Figures and data
Cryo-ET structure of the microtubule associated autoinhibited LRRK212020T.
(A) Domain organization of LRRK2. The color coding of domains is used throughout this work. (B) Front view of the focused-refined cryo-ET map (left, see Methods) of microtubule-bound LRRK212020T assembled in the presence of MLi-2 (dataset (2)), and the model (right) generated from the map. (C-F) Graphical representation of the composite LRRK2-microtubule map building process. The full, low-resolution microtubule-LRRK2 map shown in (D) was used as the template for map fitting. Regions corresponding to LRRK2 oligomers and microtubules are colored in aquamarine and light gray, respectively. Higher-resolution, focused maps including both LRRK2 (C) and microtubules (G) were fitted in (D-E) to generate the composite map (F), viewed perpendicular to (left) and along (right) the microtubule axis. The close-up view showing the two pseudo-2-fold axis in (G-1) is highlighted as the black square. (G-1) Overview (G) of the 13-pf microtubule map used for the composite map shown in (H). Fitting of a published 13-pf microtubule model (L) and an α β tubulin complex model (1) into the map are shown in close-up views. (J-M) The two pseudo-twofold interfaces mediating filament formation are highlighted, with their axis indicated by a black ellipsoid in (J) and (L). The LRRK2 monomers forming each type of dimer, WD40:WD40 (J) or COR:COR (L), are highlighted in color, with the rest of the map dimmed. The same single copy of LRRK2 is highlighted in black contours for reference. (K) Side view of the LRRK2 WD40:WD40 dimer from the composite map and its cartoon representation. The view direction is shown as eye symbols in (J). WD40:WD40 interaction surface is highlighted in the cartoon representation. (M) similar with (K), showing the side view and cartoon representation of LRRK2 COR:COR dimer from the composite map. Scale bar: 5 nm for all panels.
The filaments of autoinhibited LRRK2 are stabilized by an interaction between WD40 and ARM/ANK domains.
(A) Composite map (top) and rotated LRRK2 WD40:WD40 dimer model (bottom). The black squares indicate the location of the close-up view shown in (B). As labeled, the left and right copies of LRRK2 are named 1A and 11, respectively. (B) Side-to-side comparison of the autoinhibited LRRK2 WD40:WD40 dimer model on microtubules (color, left) and the aligned in-solution LRRK2 model (gray, right). Both models share the same orientation as the model in (A). The clashes that result from docking the insolution LRRK2 models into our map are highlighted by a black circle and are resolved in the autoinhibited LRRK2 WD40:WD40 dimer model on microtubules by a shift in the N-terminal repeats shown with an arrow. (C) The location of the close-up view shown in (D) is indicated by black squares on the LRRK2 model before (up) and after rotation (down). The original viewing angle before rotation is the same as in Fig. 1B. (D) Superimposition of the LRRK2 structure in solution (light gray) with the autoinhibited LRRK2 on microtubules (colored). The shift of the ARM/ANK repeats is shown with an arrow. Both models share the same viewing angle as the rotated model in (C) at the bottom. (E) Composite map (top) and rotated full-length LRRK2 WD40:WD40 dimer model (bottom). The black squares indicate the location of the close-up view shown in (F). The model combined the model of autoinhibited LRRK2 we built into our map and the Alpha-Fold predicted N-terminus (see Methods). Similar as in (A), the left and right copies of LRRK2 are named as copy 1 and copy 11, respectively. (F) The chimeric full-length LRRK2 model was fitted into the focused refinement map (light gray). The map was refined by creating a focused mask on the WD40 domain of LRRK2 copy 1 and the ARM/ANK domains of LRRK2 copy 11 involved in the protein-protein interface (see Methods). The Rab-binding site at the N-terminus of LRRK2 is highlighted. (G) Close-up view of the binding surfaces between ARM/ANK domains and the WD40 domain in the autoinhibited LRRK2 dimer model on microtubules. The ‘latch’ helix and the WD40 binding interfaces in the ARM domain are highlighted based on the focused refinement map. The black square in (F) highlights the focused region shown here. (H) Superimposition of the LRRK2 in-solution model (light and dark gray) and the autoinhibited LRRK2 WD40:WD40 dimer model on microtubules (colored). The two models were aligned by their WD40 domains. The view is focused on the ARM/ANK and WD40 domains of both LRRK2 copies and the rest of the models are shows as semi-transparent. Scale bar: 5 nm for all panels.
Autoinhibited LRRK2 forms short, sparse oligomers perpendicular to the microtubule axis.
