Life-cycle-coupled evolution of mitosis in close relatives of animals – Nature

Phylogenetic analysis

We generated profiles of representative proteins from mitosis-associated cellular components including the centrosome, SPB, spindle, NE and the kinetochore for early-branching animal and fungal lineages alongside model species with well-characterized mitosis. The proteomes of H. sapiens, S. rosetta, Monosiga brevicollis, Capsaspora owczarzaki, S. pombe, S. punctatus, D. discoideum and P. polycephalum were obtained from Eukprot v02.2020_06_30 (ref. ⁵⁰). The ichthyosporean proteomes were obtained from recent phylogenomic studies^19,21,35. To identify putative orthologues we first searched for homologues with human proteins using phmmer (HMMER 3.3.2, November 2020; http://hmmer.org/)⁵¹. In case of divergence or absence of human proteins, searches were also carried out with proteins from other model species including S. pombe, D. melanogaster and D. discoideum. This was followed by Hidden Markov model (HMM)-based searches with the associated PFAM (http://pfam.xfam.org)⁵² models using hmmsearch (HMMER 3.3.2, November 2020)⁵¹. For proteins in which homologues were not recovered by existing HMMs, new HMMs were generated. The multiple sequence alignment was done with MAFFT v.7.490 using ‘linsi’ optimized for local homology⁵³. The alignments were inspected and trimmed using TrimAl v.1.2 to remove the less-conserved regions⁵⁴. The trimmed alignments were used for tree inference with IQTree v.2.0.3 2020 using the model finder and ultrafast bootstraps (1,000) bootstraps^55,56,57. The trees were visualized and annotated using FigTree (http://tree.bio.ed.ac.uk/software/figtree/). This process was performed iteratively to obtain better alignments which gave trees with higher bootstrap values. The alignment was then used to generate an HMM using the hmmbuild command in HMMER. This process was performed iteratively while incorporating the newly discovered homologues in the next round. The protein sequences are provided as fasta files. For most kinetochore proteins, iterative similarity searches were performed using previously generated HMMs of which candidate genes were scrutinized on the basis of known domain and motif topologies¹⁴. Trees and phylogenetic profiles in Fig. 1 and Extended Data Fig. 1 were visualized using iTOL⁵⁸.

Ichthyosporea cultures and growth conditions

The different Ichthyophonid species (Sphaeroforma sp.^31,59, C. fragrantissima and A. appalachense³⁵ were provided by the laboratory of O.D., whereas C. limacisporum and C. perkinsii³⁵ were kindly provided by H. Suga and all the species originate from the Multicellgenome laboratory in Barcelona. The model ichthyosporean S. arctica with established protocols for cell synchronization, live cell imaging and cytoskeletal inhibitor assays was selected as a representative Ichthyophonid for this study^21,30,31. Our attempts at replicating the reported genetic transformation of C. fragrantissima⁶⁰ were unsuccessful. S. arctica cultures were maintained at 17 °C in marine broth (Difco, 37.4 g l⁻¹) and synchronized as previously described^21,30,31. Briefly, for synchronization, 1/16 marine broth was prepared by diluting marine broth in artificial seawater (Instant Ocean, 37 g l⁻¹). Cultures were diluted 1:100 in 1/16 marine broth and grown for 3 days to obtain saturated cultures. To obtain a synchronized culture, the saturated cultures were inoculated 1:50 in fresh marine broth. To obtain the 8–32 nuclear stage, cells were fixed around 28.5 h after inoculation. Other Sphaeroforma sp., S. gastrica, S. nootkatensis and S. napiecek and C. fragrantissima were maintained at 17 °C in marine broth similar to S. arctica. C. perkinsii, A. appalachense and C. limacisporum were grown at 23 °C protected from light. For experiments with C. perkinsii, 6-day-old cultures were filtered using a 5 µm filter to obtain small newborn cells which are then diluted 1:100 in C. perkinsii medium to obtain synchronous cultures. The cells were fixed at the one to eight cell stage (60–90 h after dilution) to capture the initial mitotic events. C. limacisporum cultures were grown in marine broth. A. appalachense cultures were grown in Amoebidium medium (yeast extract 3 g, peptone 5 g, water 1 l, autoclaved and aseptically filtered). For maintenance, cultures were diluted 1:1,000 every 2 weeks and restarted from a cryopreserved stock every 6 months.

