Tympanic border cells are Wnt-responsive and can act as progenitors for postnatal mouse cochlear cells

Abstract

Permanent hearing loss is caused by the irreversible damage of cochlear sensory hair cells and nonsensory supporting cells. In the postnatal cochlea, the sensory epithelium is terminally differentiated, whereas tympanic border cells (TBCs) beneath the sensory epithelium are proliferative. The functions of TBCs are poorly characterized. Using an Axin2(lacZ) Wnt reporter mouse, we found transient but robust Wnt signaling and proliferation in TBCs during the first 3 postnatal weeks, when the number of TBCs decreases. In vivo lineage tracing shows that a subset of hair cells and supporting cells is derived postnatally from Axin2-expressing TBCs. In cochlear explants, Wnt agonists stimulated the proliferation of TBCs, whereas Wnt inhibitors suppressed it. In addition, purified Axin2(lacZ) cells were clonogenic and self-renewing in culture in a Wnt-dependent manner, and were able to differentiate into hair cell-like and supporting cell-like cells. Taken together, our data indicate that Axin2-positive TBCs are Wnt responsive and can act as precursors to sensory epithelial cells in the postnatal cochlea.

Fig. 1.

Axin2 expression in tympanic border cells and their lineage origin. (A) Cryosection of X-gal-stained P3 Axin2^lacZ/+ cochlea demonstrating intense LacZ (Axin2) expression among tympanic border cells (TBCs) beneath the basilar membrane (dotted line). Myosin 7a marks outer and inner hair cells (OHCs and IHCs). (B) The location of TBCs in the cochlea in relation to the sensory epithelium (SE). (C) TBCs are proliferative and express Ki67. (D,E) Expression of jagged 1 (Jag1) and Sox2 was restricted to SE-supporting cells and absent in TBCs. (F-I) Anti-β-galactosidase antibody was used to detect Axin2-lacZ-positive TBCs (asterisks) in the P1 Axin2^lacZ/+ cochlea. Both epithelial (cytokeratin and E-cadherin) and mesenchymal markers [vimentin (Vim) and fibronectin] were absent in TBCs. (J) Whole-mount cochleae from P1 Pax2-Cre;R26R^mTmG/+ demonstrating traced mGFP-positive cells in the SE but not in TBCs. (J′) Reconstructed cross-section of z-stack images. Asterisk indicates TBCs. BC, Boettcher cells; CC, Claudius cells; HEC, Hensen’s cells; DC, Deiters’ cells; HP, habenula perforata; LER, lesser epithelial ridge; GER, greater epithelial ridge. Scale bars: 25 μm in A,J,J′; 50 μm in C-I.

Fig. 2.

Proliferation and Wnt signaling decreased with age in the postnatal cochlea. (A-C) Cryosections of cochleae from mice administered EdU once daily for 2 days. EdU-labeled TBCs decreased from P3 to P9, and further decreased from P9 to P23. (D) Quantification of EdU-positive TBCs (n=3-10). (E) The number of TBCs (between the lateral cochlear wall and habenula perforata) decreases with age, n=6 for each age. (F) Using RT-PCR and qPCR, cochleae demonstrated a significant decrease in the Wnt target genes (Axin2 and Sp5) (P<0.005 for both) between P2 and P15; experiments were carried out in triplicate. Levels of the hair cell marker (Brn3.1) remained constant (P=0.24). Levels of the supporting cell marker (Sox2) decreased (P<0.05) and those of the outer hair cell gene Prestin increased (P<0.001). (G-J) Axin2^lacZ expression in the postnatal cochlea decreased with age. Scale bars: 25 μm in A-C; 200 μm in G-J. Data are mean±s.d. Asterisk indicates statistical significance.

Fig. 3.

Lineage tracing of Axin2-positive tympanic border cells (TBC). (A-C) Schematics depicting the dynamic cytoarchitecture of the P5, P8 and P15 organ of Corti: formation of the tunnel of Corti and the inner sulcus (IS); resorption of the GER; and a decrease in TBC number. (D-F) Tamoxifen administration to P1 or P3 Axin2^CreERT2/+; R26R^mTmG/+ mice resulted in selective mGFP labeling of Axin2-positive cells. Two days post-injection (DPI), TBCs became mGFP labeled, with rare mGFP-positive medial GER cells seen. There is a robust increase in the number of mGFP-positive cells in the SE with longer post-injection periods, whereas TBCs remained mGFP positive (Table 1). (G) Corn oil alone failed to induce mGFP reporter activity. (H,I) A subset of each supporting cell subtype was mGFP positive, whereas mGFP-labeled TBCs increased in number from 59.3±10.7% at P3 to 89.7±6.4% at P8 (P<0.001, n=3-4). (J,K) mGFP-positive inner and outer hair cells expressed myosin 7a and exhibited stereocilia-like projections on the apical surface. (L-L″″) 3D-reconstructed images of traced cochlea (7 DPI). Insets in L′ depict higher-magnification views of traced OHC and IHC. L″ shows image taken at the level of hair cells’ cell bodies. L″′ and L″″ are reconstructed images. (M-O) EdU labeling shows that TBCs, and not organ of Corti cells, proliferated between P0 and P2. Six and 13 days after injection, EdU-labeled cells were found within the organ of Corti. Rare EdU-positive, GFP-positive cells were seen in the SE (supplementary material Fig. S3M-O). BC, Boettcher cells; CC, Claudius cells; PC, pillar cells; IPC, inner phalangeal cells; GER, greater epithelial ridge; LER, lesser epithelial ridge; IS, inner sulcus; OS, outer sulcus; SL, spiral limbus; SG, spiral ganglia; OC, organ of Corti; single asterisk indicates IHC; double asterisks indicate TBCs. Scale bars: 25 μm in D-J,M-O; 15 μm in K-L″″.

