Temporal and spatial assembly of inner ear hair cell ankle link condensate through phase separation

Abstract

Stereocilia are actin-based cell protrusions of inner ear hair cells and are indispensable for mechanotransduction. Ankle links connect the ankle region of developing stereocilia, playing an essential role in stereocilia development. WHRN, PDZD7, ADGRV1 and USH2A have been identified to form the so-called ankle link complex (ALC); however, the detailed mechanism underlying the temporal emergence and degeneration of ankle links remains elusive. Here we show that WHRN and PDZD7 orchestrate ADGRV1 and USH2A to assemble the ALC through liquid-liquid phase separation (LLPS). Disruption of the ALC multivalency for LLPS largely abolishes the distribution of WHRN at the ankle region of stereocilia. Interestingly, high concentration of ADGRV1 inhibits LLPS, providing a potential mechanism for ALC disassembly. Moreover, certain deafness mutations of ALC genes weaken the multivalent interactions of ALC and impair LLPS. In conclusion, our study demonstrates that LLPS mediates ALC formation, providing essential clues for understanding the pathogenesis of deafness.

Fig. 1. Characterization of the binding selectivity between WHRN with ADGRV1 and USH2A.

a The schematic diagram shows the interactions between WHRN and USH2A-CT, ADGRV1-CT. b Summaries of the binding affinities between WT or △PBM constructs of USH2A-CT, ADGRV1-CT, and different WHRN fragments. c, d Binding affinities of USH2A-CT (C) or ADGRV1-CT (D) to different WHRN fragments determined by ITC. Truncating PBM (△PBM) from USH2A-CT (C) or ADGRV1-CT (D) abolished or decreased its binding affinity with WHRN NPDZ12. N.D., not detectable. The WHRN fragments include NTD, PDZ1, PDZ2, PDZ3, NPDZ1, and NPDZ12. PBM, PDZ binding motif. NTD, N-terminal domain.

Fig. 2. Characterization of the interactions between PDZD7 with ADGRV1 and USH2A.

a The schematic diagram shows the interactions among the four USH2 proteins. b Up panel, summaries of the binding affinities between WT or △PBM constructs of USH2A-CT(S) or ADGRV1-CT and different PDZD7 fragments; bottom panel, summaries of the binding affinities between PDZD7 PDZ12 and WHRN NPDZ12, ADGRV1-CT and USH2A-CT. c Binding affinities of USH2A-CT(S) to different PDZD7 fragments determined by ITC. Truncating PBM (△PBM) from USH2A-CT abolished the binding to PDZD7 PDZ12. d Binding affinities of ADGRV1-CT to different PDZD7 fragments determined by ITC. e GST pull-down assay showing that GST-PDZD7 PDZ12 pulled down Usherin-CT(S) but not VLGR1-CT. f GST pull-down assays show that GST-WHRN NPDZ12 pulled down USH2A-CT(S) and ADGRV1-CT. N.D., not detectable. The different PDZD7 fragments include PDZ1, PDZ2, PDZ2, PDZ12, or PDZ3. PBM, PDZ binding motif. Source data are provided as a Source Data file.

Fig. 3. WHRN NTD dimerization contributes to the assembly of the ternary USH2 complex.

a The schematic diagram shows that dimerized WHRN interacts with USH2A and ADGRV1 to assemble a ternary USH2 complex. b SEC-MALS showed the molecular sizes of WHRN, WHRN/USH2A-CT(S), WHRN/ADGRV1-CT, or USH2A-CT(S)/WHRN/ADGRV1-CT. The samples for fitting molecular sizes were eluted on a Superose 12 10/300 GL or Superdex 200 increase 10/300 GL column. c Summaries of stoichiometry in b. d Fractionation coupled with FPLC showed three bands on the gel, confirming the existence of WHRN, USH2A-CT(S), and ADGRV1-CT in the peak. WHRN NPDZ12 was abbreviated as WHRN. Source data are provided as a Source Data file.