Immunostaining

The cell culture flasks were scraped and the suspension was added to 15 ml Falcon flasks to sediment for 15–30 min. The supernatant was removed and cells were transferred to 1.5 ml microfuge tubes and fixative was added for 30 min. The cells were fixed with 4% formaldehyde in 250 mM sorbitol solution, washed twice with 1× phosphate buffer saline (PBS) and resuspended in 20–30 μl of PBS. Cells were permeabilized using nine freeze–thaw cycles (liquid N₂,10 s: 42 °C, 1 min). This was followed by blocking in 3% bovine serum albumin (BSA) in PBST (1× PBS with 0.1% Tween20). Primary antibody (Tubulin-E7 antibody DSHB, NB600-936 Novus Biologicals, AA344 and AA345 (ABCD antibodies and anti-NPC proteins—MAb414 Biolegend 902901)) was used at 1:500 to 1:1,000 and incubated at 4 °C overnight or 2–5 h at 37 °C. This was followed by three washes for 10 min at room temperature and addition of the secondary antibody. Goat anti-mouse secondary antibody, Alexa Fluor 488 (Thermo A-11001), Goat anti-guinea pig secondary antibody, Alexa Fluor 568 (Thermo A-11075), Goat anti-rabbit secondary antibody, Alexa Fluor 568 (Thermo A78955) were used as secondary antibodies at 1:500 to 1:1,000. Incubation was done at 4 °C overnight or 2–5 h at 37 °C. The cells were then washed and resuspended in fresh 1× PBS for imaging. DNA was stained with Hoechst 33352 at a final concentration of 0.4 µM. For live cell imaging, cells were stained with FM-464 at a final concentration of 10 μM.

Ultrastructural expansion microscopy

U-ExM was performed as previously described⁶¹. Briefly, the cells were fixed with 4% formaldehyde in 250 mM sorbitol solution, washed twice with 1× PBS and resuspended in 20–30 μl of PBS. The fixed cells were then allowed to attach to 12 mm poly-l-lysine-coated coverslips for 1 h. This was followed by anchoring in acrylamide/ formaldehyde (1% acrylamide/ 0.7% formaldehyde) solution for at least 5 h and up to 12 h at 37 °C. A monomer solution (19% (wt/wt) sodium acrylate (Chem Cruz, AKSci 7446-81-3), 10% (wt/wt) acrylamide (Sigma-Aldrich A4058), 0.1% (wt/wt) N,N′-methylenebisacrylamide (Sigma-Aldrich M1533) in PBS) was used for gelation and gels were allowed to polymerize for 1 h at 37 °C in a moist chamber. For denaturation, gels were transferred to the denaturation buffer (50 mM Tris pH 9.0, 200 mM NaCl, 200 mM SDS, pH to 9.0) for 15 min at room temperature and then shifted to 95 °C for 1 h. Following denaturation, expansion was performed with several water exchanges as previously described⁶¹. After expansion, gel diameter was measured and used to determine the expansion factor. For all U-ExM images, scale bars indicate actual size; rescaled for gel expansion factor. Pan labelling of U-ExM was done at 1:500 with Dylight 405 (ThermoFischer, 46400) or Alexa Fluor NHS-Ester 594 (ThermoFischer, A20004) in 1× PBS or NaHCO₃ for 1.5 h or overnight. For membrane labelling, gels were stained with BODIPY TR ceramide (ThermoFischer D7540, 2 mM stock in dimethylsulfoxide) at 1:500 dilution in 1× PBS. Immunostaining was performed as mentioned above. All antibodies were prepared in 3% PBS with 0.1% Tween 20. For A. appalachense, chemical fixation did not yield good expansion, so Cryo-ExM was adapted from previously described protocols^62,63. Briefly, cells were high-pressure frozen, followed by overnight freeze substitution in acetone with 0.25% formaldehyde and 0.05% glutaraldehyde on a metal block chilled in liquid N₂ placed in dry ice. Next day, this was followed by stepwise rehydration with 100%, 90%, 75% and 50% ethanol and finally the samples were resuspended in PBS. Samples were then crosslinked in solution on a shaker overnight to maximize crosslinking, following which gels were prepared, labelled and imaged as mentioned above.