(A) Tomographic slices 4.6 nm in thickness showing LRRK2 decorated microtubules. (B) Close-up views showing subtomogram coordinates (orange spheres) and their refined x-y-z (yellow-red-blue) orientations after subtomogram analysis. The picked subtomograms are mapped back to the original tomogram, and the LRRK2 dimer groups containing different copy numbers are shown (left: 5-copy group, right: several 3-copy groups clustering together). Black squares in (A) and (D) highlight the locations of corresponding subtomogram groups. (C) Composite model showing LRRK2 oligomer groups binding to microtubules, with orientations and LRRK2 copy numbers corresponding to (B). (D) Location and orientation of picked subtomograms contributing to the final LRRK2 reconstruction, color-coded in the same way as in (B). The subtomogram picking models are aligned to their corresponding microtubules in (A). (E) Representation of the definition of the helical angle (» angle), showing the angle between (LRRK2)2 oligomers and the axis of the microtubule (above), as well as the plot showing the distribution of » angles observed from this dataset (below). (F) Picked LRRK2 subtomograms are grouped based on their relative positions and distances, and the length of (LRRK2)2 oligomers grouped together are plotted as a frequency plot. The longest LRRK2 dimer chain observed in this dataset contains six LRRK2 dimers. (G-1) Summary of the comparison between the autoinhibited LRRK2 state observed from this study (G) and the active-like LRRK2 state observed from the in situ study (1) on microtubules. The cartoon representations in (H) describe the differences of the domain architectures between the autoinhibited and the active-like LRRK2, as well as the corresponding different assemblies observed on microtubules. In situ model of active LRRK2 forming right-handed filaments is derived from the published maps, with color codes matching all previous figures. Scale bars: (A) 50 nm, (B-1) 5 nm.
Geometry-based extraction of LRRK2-microtubule subtomograms.
(A) Sample tomogram slices showing LRRK2-decorated microtubules from three cryo-ET datasets. The LRRK2 variant and kinase inhibitor used in each dataset are indicated. subtomograms that went into the final reconstruction are shown as well, overlayed on the tomogram slice in their final orientation and position. (B) Three orthogonal views of our initial LRRK2-microtubule map, with a reconstructed 3D map shown in the up-right corner. The corresponding x-y-z axis are labeled in the orthogonal views and the reconstructed map respectively. (C-E) Step-wised representation of LRRK2-microtubule subtomogram extraction workflow (see Methods). (C) Subtomograms are first regularly picked along microtubules to serve as reference points for LRRK2 subtomogram picking. Sample tomogram slice showing LRRK2-decorated microtubules from the LRRK212020T + MLi-2 + microtubules dataset (with LRRK212020T incubated with pre-assembled microtubules) is overlayed by microtubule subtomograms, in their final orientation and position. (D) The same tomogram slice overlapped with over-picked LRRK2 subtomograms regularly distributed around each microtubule subtomograms shown in (C). (E) The same tomogram slice overlapped with cleaned LRRK2 subtomogram subsets after duplication cleaning and cross-correlation based cleaning. Over-picked subtomograms and subtomograms that are poorly correlated with our initial model were cleaned out. Coordinates of subtomograms are labeled as orange spheres and their refined x-y-z orientations are shown as yellow-red-blue arrows. Scale bar: 50 nm.
Data-processing scheme for the LRRK212020T + MLi-2 + microtubule sample.
(A) Flow chart of the cryo-ET data processing procedure. (B) The gold-standard Fourier Shell Correlation (FSC) curves (0.143 cutoff) show the final resolution of the LRRK2 focused-refinement map. (C) Local resolution of the LRRK2 focused-refinement map. (D) Particle projection angle distribution plot of the map.
Data-processing scheme for the focused-refined maps from the LRRK2 + MLi-2 + microtubule sample.
(A) Flow chart of the cryo-ET data processing procedure from the microtubule-LRRK2 map to obtain the 13-pf microtubule reconstruction map. (B) The gold-standard Fourier Shell Correlation (FSC) curves (0.143 cutoff) show the final resolution of the microtubule map. (C) Particle projection angle distribution plot of the microtubule map. (D) Flow chart of the cryo-ET data processing procedure from the final LRRK2 map to obtain focused map containing the WD40:ARM/ANK interface. (E) The gold-standard FSC curves (0.143 cutoff) show the final resolution of the focused refined map. (F) Particle projection angle distribution plot of the map.
Data-processing scheme for the LRRK212020T + MLi-2 + microtubule sample with LRRK212020T and microtubule co-polymeriztion.
