The inner ear is derived from the otic placode near the hindbrain. In mice, the otic placode invaginates to form an enclosed otocyst by E10.5. By E11.5, specific molecular markers are expressed in regions that are designated for different cell lineages 24. In particular, transcriptional factor Islet1 (Isl1) marks the sensory and neuronal lineages of the inner ear at early otocyst stages (Fig. 1A) 24. Message RNA for Otoconin90 (Oto90) 25, a major protein component of murine otoconia (composite crystals that overlie the gravity sensory receptors in the vestibular macular organs), is localized to anterior and lateral regions of the otocyst (data not shown). We compared the expression of Hey2 at E11.5 to that of Isl1 and Oto90 (Fig. 1A, B, and data not shown). Hey2 is expressed in the medial region of the otic epithelium (Fig. 1B). Its expression abuts Isl1-expressing and Oto90-expressing domains at its posterior and anterior boundaries, respectively (Fig. 1A, B, and data not shown). Therefore, at early otocyst stages, the expression of Hey2 appears to be complementary to the sensory and neuronal lineages of the inner ear and excluded from cells expressing Oto90.

We further examined the expression of Hey2 in later stages in the cochlea (Figs. 1 and 2). At E14.5, the precursor cells that will give rise to hair cells and supporting cells of the organ of Corti are post-mitotic and are demarcated by the expression of a cyclin-dependent kinase inhibitor, p27/Kip1 (Figs. 1H, 2E) 2. The expression of Isl1 also marks the primordial organ of Corti (Figs. 1C, 2E) 24. At this stage, a Notch target gene Hes1 is expressed in a broad region medial to the sensory region (Figs. 1J, 2E) 1415, while the expression of Hes6 has begun at the medial boundary of the sensory primordium and marks the differentiating inner hair cells (Fig. 1E) 26. The expression of Oto90 is restricted to the roof of the cochlear duct (Fig. 1G). The expression of Hey2 appears in a region overlapping with the developing organ of Corti (Figs. 1D, 1I, 2E). Careful comparisons with different molecular markers in the adjacent sections led us to conclude that the expression domain of Hey2 may be shifted slightly to the medial side of the developing organ of Corti marked by the expression of Isl1 and p27/Kip1 (compare Fig. 1C with 1D, and compare Fig. 1H with 1I). In addition, another Hey gene 23, Hey1, is expressed in the primordial organ of Corti at this stage (Fig. 1F). In contrast to Hey2 and Hey1 and Hes1, Hes5 was not detected in the cochlea at E14.5 (data not shown).

Following the establishment of the post-mitotic primordial organ of Corti, the differentiation of hair cells and supporting cells is initiated near the base of the cochlea at the inner hair cell region 227. By E16.5, the organ of Corti in the basal to medial regions of the cochlea has been patterned into a sensory mosaic consisting of one row of inner and three rows of outer hair cells interdigitated with supporting cells (Fig. 2). From a cross section of the cochlear duct, hair cells are seen at the lumenal layer while the nuclei of supporting cells are localized to the basal layer of the cochlear epithelium (Fig. 2A–D and the diagrams in 2E), and the cytoplasmic phalangeal processes of supporting cells project into the lumenal hair cell layer to separate hair cells from each other. We compared the expression of a basic helix-loop-helix gene, Hes6, in the cochlea at E16.5 to several other molecular markers. As we have previously shown 26, Hes6 is restricted to the hair cells at the lumenal layer at this stage (Fig. 2A). In contrast, Hes5 message RNA is limited to the supporting cell nuclei (Fig. 2B) 14. Similarly, the expression of Hey2 (Fig. 2C) and Hey1 (Fig. 2D) is also seen in the supporting cells. By E17.5, the expression of Hey2 has been down-regulated starting from the base of the cochlea to the apex, while the expression of Hey1 still remains in the supporting cells (data not shown, diagramed in Fig. 2E).

The early onset of Hey2 in the developing inner ear at E11.5 and the broad expression of Hey2 in the cochlea at E14.5 suggest that its initial expression might precede hair cell and supporting cell differentiation in the cochlea. To test this possibility, we examined the expression of Hey2 in Math1^-/- animals 5, and compared its expression with that of Hes6 and Isl1 (Fig. 1D, F). In the absence of Math1, the primordial organ of Corti marked by the expressed of p27/Kip1 is established 27. However, hair cells fail to differentiate and cells within the primordial organ of Corti die 27. Hes6 marks the sensory lineage in the developing inner ear and its expression in cochlear hair cells depends on Math1 26. In Math1^-/- cochleae, Hes6 was not expressed (Fig. 1K). Isl1 is expressed early in the otocyst and its expression in the organ of Corti is independent of Math1 (Fig. 1L) 24. Similar to Isl1, Hey2 was expressed in the otocyst and its expression was maintained in the cochlear epithelium in Math1^-/- animals at E14.5 (Fig. 1M).

