signaling in prosensory domain specification and subsequent hair cell differentiation in the developing cochlea

Alain Dabdouba,1,2,3, Chandrakala Puligillaa,1,3, Jennifer M. Jonesa, Bernd Fritzschb, Kathryn S. E. Cheahc, Larysa H. Pevnyd, and Matthew W. Kelleya

aSection on Developmental Neuroscience, NIDCD, National Institutes of Health, Bethesda, MD 20892; bDepartment of Biological Sciences, University of Iowa, Iowa City, IA 52242; cDepartment of Biochemistry and Centre for Reproduction, Development and Growth, University of Hong Kong, Hong Kong; and dDepartment of Genetics, University of North Carolina, Chapel Hill, NC 27599

Edited by Gail R. Martin, University of California, San Francisco, San Francisco, CA, and approved October 2, 2008 (received for review August 18, 2008) Sox2 is a high-mobility that is one of the earliest In this study, we elucidated the developmental role of Sox2 markers of developing inner ear prosensory domains. In humans, through gain- and loss-of function studies in addition to identifica- mutations in SOX2 cause sensorineural hearing loss and a loss of tion of both upstream regulators and downstream targets. The function study in mice showed that Sox2 is required for prosensory results of these experiments enabled us to establish a Sox2 signaling formation in the cochlea. However, the specific roles of Sox2 have not cascade that includes the Notch signaling pathway upstream of Sox2 been determined. Here we illustrate a dynamic role of Sox2 as an early and the transcription factor Prox1 as a downstream target of Sox2 permissive factor in prosensory domain formation followed by a in the cochlea. Moreover, we present in vitro evidence consistent mutually antagonistic relationship with Atoh1, a bHLH nec- with a reciprocal antagonistic relationship between Sox2 and essary for hair cell development. We demonstrate that decreased Atoh1, a bHLH protein, in which Sox2 expression represses Atoh1- levels of Sox2 result in precocious hair cell differentiation and an over induced hair cell formation while expression of Atoh1 leads to production of inner hair cells and that these effects are likely medi- down-regulation of Sox2. These conclusions are supported by the ated through an antagonistic interaction between Sox2 and the bHLH demonstration of ectopic inner hair cells in cochleae from Sox2 molecule Atoh1. Using gain- and loss-of-function experiments we hypomorphic mice, suggesting that the level of Sox2 determines, at provide evidence for the molecular pathway responsible for the least in part, the number of cells that develop as inner hair cells. formation of the cochlear prosensory domain. Sox2 expression is These results suggest that Sox2 expression is required for specifi- promoted by Notch signaling and Prox1, a transcription cation of prosensory cells but that subsequent down-regulation of factor, is a downstream target of Sox2. These results demonstrate Sox2 is similarly required for a subset of those cells to differentiate crucial and diverse roles for Sox2 in the development, specification, as hair cells. and maintenance of sensory cells within the cochlea. Results Sox2 Is Expressed in the Prosensory Region of the Cochlea. As a first development ͉ organ of Corti ͉ inner ear ͉ HMG box ͉ bHLH step toward understanding the role of Sox2, we examined its expression in the developing cochlea. Within the cochlear duct at he sensory epithelium of the mammalian cochlea (the organ of E12.5, Sox2 is expressed in a band of cells located in the medial half TCorti) develops from a pool of prosensory cells derived from of the duct that appear to correspond with the prosensory domain, the ventral region of the otocyst. Proper development of the cochlea the population of cells that will give rise to the organ of Corti (Fig. requires that cochlear progenitor cells transition through states of 1A). This was confirmed by comparing the expression of Sox2 with developmental competence, coordinated cell cycle exit, and cell p27kip1, a known marker of the prosensory domain (4) at E14 fate specification and differentiation to generate the distinctly fated (supporting information (SI) Fig. S1). At E16, the band of Sox2 cell populations within the highly ordered mosaic of the organ of expression within the cochlear duct correlates with the position of Corti (1). The signal transduction pathways that coordinate co- the developing organ of Corti (Fig. 1B). In addition, Sox2 is chlear prosensory specification are only beginning to be identified. expressed in cells in Kolliker’s organ (KO) located adjacent to the Moreover, since these signal transduction pathways are unlikely to medial edge of the developing sensory epithelium. Within the organ be linear cascades, it will also be necessary to determine how of Corti, Sox2 expression begins to be down-regulated in cells that different pathways are organized into complex signaling networks will differentiate as hair cells while it is maintained in cells that will that ultimately generate precise and unique responses within indi- develop as supporting cells—inner phalangeal cells, inner pillar vidual cochlear prosensory cells. cells, outer pillar cells, Deiters’ cells, and Hensens’ cells (Fig. 1B). Sox (SRY related HMG box) are a group of transcrip- To confirm that down-regulation of Sox2 correlates with hair cell tion factors that regulate diverse developmental processes; for instance, Sox2 is a universal marker of stem cells and is also expressed in neural progenitor cells at different stages of central Author contributions: A.D., C.P., and M.W.K. designed research; A.D., C.P., and J.M.J. performed research; A.D., C.P., B.F., K.S.E.C., and L.H.P. contributed new reagents/analytic nervous system development. Sox2, along with Sox1 and Sox3, tools; A.D. and C.P. analyzed data; and A.D., C.P., and M.W.K. wrote the paper. comprise the SoxB1 group. Members of this group are thought to The authors declare no conflict of interest. maintain neural precursor cells in a progenitor state by inhibiting This article is a PNAS Direct Submission. bHLH-mediated neuronal differentiation (2). Reciprocally, bHLH 1A.D. and C.P. contributed equally to this work. proteins must suppress expression and activity of SoxB1 proteins to 2Present address: Department of Surgery, Division of Otolaryngology, School of Medicine, induce cellular differentiation. A recent loss-of-function study University of California San Diego, 9500 Gilman Drive, 0666, La Jolla, CA 92093. demonstrated that Sox2 is required for development of sensory 3To whom correspondence may be addressed. E-mail: [email protected] or epithelia, including the organ of Corti, within the inner ear (3). [email protected]. However, despite its absolute necessity for the formation of inner This article contains supporting information online at www.pnas.org/cgi/content/full/ ear sensory epithelia, the specific role of Sox2 remains mostly 0808175105/DCSupplemental. unknown. © 2008 by The National Academy of Sciences of the USA

