How myosin VI traps its off-state, is activated and dimerizes

Abstract

Myosin VI (Myo6) is the only minus-end directed nanomotor on actin, allowing it to uniquely contribute to numerous cellular functions. As for other nanomotors, the proper functioning of Myo6 relies on precise spatiotemporal control of motor activity via a poorly defined off-state and interactions with partners. Our structural, functional, and cellular studies reveal key features of myosin regulation and indicate that not all partners can activate Myo6. TOM1 and Dab2 cannot bind the off-state, while GIPC1 binds Myo6, releases its auto-inhibition and triggers proximal dimerization. Myo6 partners thus differentially recruit Myo6. We solved a crystal structure of the proximal dimerization domain, and show that its disruption compromises endocytosis in HeLa cells, emphasizing the importance of Myo6 dimerization. Finally, we show that the L926Q deafness mutation disrupts Myo6 auto-inhibition and indirectly impairs proximal dimerization. Our study thus demonstrates the importance of partners in the control of Myo6 auto-inhibition, localization, and activation.

Fig. 1. Importance of ADP.P_i for the compact, back-folded Myo6 conformation.

a Schematic representation of FLMyo6 with the Motor domain (MD, gray), CaM binding sites (Ins2/IQ, purple/red), CaM (lilac/pink), 3-helix bundle (3HB, blue), single alpha helix (SAH, green), distal Tail (DT, orange) and CBD (brown). Residue numbers correspond to human Myo6, Uniprot entry Q9UM54-2. b Dimensionless Kratky plot representation from SEC-SAXS. FLMyo6 in the presence of ADP.AlF₄ (a widely used ADP.P_i analog that stabilizes the pre-powerstroke of Myo6) (green) results in a bell-shaped spectrum with a maximum close to the intersection of the dashed lines (√3:1.104), typical of a globular protein. The spectrum for FLMyo6 in NF/high salt (black) suggests a much more elongated shape. Source data are provided as a Source Data file. c Representation of the ab initio SAXS envelope of Myo6 in ADP.AlF4 condition (green) with MD^Ins2-IQ-3HB docked. Myo6 adopts a compact conformation that requires Myo6 to fold back after the 3HB domain (see “Methods” and Supplementary Fig. 3A, B). d Scheme representing the interactions stabilizing the Myo6 back-folded state. e Example of a negative staining micrograph of Jo-Myo6-In in ADP.VO₄ (representative of 25 grids prepared with 2 different protein batches) with selected 2D classes overlayed (from left to right: 8630; 9284; 9261; 7822 and 7179 particles averaged, respectively). f EM density for Jo-Myo6-In (gray mesh) obtained by negative staining. Myo6 fragments and Jo-In were manually docked inside the negative staining 3D reconstruction (see “Methods”). Negative staining 3D reconstruction and the ab initio SAXS envelope exhibit similar overall size and shape (Supplementary Fig. 3C). g (Top) Crystal structure of the Myo6 C-terminus (CBD^c) (PDB: 3H8D). Star: highly conserved and exposed loop between the β_A and β_B strands. (Bottom) Alignment of Myo6 CBD^c domain (aa 1143 to 1262 in Q9UM54-2) from different species. Strictly conserved and similar residues are shown in blue and red, respectively. Stars: residues implicated in binding to the Myo6 Head (Table 1). h CBD^c (brown) added to the negative staining-based model pictured in (f), (see “Methods”). The distances between Jo C-terminus and Myo6 N-terminus; and between Myo6 C-terminus and In N-terminus are indicated.

Fig. 2. Role of the proximal Myo6 sequence in the stabilization of the off-state.

a Model of Myo6 opening/back-folding. Back-folding requires the SAH to fold back on the 3HB. The L926 residue (red cross) leads to deafness when mutated into Gln. (Insert) Mutations of the apolar residues at the N-terminus of the SAH to turn Myo6 into a constitutive monomer (SAHmimic). b Dimensionless Kratky plot representation from SEC-SAXS. FLMyo6 in NF/high salt is pictured in black. In the presence of ADP.AlF₄ (ADP.P_i analog), FLMyo6 (L926Q) (yellow) and FLMyo6 (SAHmimic) (light blue) spectrums correspond to an elongated shape, as opposed to FLMyo6 WT (green). c Rg of FLMyo6 WT, L926Q and SAHmimic determined by SEC-SAXS experiments (n = 1) in the presence of ADP.AlF₄ (ADP.P_i analog) and FLMyo6 in NF/high salt. Rg values were extracted from linear fits of the Guinier plots shown in Supplementary Fig. 1C using primusqt (ATSAS suite). Mean ± SD. d Actin-activated ATPase rate of FLMyo6 WT, L926Q, SAHmimic and MD^Ins2 (n = 6). Mean ± SD. (b-d) Source data are provided as a Source Data file.