In summary, Hey 2 is expressed in the otocyst prior to the formation of the cochlear duct; its onset in the cochlea is independent of Math1; and its expression in the cochlea is dynamic, partially overlapping with Hes1 and Hes5 in the developing cochlea (Fig. 2E). An additional Hey gene, Hey1, is also expressed in an overlapping pattern with Hey2.

To examine whether the dynamic expression of Hey2 in the inner ear is required for inner ear development, we examined the loss-of-function of Hey2 in mice. Hey2^-/- animals die around postnatal day 5 (P5) due to massive postnatal cardiac hypertrophy and isolated ventricular septal defects 282930. We therefore examined inner ears from Hey2^-/- animals from E13.5 to P2. Morphologically, no visible abnormality was observed in inner ears isolated from Hey2^-/- animals at all the stages examined (data not shown). Upon molecular examinations using cell type-specific antibodies, we consistently observed a minor patterning defect in the organ of Corti from Hey2^-/- animals (Figs. 3 and 4). In comparison to the normal three rows of outer hair cells in wild-type littermates (Fig. 3A–C), cochlear whole mount preparations from Hey2^-/- animals stained for hair cell-specific marker Myosin6 showed segments of an additional row of outer hair cells (Figs. 3D–F, 4, mis-patterned OHCs = 43.6 ± 5.6, n = 3). The apparent patterning defect in outer hair cells in Hey2^-/- animals is reminiscent of that observed in Hes5^-/- animals 1415, and the number of additional outer hair cells in Hey2^-/- animals is comparable to that observed in Hes5^-/- (Fig. 3G–I, mis-patterned OHCs = 47.6 ± 9.8, n = 3) animals among littermates (Figs. 3D–I, 4, p = 0.574).

Hey2 belongs to the same bHLH gene sub-family as Hes5, which can function as downstream targets for Notch signaling 31. The similarity in patterning defect observed in Hey2^-/- and Hes5^-/- animals suggests that Hey2 may play a similar role as Hes5 in the organ of Corti. The partially overlapping expression of Hey2 and Hes5 (Fig. 2B, C) during terminal differentiation of the organ of Corti and the weak patterning defect in Hey2^-/- or Hes5^-/- animals further suggested that Hey2 may function redundantly with Hes5 for patterning the organ of Corti. To test this possibility, we crossed animals carrying mutant alleles for both Hey2 and Hes5, and compared Hey2^-/- (Fig. 3D–F), Hes5^-/- (Fig. 3G–I), and Hey2^-/-; Hes5^-/- (Fig. 3J–L) littermates, and quantified the number of mis-patterned outer hair cells in these animals (Fig. 4). We found a statistically significant increase in the numbers of mis-patterned outer hair cells in Hey2^-/-; Hes5^-/- animals (mis-patterned OHCs = 101.3 ± 29.1, n = 3) in comparison to Hey2^-/- (p = 0.028) or Hes5^-/- (p = 0.039) animals (Fig. 4). The effect of genetic inactivation of Hey2 and Hes5 on the patterning of outer hair cells appeared to be additive (Fig. 4).

In addition to the presence of segments of additional row of outer hair cells, we also observed variable amounts of inner hair cell doublets in Hey2^-/- animals (Fig. 3D–F), in contrast to a single row of inner hair cells in control animals (Fig. 3A–C). However, statistic analyses showed that there is no significant increase in the number of inner hair cell doublets in Hey2^-/- animals (IHC doublets = 4 ± 3, n = 3) in comparison to wild-type (IHC doublets = 3 ± 2, n = 3) control littermates (Fig. 4, p = 0.768). Since Hey2 shows a partial overlapping expression with a homologous gene, Hes1, at the onset of terminal differentiation of the organ of Corti (Fig. 1I, J) and the loss of Hes1 results in an increase in inner hair cell doublets and, to a lesser degree, an additional row of outer hair cells (Fig. 3M–O) 1415, we sought to determine whether Hey2 plays a redundant role to Hes1 in development of the organ of Corti.

