StableMARK-decorated microtubules in cells have expanded lattices

Abstract

Microtubules are crucial in cells and are regulated by various mechanisms like posttranslational modifications, microtubule-associated proteins, and tubulin isoforms. Recently, the conformation of the microtubule lattice has also emerged as a potential regulatory factor, but it has remained unclear to what extent different lattices co-exist within the cell. Using cryo-electron tomography, we find that, while most microtubules have a compacted lattice (∼41 Å monomer spacing), approximately a quarter of the microtubules displayed more expanded lattice spacings. The addition of the microtubule-stabilizing agent Taxol increased the lattice spacing of all microtubules, consistent with results on reconstituted microtubules. Furthermore, correlative cryo-light and electron microscopy revealed that the stable subset of microtubules labeled by StableMARK, a marker for stable microtubules, predominantly displayed a more expanded lattice spacing (∼41.9 Å), further suggesting a close connection between lattice expansion and microtubule stability. The coexistence of different lattices and their correlation with stability implicate lattice spacing as an important factor in establishing specific microtubule subsets.

Figure S1.

Summary of previously reported lattice spacings under various conditions. The figure shows at the top the range of the reported lattice spacings for the GDP, Taxol, and GMPCPP lattices in vitro. Underneath, along the same scale, proteins for which binding in relationship to the microtubule lattice spacing has been investigated are depicted. Their placing is based either on direct (triangle) or relative (tilde) lattice measurements. References in Table S1.

Figure 1.

A subset of microtubules has an expanded lattice in cells. (A) Tomogram slice (thickness: 10 nm) showing two selected microtubule (MT) backbones in an untreated U2OS cell (red and pink). (B and C) Power spectra of the masked and transformed MT segments from the red MT (B) and pink MT (C) shown in A. (D) Overlay of the layer line plots of the power spectra of the MT segments from the compacted (red) and expanded (pink) MTs in A. Arrows indicate the location of the layer line peaks and their related lattice spacing. (E) Violin plot showing the distribution of lattice spacings in untreated U2OS cells (N = 31, 12 tomograms, 7 cells), from microtubules assembled in vitro from GTP-bound soluble tubulin yielding dynamic microtubules (N = 40, 6 tomograms), in the presence of Taxol (N = 32, 3 tomograms), or from GMPCPP-bound soluble tubulin (N = 33, 14 tomograms). Horizontal lines correspond to the discrete spatial frequency values in reciprocal space. (F) Simplified cartoon showing the long-range effect of a compacted or an expanded MT lattice. Scale bar: 100 nm (A).

Figure S2.

In situ layer line analysis workflow. (A) Graphical depiction of the steps performed during the layer line analysis. (Step 1) Microtubule (MT) backbone is traced with evenly spaced coordinates. (Step 2) Using the backbone coordinates, a mask around the microtubule is generated and (Step 3) particles are cropped and transformed to ensure the MT axis is perpendicular for viewing direction. From the cropped particles (box size 190), particles are picked (pink squares) so that after re-cropping, the new particles with a box size of 1,030 (Step 4) cover the whole microtubule with minimal overlap. (Step 5) 2D power spectrum of each particle is calculated. (Step 6) Power spectra of all particles from the same microtubule are summed and the final power spectrum is used to localize the layer lines and thereby calculate the lattice spacing. (B) Boxplot of the signal-to-noise ratio (SNR) of the layer line analysis when using different particle box sizes (N = 3). Scale bars: 100 nm (step 1, 2), 10 nm (step 3), 50 nm (step 4).

Figure S3.

Representative in vitro microtubule data. (A–E) Tomogram slices (0.87 nm thick) showing representative images of microtubules assembled in vitro (A) from GTP-bound soluble tubulin yielding dynamic microtubules, (B and C) in the presence of Taxol, either labeled with a fluorophore and biotin (B) or unlabeled (C) (see Materials and methods), or (D and E) from GMPCPP-bound soluble tubulin, either labeled with a fluorophore and biotin (D) or unlabeled (E) (see Materials and methods). Scale bars: 100 nm (A–E). (F) Violin plots showing the lattice spacing distribution of unlabeled in vitro Taxol (N = 10, 2 tomograms) and unlabeled in vitro GMPCPP-bound microtubules (N = 10, 2 tomograms). Horizontal lines correspond to the discrete spatial frequency values in reciprocal space.

Figure 2.

Taxol treatment induces a hyperexpanded lattice within cells. (A) Tomogram slice showing a representative image of Taxol-treated microtubules in WT U2OS cells. Scale bar: 100 nm. (B) Violin plot showing the lattice spacing distribution in Taxol treated cells (N = 30, 6 tomograms, 5 cells) and in untreated cells (N = 31, 12 tomograms, 7 cells, same data as Fig. 1 E, included for comparison). Horizontal lines correspond to the discrete spatial frequency values in reciprocal space. Taxol distribution is significantly different from the untreated distribution (****P value <0.0001, unpaired t test based permutation test). (C) Microtubule average shows that Taxol-treated microtubules consist of 13 PFs. Central volume slices (28 nm thick) from top (left) and side (right) views. Scale bar: 5 nm.

Figure S4.

Data analysis components of the in situ Taxol microtubule averaging workflow. (A) Front view and side view (90° rotation) of the hollow tube reference used for the initial 3D refinement to determine the average PF number. (B) Front view and side view (90° rotation) of the 13 PF reference used during subsequent 3D classification and 3D refinement steps. (C) FSC curves of the in situ Taxol microtubule subtomogram average are shown in Fig. 2.

Figure 3.