18396–18401 ͉ PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0808175105 Downloaded by guest on September 28, 2021 A B C

D E F

Fig. 1. Sox2 is expressed in the prosensory domain and inhibits hair cell formation. (A-D) Immunohistochemistry using anti-Sox2 (red in A,B, and D and brown in C) and actin (green in A, B, and D) on mid-modiolar cross-sections of WT cochleae shows Sox2 expression at E12.5, E16, and P0. Sox2 is broadly expressed in the prosensory cells at E12.5 (A); however, by E16 Sox2 levels are downregulated in cells that will subsequently acquire a hair cell fate (B, arrows). (C) Co-localization of Sox2 (brown) and Atoh1 (␤-gal staining in blue) in the organ of Corti of Atoh1LacZ/ϩ mice at E16.5 demonstrates downregulation of Sox2 in Atoh1-positive hair cells. (D) By P0 the expression of Sox2 is restricted to supporting cells (Inset: organ of Corti) with weak expression in a subset of cells within KO. Sox2 overexpression inhibits hair cell formation. (E) Low-magnification confocal image of a cochlear explant culture transfected with Sox2.nucEGFP. Explants are comprised of three regions, KO, the SE, and the LER, see Experimental Procedures for details. Transfected cells are typically present in all three regions (arrows in KO and LER, and arrowhead in SE). (F) High-magnification confocal image of a cochlear prosensory cell transfected with Sox2.nucEGFP (arrowhead). The transfected cell is negative for the hair cell marker Myo6 (red); (Inset) Z-stack confocal cross-section of the cell illustrated in F. Although the cell is located in the hair cell layer, it is negative for Myo6. KO, Kolliker’s organ; LER, lesser epithelial ridge; SE, sensory epithelium; OC, organ of Corti; IHC, inner hair cell; O1-O3, outer hair cells; IPh, inner phalangeal cell; IP, inner pillar cell; OP, outer pillar cell; D1-D3, Deiters’ cells; HeC, Hensens’ cells. (Scale bars, A, B, D,20␮m; C,10␮m; E,50␮m; F,10␮m.)

differentiation, expression of Sox2, and the bHLH transcription investigate whether a similar relationship exists between Sox2 and factor, Atoh1 was compared at E16.5. Sox2 (brown) in Atoh1- Atoh1. As a first step, P19 embryonic carcinoma cells, which express positive hair cells (blue) is downregulated (Fig. 1C). At P0, expres- Sox2 endogenously (Fig. 2 A and B), were transfected with sion of Sox2 is restricted to supporting cells and to a group of cells Atoh1.EGFP containing an IRES sequence (Fig. 2C) and assayed within KO (Fig. 1D) (5). for expression of Sox2 after 24 h. Transfected cells were consistently Sox2-negative suggesting that Atoh1 antagonizes Sox2 expression

Forced Expression of Sox2 Inhibits Hair Cell Formation. Previous (97% of Atoh1-transfected cells were negative for Sox2 expression; BIOLOGY

