Abstract
In fission yeast, the length of interphase microtubule (iMT) arrays are adapted to cell length so as to maintain cell polarity and to help centre the nucleus and cell division ring. Here we show that length regulation of iMTs is dictated by spatially-regulated competition between MT-stabilising Tea2/Tip1/Mal3 (Kinesin-7) and MT-destabilising Klp5/Klp6/Mcp1 (Kinesin-8) complexes at iMT plus tips. During MT growth, the Tea2/Tip1/Mal3 complex remains bound to the plus tips of iMT bundles and restricts access to the plus tips by Klp5/Klp6/Mcp1, which accumulates behind it. At cell ends, Klp5/Klp6/Mcp1 invades the space occupied by the Tea2/Tip1/Tea1 kinesin complex triggering its displacement from iMT plus tips and MT catastrophe. These data show that in vivo, whilst the “antenna model” for iMT length- and age-dependent catastrophase accumulation has validity, length control is an emergent property reflecting spatially-regulated competition between multiple complexes at the MT plus tip.
Introduction
Microtubule (MT) length control is important for multiple cellular processes including vesicle transport, mitotic spindle size, chromosome bi-orientation and cilia function (Kuan & Betterton, 2013; Niwa et al., 2012; Wang et al., 2016; Wordeman & Stumpff, 2009). In the fission yeast, Schizosaccharomyces pombe, arrays of interphase microtubules (iMT) grow along the long axis of the cell and undergo catastrophe at cell ends. Interaction of iMTs with the cell end cortex is required to maintain cell polarity and correctly position the nucleus and cell division ring (Huang et al., 2007; Martin, 2009; Tran et al., 2001). The maintenance of cell polarity and positioning of the division ring requires transport of the Kelch-repeat protein Tea1 and SH3-domain protein Tea4 on the plus tips of iMTs by association with Tea2 (Kinesin-7) and their deposition at cell ends following interaction of iMTs with the cell end cortex (Martin, 2009; Martin et al., 2005; Mata & Nurse, 1997; Sawin & Snaith, 2004; Snaith et al., 2005; Tatebe et al., 2005). Reconstitution and live-cell imaging experiments reveal that association of Tea2 (Kinesin-7) to the growing plus tips of MTs requires two other components, Mal3 (EB1 homologue) and Tip1 (Clip170 homologue) (Akhmanova & Steinmetz, 2008; Bieling et al., 2007; Busch & Brunner, 2004; Busch et al., 2004). The presence of the Tea2/Tip1/Mal3 complex, but not its cargo, at the MT plus tip also prevents premature MT catastrophe in the cytoplasm (Verde et al., 1995). Members of the Kinesin-8 family have attracted particular attention as regulators of MT length because they are both highly processive motors and undergo a conformational switch at the MT plus tip that promotes MT disassembly (Arellano-Santoyo et al., 2017; Su et al., 2012; Su et al., 2011; Tischer et al., 2009; Varga et al., 2006; Varga et al., 2009). These features have given rise to the “antenna model” for MT length control whereby more Kinesin-8 accumulates at the plus tips of longer MTs, thus increasing the likelihood of catastrophe and MT shrinkage (Leduc et al., 2012; Varga et al., 2006; Varga et al., 2009). Fission yeast contains two Kinesin-8 motors, Klp5 and Klp6, that function as an obligate heterodimer. Deletion of either gene causes numerous mitotic defects and overgrowth of interphase MTs, which result in defective nuclear positioning and loss of cell polarity, particularly in longer cells (Garcia et al., 2002; Grissom et al., 2009; Sanchez-Perez et al., 2005; West et al., 2002; West et al., 2001). Timely shrinkage of iMTs also requires Mcp1, a +TIP that is distantly related to the Ase1/PRC1/MAP65 family of anti-parallel MT binding proteins (Akhmanova & Steinmetz, 2010; Pardo & Nurse, 2005; Zheng et al., 2014). However, the precise role and relationship of these factors in controlling interphase MT length is unknown.