Fig. 4. The assembly of the quaternary USH2 complex requires the dimerized USH2A-CT.

a The schematic diagram shows that PDZD7 is recruited by dimerized USH2A to assemble the quaternary USH2 protein complex. b SEC-MALS showing the molecular size of USH2A-CT, summarized in Fig. 5c. c GST pull-down assay showing that GST-WHRN pulled down PDZD7 in the presence of USH2A-CT and quaternary protein complex was successfully assembled. d Co-sedimentation assay showed that WHRN, PDZD7, ADGRV1-CT, and USH2A-CT were enriched in the pellet. WHRN NPDZ12 and PDZD7 PDZ12 were abbreviated as WHRN, and PDZD7, respectively. Source data are provided as a Source Data file.

Fig. 5. Liquid-Liquid Phase Separation (LLPS) of quaternary USH2 protein condensate.

a Co-sedimentation-based assay showing that proteins were much more enriched in the pellet in WHRN/USH2A-CT than the isolated group. b Fluorescence images showing that droplets formed in WHRN/USH2A-CT but not the isolated group (20 μM each). WHRN and USH2A-CT were labeled with Alexa 488 and Cy3, respectively. Labeled proteins were added at a ratio of 1%. Scale bar: 5 μm. c Summaries of the stoichiometry of WHRN, WHRN NTD, WHRN △NTD, USH2A-CT, or USH2A-CT(S). d Co-sedimentation-based assay showing that the pellet enrichments of the two proteins in WHRN/USH2A-CT were deceased when USH2A-CT △PBM was introduced. Proteins were mixed at the final concentration of 20 μM each. e Co-sedimentation-based assays showed that the pellet enrichments of the two proteins in WHRN/USH2A-CT were decreased when WHRN △NTD or USH2A- CT(S) was introduced. Proteins were mixed at the final concentration of 20 μM each. f Fluorescence images show that the droplets formed by WHRN/USH2A-CT decreased in size or number when USH2A-CT △PBM, USH2A-CT(S), or WHRN △NTD was introduced. Proteins were mixed at the final concentration of 20 μM each. Scale bar: 10 μm. g Fluorescence images showed that the droplets were colocalized in highly enriched WHRN, USH2A-CT, PDZD7, and ADGRV1-CT. WHRN, USH2A-CT, PDZD7, and ADGRV1-CT were labeled with Alexa 488, Cy3, Cy5, and Alexa-405, respectively. Labeled proteins were added at a ratio of 1%. Scale bar: 5 μm. h Kinetic quantification of the FRAP assay showing the recovery rates of WHRN, USH2A-CT, ADGRV1-CT, and PDZD7. Mean ± SD, n = 3–5. WHRN NPDZ12 and PDZD7 were abbreviated as WHRN and PDZD7, respectively. △PBM or △NTD, the truncation of PBM or NTD. PBM, PDZ binding motif. NTD, N-terminal domain. Source data are provided as a Source Data file.

Fig. 6. High concentration of ADGRV1-CT disassembles the USH2 condensates formed by phase separation.

a Co-sedimentation-based assay showing that the pellet enrichments of both WHRN and USH2A-CT in WHRN/USH2A-CT (20 μM each) were decreased by ADGRV1-CT concentration-dependently at 0–60 μM. b Fluorescence images showing the droplets formed in WHRN/USH2A-CT decreased in both number and size by ADGRV1-CT concentration-dependently at 0–60 μM. WHRN and USH2A-CT were labeled with Alexa 488 and Cy3, respectively. Scale bar: 5 μm. c Co-sedimentation-based assay showed that the decrease in the pellet enrichment of WHRN, USH2A-CT, or ADGRV1-CT was not much changed when PBM was truncated (△PBM) from ADGRV1-CT. d Fluorescence images show that the amounts and sizes of the droplets formed by WHRN/USH2A-CT was still decreased by ADGRV1-CT when its PBM was truncated (△PBM). Scale bar: 5 μm. e Left panel, stereoview of the predicted structure of WHRN NTD drawn in the ribbon diagram; middle panel, surface representation showing the existence of an exposed hydrophobic pocket in NTD (highlighted by a black circle), which is predicted to serve as the binding site for the NBM of ADGRV1-CT.; right panel, surface representation of the predicted three-dimensional structure of WHRN NTD in complex with the ADGRV1-CT NBM. The hydrophobic, positively charged, negatively charged, and uncharged polar amino acid residues are shown in yellow, blue, red, and gray, respectively; ADGRV1-CT NBM is drawn in the ribbon model in cyan. WHRN NPDZ12 was abbreviated as WHRN. PBM, PDZ binding motif. NBM, NTD binding motif. Source data are provided as a Source Data file.

