Math1: An Essential Gene for the Generation of Inner Ear Hair Cells

See allHide authors and affiliations

Science  11 Jun 1999:
Vol. 284, Issue 5421, pp. 1837-1841
DOI: 10.1126/science.284.5421.1837


The mammalian inner ear contains the cochlea and vestibular organs, which are responsible for hearing and balance, respectively. The epithelia of these sensory organs contain hair cells that function as mechanoreceptors to transduce sound and head motion. The molecular mechanisms underlying hair cell development and differentiation are poorly understood. Math1, a mouse homolog of theDrosophila proneural gene atonal, is expressed in inner ear sensory epithelia. Embryonic Math1-null mice failed to generate cochlear and vestibular hair cells. This gene is thus required for the genesis of hair cells.

The inner ear initially forms as a thickening of the ectoderm, termed the otic placode, between rhombomeres 5 and 6 in the hindbrain. The otic placode gives rise to neurons of the VIIIth cranial nerve and invaginates to become the otocyst, from which the inner ear will develop. The mature mammalian inner ear comprises one auditory and five vestibular organs: the utricle, the saccule, and three semicircular canals (Fig. 1A). The sensory epithelia of these organs consist of mechanoreceptive hair cells, supporting cells, and nerve endings. Hair cells and supporting cells proliferate and differentiate within the sensory epithelia, with peak mitoses between embryonic day 13 (E13) and E18 in mice (1). Although several genes have been implicated in the development of the inner ear (2), none have been shown to be required specifically for the genesis of hair cells.

Figure 1

(left). Inner ear β-Gal staining (blue) of (B, D, E, and G) Math1 +/ β-Galand (C, F, and H)Math1 β-Gal/β-Gal embryos. (A) Schematic of mature inner ear. Blue indicates location of sensory epithelium, cochlea (C), saccule (S), utricle (U), and semicircular canal ampullae (SCA). (B and C) β-Gal staining in the otic vesicle at E12.5 demonstrates that Math1 is expressed before the differentiation of inner ear hair cells; (D) staining of an E14.5 inner ear demonstrates staining throughout the sensory organ epithelia. (E and F) Surface views at the middle turn of an E18.5 cochlea. There is positive staining of the three outer and one inner hair cell rows in (E) and nonspecific staining in (F). (G and H) Sections through the basal region of an E18.5 cochlea: scala vestibuli (SV) and scala tympani (ST). Insets show high magnification of the middle region of the organ of Corti and demonstrate the absence of hair cells. (G) Hair cells are deeply stained; (H) in the null mutant, hair cells are absent. Original magnifications are as follows: (B and C), ×100; (D), ×50; (E and F), ×630; (G and H), ×160; and insets in (G) and (H), ×400.

Hair cells and supporting cells arise from a common progenitor (3), but the mechanisms that specify fate are unknown. During myogenesis and neurogenesis, cell fate specification requires basic helix-loop-helix (bHLH) transcription factors (4, 5).Math1 [mouse atonal (ato) homolog 1], a member of the bHLH family, is expressed in the hindbrain, dorsal spinal cord, and external germinal layer of the cerebellum (6–8). Mice that are heterozygous for a targeted deletion of Math1(Math1 +/–) are viable and appear normal, butMath1-null mice (Math1 –/–) die shortly after birth and lack cerebellar granule neurons (8). To detect subtle Math1 expression patterns missed by RNA in situ hybridization, we generated a secondMath1-null allele (Math1 β-Gal/β-Gal) by replacing theMath1 coding region with β-galactosidase (β-Gal) (9, 10). Math1 β-Gal/β-Gal mice showed all the phenotypic features reported in theMath1 –/– mice (8). β-Gal expression in the cerebellum and dorsal spinal cord was identical to that of Math1 (10). β-Gal was also expressed throughout the prospective sensory epithelia of the otic vesicle at E12.5 of both Math1 +/β-Galand Math1 β-Gal/β-Gal embryos (Fig. 1, B and C). Expression was evident throughout the sensory epithelia of the utricle, saccule, semicircular canals, and cochlea at E14.5 (Fig. 1D). By E18.5, β-Gal was restricted to the hair cells of the developing sensory epithelia in Math1 +/ β-Galmice (Fig. 1, E and G) (11). However,Math1 β-Gal/β-Gal mice retained β-Gal expression in some supporting cell layers of the sensory epithelium. In the cochlea, β-Gal expression was absent in the basal turn (Fig. 1H), greatly decreased in the middle turn (Fig. 1F), and similar to heterozygotes in the apical turn at E18.5 (12), consistent with the basal to apical developmental gradient of the cochlea. Gross morphological analysis of the inner ear of Math1 β-Gal/β-Gal mice at E18.5, 1 day before full gestation, revealed no obvious defects in overall structure compared with wild-type (wt) littermates. The branches of the VIIIth cranial nerve were present and reached the epithelia (12).