Results and Discussion
We find that, like Klp5 and Klp6, the intensity of Mcp1 increases at plus tips of iMTs as they grow and dwell at cell ends and decreases as iMTs undergo shrinkage (Figure 1A & Figure 1 - figure supplement 1A). Importantly, binding of Mcp1 to iMT plus tips requires the motor activity of Klp5/Klp6 indicating that Mcp1 is a cargo of the Klp5/Klp6 complex (Figure 1 - figure supplement 1B-C; Zheng et al., 2014). Indeed, we find that Mcp1 binds weakly to Klp5 in co-immunoprecipitates from cell extracts (Figure 1 - figure supplement 1D). Importantly, the dwell time of iMTs at the cell end is extended in the absence of both Klp5 and Klp6 to the same extent as in the absence of Mcp1 and this effect is not additive, indicating that Mcp1 controls destabilisation of iMTs via its association with the Klp5/Klp6 complex (Figure 1B). Notably though, unlike deletion of either Klp5 or Klp6, loss of Mcp1 does not cause cell polarity defects in elongated cdc25-22 cells (Figure 1 - figure supplement 2: West et al., 2001) and does not influence mitotic timing or accuracy of chromosome segregation (Figure 1 – figure supplement 3A-E). These functions may instead be due to association of Klp5/Klp6 to PP1, type-1-phosphatase (Dis2) (Meadows et al., 2011). Consistently, Mcp1 is not required for Klp5 and Klp6 to bind the mitotic spindle or kinetochores during mitosis and is not present in the nucleus during mitosis (Figure 1 - figure supplement 3F-G). We conclude, therefore, that Mcp1 is an interphase-specific regulator of Kinesin-8 mediated iMT length control in fission yeast.
To further understand the role of Mcp1 in Kinesin-8 function, we examined Klp5/Klp6 motility and accumulation at iMT plus tips in control and Δmcp1 cells by dual camera live-cell imaging (Figure 1C). We find that Klp5/Klp6 walks along the lattice of iMTs at 134 ± 28 nm/s, compared to average iMT growth speed of 69 ± 20 nm/s and accumulates at iMT plus ends, particularly as iMTs dwell at the cell end, and then dissociates following MT catastrophe (Figure 1C-E). In fact, Klp5/Klp6 walks faster (168 ± 37 nm/s) along the lattice of iMTs in the absence of Mcp1, although the growth speed of iMTs is unaffected. These data indicate that Mcp1 is required for the iMT destabilising function of Klp5/Klp6, perhaps by promoting its association to curved tubulin, but not for its processive motility. Instead it is likely that, like KIP3 and KIF18A, the highly processive motility of Klp5/Klp6 depends on MT binding site(s) in the C-terminal tail of those proteins (Mayr et al., 2011; Stumpff et al., 2011; Su et al., 2011; Weaver et al., 2011). To quantify the intensity of Klp5/Klp6 at plus tips, mixed populations of cells lacking Mcp1 and control cells were imaged in the same field of view. Although the absence of Mcp1 does not influence Klp5 or Klp6 stability, as judged by western blot of whole cell extracts (Figure 1 - figure supplement 1E), nor its intensity in the nucleus, Klp5/Klp6 accumulated to approximately half the level at iMT plus tips in the absence of Mcp1, even though iMTs dwell for longer at cell ends (Figure 1F). This may reflect a role for Mcp1 in enhancing the affinity of Klp5/Klp6 for the iMT lattice, resulting in additional runs to plus tips, or a role for Mcp1 in either retaining Klp5/Klp6 at plus tips or controlling the oligomeric status of the Klp5/Klp6 complex. Further work is needed to distinguish between these possibilities.
We next examined the functional relationship between the Klp5/Klp6/Mcp1 and Tea2/Tip1/Mal3 kinesin complexes. In contrast to Klp5/Klp6/Mcp1, which accumulates in both a MT length- and dwell time-dependent manner, binding of Tea2 kinesin to plus tips is independent of MT length (Figure 2A) but, like Klp5/Klp6/Mcp1, Tea2 dissociates from plus tips following MT catastrophe. To our surprise, we find that Tea2 remains bound, at the same average intensity, to stalled MT plus tips in the absence of Klp6 or Mcp1 (Figure 2B-C & Figure 2 – figure supplement 1A-B), indicating that the Klp5/Klp6/Mcp1 complex is required for timely dissociation of the Tea2 complex from plus tips. In the absence of Tea2, iMTs dwell at cell ends for less time than in control cells and undergo frequent MT catastrophe in the cytoplasm before iMTs reach the cell end, as previously observed (Figure 2D-F: Verde et al., 1995). Importantly, we find that deletion of klp5 and klp6 or mcp1 in Δtea2 cells largely restores MT catastrophe near the cell end and increases MT dwell time, although not quite to that observed in control cells (Figure 2E-F). A similar effect is seen in the absence of Tip1, but not in the absence of Tea1 (Figure 2 - figure supplement 2). This suggests that, in the absence of the Tea2/Tip1/Mal3 complex, but not its cargo, Klp5/Klp6/Mcp1 induces premature MT catastrophe in the cytoplasm. Consistently, Klp5/Klp6 accumulates at iMT plus tips in the absence of Tea2, albeit to a lower intensity than in control cells, as iMTs rarely last long enough to reach the cell end (Figure 2 - figure supplement 3A-B). In the absence of both Tea2 and Mcp1, Klp5/Klp6 accumulates at the plus tips of iMTs that reach the cell end, but at greatly reduced levels compared to control cells (Figure 2 - figure supplement 3C-D). In this situation, iMT catastrophe may purely be reliant on the force exerted by interaction of the plus tip with the cortex at the cell end (Tischer et al., 2009).