Light microscopy

For immunolabelled cells, we used poly-l-lysine-coated Ibidi chamber slides (eight-well, Ibidi 80826). The wells were filled with 1× PBS and 0.4 µM Hoechst 33342 (ThermoFischer 62249) was added. Immunostained cells were added to wells and allowed to settle for an hour before imaging. Imaging was done on the Zeiss LSM 880 using the Airyscan Fast mode using the Zen software with the Plan-Apochromat 63×/1.4 Oil DIC M27 objective. For staining efficiency, sample overviews were imaged in LSM mode using the tilescan function with the Plan-Apochromat 63×/1.4 Oil DIC M27 objective. For immunolabelled U-ExM gels, we also used poly-l-lysine-coated Ibidi chamber slides (two-well, Ibidi 80286; four-well, Ibidi). Gels were cut to an appropriate size to fit the Ibidi chambers and added onto the wells. The gels were overlaid with water to prevent drying or shrinkage during imaging. The gels were imaged using the Zeiss LSM 880 with the Airy fast mode using a Plan-Apochromat 63×/1.4 Oil DIC M27 or an upright Leica SP8 confocal microscope with an HC PL APO 40×/1.25 glycerol objective or Nikon-CSU-W1 Sora with a SR P-Apochromat IR AC 60× WI/ 1.27 objective.

Analysis of S. arctica MTOC duplication

In hydroxyurea inhibition assays (Extended Data Fig. 3), synchronized S. arctica cultures were grown at 17 °C as mentioned above in marine broth for 16 h. At this timepoint, one sample was fixed (16 h control) and 50 mM hydroxyurea was added to the second sample and grown for a further 8 h (hydroxyurea treated) and fixed at 24 h after inoculation. An extra untreated control sample was fixed at 24 h. Cells were immunostained for tubulin and DNA as above and cells were imaged. Interphase cells were classified as one or two MTOCs. All mitotic cells were counted as having one MTOC.

Effect of inhibitor treatments on S. arctica mitosis

For microtubule perturbation in mitosis (Extended Data Fig. 6), acute treatment of low-concentration carbendazim (378674; Sigma) was used. S. arctica cells were grown as described above up to 28 h after synchronization, followed by the addition of 0.5 μg ml⁻¹ of MBC for 15 min and collected and fixed as above. For analysing the impact of microtubule depolymerization on NPC arrays in S. arctica (Extended Data Fig. 7), 24 h after synchronization, cells were treated with 25 μg ml⁻¹ of MBC for 4 h and collected and fixed as above. For impact of lipid depletion on S. arctica nuclei and mitosis (Extended Data Fig. 6), cells were treated with 25 μg ml⁻¹ cerulenin (CAS 17397-89-6; Santa Cruz Biotechnology) 6 h after synchronization. Cells were incubated for another 24 h before they were fixed.

Analysis of mitosis in multinucleate C. perkinsii cells

In the cytokinesis blocking experiments (Extended Data Fig. 10), C. perkinsii synchronization was achieved through filtration. Subsequently, the cells were cultured at 23 °C for 45 h, followed by the addition of either dimethylsulfoxide or blebbistatin (reference 72402; StemCell Technologies) at a concentration of 1 mM. The cells were further incubated in the presence of the inhibitor for 30 h before fixation. Subsequent to fixation, U-ExM was used to visualize nuclei and microtubule spindles, facilitating the enumeration of cells exhibiting multipolar spindles. Cells were pan labelled with NHS-ester to demarcate cell boundaries.