results had demonstrated that Sox2 expression is necessary for n ϭ 4, t ϭ 63) (Fig. 2 A-D). DEVELOPMENTAL development of all of the cell types within the organ of Corti, To examine whether Sox2 antagonizes the effects of Atoh1, cells including mechanosensory hair cells and non-sensory supporting within KO or the lesser epithelial ridge (LER, located lateral to the cells (3). To determine whether the observed down-regulation of sensory epithelium) were transfected with either Atoh1 alone or a Sox2 is necessary for hair cell formation, prosensory cells were combination of Atoh1 and Sox2. Transfection of Atoh1.EGFP transfected with Sox2.nucEGFP at E13 resulting in maintenance of resulted in expression of the hair cell marker Myo6 in 97.2% of expression of Sox2. The expression vector also contained an inter- transfected cells (Fig. 2 E and F and Table S1) (7, 8) while cells nal ribosome entry site (IRES) sequence which resulted in the transfected with Sox2.nucEFGP did not express hair cell markers generation of independent transcripts for nuclear-localized en- (Fig. 2G and Table S1). To confirm that transfection with hanced green fluorescent protein (nuclear EGFP) under the con- Atoh1.EGFP results in the induction of a hair cell fate, transfected trol of the same promoter. Cochlear explants were maintained for cells were also assayed for the expression of a second hair cell 6 days in vitro and cells that maintained expression of Sox2 in the marker, Myosin7a (6) and for the presence of a stereociliary bundle. sensory epithelium were analyzed using expression of Myosin6 Nearly all Atoh1.EGFP-transfected cells expressed Myosin7a and (Myo6) as a marker for hair cell development. While Myo6 is had a stereociliary bundle (Fig. S2). In contrast, only 50.1% of cells expressed in a number of different tissues throughout the body, co-transfected with Atoh1.EGFP and Sox2.nucEGFP developed as within the inner ear, it is expressed in all hair cells and in no other hair cells as compared with 97.2% when Atoh1 was singly trans- cell types (6). Quantification of transfected cells indicated that fected (Fig. 2 H-J and Table S1). These results suggest that even 99.4% of cells transfected with Sox2 in the sensory epithelium were though Sox2 is necessary for the expression of all or most aspects inhibited from developing as hair cells (Fig. 1 E and F and Table S1) of the prosensory domain, Sox2 also acts as an antagonist by suggesting that maintenance of high levels of Sox2 expression inhibiting the ability of Atoh1 to induce hair cell formation. If this inhibits hair cell formation. In contrast, 50.1% of prosensory cells is the case, then reducing, but not eliminating, Sox2 activity might transfected with a control vector expressing only nuclear EGFP lead to premature formation and over-production of hair cells due developed as hair cells (Table S1). Overexpression of Sox2 had no to a decrease in the antagonistic effect on Atoh1 function. To test effect on support cell fate (data not shown) demonstrating a specific this hypothesis, we examined cochleae from hypomorphic inhibition of hair cell fate by Sox2. Sox2EGFP/LP mice which express Sox2 at approximately 30% of normal levels (9). At E15.5, wild-type cochleae contain only a single An Antagonistic Relationship Between Sox2 and Atoh1. Following a row of Myo6 positive inner hair cells (Fig. 2K). This single row of period of co-expression with Atoh1, Sox2 becomes down-regulated inner hair cells extends along the basal half of the cochlear duct. in progenitor cells that will develop as hair cells. The timing of Sox2 This pattern is consistent with the known basal-to-apical gradient down-regulation correlates with increased expression of Atoh1 in of hair cell differentiation. In contrast, in cochleae from E15.5 the same cells. This pattern of expression, along with the demon- Sox2EGFP/LP littermates, a single row of inner hair cells and two rows strated existence of an antagonistic relationship between SoxB1 and of outer hair cells were clearly evident in the basal half of the bHLH transcription factors in the CNS (2), prompted us to cochlear duct (Fig. 2L). Moreover, by E18 a significant increase in

Dabdoub et al. PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 ͉ 18397 Downloaded by guest on September 28, 2021 the number of inner hair cells was observed along the entire length ABC D of the basal-to-apical axis of the cochlea in Sox2EGFP/LP mice (P Ͻ 0.001) (Fig. 2 M-O). The increase in inner hair cell number was not at the expense of support cells as marked my Prox1 or Sox2 (data not shown). No significant change was observed in the length of the sensory epithelium or in the number of outer hair cells. Moreover, E F G the epithelium was continuous along its entire length. A comparison between cochleae from animals with different levels of decreased Sox2 activity demonstrated a dose-dependent effect on the number of additional inner hair cells (Fig. 2 O).