We next considered how the antagonistic activities of the MT-stabilising Tea2/Tip1/Mal3 and MT-destabilising Klp5/Klp6/Mcp1 complexes are coordinated in space and time. One possibility is that they compete for the same physical space at the iMT plus tip. As Klp5/Klp6/Mcp1 concentration increases it may physically displace the Tea2/Tip1/Mal3 complex from the plus tip and induce MT catastrophe. An alternative, but not necessarily exclusive, possibility is that the Tea2/Tip1/Mal3 complex may occlude the Klp5/Klp6/Mcp1 complex from the plus tip until the iMT encounters the cell end. To test these models, we monitored the relative positions and levels of components of the Tea2/Tip1/Mal3 and Klp5/Klp6/Mcp1 complexes at the iMT plus tip. Importantly, dual imaging of tea2-GFP tip1-tdTomato cells shows that Tea2 and Tip1 occupy the same region of the MT plus tip and remain at the same intensity both before and after MT contact with the cell end cortex, consistent with the fact that these proteins are components of the same complex (Figure 3 - figure supplement 1A-B). By contrast, dual imaging of klp5-mNG klp6-mNG tip1-tdTomato cells reveals that Klp5/Klp6 accumulates behind Tip1 as the iMT grows and dwells at the cell end, as judged by maximal intensity measurements of the fluorescence (Figure 3A-B & Figure 3 - figure supplement 2). Importantly, the Tea2/Tip1/Mal3 and Klp5/Klp6/Mcp1 complexes converge at the plus tip approximately 10 seconds before MT shrinkage is detected. This behaviour is also observed in the absence of Mcp1, although in this case Klp5/Klp6 remains behind the Tea2/Tip1/Mal3 complex at the plus tip for longer as the MT dwells at the cell end (Figure 3C & Figure 3 - figure supplement 3).
These results indicate that the “antenna model” for MT age- and length-dependent accumulation of Kinesin-8 has validity but cannot solely explain the position-dependent interphase function of Klp5/Klp6 in fission yeast. The Klp6/Klp6/Mcp1 complex accumulates at MT plus tips over time, but Kinesin-8 walks only marginally faster than iMT polymerisation, so that its accumulation is largely dependent on MT growth slowing at cell ends rather than on MT length. More importantly, our data suggest a new model to explain iMT length control in fission yeast, in which the MT-stabilising Tea2/Tip1/Mal3 Kinesin complex binds ahead of the MT-destabilising Klp5/Klp6/Mcp1 complex and restricts its access to the iMT plus tip until the MT reaches the cell end (Figure 4). The earliest events in the sequence that results in MT catastrophe remain unclear, but we do see clearly that a slowing of iMT plus tip growth precedes any measurable change in the Tea2 complex signal. One possibility is that slowing down of iMT growth near cell ends allows Kinesin-8 to build up to a critical concentration that enables it to invade the space occupied by Tea2 and/or displace it from plus tips. Regardless, our work emphasises that iMT length control in cells is a complex emergent property reflecting competition and collaboration between multiple factors that influence tubulin dynamics. In the case of fission yeast, MT-destabilising Kinesin-8 motors accumulate at plus ends in a manner that is dependent on MT length and lifetime (as in the antenna model), but access to the MT plus end is controlled in time and space by an engagement mechanism that slows growth and favours unloading of the Tea2 complex. These two mechanisms act as separable layers of control that overlie the fundamental mechano-sensitive dynamics of tubulin itself at iMT plus ends.