Image analysis

For immunostaining efficiency measurements (Extended Data Fig. 5a), immunostained (immunofluorescence) and expanded (U-ExM) samples were stained for microtubules (E7 antibody, DSHB). Cells were imaged using confocal microscopy in tilescan mode. Hoechst 33342 or NHS-ester was used as a reference to determine the percentage of immunostained cells.

For NPC density measurements (Fig. 2f), NPC density was determined using nuclei from U-ExM gels labelled with MAb414 (Biolegend 902901). In S. arctica NPC densities are different around the MTOC as compared to the rest of the nucleus. Thus, two regions of interest (ROIs) were selected per nucleus, one in the radial arrays in the vicinity of the MTOC and a second one away from it (marked in the graph as NE). Each ROI was a 5 µm cube. The nuclei were classified as interphase or mitotic, on the basis of nuclear shape and presence or absence of intranuclear microtubules. The images were thresholded and binarized. The three-dimensional (3D) object counter plugin was used to obtain NPC counts. The counts were divided by cube volume to obtain NPC density. The measurements were corrected for the gel expansion factor to obtain actual NPC density per µm³.

For SPB dimensions (Extended Data Fig. 3b), analysis of SPB dimension was done using pan-labelled U-ExM gels. The images were cropped to a 5 μm region around the SPB, thresholded and binarized. The ‘Analyse particle’ function was used to obtain SPB shape measurements, including width and height (Extended Data Fig. 3b). The measurements were corrected for gel expansion factor to obtain actual width and height.

For S. arctica MTOC–MTOC distance (Fig. 2b), the distance was determined using tubulin-labelled U-ExM gels. The images were thresholded using morphological filtering and binarized. The structure was then skeletonized and the Analyse Skeleton plugin (https://github.com/fiji/AnalyzeSkeleton) was used to determine spindle length. In cases for which this was not possible, MTOC positions were marked manually and Euclidean distance was calculated between the two points.

C. perkinsii centriole diameter was measured in Fiji from TEM tomography images. Both longitudinal and transversely placed centrioles were used for the analysis. Centrioles were placed at varying angles to the sectioning and imaging plane.

We performed image analysis using Fiji software^64,65. All figures were assembled with Illustrator 2022 CC 2020 (Adobe). Graphs were generated using GraphPad Prism 9. The 3D reconstructions of Supplementary Videos 3, 4, 7 and 8 were done in Imaris v.992.

Live cell imaging

Light-sheet microscopy in Supplementary Video 2 was performed using the LS1 Live 246 light-sheet microscope system (Viventis) as previously described using a 25 × 247 1.1 NA objective (CFI75 Apo 25XW; Nikon) and an sCMOS camera (Zyla 4.1 andor)³¹. Light-sheet imaging was conducted in a room specifically cooled at 17 °C using an air-conditioning unit.

Electron microscopy

A combination of three different sample preparation techniques was tried for the different ichthyosporean species and the one that yielded better results for each sample was chosen for further imaging and presented here.