As reported previously, Hes1^-/- animals show progressively increased lethality after E12.5 31. We found that Hes1^-/-; Hey2^-/- animals have a complete penetrance in early embryonic lethality. All of the Hes1^-/-; Hey2^-/- animals died before E13.5, prior to the terminal differentiation of the organ of Corti. In addition, in littermate crosses between Hes1^+/-; Hey2^+/- double heterozygotes and Hey2^+/- single heterozygotes or within Hes1^+/-; Hey2^+/-double heterozygotes, eight Hes1^+/-; Hey2^-/- embryos were expected, none of which were recovered after E13.5, indicating a strong genetic interaction between Hey2 and Hes1 for early development and survival of mouse embryos.

We were able to recover Hes1^-/- and Hes1^-/-; Hey2^+/- embryos and analyzed organs of Corti isolated from these animals (Figs. 3, 4). In comparison to Hes1^-/- animals that showed frequent inner hair cell doublets (Figs. 3M–O, 4, IHC doublets = 56.7 ± 1.5, n = 3), the loss of a single copy of Hey2 in the Hes1^-/- background (Figs. 3P–R, 4, IHC doublets = 108.3 ± 15.2, n = 3) led to a 2 fold-increase in the number of inner hair cell doublets (Fig. 4, p = 0.0003). Since Hey2^+/- had no more inner hair cell doublets than control samples (Fig. 4, IHC doublets = 1.6 ± 1.4, n = 3), the increase in inner hair cell doublets in Hes1^-/-; Hey2^+/- animals appeared to be synergistic rather than additive. These data together indicate that Hey2 genetically interacts with Hes1 for early development of mouse embryos and for patterning inner hair cells in the organ of Corti.

The mis-patterning of hair cells in Hes1 and Hes5 mutants was observed previously 1415. It was thought that the conversion of supporting cells to hair cells might account for the abnormality observed in Hes1 and Hes5 mutants, as predicted from a simple lateral inhibition model 1415. However, the underlying molecular mechanism has not been tested. We observed enhanced phenotypes in Hes1 and Hey2 and in Hes5 and Hey2 compound mutants (Figs, 3 and 4), and explored the molecular roles of Hes family genes in patterning the organ of Corti.

We examined the cellular architect of the organ of Corti in control and Hey2 compound mutants in whole mount preparations and cross sections of the organ of Corti (Fig. 5). The precise rows of hair cells (Fig. 5A, C, D) are mirrored by corresponding parallel rows of supporting cells in the organ of Corti (Fig. 5B–D) 3233. As reported previously 34 an antibody against a homeodomain transcriptional factor, Prox1, distinctively labeled outer pillar cells (OP) between the inner and outer hair cells, and four rows of supporting cells that alternate with the three rows of outer hair cells (Fig. 5B–D). In Hes5^-/-;Hey2^-/- samples, the ectopic row of outer hair cells (Fig. 5E, G, H) were accompanied by an additional row of supporting cells (Fig. 5F–H). In Hes1^-/-;Hey2^+/- animals, the ectopic outer hair cells were also associated with additional supporting cells (Fig. 5I–N).

Since inner hair cell doublets were observed in Hes1^-/-;Hey2^+/- animals, we also attempted to examine the arrangement of inner hair cells and its surrounding supporting cells in the mutant and wild-type cochleae (Fig. 5I–N). An antibody against p27/Kip1 marks all the supporting cells at this stage (Fig. 5J). However, the supporting cells surrounding inner hair cells labeled by p27/Kip1 at E18.5, the latest surviving stage for Hes1 and Hey2 compound mutants, showed variable arrangements in control cochleae (Fig. 5I, K). We could not determine whether the number of supporting cells surrounding inner hair cells changes accordingly in Hes1 and Hey2 compound mutants (Fig. 5L–N). Nevertheless, it is apparent that the ectopic inner hair cells in Hes1^-/-;Hey2^+/- animals do not contact other inner hair cells and are interdigitated by non-hair cells as shown in whole mount preparations of the mutant organ of Corti (Fig. 3P–R), implying a preserved mosaic arrangement of inner hair cells and supporting cells in Hes1^-/-;Hey2^+/- animals.