Sox2 Induces Prox1 Expression in the Cochlea. The initial onset and HI J pattern of expression for Sox2 coincide with, but precedes and is slightly larger than, that of Prox1, a homeodomain transcription factor that is expressed in the developing sensory epithelium and also becomes restricted to supporting cell nuclei by P0 (10, 11) (Fig. 3 A and A’). Since expression of Prox1 overlaps with Sox2 in the lateral region of the cochlear duct (Fig. 3B), we speculated that Sox2 might be required for Prox1 expression. To examine this hypothesis, K L expression of Prox1 was determined in cochleae from Sox2Lcc/Lcc (light coat and circling) mice (3). Consistent with this hypothesis, Prox1 expression was absent in cochleae from Sox2Lcc/Lcc mice at both E15.5, a time point just after the normal onset of Prox1 expression (Fig. 3 C and D), and P0 (Fig. S3). To determine whether Sox2 induces Prox1 expression, Sox2 was ectopically expressed in cells within cochlear explants. Forced expression of Sox2.nucEGFP M induced expression of Prox1 in 100% of transfected cells (n ϭ 4, t ϭ 103) located throughout the cochlea including KO (Fig. 3 E-G) and the LER (Fig. 3 H-J). In addition, cells transfected with Sox2.nucEGFP in the developing sensory epithelium show an apparent increase in Prox1 expression (Fig. S4). In contrast, cells transfected with nuc.EGFP alone did not express Prox1 (data not shown). To ascertain if down-regulation of Prox1 is required for hair N cell formation, cochlear prosensory cells were transfected with Prox1.nucEGFP; 95.5% of cells that maintained expression of Prox1 were inhibited from developing as hair cells as assayed by Myo6 (Fig. 3 K and L and Table S1), suggesting that the inhibitory effects of Sox2 could be mediated, at least in part, by Prox1. To test this hypothesis directly, non-prosensory cells located in KO or the LER were co-transfected with Atoh1.EGFP and Prox1.nucEGFP as described previously. Only 4.5% of co-transfected cells developed O as hair cells (Fig. 3M and Table S1), strongly supporting the hypothesis that Prox1 plays a key role in the antagonistic effects of Sox2.

Notch Signaling Regulates Expression of Sox2. Jagged1-mediated activation of the Notch pathway has been shown to be required for specification of most inner ear prosensory domains (12). To de- termine whether expression of Sox2 is dependent on Notch signal- ing, activation of the Notch pathway was inhibited in cochlear explant cultures beginning at E13 by exposure to the ␥-secretase inhibitor DAPT (N-[N-(3,5-difluorophenacetyl)-L-alanyl]-S- phenylglycine t-butyl ester), which blocks Notch signaling. DAPT treatment in cochlear explants leads to a reduction in the expression Fig. 2. An antagonistic relationship between Sox2 and Atoh1. (A-D) P19 cells, of both Sox2 (Fig. 4 A and B) and consequently Prox1 (Fig. 4 C and labeled with DAPI (blue in A), endogenously express Sox2 (red in B). However, the D), suggesting that Notch signaling is required to maintain Sox2 and cell transfected with Atoh1.EGFP (asterisk in A-D) is negative for Sox2 expression Prox1 expression within prosensory cells. However DAPT treat- (B-D). Sox2 antagonizes Atoh1 in vitro.(E-G) Cells within KO or the LER trans- ment also results in a significant increase in the number of hair cells fected with Atoh1.EGFP alone are positive for Myo6 (red) indicating develop- as previously reported (13) (Fig. 4 E and F) suggesting that the loss ment as hair cells (E-F) while Sox2-transfected cells never develop into hair cells (G). (H-J) Cells within the LER that are co-transfected with Sox2.nucEGFP and of Sox2 and Prox1 could be a result of changes in cell fate rather Atoh1.EGFP. One cell is positive for Myo6 (red), while the other is not, illustrating the antagonistic interaction between Sox2 and Atoh1. Early hair cell differenti- ation and formation of extra hair cells in Sox2 hypomorphic cochlea. (K-O) Progressive decrease in the level of Sox2 activity result in an increase in the Cross-sections of the basal turn of the cochlea from Sox2ϩ/ϩ and Sox2EGFP/LP mice number of extra inner hair cells (stars in N). (O) Quantification of extra inner at E15.5. (K) Anti-Myo6 labeling shows the presence of a single inner hair cell in hair cells in Sox2ϩ/ϩ (n ϭ 5), Sox2EGFP/ϩ (n ϭ 5), and Sox2EGFP/LP (n ϭ 3 from 2 the Sox2ϩ/ϩ cochlea (arrowhead). (L) In contrast, the Sox2EGFP/LP cochlea contains mice) littermate cochleae indicates a significant increase in extra inner hair a single inner hair cell (arrowhead) and two outer hair cells (arrows). (M and N) cells in Sox2EGFP/LP cochlea (P Ͻ 0.001). Error bars are S.E.M. IHC, inner hair cell; Analysis of hair cells (labeled with anti-Myo6) in cochleae from E18 littermates. O1-O3, outer hair cells. (Scale bars, A-J,10␮m; K-N,20␮m.)