Correlation of FM to SEM data using an integrated cryo-FM. (A) Cartoon describing the FM-SEM correlation with FM data obtained after milling. Correlation is confirmed using extracellular beads. (B) FIB image of an intact U2OS cell (9° tilted side view). (C) Untilted SEM image of the same grid square as shown in B. The beads used to confirm FM-SEM correlation are indicated with white and black arrows in C and D, respectively. (D) Untilted SEM image of the polished lamella of the cell shown in C. (E) Scatterplot of correlation errors from leave-one-out calculations; each dataset has a unique color, grey circles mark the 1xSD and 2xSD boundaries (10 datasets, 51 beads), dx = difference in x, dy = difference in y. (F) Boxplot showing the distribution of scaling factors (mean = 0.584, standard deviation = 0.005, N = 10). (G) Scaled FM image of the extracellular beads used to guide FM-SEM overlay, beads used to confirm FM-SEM correlation are indicated with black arrows, similar to C and D. (H) Scaled FM image of fBSA-Au⁵ beads used for subsequent FM-TEM correlation (see Fig. 4). (I) Scaled FM image of the StableMARK signal. Scalebars: 10 µm (B–D and G–I).

Figure S5.

Correlation of FM to SEM data for targeted cryo-FIB milling. (A) Cartoon describing the FM-SEM correlation performed using extracellular beads. FM data was obtained prior to milling using the CorrSight. (B) MIP of a StableMARK z-stack; beads used for 3D correlation are indicated with arrows. (C) MIP of a z-stack with fBSA-Au⁵ beads used for the subsequent FM-TEM correlation (see Fig. S6). (D) Untilted SEM image of the same grid square as shown in B and C. (E) Correlated StableMARK and fBSA-Au⁵ overlayed with the SEM image. (F) Correlated StableMARK and fBSA-Au⁵ overlayed with the untilted SEM image of the polished lamella (milled at a 9° angle). (G) Scatterplot of correlation errors from leave-one-out calculations; each dataset has a unique color, grey circles mark the 1xSD and 2xSD boundaries (12 datasets, 98 beads). Scale bars: 10 µm (B–F).

Figure 4.

The lattice of StableMARK-positive microtubules is expanded compared to the GDP-compacted lattice. (A) Cartoon describing the FM-TEM correlation performed using intracellular beads. (B) TEM overview image of a lamella; dotted line indicates outline of the lamella, red squares the location of the fBSA-Au⁵ beads. (C–E) Zoom-in of the fBSA-Au⁵ beads found both in the TEM lamella in B and in the correlated fBSA-Au⁵ FM data in E. (E) Correlated fBSA-Au⁵ FM data; red squares indicate fBSA-Au⁵ location. (F) Correlated StableMARK-bound microtubules. (G) Overlay of TEM lamella and correlated FM data of both fBSA-Au⁵ (pink) and StableMARK-bound microtubules (blue). (H and I) Two-step zoom of two microtubules overlapping with the StableMARK-bound FM data. (J) Violin plot showing the distribution of lattice spacings of the StableMARK subset of microtubules (N = 43, 25 tomograms, 23 cells) and in untreated cells (N = 31, 12 tomograms, 7 cells, same data as Fig. 1 E, included for comparison). Horizontal lines correspond to the discrete spatial frequency values in reciprocal space. StableMARK distribution is significantly different from the untreated distribution (****P value <0.0001, unpaired t test based permutation test). Scale bars: 1 µm (B and E–G), 100 nm (C, D, and I), 400 nm (H).

Figure S6.

FM to TEM correlation using whole-cell FM data. (A) TEM overview image of a lamella; dotted line indicates the outline of the lamella, red squares the location of the fBSA-Au⁵ beads. (B and C) Zoom-in of the fBSA-Au⁵ beads found both in the TEM lamella in A and in the correlated fBSA-Au⁵ FM data in D. (D) Correlated fBSA-Au⁵ FM data; red squares indicate fBSA-Au⁵ location. (E) Correlated StableMARK-bound microtubules. (F) Overlay of TEM lamella and correlated FM data of both fBSA-Au⁵ (pink) and StableMARK-bound microtubules (blue). (G and H) Two-step zoom of two microtubules overlapping with the StableMARK-bound FM data. (I) Violin plots showing the lattice spacing distribution of StableMARK-bound microtubules, split between microtubules correlated using FM data from CorrSight (light blue, N = 12, 6 tomograms, 6 cells) or FM data from Meteor (dark blue, N = 31, 19 tomograms, 17 cells). The two distributions are not significantly different (P value = 0.091, unpaired t test based permutation test). Horizontal lines correspond to the discrete spatial frequency values in reciprocal space. Scale bars: 1 µm (A and D–F), 100 nm (B, C, and H), 400 nm (G).

Figure S7.

Expanded and compacted microtubules can be found within one lamella. (A) TEM overview image of a lamella. (B) Correlated StableMARK FM data (Meteor). (C) Overlay of TEM and FM data; black squares indicate the location of tomograms collected at a spot with and without StableMARK signal. (D and E) Tomogram slices (∼4 nm thickness) of an expanded microtubule with StableMARK signal (D) and of a compacted microtubule devoid of StableMARK signal (E). (F) Violin plots showing the lattice spacing distribution of uncorrelated microtubules (N = 12, 10 tomograms, 8 cells), StableMARK-bound microtubules, and untreated microtubules (same data as Fig. 4 J, included for comparison). The StableMARK distribution is significantly different from the uncorrelated distribution (**P value <0.01, unpaired t test based permutation test). Horizontal lines correspond to the discrete spatial frequency values in reciprocal space. Scale bars: 1 µm (A–C), 100 nm (D and E).