The mirror increase in supporting cells in Hes5^-/-;Hey2^-/- and Hes1^-/-;Hey2^+/- animals was not predicted from a simple lateral inhibition model 11, but consistent with an emerging complex role for the Notch signaling pathway. Indeed, the Hes homologous genes Hes1 and Hes5 can regulate sequential stages during neurogenesis in several nervous systems 313536. Initially, they define the domain undergoing neurogenesis and, subsequently, control the number of neural progenitors within the neurogenic domain. Notch signaling can affect cell division and/or cell fate determination to regulate the pool of progenitor cells at both stages 36373839. In the cochlea, mutations in Notch receptor or ligands lead to perturbations in cell division 78. Cell proliferation was observed in Notch or Notch ligand mutants after E14.5, when cells in the organ of Corti have withdrawn from the cell cycle in wild-type animals 8. To determine whether prolonged cell division contributes to mis-patterning and ectopic hair cells and supporting cells in Hey2, Hes1, and Hes5 mutants, we analyzed the incorporation of a nucleotide analog, BrdU, in the organ of Corti from E14.5–E18.5 (Fig. 6, and data not shown).

BrdU was injected into timed-pregnant mice at E14.5, and embryos were collected the same day after three consecutive injections (Fig. 6A–D). In control mice, the organ of Corti was post-mitotic as demonstrated by the lack of BrdU incorporation in the region at E14.5 (Fig. 6A, B). No proliferating cells were detected in the primordial organ of Corti in Hes1^-/-;Hey2^+/- embryos at E14.5 (Fig. 6C, D). Furthermore, to facilitate the identification of proliferating cells in the organ of Corti after E14.5, embryos injected with BrdU at E14.5 for three times were not collected until E18.5, allowing a longer chasing period for the incorporation of BrdU (Fig. 6E–J). Hair cells generated after E14.5 would be readily detected in embryos harvested at E18.5. The control E18.5 embryos had no BrdU⁺ cells in the organ of Corti (Fig. 6E–G). In Hes1^-/-;Hey2^+/- E18.5 embryos, no BrdU⁺ cells were detected (Fig. 6H–J). Similar results were obtained with Hes5 and Hey2 compound mutants (data not shown), indicating that no detectable extension of cell proliferation contributes to the ectopic hair cells and supporting cells in Hes1^-/-;Hey2^+/- (Fig. 6) or Hes5^-/-;Hey2^-/- (data not shown) mutants. Since Hes1^-/-;Hes5^-/- or Hes1^-/-;Hey2^-/- embryos die early in development, we could not examine perturbation of cell division in these mutants.

The animals used are Hes1^+/- (a gift from R. Kageyama) 31, Hes5^+/- (a gift from R. Kageyama) 31, Hey2^+/- 29, Math1/GFP 43, and Math1^+/- 5 mice. Hes1, Hey2 or Hes5, Hey2 compound mutant embryos were obtained from intercrosses of Hes1^+/-; Hey2^+/- or Hes5^+/-; Hey2^+/- doubly heterozygous mice. Hey2^+/- colonies were maintained in C57/B6 background. Hes1^+/- and Hes5^+/- mice in the CD1 background were backcrossed with C57/B6 mice for more than 8 generations for this study. The Hes1 and Hey2, and the Hes5 and Hey2 breeds are of a homogenous black coat color. Math1^-/- animals were generated by crossing Math1^+/- animals. Genotyping were carried out by PCR with tail DNA using the following oligonucleotides: Hes1, 5'-TGG GAT GEG GGA CAT GCG GG-3', 5'-TCA CCT CGT TCA TGA CTC G-3'; 5'-GCA GCG CAT CGC CTT CTA TC-3'; Hes5, 5'-GCT GGG GGC CGC TGG AAG TGG-3', 5'-CCG CTC CGC TCC GCT CGC TAA-3', 5'-GCA GCG CAT CGC CTT CTA TC-3'; and Hey2, 5'-CAC TAA GAA CTA GCG ATC TGG-3', 5' CTC AGG GGA TTT TGA AAG C-3', 5'GCA CGA GAC TAG TGA GAC GTG.

Animal care and use was in accordance with NIH guidelines and was approved by the Animal Care and Use Committee of Emory University and the University of California, Irvine. The morning after the mating of the animals is designated as embryonic day 0.5 (E0.5). The number of animals used for each genotype at each stage is least three and indicated where data are presented.

