of each transcript’s expression pattern with transgenic GFP expression in mantle cells. (Right) Nuclei are labeled by DAPI. White arrowheads indicate colocalization of in situ labeling and anti-GFP immunofluorescence. (Scale bar: 10 μm.)

Fig. 4.

Transcriptional response of mantle cells to hair-cell destruction. (A) Extirpation of hair cells by the ototoxic chemical CuSO₄ evokes the dynamic expression of select candidate genes in sorted mCh⁺ cells. Microarray results from samples dissected 1, 3, 5, and 11 h after treatment were normalized to those from untreated control samples (Cont.). (Left) Unsupervised hierarchical clustering of all of the genes whose expression was up- or down-regulated by at least a factor of 2 at some time after treatment, except genes whose expression also changed in NF cells. (Right) Genes of interest were selected on the basis of the degree and significance of differential expression as well as the availability of reliable annotation. The genes listed in blue were primarily down-regulated following hair-cell ablation, whereas those shown in red were principally up-regulated. The genes labeled in black were previously reported to be up-regulated in the sound-damaged zebrafish inner ear (42). (B) In situ hybridization supports the results from microarrays and reverse-transcription qPCR analyses: Both fndc7 and tspan1 are enriched in mantle cells and demonstrate reduced expression following CuSO₄ treatment. The first column displays the results of chromogenic in situ hybridization; the second and third columns compare FISH (Fast Red) with GFP immunoreactivity (green) in Et20 larvae. The merged images in the fourth column include nuclear staining with DAPI (blue). (Scale bars: 10 μm.)