(A) Flow chart of the cryo-ET data processing procedure. (B) The gold-standard Fourier Shell Correlation (FSC) curves (0.143 cutoff) show the final resolution of the LRRK2 focused-refinement map. (C) Local resolution of the LRRK2 focused-refinement map. (D) Particle projection angle distribution plot of the map.
Data-processing scheme for the LRRK2 + MLi-2 + microtubule sample.
(A) Flow chart of the cryo-ET data processing procedure. (B) The gold-standard Fourier Shell Correlation (FSC) curves (0.143 cutoff) show the final resolution of the LRRK2 focused-refinement map. (C) Local resolution of the LRRK2 focused-refinement map. (D) Particle projection angle distribution plot of the map.
Data-processing scheme for the LRRK2 + GZD-824 + microtubule sample.
(A) Flow chart of the cryo-ET data processing procedure. (B) The gold-standard Fourier Shell Correlation (FSC) curves (0.143 cutoff) show the final resolution of the LRRK2 focused-refinement map. (C) Local resolution of the LRRK2 focused-refinement map. (D) Particle projection angle distribution plot of the map.
Comparison of the LRRK2 kinase domain conformations in the in-solution and microtubule-bound structures.
(A) Models of insolution LRRK212020T bound to kinase inhibitors were fitted into our maps of LRRK2 bound to microtubules. The color scheme is the same one introduced in Fig. 1. The published LRRK2 models that were fitted into each map contain the same kinase inhibitor used in the sample giving rise to the map, as noted below each image. (B) Close-up view focusing on the fitting of the kinase from the published LRRK2-inhibitor complexes into each of our LRRK2 maps. The motifs that undergo major conformational changes, the ‘G-loop’ and the ‘DYG’ motif, are highlighted and colored in olive green. The inhibitors are shown in black.
Surface electrostatic potential at the WD40:ARM/ANK interaction interfaces in the LRRK2 WD40 dimer observed in the LRRK2-microtubule complex.
(A) The location of the close-up view shown in (B) is indicated by the black square on the composite map (left). How panel (B) is generated by rotating from the composite map is described step-by-step. (B) The rotated LRRK2 COR:COR dimer map (colored) derived from the autoinhibited LRRK2 map on microtubules. The viewing angle is the same as in Fig. 1J. (C) The LRRK2 in-solution model (light and dark gray) and the autoinhibited LRRK2 model on microtubules (colored) are fitted into the map shown in (B) and superimposed to each other for comparison. The view is focused on the COR domains of both LRRK2 models and the rest of the models are set as transparent in the background. (D) The location of the close-up view shown in (E) is indicated by the black square on the composite map (left) and the LRRK2 chimeric WD40:WD40 dimer model (right). How panel (E) is generated by rotating from the composite map is described step-by-step. (E) Close-up view of the interaction surface between the ARM/ANK domains and the WD40 domain. The view shown here is identical to Fig. 2G and serves as the reference view. (F) The corresponding surface electrostatic potential map aligned to (E). (G-H) Surface electrostatic potential view of the ARM/ANK domain (G) and the WD40 domain (H). Both hydrophobic surfaces and charged surfaces are highlighted. (1) ribbon representation of (E) showing the PD associated LRRK2 mutations identified from patients.
Geometrical analysis of LRRK2 subtomograms for the different LRRK2-microtubule datasets.
(A) As in Fig. 3B, subtomogram coordinates (orange spheres) and their refined x-y-z (yellow-red-blue) orientations are mapped back to the original tomogram (left) to illustrate the analyzed geometrical parameters (right). Distances between adjacent particles, helical angles and (LRRK2)2 group chain length is labeled on the cartoon. (B) Distribution of distances between adjacent LRRK2 subtomograms. Only subtomogram pairs that fall into the same (LRRK2)2 group are analyzed. The analysis was applied to the same dataset and the same set of (LRRK2)2 groups as in Fig. 3 E-F. LRRK2 subtomograms are grouped based on their relative positions and distances. (C) Distribution of distances, helical angles and group chain length for the LRRK2 + MLi-2 + microtubule dataset. (D-E) Same as in C, for the LRRK2 + MLi-2 + microtubule dataset (D), and the LRRK2 + GZD-824 + microtubule dataset (E). (F) Distribution of (LRRK2)2 subtomogram number along each microtubule (left) and along a fixed length (1 nm) of microtubules (right). The mean ± s.d. and P values (one-way ANOVA, ns:P > 0.05; *:P f 0.05; **:P f 0.01; ***:P f 0.001; ****:P f 0.0001) of the indicated comparisons are shown (n = number of microtubules for each category).