18398 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0808175105 Dabdoub et al. Downloaded by guest on September 28, 2021 A A’ A B

B

CD C D

EFG

E F

H IJ

G

KL M BIOLOGY

H DEVELOPMENTAL

Fig. 3. Sox2 regulates Prox1 expression. (A and A’) Labeling of Prox1 (red) and actin (green) on a mid-modiolar cross-section from a WT cochlea at P0 shows expression of Prox1 in a subset of support cell nuclei; IP, OP, and D1-D3. (B) Double-immunolabeling using anti-Prox1 (red) and anti-Sox2 (green) in- dicates overlap of Prox1 and Sox2 expression in the lateral region including PC and DC nuclei. (C and D) Prox1 is absent in cochleae from Sox2Lcc/Lcc mice. (C) Anti-Prox1 (red) staining labels nuclei of prosensory cells (circle) at E15.5. (D) In contrast with WT, Prox1 staining is absent in Sox2Lcc/Lcc mutant cochlea at E15.5. DAPI (blue) staining in C and D shows cell nuclei. Sox2 induces ectopic Fig. 4. Notch signaling is required for Sox2 expression. (A-F) Inhibition of expression of Prox1.(E-J) Sox2.nucEGFP transfected in cells within KO (E-G), or Notch signaling using the ␥-secretase inhibitor, DAPT on cochlear explant the LER (H-J) are positive for Prox1 (red). (K and L) Forced expression of Prox1 cultures beginning at E13 results in the reduction of Sox2 (A and B) and Prox1 inhibits hair cell formation. Cochlear prosensory cells were transfected with (C and D) expression. (E and F) In contrast, Myo6 labeling indicates an increase Prox1.nucEGFP.(K) Low-magnification view of confocal image of explant in the number of inner and outer hair cells in DAPT-treated explant cultures. culture illustrating Prox1.nucEGFP transfection (green) in KO and SE (arrow- (G) Forced expression of NICD.EGFP in cells within the LER induces Sox2 heads). (L) High-magnification view of the SE demonstrating that (yellowish red nuclei) expression. (H) Notch1 expression (brown) is present in Prox1.nucEGFP expressing cells (arrowheads) are negative for the hair cell E15.5 Sox2Lcc/Lcc mutant cochleae. (Scale bars, 20 ␮m.) marker Myo6 (red). (M) Co-transfection with Prox1.nucEGFP and Atoh1.EGFP results in inhibition of Atoh1. Despite expression of Atoh1, the cell is negative for Myo6 (red). IHC, inner hair cell; IP, inner pillar cell; OP, outer pillar cell; D1-D3, Deiters’ cell; O1-O3, outer hair cells; IPh, inner phalangeal cell; HeC, not shown). However, no NICD.EGFP-transfected cells were ob- Hensens’ cell; KO, Kolliker’s organ; LER, lesser epithelial ridge; SE, sensory served to express Prox1 (data not shown). These results suggest that epithelium. (Scale bars, A,20␮m; A’-B,10␮m; C-K,20␮m, L,10␮m, M,20␮m.) activation of the Notch signaling pathway acts to induce ectopic expression of Sox2, but that additional factors are required for the subsequent activation of Prox1 in the same cells. This result is also than a direct effect of the Notch pathway on Sox2/Prox1 expression consistent with the observation that Prox1 is only expressed in a in prosensory cells. Therefore, to determine whether activation of subset of prosensory cells. Furthermore, NICD overexpression in Notch is sufficient to induce expression of Sox2 and/or Prox1, the sensory epithelium was inhibitory to hair cell formation (n ϭ 5, Notch1 intracellular domain (NICD) was ectopically expressed in t ϭ 35; data not shown) similar to results obtained for Sox2 cells within KO or the LER. An average of 96.4% of NICD.EGFP- overexpression. To confirm that Notch signaling acts upstream of transfected cells were positive for Sox2 expression (Fig. 4G; n ϭ 3, Sox2, we examined Notch1 expression in Sox2Lcc/Lcc mice at E15.5. t ϭ 51). None of the cells transfected with the control vector As expected, expression of Notch1 is present in these mutants (Fig. expressing EGFP alone were positive for Sox2 (n ϭ 3, t ϭ 47; data 4H). Taken together, our data suggest that while the Notch signal-