Fig. 4.

Purified Axin2^hi cells display robust proliferative capacity. (A) P0-P2 Axin2^lacZ/+ cochleae were dissociated and labeled with propidium iodide (PI) and CUG to allow for isolation of PI-negative (viable), CUG-positive cells. The flow cytometry plot depicts gates used for isolation of the Axin2^hi (red) and Axin2^lo (green) cells. (B-H) Isolated CUG-positive cells robustly expressed the Wnt target genes Axin2 and Sp5, and not the hair cell marker Brn3.1 or the supporting cell markers p27^Kip1 and Sox2 (P<0.01 for all). Experiments were carried out in triplicate. Immunostaining of CUG-positive cells shows 93.4±0.6% β-gal-positive, 74.9±2.6% Brn4-positive, 0.0% myosin 7a-positive hair cells (0.0%) and rare (<0.1%) (Sox2-, jagged 1-, p27^Kip1-, E-cadherin or Prox1-positive) supporting cells. (I,J) After 10 days in vitro, Axin2^hi cells formed cytokeratin-positive colonies more frequently than Axin2^lo cells (88.2±19.4 versus 22.0±6.2, P<0.0001, n=5). (K-M″) Axin2^hi cells isolated from both Actin-GFP-positive Axin2^lacZ/+ cochleae and Actin-GFP^-negative Axin2^lacZ/+ cochleae were mixed 1:1 and cultured. After 10 days, 94% of cytokeratin-positive colonies were monochromatic, suggesting Axin2^hi colonies were highly clonal (n=3). Scale bars: 25 μm. Data are mean±s.d. Asterisks indicate statistical significance.

Fig. 5.

Purified Axin2^hi cells acquire hair cell and supporting cell phenotypes. (A-D′) Axin2^hi cells formed colonies uniformly expressed cytokeratin and E-cadherin, whereas 77.5±3.2%, 74.3±3.5% and 15.1±3.2% of all cytokeratin-positive colonies expressed Sox2, jagged 1 and Prox1, respectively (n=3 for each, median=22 Sox2-positive cells, 22 jagged 1-positive cells and 4 Prox1-positive cells per colony). (E-E″′) In the presence of EdU, myosin 7a-positive, EdU-positive hair cell-like cells were observed on cytokeratin-positive colonies (arrow) (n=4). (F) Myosin 7a-positive hair cells were observed more frequently among Axin2^hi (median indicates one myosin 7a-positive cell per colony) than Axin2^lo colonies (P<0.005, n=4). (G) All myosin 7a-positive hair cell-like cells also expressed parvalbumin. (H-H″′) A polarized pattern of filamentous actin on the hair cell-like cells was observed. (I) Two parvalbumin-positive hair cell-like cells with espin expression localized to one pole of the cell. (J,K) Cytokeratin-positive colonies containing actively proliferating Ki67-labeled cells (arrowheads) were noted more commonly in Axin2^hi than in the Axin2^lo colonies (P<0.001, n=3). (L) Within each Axin2^hi colony, a subset of cells expressed the M-phase marker phosphohistone 3 (PH3). (M) There were more Ki67-positive cells in Axin2^hi than Axin2^lo colonies (P<0.0001, n=3). Scale bars: 25 μm in A-E″′,I,L; 50 μm in G,J; 5 μm in H-H″′. Data are mean±s.d. Asterisks indicate statistical significance.

Fig. 6.

Wnt3a protein acts as a proliferation stimulant on Axin2-positive tympanic border cells and their descendants. (A-E) When cultured with Wnt3a for 10 days, only Axin2^hi colonies significantly upregulated Ki67 expression (P<0.05). Conversely, the soluble Wnt antagonist Fz8CRD effectively suppressed Ki67-positive colonies in comparison with IgG-only control (P<0.005, n=3-5). (F,G) Wnt3a-treated Axin2^hi colonies robustly expressed Axin2 when compared with vehicle-only control. (H-J) Axin2^hi and Axin2^lo cells from Actin-DsRed-positive Axin2^lacZ/+ cochleae were cultured and DsRed-positive colonies were examined. Axin2^hi cells had higher colony forming capacity than Axin2^lo cells (P<0.01, one-way ANOVA). Wnt3a treatment significantly increased the Axin2^hi and not the Axin2^lo colonies (P<0.05, one-way ANOVA, n=3). (K-L″′) Third generation colonies from Wnt3a-treated Axin2^hi cells expressed myosin 7a and Sox2 (supplementary material Fig. S5F). Scale bars: 25 μm in A-D,F,G,K-L″′; 50 μm in H-I′. Data are mean±s.d. Asterisks indicate statistical significance.

Fig. 7.

Wnt activity modulates proliferation and Axin2 expression in cochlear explants. (A-D) 3D-reconstructed images show that Wnt3a and R-spondin 1 significantly (P<0.001 for both, n=3) and selectively increased EdU labeling among TBCs beneath the sensory epithelium in cultured wild-type cochleae. (E-G) Wnt inhibition with Fz8CRD significantly reduced EdU-labeled TBCs (P<0.0001 for both the 36- and 72-hour treatments, n=3-4) in comparison with IgG control. (H,I) Fz8CRD suppressed LacZ (Axin2) expression in cultured Axin2^lacZ/+ cochleae. (J) ELISA studies show that Wnt3a significantly enhanced β-galactosidase activity in cultured Axin2^lacZ/+ cochleae (P<0.01), whereas Fz8CRD reduced it (P<0.005, n=3). Scale bars: 15 μm in A-C,E,F; 100 μm in H,I. Data are mean±s.d. Asterisks indicate statistical significance.