BrdU (Sigma) was dissolved at 5 mg/ml in 1× PBS and injected at 50 μg per gram of body weight. Mice were injected intraperitioneally three times at 2 hour intervals starting at E14.5 and sacrificed at the stages specified in the text.

Inner ears were dissected from E11.5, E14.5, and E16.5 embryos and immediately fixed in 4 % paraformaldehyde (PFA) in PBS for 1 hour. Fixed inner ears were cryoprotected in 20% sucrose in PBS and embedded in OCT for cryostat sectioning. Sections (12–16 μm) were dried at room temperature for 1 hour and processed for in situ hybridization with digoxigenin-labeled probes 24. Briefly, the inner ear sections were post-fixed in ice-cold 4% PFA for 10 minutes, treated with protease K (20 μg/ml) for 2–3 minutes, rinsed with DEPC-treated PBS, acetylated in 0.1 M TEA (Triethanolamine) freshly-made by adding 125 μl of acetic anhydride to 50 ml 0.1 M TEA, pH7.5, re-fixed in 4% PFA, and dehydrated by a series of EtOH (70%, 95%). The sections were then incubated with a hybridization solution (50% deionized Formamide, 10% Dextran Sulfate, 1× Denhardt's, 0.25 mg yeast tRNA, 0.3 M NaCl, 20 mM Tris-HCl, pH8.0, 5 mM EDTA, 10 mM NaPO4, pH7.2, 1% Sarcosyl) for 1 hour at 55° before incubated with the hybridization solution containing a probe (20 μg/ml) overnight at 55° in a sealed humidifier chamber. After the hybridization, slides were washed in pre-warmed washing solution (50% Formamide, 2×SSC) at 65° for three times, washed once with PBT (PBS, 0.1% Tween-20) at room temperature, blocked with 10% sheep serum in PBT for 1 hour at room temperature, incubated with alkaline phosphatase-conjugated anti-digoxigenin antibody in PBT, 1% sheep serum at 4° overnight. Following antibody incubation, slides were washed in NTMT (0.1 M NaCl, 0.1 M Tris-HCL, pH9.5, 0.05 M MgCl₂, 0.1% Tween-20 containing Levamisole (0.5 mg/ml) for four times, and incubated with BM-purple AP substrate in NTMT without Levamisole for up to one week. Slides were washed in PBS and when appropriate color developed. Probes for Hes1, Hes5, Hey1, Hey2, Oto90, Hes6, and Isl1 were prepared using cDNAs cloned from the cochlea.

Immunostaining of cochlear sections were performed as described previously 24. Primary antibodies used for this study are: anti-BrdU (Chemicon, mouse monoclonal, 1:100); anti-p27/Kip1 (BD Transduction Laboratories, mouse monoclonal, 1:200) anti-Myosin6 (Proteus BioSciences, rabbit polyclonal, 1:200); anti-Prox1 (Chemicon, rabbit polyclonal, 1:1000). Secondary antibodies are: FITC-conjugated donkey anti-rabbit (Jackson Lab, 1:1000); Rhodamine-conjugated donkey anti-goat (Jackson Lab, 1: 1000); Cy5-conjugated donkey anti-mouse (Jackson Lab, 1:1000). For p27/Kip1 and BrdU staining, sections were steamed in 10 mM NaCitrate for 15 minutes prior to being subjected to the staining procedure. For Prox1 staining, the samples were treated with 1% SDS followed by blocking and antibody incubations as described previously 24. For hair cell staining of whole mount preparations of the cochlea, cochleae ducts were opened to expose the developing sensory epithelia prior to the staining procedure 40.

Digital images were captured on an LSM 510 (Carl Zeiss, Germany) confocal laser scanning microscope and an inverted fluorescent microscope (Olympus, Japan). Images for entire cochlear ducts were composed using Photoshop (Adobe System, USA). The number of inner or outer hair cells in each composed cochlea was counted by scoring Myosin6⁺ cells. The most distal tip of the apex region of the cochlea (10% of the length) was excluded from quantification since patterning of hair cells is incomplete in this region, and care was taken to use similarly 90% of the cochlear duct for each sample. Mis-patterned outer hair cells appearing as segments of an additional row of cells and inner hair cell doublets were counted. The number of mis-patterned inner and outer hair cells was directly plotted or the numbers of mis-patterned hair cells were calculated against normally patterned hair cells and plotted. Data collected from each experimental group with 3 samples from different animals are expressed as mean ± s.e.m. Student's t-test was used for statistical analysis.