Sample preparation 1

This applies to Figs. 1c, 2g (bottom) and 3f. For TEM tomography of S. arctica and C. perkinsii cells, samples were concentrated by sedimentation and high-pressure frozen with the HPM010 (Abra Fluid) using 200-μm-deep, 3-mm-wide aluminium planchettes (Wohlwend GmbH). Freeze substitution (FS) was done using the AFS2 machine (Leica microsystems) in a cocktail containing 1% OsO₄, 0.1% uranyl acetate and 5% water in acetone. The samples were incubated as follows: 73 h at −90 °C, temperature increased to −30 °C at a rate of 5 °C h⁻¹, 5 h at −30 °C, temperature increased to 0 °C at a rate of 5 °C h⁻¹, 4 × 0.5 h rinses in water-free acetone at 0 °C. This was followed by Epon 812 (Serva) infiltration without BDMA (25%—3 h at 0 °C, 50% overnight at 0 °C; 50%—4 h at room temperature; 75%—4 h, 75% overnight, 100% 4 h (×2) and 100% overnight. This was followed by exchange with 100% Epon 812 with BDMA (4 h × 2, followed by overnight). After this the samples were polymerized in the oven at 66 °C for over 2 days. The samples were then cut using an ultramicrotome (Leica UC7) in 70 nm sections screened by two-dimensional (2D) TEM (Jeol 1400 Flash) to assess sample preparation. For TEM tomography (300 nm sections), sections were poststained with 2% uranyl acetate in 70% methanol (5 min, room temperature) and in Reynolds lead citrate (2 min, room temperature).

Sample preparation 2

This applies to Figs. 1d and 2g (top) and Supplementary Videos 1 and 5. For serial tomography and FIB-SEM of S. arctica cells, samples were concentrated and high-pressure frozen as mentioned above. Freeze substitution was done in the AFS2 machine (Leica microsystems) in a cocktail containing 1% OsO₄, 0.5% uranyl acetate and 5% water in acetone. The samples were incubated as follows: 79 h at −90 °C, temperature increased to −60 °C at a rate of 2 °C h⁻¹; 10 h at −60 °C, temperature increased to −30 °C at a rate of 2 °C h⁻¹; 10 h at −30 °C, temperature increased to 0 °C at a rate of 5 °C h⁻¹; 1 h at 0 °C. After this, the samples were rinsed in acetone and further incubated in 0.1% thiocarbohydrazide, 10% water in acetone for 30 min at room temperature, followed by 1% OsO₄ in acetone. After rinsing, the samples were infiltrated in Durcupan ATM (Sigma) and finally polymerized in a 60 °C oven for 72 h. The OsO₄ step and the infiltration were performed in a Biowave (Ted Pella). The samples were then sectioned using an ultramicrotome (Leica UC7) for serial section TEM tomography (300 nm sections). The sections were poststained with 2% uranyl acetate in 70% methanol (5 min, room temperature) and in Reynolds lead citrate (2 min, room temperature). Tomograms were acquired with a Tecnai F30 (ThermoFisher Scientific) using SerialEM³⁹ and reconstructed and joined with Imod Etomo. After this, a 70 nm section was collected and screened by 2D TEM (Jeol 1400 Flash) to target interphase and mitotic cells for FIB-SEM analysis. The samples were then mounted on a SEM stub using silver conductive epoxy resin (Ted Pella), gold sputter coated (Quorum Q150RS) and imaged by FIB-SEM. The acquisition was performed using a Crossbeam 540 or 550 (Zeiss) following the Atlas 3D workflow. SEM imaging was done with an acceleration voltage of 1.5 kV and a current of 700 pA using an ESB detector (1,100 V grid). Images were acquired at 5 × 5 nm² pixel size and 8 nm slices were removed at each imaging cycle. FIB milling was performed at 700 pA current. For segmentation and visualization we used 3DMod and Amira (v.2019.3 or 2020.1; ThermoFisher Scientific).

Sample preparation 3

This applies to Extended Data Figs. 9 and 10. For TEM tomography and FIB-SEM of A. appalachense and C. limacisporum cells, we adapted a protocol shown to be compatible with FIB-SEM in ref. ⁶⁶. We concentrated and high-pressure froze the samples and FS them with 0.1% uranyl acetate in dry acetone. After 72 h of incubation at −90 °C, the temperature was increased to −45 °C at a speed of 2 °C h⁻¹ and then the samples were incubated in the uranyl acetate solution for an extra 10 h at −45 °C. The samples were then rinsed with pure acetone before infiltration with increasing concentrations of the resin Lowicryl HM20 (Polysciences), while increasing the temperature to −25 °C. The blocks were polymerized with ultraviolet for 48 h at −25 °C. Finally, the samples were mounted on stubs, prepared and imaged by FIB-SEM as described above.