Dabdoub et al. PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 ͉ 18399 Downloaded by guest on September 28, 2021 Atoh1 and unknown factors regulate the initiation of Atoh1 ex- A B pression or additional factors are required along with Sox2 to induce Atoh1 expression. However, following the onset of Atoh1 expression, our data demonstrate that Sox2 levels become down- regulated in differentiating hair cells and that an antagonistic interaction between Sox2 and Atoh1 plays a role in this down- regulation. Similar effects have been observed in the CNS where C D SoxB1 generally become down-regulated upon differentia- tion (2) and constitutive expression of SoxB1 genes inhibits differ- entiation (2, 16). In addition, our data indicate that Sox2 inhibits inner hair cell formation in a dose dependent manner, suggesting that the relative ratio of Sox2 to Atoh1 is important for the regulation of inner, and possibly outer, hair cell fate. Consistent with this hypothesis is the demonstration of prema- E F ture hair cell differentiation and the formation of extra inner hair cells in cochleae from Sox2 hypomorphic mice, suggesting that under these circumstances, while sufficient Sox2 is expressed for formation of the prosensory domain, the amount of Sox2 is insufficient for its subsequent role as an Atoh1 antagonist. In addition, while forced expression of Atoh1 alone results in a strong Fig. 5. The prosensory domain is present in Atoh1 mutants. Sox2, Jagged1, induction of expression of the hair cell markers Myosin6 and and Prox1 were used as markers for the prosensory domain. Immunolabeling Myosin7a and the development of a stereociliary bundle in non- was done on apical cochlear sections at E16.5. (A and B) Sox2 immunolabeling sensory cells, co-expression of Sox2 along with Atoh1 significantly ϩ (brown) in Atoh1LacZ/ and Atoh1LacZ/LacZ (mutant) cochleae shows no change reduces the number of transfected cells that express hair cell in Sox2 expression. The X-gal staining (blue) illustrates reporter activity for markers. These data suggest that the capacity of endogenous Atoh1 Atoh1.(C-F) Expression of Jagged1 (red in C and D) and Prox1 (red in E and F) are also present in the Atoh1 mutants. OC, organ of Corti; SG, spiral ganglion. to direct the commitment of hair cells from prosensory cells seems (Scale bars, 20 ␮m.) to be based, at least in part, on its ability to repress Sox2. Furthermore, we demonstrate that Prox1 is a downstream target of Sox2 and that forced expression of Prox1 resulted in the ing pathway is independent and upstream of Sox2, it plays a role in inhibition of hair cell development. Similarly, co-transfection of the regulation of Sox2 expression. However, our results do not Prox1 and Atoh1 resulted in inhibition of the ability of Atoh1 to exclude an additional role for Notch signaling downstream of Sox2. induce expression of the hair cell marker Myo6 in non-sensory cells suggesting that the inhibitory action of Sox2 is mediated through its Specification of the Prosensory Domain Is Independent of Atoh1. The ability to induce Prox1 expression. However, endogenous Prox1 results described above demonstrate the molecular signaling cas- expression in the cochlea is restricted to the lateral domain of the cade that specifies prosensory cells within the organ of Corti. Since sensory epithelium—the outer hair cell region (10, 11)—while data EGFP/LP it has been reported that hair cells and supporting cells are absent from the Sox2 hypomorphic mice indicated that the inner in Atoh1Ϫ/Ϫ cochlea (7, 14), we wanted to examine if the formation hair cell (Prox1-negative) region is most affected. These results of the prosensory domain is affected in these mutants. The prosen- suggest that while Sox2 acts to inhibit Atoh1 through Prox1 in the sory marker Sox2 is expressed at E16.5 in the Atoh1Ϫ/Ϫ cochlea (Fig. outer hair cell region, it also acts, either directly or through an 5 A and B) (3). Since we demonstrated that Notch signaling is unidentified intermediary targets, to inhibit Atoh1 in the inner hair upstream of Sox2, we examined the expression of the Notch ligand, cell region. Jagged1 in the Atoh1Ϫ/Ϫ mice. Jagged1 is expressed in the area that While the factors that determine the specific number of cells that is topographically equivalent to the organ of Corti (Fig. 5 C and D), will develop as hair cells are not known, the relatively invariant number of hair cells within the mammalian cochlea strongly sug- suggesting that at least the initial phase of Notch signaling is gests a highly specific regulatory mechanism. One possible factor unaffected in these mutants. Finally, since we provide evidence that could be the expression of Id (Inhibitors of differentiation) genes Sox2 is upstream of Prox1 we examined the expression of Prox1. which have been shown to antagonize bHLH function and are Prox1 is expressed in the Atoh1Ϫ/Ϫ cochlea at E16.5 similar to its specifically down-regulated as cells begin to develop as hair cells (8). expression in Atoh1ϩ/Ϫ cochlea (Fig. 5 E and F). In fact, the Ϫ/Ϫ These data suggest that down-regulation of Id expression in a subset expression pattern of Prox1 at E16.5 in Atoh1 cochlea is similar of progenitor cells could allow Atoh1 activity to increase sufficiently to its expression in undifferentiated sensory epithelia at E14.5 of to effectively antagonize and down-regulate Sox2, leading to in- WT cochleae (10). These data suggest that the prosensory domain duction of hair cell formation. in the cochlea is specified independent of Atoh1 and the Notch- The working definition of the cochlear prosensory domain is a Sox2-Prox1 signaling cascade is intact in the Atoh1 mutants. population of cells with the unique ability to develop as either hair cells or supporting cells. However, the factors that specify this Discussion domain have not been determined. Here we show that the prosen- Sox2 is one of the earliest transcription factors expressed in the sory domain, marked to differing degrees by Jagged1, Sox2, and prosensory domain of the otocyst. Mutations in SOX2 lead to Prox1, forms normally in the absence of Atoh1. These results are sensorineural deafness in humans (15) and to deafness and absence consistent with the suggestion that Atoh1 acts as a pro-hair cell of sensory structures in the inner ears of mice (3). We show here rather than a prosensory (7). In contrast, both Jagged1 and that the role of Sox2 is complex and includes both inductive and Sox2 are expressed in patterns that are consistent with prosensory antagonistic effects. Our results suggest that the initial expression specification and disruption of either results in the absence of most of Sox2 plays a direct role in establishing the prosensory domain or all prosensory cells (3, 12). Moreover, Sox2 expression is within the cochlea and that Sox2 is permissive for the development downregulated in the cochleae of Jagged1 conditional mutants (12), of hair cells. The inability of Sox2, or other members of its signaling while Jagged1 expression persists in Sox2Lcc/Lcc mice (A. Pelling and cascade (Notch signaling, and Prox1), to induce Atoh1 expression K.S.E.C., unpublished data), suggesting that Jagged1 acts upstream suggests that either Sox2 functions in a separate pathway from of Sox2. This conclusion is supported by our results demonstrating