Materials and Methods
Cell Culture and Strain Construction
Media and the growth and maintenance of strains were as described previously (Moreno et al., 1991). All experiments were performed at 29.5°C unless stated. A full list of strains and oligonucleotides used in this study can be found in Supplementary File 1 and Supplementary File 2 respectively.
Deletion of mcp1 with natMX6, hphMX6 and kanMX6, C-terminal tagging of mcp1 with GFP, and C-terminal tagging of klp5 and klp6 with mNeonGreen were carried out by two-step PCR gene targeting as described previously (Krawchuk & Wahls, 1999). The mcp1, klp5 and klp6 genes were also C-terminally tagged with RFP, mCherry or TdTomato. However, the fluorescence intensity of the resultant strains was too low and bleached too quickly to enable time-lapse live-cell imaging. N-terminally tagged lys1::nmt1-GFP-mcp1 was constructed by amplifying the mcp1 ORF flanked by attB1 sites. The resulting PCR product was cloned into pDONR221TM and then shuttled to pLYS1U-HFG1c using Gateway LR clonase II enzyme mix (Invitrogen, USA). Following linearisation with NruI this construct was integrated into S. pombe strains at the lys1 locus and confirmed with PCR.
Live-Cell Fluorescence Microscopy
Live-cell analyses were performed in imaging chambers (CoverWell PCI-2.5, Grace Bio-Labs) filled with 1 mL of 1% agarose in minimal medium and sealed with a 22 × 22 mm glass coverslip.
Figure 1B, Figure 1 - figure supplement 1A-C, Figure 2A, Figure 2E, and Figure 2 - figure supplement 2 used a single-camera Andor spinning disk confocal system comprising of a Nikon Ti-Eclipse microscope base equipped with a Plan Apo Vc 100x/1.40 N.A. objective lens, a Yokogawa CSU-XI confocal unit, an Andor iXon Ultra EMCCD camera with 2x magnification adapter and Andor iQ3 software. 488 nm and 561 nm solid state diode (power 50/100 mW) lasers were used for excitation. All images were processed and analysed using ImageJ (Fiji Life-Line version). Temperature was maintained at 29.5°C by a cage incubator controlled by OKO-touch (OKOLAB, Italy)
Figure 1C-F, Figure 1 - figure supplement 3F-G, Figure 2B-D,F, Figure 2 - figure supplement 1 & 3, Figure 3, Figure 3 - figure supplement 1-3 were generated using an Andor TuCam spinning disk confocal system comprising of a Nikon Ti-Eclipse microscope base equipped with a 100x/1.45 NA objective lens (Nikon CFI Plan Apo Lambda), a Nikon PFS (perfect focus system), a Yokogawa CSU-X confocal unit, and 2 Andor iXon Ultra EMCCD cameras with 2x magnification adapter and Andor TuCam two camera imaging adapter (beam splitter: Semrock DiR561). The final pixel size with 100x lens is 69 nm/pixel (measured). The light source is an Andor ILE laser unit equipped with solid state 50 mW lasers (488 & 561 nm). Temperature was maintained at 29.5°C by an OKO Lab hated enclosure.
Registration of Images in the TuCam System
Registration of the 2 channels recorded simultaneously by 2 separate cameras was performed after acquisition using a reference image and ImageJ plugins for both image registration and transformation. Before every experiment, 3 2-channel reference images of 0.5 µm beads (calibration slide, Tetraspec Molecular Probes) in both the mNeonGreen/GFP and TdTomato/mCherry channels were captured. Following the experiment, reference images were again collected to compare for changes in calibration during the imaging session. Next, the channels in one of the reference images were registered using similarity transformation in the ImageJ plugin RVSS (Register Virtual Stack Slices) which performs registration using translation, rotation and isotropic scaling. This file was further tested on the remaining 2-channel reference images using the ImageJ plugin TVSS (Transform Virtual Stack Slices) and the transformation file saved for further use. Registered images were inspected to confirm correct alignment of both channels across the whole field of view. Finally, the transformation file was used to register the channels of the corresponding batch of experimental data.