To examine the sensory epithelia, we excised utricles and cochleas of wt,Math1+ / β-Gal, andMath1 β-Gal/β-Gal mice and viewed the sensory epithelia with Nomarski optics. Hair bundles were present in both organs of wt mice and heterozygotes but were completely absent inMath1-null littermates. Scanning electron microscopy (SEM) of the cochlea and vestibular organs confirmed the absence of hair bundles in null mice (Fig. 2) (13). To determine whether a lack of hair bundles reflects the absence of hair cells, we examined cross sections of the sensory epithelia of all inner ear organs, using both light microscopy (LM) and transmission electron microscopy (TEM). Sensory epithelia in null mice were considerably thinner, lacked the normal stratification of cell nuclei, and stained uniformly. All of this is consistent with the absence of hair cells (Fig. 3). TEM clearly distinguishes between hair cells and supporting cells in normal utricles: Hair cells have hair bundles, less electron-dense cytoplasm, more apical nuclei, and no secretory granules (14). The sensory epithelia of the null mutants lacked hair cells entirely but did have supporting cells with a normal appearance, including electron-dense cytoplasm, basal nuclei, and secretory granules (Fig. 4, A and B). In the cochlea, TEM analysis demonstrated the loss of hair cells as well (Fig. 4, C and D). In addition, there was no evidence of apoptotic cell death in the sensory epithelium by LM and TEM analysis.

Figure 2

(right).Scanning electron micrographs of E18.5 inner ear sensory epithelia in wt and Math1 β-Gal/β-Gal mice. (A,C, and E) Wild-type epithelia; (B,D, and F) null epithelia. (A and B) The organ of Corti of the cochlea at the middle turn. In wt mice, there are three rows of outer hair cells (1, 2, and 3), one row of inner hair cells (I), all with hair bundles (HB). The tectorial membrane (TM), an accessory structure of the cochlea, is seen at the bottom. Above the hair cells are nonspecialized epithelial cells with rudimentary kinocilia (RK). (B) In null mutants, there are only nonspecialized epithelial cells. (C and D) Crista ampullaris of a vertical semicircular canal. The null crista is similar to the wt crista in overall shape, including the septum (eminentia) cruciatum (EC), but is smaller and lacks hair cells. (E and F) The macula of the utricle. The null macula is smaller than the wt macula and lacks hair cells. Scale bars represent 10 μm in (A) and (B), 50 μm in (C) and (D), and 100 μm in (E) and (F).

Figure 3

Light micrographs of semithin transverse sections of inner ear sensory epithelia in (A, C, and E) wt and (B, D, and F)Math1 β-Gal/β-Gal mice at E18.5. (A) In wt cochlea at the middle turn, three outer hair cells (1, 2, and 3) and one inner hair cell (I) are present, whereas (B) the null mutant cochlea has only epithelial cells in the same region. The tectorial membrane (TM) is seen at the top of (A) and (B). Hair cells (HC) and supporting cells (SC) are present in (C) the wt crista ampullaris and (E) utricular macula, but only supporting cells are present in (D) and (F) null mice. The crista was cut obliquely, accounting for the multiple layers of hair cells in (C). The otolithic membrane (OM), an accessory structure of the utricle, is present in both (E) wt and (F) null mice. Scale bars represent 100 μm in (A) and (B), 50 μm in (C) and (D), and 25 μm in (E) and (F).

Figure 4

(left). Transmission electron micrographs of E18.5 (A and B) utricular macula and (C and D) the middle turn of the cochlea in wt and null Math1 β-Gal/β-Galmice. Hair cells (HC) and supporting cells (SC) are present in (A) wt utricular macula, but only supporting cells are present in (B) null mice. Hair cells have hair bundles (HB); supporting cells have microvilli (MV). Hair cells are less electron dense and have more apical nuclei than supporting cells, but only the latter have secretory granules (SG). Some immature hair cells (IM) are evident in the wt mice. (C) In the wt cochlea, several cell types can be observed: outer hair cells (1, 2, and 3), inner hair cells (I), Deiter's cells (D), Hensen's cells (H), and inner (IP) and outer pillar (OP) cells. Radial afferents (RA) are seen approaching the base of the inner hair cell. The section is slightly oblique, and parts of two adjacent inner hair cells can be seen. (D) In the null cochlea, there are no obvious distinctions between cell types, with the possible exception of the cell at the far left, which may be a Hensen's cell. Both wt and null cochlea have a tectorial membrane (TM), which covers the supporting cells on the right of (C and D). Scale bars represent 10 μm.