Twist analysis

To calculate spindle twist for C. perkinsii, Fiji Software (ImageJ)⁶⁴ was used to analyse microscopy images of horizontal spindles. Only images with both spindle poles in the same plane or in two consecutive planes apart in each direction of the z-stack were included in the analysis to prevent spindle tilt from affecting the spindle twist calculation. Horizontal spindles were transformed into a vertical orientation (end-on view) using a previously developed code written in R programming language in RStudio⁶⁷. In the transformed stack, microtubule bundles and poles appear as blobs. The spindle poles are tracked manually using the Multipoint tool in ImageJ. Next, we used the previously developed optical flow method to calculate the twist⁶⁸ and presented the absolute values in the graph. The tracing of bundles and twist calculations were previously written in Python programming language using PyCharm IDE, with external libraries such as NumPy, scikit-image, Matplotlib, PIL, OpenCV and SciPy. The code and instructions are available at GitLab: https://gitlab.com/IBarisic/detecting-microtubules-helicity-in-microscopic-3d-images. Twist values for RPE1 cells expressing CENP-A-GFP and centrin1-GFP were taken from ref. ⁶⁸.

Analysis of spindle length, width and interkinetochore distance

To measure spindle length, width and interkinetochore distance, the Line tool in Fiji Software (ImageJ)⁶⁴ was used. Length was measured by drawing a line from pole to pole of the spindle. In C. perkinsii the pole positions were determined visually as the outermost points of the spindle along the central spindle axis. In RPE1 cells expressing CENP-A-GFP and centrin1-GFP, length was measured by using the images from ref. ²⁷ and a line was drawn from one centrosome to the other. Width in C. perkinsii was measured by drawing a line across the equatorial plane of the spindle, with the line ending at the outer edges of the spindle. Width in RPE1 cells expressing CENP-A-GFP and centrin1-GFP was measured by drawing a line across the equatorial plane of the spindle, with the line ending at the outer kinetochore pairs. Interkinetochore distance in C. perkinsii was measured by using the Rectangle tool in Fiji Software which was drawn between endings of k-fibres at the spindle midzone. Interkinetochore distance in RPE1 cells expressing CENP-A-GFP and centrin1-GFP was taken from ref. ²⁷. It was not possible to measure the interkinetochore distance in S. arctica because of the tight microtubule bundles of the spindles and it was not possible to distinguish kinetochore microtubules.

Analysis of the bridging fibre intensity

To measure the intensity of bridging fibres⁶⁹ we used the Square tool (ImageJ)⁶⁴. In C. perkinsii, the position of the square when measuring bridging fibre intensity was on the fibre located between the endings of kinetochore fibres. As we did not have labelled kinetochores and could not determine where a single kinetochore fibre is, we put squares close to the end of kinetochore fibres to obtain values of kinetochore fibres together with bridging fibres (I_bk) (Extended Data Fig. 7a). Background was measured and subtracted as follows: I_b = I_b+bcg − I_bcg for bridging fibres and I_bk = I_bk+bcg − I_bcg for bridging fibres together with kinetochore fibres. The values for RPE1 cells expressing CENP-A-GFP and centrin1-GFP were taken from ref. ²⁷.

Statistics and reproducibility

Results are reported as mean ± s.d. Statistical parameters including the numbers of cells, nuclei or MTOCs analysed, n and statistical significance are reported in the figure legends. Statistical significance was calculated by Mann–Whitney U-test, Kruskal–Wallis or Student’s t-tests. Asterisks in graphs indicate the statistical significance (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). We performed statistical analysis in GraphPad Prism 9. All light microscopy and electron microscopy images are representative images obtained from three and two independent experiments, respectively.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.