18400 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0808175105 Dabdoub et al. Downloaded by guest on September 28, 2021 that Notch signaling is upstream of Sox2. These results, along with known to play early roles in inner ear development (22–24), also the results of forced expression of Notch-ICD in chick inner ear (17), converge to activate Sox2 expression as they do in developing neural suggest that activation of one of the Notch receptors acts to induce plate (25). prosensory identity. Here we demonstrate that this probably occurs through induction of Sox2. This is in contrast to the eye and Experimental Procedures neocortex, where it has been shown that Sox2 appears to act Organotypic Explant Cochlear Cultures. Cochlear cultures from E13 embryos upstream of Notch signaling (9, 18). However, the demonstration were established as described previously (26) (see SI Methods for more details). To that inhibition of Notch signaling within an existing prosensory inhibit notch signaling, the ␥-secretase inhibitor N-[(3,5-Difluorophenyl)acetyl]- domain leads to down-regulation of Sox2 expression suggests that L-alanyl-2-phenyl]glycine-1,1-dimethylethyl ester (DAPT) was added to explants the role of Notch in Sox2 expression extends beyond the period of at a concentration of 50 ␮M and replenished every day for six days. At the Sox2 induction. Moreover, we hypothesize that the extra hair cells conclusion of each experiment, explants were fixed in 4% paraformaldehyde for observed in Notch1 conditional mutant cochleae (19) might occur, 30 min and analyzed by immunohistochemistry. at least in part, because the loss of Notch1 leads to a decrease in the level of Sox2 within the existing prosensory domain. Since Sox2 acts Cell Cultures. Transfections in P19 cells were carried out using Lipofectamine2000 to antagonize Atoh1, the loss of Sox2 could result in increased reagent (Invitrogen) according to manufacturer’s instructions. Cells were fixed in activity of Atoh1 and, as a result, to an increase in the number of 4% PFA for 20 min and analyzed by immunohistochemistry using indicated cells that develop as hair cells, similar to the phenotype in Sox2 antibodies. Cell nuclei were counterstained with DAPI nuclear stain (Molecular hypomorphic mice. Probes, 1:5000). Our gain-of-function experiments using KO and the LER as Immunohistochemistry. Immunocytochemistry was performed on cochleae and model cell populations to examine the signaling pathways that cochlear explant cultures as described previously (7, 8) (see SI Methods for more specify prosensory cell fate within the otocyst suggest a signaling details). cascade in which Sox2 is expressed in response to Jagged1-mediated activation of Notch, and that Sox2 in turn induces expression of Expression Constructs. See SI Methods for details. Prox1. However, it is important to note that while forced expression of either NICD or Sox2 leads to induction of Sox2 or Prox1 Electroporation. Individual cells in cochlear explants were transfected using respectively, data on the endogenous expression patterns of acti- square-wave electroporation as described previously (8) (See SI Methods for more vated Notch, Sox2, and Prox1 (1) which are not exactly overlapping, details). suggest that additional unidentified modifiers act to modulate the expression of each target. Quantification of Changes in Cell Fate. See SI Methods for details. In summary, the results presented here demonstrate that Sox2 is centrally situated in inner ear development functioning both as a Generation of Sox2Lcc/Lcc, Atoh1lacZ/lacZ, and Sox2LP/EGFP Mice. Sox2Lcc/Lcc (3), node, receiving signals from several inputs, including Notch signal- Atoh1lacZ/lacZ (14), and Sox2LP/EGFP (9) mutants were generated by crossing ing, and as a junction, directing signals to different effectors that Sox2ϩ/Lcc, Atoh1ϩ/LacZ, and Sox2ϩ/EGFP x Sox2ϩ/LP mice respectively. apparently play unique roles in sensory development. Additional

work to identify other molecular components of the Sox2 pathway ACKNOWLEDGMENTS. We thank Drs. A. Kiernan, A. Pelling, and T. Friedman for BIOLOGY reading the manuscript and C. Woods, CW. Kramer, T. Dennison, and Dr. EC. should identify crucial co-factors (20, 21) that determine the DEVELOPMENTAL different molecular effects of Sox2 in different cells and at different Driver for technical assistance. K.S.E.C. was supported by the Research Grants Council of Hong Kong HKU7385/02M, B.F. by National Institutes of Health Grant developmental time points during inner ear development. In ad- R01 DC005590, L.H.P. by National Institutes of Health Grant R01 EYO1861. This dition to the upstream Notch signaling that we describe, it would be work was supported by the National Institute on Deafness and Other Commu- interesting to investigate whether Wnt and/or FGF signaling, both nication Disorders intramural program.