Fig. 3. GIPC1 can bind to and activate the back-folded form of Myo6, while Dab2 and TOM1 can only bind Myo6 once the motor has been primed open.

a EM density for the Jo-Myo6-In (gray mesh) obtained by negative staining, as in Fig. 1h and Supplementary Movie 1. The WWY motif (red spheres) of CBD^c is buried. The CBDⁿ fragment (beige) (PDB: 5V6E) is positioned in the remaining, uninterpreted part of the density so that the RRL motif (red spheres) on CBDⁿ and the I1072 (blue sphere) proposed to mediate interaction between ubiquitin and Myo6 are both exposed. Note that no experimental model exists for 36 missing residues between the CBDⁿ and CBD^c (dashed lines) and that the position of the CBDⁿ is consistent with the cross-links found between CBDⁿ and the rest of the Myo6 molecule through cross-linking mass spectrometry of the purified FLMyo6 with disuccinimidyl sulfoxide (DSSO) (Supplementary Text, Supplementary Fig. 7A–C, Supplementary Table 1). The placement of elements of the Myo6 Tail within the model improved the fitting between our atomic model and the SAXS data (Supplementary Figs. 3D–F, 5B, C). b Fitting of CBD^c-TOM1 structure (PDB: 6J56) with CBD^c (brown) in the model presented in Fig. 1h. TOM1 (yellow) binding would result in clashes with SAH (green) and CaM (lilac). c Fitting of CBDⁿ (beige)-GIPC1 (light blue) structure (PDB: 5V6E) as for CBDⁿ alone. GIPC1 binding seems compatible with Myo6 back-folded conformation. d Elutions of anti-His pull-down assays (FLMyo6 against ^HisTOM1 and ^HisGIPC1) revealed using SYPRO (Input and last wash pictured in Supplementary Fig. 9). Crosses: quantification of retained Myo6 (Image-Lab software, Bio-Rad) followed by stoichiometric normalization based on partner concentration (n = 4 for WT and n = 2 for SAHmimic). + means less than 10% Myo6 retained; +++ means more than 20% Myo6 retained. e ATPase rates of FLMyo6 (WT) and FLMyo6 (SAHmimic) with 40 µM F-actin and increasing concentrations of GIPC1, TOM1 or Dab2 (n = 6). Purple line: ATPase rate of MD^Ins2 at 40 µM actin (n = 6) for reference. Mean ± SD. d, e Source data are provided as a Source Data file.

Fig. 4. GIPC1 recruits Myo6 to melanosomes independently of Myo6 closure; Dab2 and TOM1 can only recruit Myo6 after the motor has been primed open.

a Representative fixed MNT-1 cells co-expressing different ^GFPMyo6 (I1072A), ^mCherryMST and ^iRFPVAMP7 constructs. b Representative fixed MNT-1 cells co-expressing different ^GFPMyo6 (I1072A) constructs with ^mCherryMST-GIPC1 and ^iRFPVAMP7. c Representative fixed MNT-1 cells co-expressing different ^GFPMyo6 (I1072A) constructs with ^mCherryMST-TOM1 and ^iRFPVAMP7. d Representative fixed MNT-1 cells co-expressing different ^GFPMyo6 (I1072A) constructs with ^mCherryMST-Dab2 and ^iRFPVAMP7. (a–d) Green: Myo6 GFP; Cyan: ^irRFPVAMP7; Magenta: ^mCherryMST partner. From left to right: entire cell, 3 channels merged; 8x zoom on boxed region: ^GFPMyo6 / ^mCherryMST-partners merged, then individual channels. Scale bars: 10 µm. Arrowheads: recruitment of Myo6 on melanosomes. e Myo6-positive melanosomes quantification of different ^GFPMyo6 mutants when different ^mcherryMST tagged partners are expressed (n = 3, total cell number~30). Myo6-positive melanosomes are expressed in percentage and normalized to the total number of VAMP7-positive melanosomes. Cells were fixed 48 h post-transfection then imaged and processed for quantification. Data are presented as the mean ± SEM. Significant stars: ***, p < 0.001; **, p < 0.01; *, p < 0.05; n.s., not significant (two-sided unpaired t-test with Welch’s correction), for each ^GFPMyo6 construct, significance of experiments with partners compared to the control without partner (in black on the graph). P-values are the following: FLMyo6 (I1072A)/GIPC1: p < 0.0001, FLMyo6 (I1072A)/Dab2: p = 0.698, FLMyo6 (I1072A)/TOM1: p = 0.0071, Jo-Myo6-In (I1072A)/GIPC1: p < 0.0001, Jo-Myo6-In (I1072A)/Dab2: p = 0.344, Jo-Myo6-In (I1072A) /TOM1: p = 0.5005, FLMyo6 (SAHmimic.I1072A)/GIPC1, TOM1 or Dab2: p < 0.0001, FLMyo6 (L926Q.I1072A)/GIPC1 or Dab2: p < 0.0001, FLMyo6 (L926Q.I1072A)/TOM1: p = 0.003. Source data are provided as a Source Data file.