The lack of hair cells at E18.5 may be due to a lack of sensory cell progenitors, the inability of progenitors to differentiate into hair cells, or the inability of hair cells to maintain the differentiated state (15, 16). The first possibility is unlikely because progenitors give rise to both hair cells and supporting cells (3), and supporting cells were present inMath1 β-Gal/β-Gal mice. To test the remaining possibilities, we examined the expression of the hair cell–specific markers, myosin VI and calretinin. Both are expressed in differentiating hair cells (before hair bundle formation) and in mature vestibular and auditory hair cells, but not in supporting cells (17–19). Myosin VI is first detected at E13.5 in the developing inner ear (16), whereas calretinin is expressed 2 days later at E15.5 (18). Antibodies to myosin VI and calretinin were used to detect expression of these hair cell markers in wt and Math1 β-Gal/β-Gallittermates. Myosin VI–positive cells were clearly visible in the developing otic vesicles of wt embryos but were completely absent in Math1 β-Gal/β-Galembryos (Fig. 5, A and B). Myosin VI staining was absent in the cochlea of E18.5 Math1-null mutants (12). Calretinin-positive cells were observed in the sensory epithelia of all E15.5 (12), E16.5 (Fig. 5, C through F), and E18.5 (Fig. 5, G and H) wt mice, but they were not observed in the sensory epithelia of Math1 β-Gal/β-Gal mice (Fig. 5, C through H) (20). Given the basal to apical developmental gradient of the cochlea, we examined sections from basal, middle, and apical turns. Hair cells were lacking in all sections; representative sections from the middle turn are shown in Figs. 2 through 4. These data demonstrate that hair cells never developed within the sensory epithelia ofMath1 β-Gal/β-Gal mice. The presence of the tectorial and otolithic membranes [secreted in part by the supporting cells (21, 22)], together with the TEM results, indicates that the sensory epithelia ofMath1 β-Gal/β-Gal mice have functional supporting cells. Preliminary data from TEM and LM cross sections show apparent overcrowding of nuclei in the basal layers of the sensory epithelia of null mice, consistent with an increase in the number of supporting cells in the sensory epithelia (Fig. 4, C and D). This finding, together with the absence of apoptosis, raises the possibility that, in Math1 β-Gal/β-Galmice, there is a fate switch from hair cells to supporting cells.

Figure 5

(right).(A through F) Confocal and (G andH) immunofluorescence microscopy of (A, C, E, and G) wt and (B, D, F, and H) Math1 β-Gal/β-Gal embryos. (A and B) Myosin VI immunostaining of an E13.5 vestibule. (C through F) Calretinin immunostaining of E16.5 utricles. (G and H) Calretinin immunostaining of E18.5 crista ampullaris. The crista is cut at an oblique angle, which accounts for the multiple layers of hair cells in (G). Immunostaining (green) with myosin VI and calretinin is evident in immature and mature hair cells of (A, C, E, and G) wt epithelia but is clearly absent in (B, D, F, and H)Math1 β-Gal/β-Gal epithelia. Boxed areas in (C) and (D) indicate regions magnified in (E) and (F). Sections were counterstained with (A and B) TOTO-3 iodide or (C through F) propidium iodide for confocal microscopy and were counterstained with (G and H) DAPI for immunofluorescent microscopy. Scale bars represent 50 μm in (A) and (B), 100 μm in (C) and (D), and 15 μm in (E) and (F); original magnification is ×200 in (G) and (H).

Math1 is essential for hair cell development in the inner ear. Its expression pattern and in vivo function are akin to those ofMath1's proneural homolog, ato (23).ato is expressed in a ring of epithelial cells within the antennal disc of Drosophila. Some of these cells will subsequently give rise to mechanoreceptors in the Johnston organ, which is necessary for hearing (24) and negative geotaxis (25). Mechanoreceptor progenitor cells are affected in ato mutants (23), whereas only the mechanoreceptors, and not their progenitors, are absent in Math1-null mice.

To our knowledge, Math1 is the first gene shown to be required for the specification of hair cells. We propose thatMath1 acts as a “pro–hair cell gene” in the developing sensory epithelia. Three recent studies provide evidence supporting a lateral inhibition model for the determination of hair cells and supporting cells (26), in which the interplay of Delta and Notch homologs results in the selection of individual hair cells from clusters of competent cells. Such a model entails that the sensory epithelia express a pro–hair cell gene, whose function is essential for hair cell fate specification. Math1 is a likely candidate for such a gene.

Finally, because ectopic expression of ato in the fruit fly (23) [and its homolog Xath1 inXenopus (27)] can recruit epithelial cells into specific neuronal fates, it should be possible to test whether expression of Math1 in inner ear epithelia recruits hair cells. The potential clinical applications of such studies cannot be overlooked, because loss of hair cells through disease, trauma, and aging is a common cause of deafness and vestibular dysfunction.

  • * Present address: Department of Cell and Animal Biology, Alexander Silberman Institute of Life Sciences, Hebrew University of Jerusalem, Givat-Ram, Jerusalem 91904, Israel.

  • To whom correspondence should be addressed. E-mail: hzoghbi{at}


View Abstract

Navigate This Article