Fig. 7. NTD is necessary for WHRN to target the ankle region of stereocilia.

Cochlear explants from P4 wild-type mice were injectoporated with expression vectors to express EGFP-tagged full-length WHRN (a) or WHRN lacking NTD (b) in cochlear hair cells. Immunostaining with an anti-ADGRV1 antibody was performed to locate ADGRV1 (red). Phalloidin staining was performed to indicate the stereocilia (blue). Scale bars: 2 μm. NTD, N-terminal domain.

Fig. 8. Deafness-related mutations on WHRN affect the LLPS capacities of USH2 proteins.

a The predicted WHRN NPDZ1 structure is superimposed with the resolved structure of Harmonin NPDZ1/SANS SAM-PBM. Superimposition shows that A64, T110, and R223 on WHRN NPDZ1 correspond to M32, T78, and L169 on Harmonin NPDZ1, respectively. The side chain of WHRN A64 points outwards. A64D may disturb WHRN dimerization by disrupting WHRN NTD oligomerization. WHRN T110 localizes to the αE helix of NTD. We speculate that the T110A may affect WHRN’s interaction with USH2A by disrupting the overall conformation of WHRN. WHRN R223 localizes on the mini-domain. We speculate that R223H disrupts the overall folding of the mini-domain and further leads to the misfolding of NPDZ1 on WHRN. b Summaries of the binding affinities between WT or deafness-related mutants of WHRN, ADGRV1-CT, and USH2A-CT. N.D. not detectable. c Up panel, a summary of the SEC-MALS result in the bottom panel; bottom panel, SEC-MALS assay showing that A64D disrupted the dimer capacity of WHRN. d Fluorescence images show that A64D and R223H decreased the size and number of the droplets formed by quaternary USH2 protein phase separation. Scale bar: 10 μm. e Up panel: Co-sedimentation assay showing that both A64D and R223H decreased the pellet enrichments of WHRN/USH2A-CT/PDZD7/ADGRV1-CT. Bottom panel: Quantifications of each protein recovered from the condensed phase (pellet) in the co-sedimentation assays described in the up panel. Mean ± SD, n = 3. ***p < 0.001 using one-way ANOVA with Dunnett’s multiple comparison test. WHRN NPDZ12 and PDZD7 PDZ12 were abbreviated as WHRN and PDZD7, respectively. Source data are provided as a Source Data file.

Fig. 9. Deafness-related mutation on ADGRV1-CT affects the LLPS capacities of USH2 proteins.

a GST pull-down assay showing that though the pulled-down USH2A-CT and PDZD7 remained unaffected, ADGRV1-CT Y6236X was not pulled down by GST-WHRN and the assembly of quaternary USH2 protein complex was disrupted. b Co-sedimentation assay showing that ADGRV1-CT Y6236X was not enriched in the protein pellets, and the protein pellet enrichments of WHRN, USH2A-CT, and PDZD7 increased in the presence of Y6236X when compared to the WT. c Fluorescence images show that ADGRV1-CT Y6236X was absent in the droplets formed by USH2 protein phase separation. Scale bar: 10 μm. WHRN NPDZ12 and PDZD7 PDZ12 were abbreviated as WHRN and PDZD7, respectively. Source data are provided as a Source Data file.

Fig. 10. Deafness-related mutation on PDZD7 affects the LLPS capacities of USH2 proteins.

a PDZD7 PDZ1 is structurally superimposed with Harmonin PDZ1/SANS PBM. Superimposition shows that G103 on PDZD7 PDZ1 corresponds to G104 on Harmonin PDZ1. G103 localizes on the binding groove and may be involved in the interaction with USH2A. b Summaries of the binding affinities between WT or deafness-related mutants of PDZD7 and USH2A-CT(S). N.D., not detectable. c GST pull-down assay showing that GST-WHRN could not pull down PDZD7 G103R in the presence of USH2A-CT. d Fluorescence images show that PDZD7 G103R was absent in the droplets formed by USH2 protein phase separation. Scale bar: 10 μm. WHRN NPDZ12 and PDZD7 PDZ12 were abbreviated as WHRN and PDZD7, respectively. PBM, PDZ binding motif. Source data are provided as a Source Data file.