Fig. 5. Myo6 can form an antiparallel dimer through residues 875–940 which allow large steps.

a (Left) X-ray structure of mouse Myo6 875–940 antiparallel dimer colored according to B-factor from 18.6 Ų (dark blue) to 150.8 Ų (red). (Right) Key residues for dimer stabilization. Apolar contacts are mediated by residues pictured in green. Dotted blue line: polar contacts. Residues mutated in our triple mutant (T888D.R892E.V903D) are underlined. b Close-up of the dimerization interface of Myo6 875–940 in the electronic density. c Triple helix bundle (PDB: 2LD3) domain. T888, R892, V903, T845C and A880C pictured as sticks are surface residues. d SEC-MALS profiles of Myo6 875–940 WT (red) and T888D.R892E.V903D mutant (blue), following injection of 50 μL at 10 mg/mL in 10 mM Tris-HCl pH 7.5; 50 mM NaCl; 5 mM NaN₃; 0.5 mM TCEP. Thin lines: static light scattering; thick lines: measured molecular mass. WT elutes as dimers (32 μM concentration at the peak, measured by the in-line refractometer) and T888D.R892E.V903D mutant elutes as monomers (43 μM at the peak). e Model of active FLMyo6 dimer (see “Methods”). f ATPase rates (mean ± SD) of FLMyo6 WT (green), T888D.Q892E.V903D (gray), SAHmimic (blue) and L926Q (yellow) at 40 µM F-actin and increasing concentrations of GIPC1 (n = 6). ATPase rates of MD^Ins2 and zippered dimer without partner (n = 6) plotted as purple and red thick lines (respectively) as references for monomeric and dimeric Myo6. g Fluorescence intensity of internalized transferrin was measured for each condition after treatment with genistein (cells examined over 2 independent experiments: WT = 62, KO = 58, KO+ FLMyo6 (WT) = 79, KO+ FLMyo6 (T888D.R892E.V903D) = 66) (p < 0.001, one-way ANOVA; Tukey post-hoc comparisons; one-sided). P-values are the following: WT vs KO: p < 0.0001, WT vs KO+ FLMyo6 (WT): p = 0.3363, WT vs KO+ FLMyo6 (T888D.R892E.V903D): p < 0.0001 (***), KO vs KO+ FLMyo6 (WT): p = 0.0001, KO vs KO+ FLMyo6 (T888D.R892E.V903D): p = 0.5228, KO+ FLMyo6 (WT) vs KO+ FLMyo6 (T888D.R892E.V903D): p = 0.0146 ($).Whisker boxes (10–90 percentile with 2nd and 3rd quartiles within the box; white dot indicates the median) encased within a violin plot (generated with BoxPlotR). (d, f, g) Source data are provided as a Source Data file.

Fig. 6. Importance of a folded monomer for regulation.

a When auto-inhibited, Myo6 can diffuse across actin-rich regions and interacts weakly with F-actin. These weak actin interactions (~7 µM apparent affinity, estimated in Supplementary Fig. 4E) result in facilitated diffusion and in increasing the Myo6 concentration in actin-rich regions of the cell. Once recruited by a partner, Myo6 is activated and starts performing its cellular function. b Scheme representing possible activation mechanisms for Myo6. Myo6 domains are color-coded: Myo6 MD (gray), Ins2/CaM (purple), IQ/CaM (red/pink), 3HB in blue, SAH (green), DT (orange), CBD (brown), and the partner binding sites (garnet). The binding site (WWY) for Dab2 and TOM1 is blocked, preventing recruitment of Myo6 without a prior unfolding signal prior to unblock their binding. GIPC1 can bind the accessible RRL motif resulting in Myo6 recruitment and opening. Other signals can act as unfolding factors such as Ca²⁺, which can allow TOM1 to bind to Myo6. Such an activation cascade was previously proposed. Once unfolded, Myo6 potentially acts as a monomer, as previously proposed upon TOM1 binding; or it can dimerize through proximal dimerization, as demonstrated in this study with GIPC1 binding; or it dimerizes through distal dimerization upon Dab2 binding, which may lead to proximal dimerization.