Kymograph Generation and Analysis
Kymographs were produced from data generated on the TuCam system. Mid-log phase cells were imaged for both green (mNeonGreen and GFP) and red (TdTomato and mCherry) tags simultaneously with 150 ms exposure every 200 ms for 2 minutes using 6% laser power for both channels. PFS was set to ON for all time points. Camera noise was suppressed by filtering all images using the Mean Filter in ImageJ set to 1×1. A single focal plane was used to increase temporal resolution. Only microtubules that remained in this plane throughout image acquisition were used to generate kymographs with the ImageJ plugin KymoResliceWide set to generate average maximum intensities for each frame perpendicular to a manually drawn line 10 pixels wide with spline fit.
Kymographs were analysed for fluorescence levels at the MT plus tip relative to MT length using a manually drawn line. MT shrinkage was defined as the first frame in which unambiguous MT shrinkage was first observed. MT growth speed was measured by calculating the gradient of mCherry through time during the portion of the kymograph where MTs were not in contact with the cortex. Klp5/Klp6 speed was measured by calculating the gradient of lines tracking mNeonGreen puncta on MT lattices. For the analyses presented in Figure 3 and Figure 3 - figure supplement 1 & 3 relating to maximal intensity pixel values, both channels from kymographs were saved as text images giving the intensity of every pixel for both channels. The pixel with maximal intensity was then determined for each channel at every time point and its position and fluorescence level recorded. The numbered plots that correspond to the numbered 20 pixels (4 s) wide segments of the kymographs show averaged fluorescence levels for each channel (relative to 100% maximum intensity for each trace) and position. These plots were generated using the Gel Analyzer tool in ImageJ.
Relative Fluorescence Intensity Quantification
To compare the levels of the fluorescently-tagged proteins in two strain backgrounds, we performed mixing experiments to allow quantification within the same field of view on the TuCam system. Mid-log phase cells were excited for both green (mNeonGreen and GFP) and red (mCherry) tags simultaneously with 150 ms exposure for each of 11 Z sections (0.3 µm apart) every 2 s for 4 minutes using 6% laser power for both channels. PFS was initialised every 5 time points. Camera noise was suppressed by filtering all images using the Mean Filter in ImageJ set to 1×1 and Z stacks were flattened by maximum intensity projection. Fluorescence levels were quantified using a 5 x 5 pixel oval region of interest and determining the mean value. For plus tip values this signal had to be associated with the end of an iMT and was recorded in the frame immediately prior to MT shrinkage, for nuclear signals measurements were taken in the first frame. Background subtraction was performed against the cytoplasmic signal in all instances. Mitotic cells were excluded from the analysis.
iMT Dwell Time Analysis
Cells expressing GFP-atb2 were grown until mid-log phase and then microtubules imaged live using the single-camera Andor spinning disk confocal system. 8 z-stacks (0.5 μm apart) were taken with exposure times of 100 ms every 5 s for 10 minutes for GFP. These time-lapse series were then flattened in the Z dimension by maximum intensity projection in ImageJ. Dwell time was defined as the length of time the iMT plus end remained in the final 1.1 μm of a cell, with the cell perimeter defined by background GFP fluorescence. 1.1 μm is the mean length (plus one standard deviation) for the curved end of wild type cells and accounts for 84% of shrinkage events. Dwell times were measured for the full 10 minutes of each movie. ∼100 dwell times were measured for each condition.
Imaging of plus tip proteins relative to iMT length
For Figure 1 - figure supplement 1A-C and Figure 2A, mid-log phase cells expressing fluorescently-tagged Klp5, Klp6, Tea2 or Mcp1, together with fluorescently-tagged MTs, were imaged live on the single-camera Andor system. 8 z-stacks (0.5 μm apart) were taken with green fluorophores (GFP and mNeonGreen) and red fluorophores (mCherry) excited for 200 ms every 4.8 s for no more than 5 minutes. Movement of Mcp1 puncta over time (Figure 1 – figure supplement 1B-C) were tracked using the ImageJ plugin MTrackJ. Klp5/Klp6 and Tea2 levels at plus tips relative to MT length (Figure 2A) were analysed manually with mNeonGreen and GFP levels background substituted and normalised against the mCherry mean intensity of iMTs. iMT length was measured by using the line tool in ImageJ with the anti-parallel MT bundle used to define the minus end. Measurements were performed over 3 independent experiments.