1. Kelley MW (2006) Regulation of cell fate in the sensory epithelia of the inner ear. Nat Rev 15. Hagstrom SA, et al. (2005) SOX2 mutation causes anophthalmia, hearing loss, and brain Neurosci 7:837–849. anomalies. Am J Med Genet A 138:95–98. 2. Bylund M, Andersson E, Novitch BG, Muhr J (2003) Vertebrate neurogenesis is counter- 16. Graham V, Khudyakov J, Ellis P, Pevny L (2003) SOX2 functions to maintain neural pro- acted by Sox1–3 activity. Nat Neurosci 6:1162–1168. genitor identity. Neuron 39:749–765. 3. Kiernan AE, et al. (2005) Sox2 is required for sensory organ development in the mammalian 17. Daudet N, Lewis J (2005) Two contrasting roles for Notch activity in chick inner ear inner ear. Nature 434:1031–1035. development: Specification of prosensory patches and lateral inhibition of hair-cell dif- 4. Chen P, Segil N (1999) p27(Kip1) links cell proliferation to morphogenesis in the developing ferentiation. Development 132:541–551. organ of Corti. Development 126:1581–1590. 18. Bani-Yaghoub M, et al. (2006) Role of Sox2 in the development of the mouse neocortex. 5. Hume CR, Bratt DL, Oesterle EC (2007) Expression of LHX3 and SOX2 during mouse inner Dev Biol 295:52–66. ear development. Gene Expr Patterns 7:798–807. 19. Kiernan AE, Cordes R, Kopan R, Gossler A, Gridley T (2005) The Notch ligands DLL1 and 6. Avraham KB, et al. (1995) The mouse Snell’s waltzer deafness gene encodes an uncon- JAG2 act synergistically to regulate hair cell development in the mammalian inner ear. ventional myosin required for structural integrity of inner hair cells. Nat Genet 11:369–375. Development 132:4353–4362. 7. Woods C, Montcouquiol M, Kelley MW (2004) Math1 regulates development of the 20. Ambrosetti DC, Scholer HR, Dailey L, Basilico C (2000) Modulation of the activity of multiple sensory epithelium in the mammalian cochlea. Nat Neurosci 7:1310–1318. transcriptional activation domains by the DNA binding domains mediates the synergistic 8. Jones JM, Montcouquiol M, Dabdoub A, Woods C, Kelley MW (2006) Inhibitors of differ- action of Sox2 and Oct-3 on the fibroblast growth factor-4 enhancer. J Biol Chem entiation and DNA binding (Ids) regulate Math1 and hair cell formation during the 275:23387–23397. development of the organ of Corti. J Neurosci 26:550–558. 21. Dailey L, Basilico C (2001) Coevolution of HMG domains and homeodomains and the 9. Taranova OV, et al. (2006) SOX2 is a dose-dependent regulator of retinal neural progenitor generation of transcriptional regulation by Sox/POU complexes. J Cell Physiol 186:315–328. competence. Genes Dev 20:1187–1202. 22. Riccomagno MM, Takada S, Epstein DJ (2005) Wnt-dependent regulation of inner ear 10. Bermingham-McDonogh O, et al. (2006) Expression of Prox1 during mouse cochlear morphogenesis is balanced by the opposing and supporting roles of Shh. Genes Dev development. J Comp Neurol 496:172–186. 19:1612–1623. 11. Puligilla C, et al. (2007) Disruption of fibroblast growth factor 3 signaling results 23. Ohyama T, Mohamed OA, Taketo MM, Dufort D, Groves AK (2006) Wnt signals mediate in defects in cellular differentiation, neuronal patterning, and hearing impairment. Dev a fate decision between otic placode and epidermis. Development 133:865–875. Dyn 236:1905–1917. 24. Schimmang T (2007) Expression and functions of FGF ligands during early otic develop- 12. Kiernan AE, Xu J, Gridley T (2006) The Notch ligand JAG1 is required for sensory progenitor ment. Int J Dev Biol 51:473–481. development in the mammalian inner ear. PloS Genet 2:27–38. 25. Takemoto T, Uchikawa M, Kamachi Y, Kondoh H (2006) Convergence of Wnt and FGF- 13. Takebayashi S, et al. (2007) Multiple roles of Notch signaling in cochlear development. Dev signals in the genesis of posterior neural plate through activation of the Sox2 enhancer Biol 307:165–178. N-1. Development 133:297–306. 14. Bermingham NA, et al. (1999) Math1: An essential gene for the generation of inner ear hair 26. Dabdoub A, et al. (2003) Wnt signaling mediates reorientation of outer hair cell stereo- cells. Science 284:1837–1841. ciliary bundles in the mammalian cochlea. Development 130:2375–2384.

Dabdoub et al. PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 ͉ 18401 Downloaded by guest on September 28, 2021