Measurement of the Pre-anaphase Mitotic Index
Mid-log phase fta3-GFP sid4-TdTomato, or cdc13-GFP sid4-TdTomato cells were grown at 30°C and, following fixation in 3.7% formaldehyde for 10 minutes, mounted in medium containing DAPI to label DNA. Stacks of 18 z sections (0.2 μm apart) were taken on a wide field microscope system using a Nikon TE-2000 base with a 100x/1.49 N.A. objective lens equipped with a Photometrics Coolsnap-HQ2 liquid cooled CCD) camera and analysed using MetaMorph (version 7.5.2.0, MAG, Biosystems Software). The percentage of cells with Cdc13-GFP on the spindle pole bodies and mitotic spindle or the percentage of cells with kinetochores between separated poles prior to anaphase was determined. Exposure times of 1 s were used for GFP and TdTomato and 0.25 s for DAPI. For each experiment, at least 500 cells were counted, and each experiment was conducted three times.
Analysis of Cell Curvature
Since polarity defects are emphasised in longer cells, cdc25-22 cells were grown at 28°C until mid-log phase at which point they were shifted to 35.5°C for 6 hours to arrest at G2/M and then imaged live on the Nikon wide field system. 18 z-stacks (0.2 µm apart) were taken with a 100 ms exposure time for bright-field. Each experiment analysed at least 280 cells, was repeated 3 times and excluded cells that were less than 14 µm long. The z-stack containing the middle of the cell was used for analysis, whereby a line was drawn manually between the two cell ends in the middle of the cell using the line tool in ImageJ. The length of this line (L, length) was then compared to that of a straight line between the cell ends (E, Euclidean distance) using the measure function of ImageJ. The ratio between these values, converted to a percentage, gives a measure of the curvature of each cell. The distribution of cell lengths was unaltered between backgrounds.
Biochemistry
Cells were lysed in buffer containing 50 mM HEPES (pH 7.6), 75 mM KCl, 1 mM MgCl2, 1 mM EGTA, 0.1% Triton X-100, 1 mM DTT, 1 mM PMSF and complete EDTA-free protease inhibitor cocktail (Roche). Protein concentration was quantified by Bradford assay (Biorad, USA). For co-immunoprecipitation, 2 mg protein extract was incubated with 5 μL (∼ 5 μg) rabbit αGFP (Immune Systems) (Immune sera, I) or 2.5 µL (∼5 μg) normal rabbit serum (Pre-Immune sera, PI) for 30 minutes and subsequently with 20 µL protein A-sepharose beads (GE Healthcare, USA) for 45 minutes. Following centrifugation at 4°C, beads were washed 3 times in lysis buffer before being heated at 100°C in SDS-sample buffer for 6 minutes.
Precipitates using immune (I) or pre-immune (PI) sera and whole cell extracts (WCE) were migrated using SDS-PAGE and transferred to nitrocellulose membrane. GFP was detected using anti-GFP sheep polyclonal (1/5000, Kevin Hardwick) and HRP-conjugated anti-sheep secondary antibody (1/10000, Jackson ImmunoResearch, USA). Myc was detected using anti-Myc mouse monoclonal (1/500, Abcam, UK) and HRP-conjugated anti-mouse secondary antibody (1/10000, GE Health Care, USA). Tubulin was detected by anti-TAT1 mouse monoclonal (1/1000, Keith Gull) and HRP-conjugated anti-mouse secondary antibody (1/10000, GE Health Care, USA). ECL detection was performed using Amersham ECL system (GE Healthcare).
Statistics
Since some data sets were not normally distributed, as ascertained by the Shapiro-Wilk test, the nonparametric two-sample Kolmogorov-Smirnov test was used throughout to determine the probability that two data-sets come from different distributions. Text files were manipulated in Excel (version 15.33, Microsoft) and data were analysed by R (version 3.3.3, R Core Team) and RStudio (version 1.0.136, RStudio, Inc).
Funding
This work was funded by a Medical Research Council UK programme grant (MR/K001000/1) to J.B.A.M, and Wellcome Trust Senior Investigator Awards to R.A.C. (103895/Z/14/Z) and M.K.B. (WT101885MA). J.C.M. was funded by a University of Warwick, Institute of Advanced Study, Global Research Fellowship. L.J.M. was funded by a University of Warwick, Chancellor’s Scholarship and enrolled in their Medical Research Council UK funded Doctoral Training Partnership.
Competing Interests
There are no competing interests to declare.
Acknowledgements
The authors are grateful to Kevin Hardwick and Keith Gull for their gifts of antibodies and to Fred Chang, Ken Sawin, Takashi Toda, Xiangwei He, Mitsuhiro Yanagida, Jean-Paul Javerzat and Paul Nurse for providing fission yeast strains.