Loss of Kinesin-8 improves the robustness of the acentrosomal spindle

Chromosome segregation in female meiosis is inherently error-prone, among other reasons, because the acentrosomal spindle assembles and segregates chromosomes without the major microtubule-organizing centres in eukaryotes, the centrosomes, which causes high rate of aneuploidy. The molecular basis underlying formation of acentrosomal spindles is not as well-understood as that of centrosomal spindles and, consequently, strategies to improve spindle robustness are difficult to address. Recently, we noticed during fission yeast meiosis the formation of unexpected microtubules arrays, independent of the spindle pole bodies (yeast centrosome equivalent), with ability to segregate chromosomes. Here, we establish such microtubules formation as bonafide self-assembled spindles that depend on the canonical microtubule crosslinker Ase1/PRC1, share similar structural polarity and harbour the microtubule polymerase Alp14/XMAP215, while being independent of conventional γ-tubulin-mediated nucleation mechanisms. Remarkably, acentrosomal spindle robustness was reinforced by deletion of the Klp6/Kinesin-8, which, consequently, led to a reduced meiotic aneuploidy rate. Our results enlighten the molecular basis of acentrosomal meiosis, a crucial event in understanding gametogenesis.

To further substantiate the idea that self-assembled spindles in bqt1∆ sad1.2 meiosis share the 126 canonical features of a proper spindle, we carried out a characterization of the elements controlling its 127 dynamics. We previously observed that the microtubule crosslinker protein Ase1/PRC1 localizes to the body 128 of self-assembled spindles in very similar patterns to those seen for SPB-mediated spindles, including at the 129 spindle midzone (Pineda-Santaella and Fernández-Álvarez, 2019). Ase1 location at the midzone in self-MII ( Fig. 2 C, 105', yellow arrowhead). Strikingly, ase1∆ self-assembled spindles displayed similar behaviour compared to that of ase1∆ SPB-mediated spindles (Fig. 2 D 50' to 60', arrowheads). Congruently, for both 142 SPB-mediated and self-assembled spindles, the maximum spindle length is significantly shorter upon ase1 143 deletion ( Fig. 2 E), indicating that spindle breakage takes place in a premature manner during elongation, 144 whereas in ase1 + settings the spindle keeps elongating until normal disassembly occurs with a longer length.

145
These results establish that Ase1/PRC1 is essential for the maintenance of the structural integrity of self-146 assembled spindles in meiosis.

148
Microtubules arrangement in self-assembled spindles shows minus ends at the poles 149 We have shown that self-assembled spindles are organised around chromosomes and grow with a 150 polarity that is not determined by the SPB-LINC complex; for this reason, we wanted to establish if these 151 spindles are characterised by a normal or inverted polarity. The fact that Ase1 is required for the dynamics of

F-actin network is dispensable for self-assembly spindle formation 167
To further characterise the formation and dynamics of self-assembled spindles, we wanted to explore 168 the possible role of other cytoskeleton components. In this context, F-actin plays a role in chromosome 169 segregation in oocytes (Mogessie and Schuh, 2017). In fission yeast, F-actin network is important for proper 170 spindle orientation in mitosis (Gachet et al., 2001(Gachet et al., , 2006, but its role for meiotic spindle formation and 171 behaviour has not been yet disclosed. Given this uncertainty, we decided to explore whether F-actin network 172 is important for formation and dynamics of SPB-dependent and self-assembled spindles during fission yeast 173 meiosis. For this purpose, F-actin network was disrupted using the actin-depolymerizing drug Latrunculin A 174 (LatA), specifically during meiosis (see Methods) and then spindle behaviour was analysed. In order to follow chromosomes and microtubules. In the absence of LatA, F-actin can be observed as cables (Fig. S1 A, red arrowheads) and numerous patches which fluctuate throughout the cell body during prophase and MI ( Fig. S1 178 A, -100' to 35'), then it collapses and assembles into four meiotic actin rings surrounding post-MI nuclei during 179 MII ( Fig. S1 A, 85' to 90') and these eventually contract (Fig. S1 A, 95') and disassemble, congruent with 180 previous description in meiosis (Yan and Balasubramanian, 2012). In this setting, SPB-dependent spindles 181 display normal assembly, elongation and disassembly dynamics along with symmetrical segregation of 182 chromosomes into four equal-size masses (Fig. S1 A, 0' to 95'). Under addition of 4 µM LatA, F-actin patches 183 signal is significantly reduced (Fig. S1 B and C, yellow rectangles, 'actin fading' in down panels) (Fig. S1 C), 184 although it is recovered later on as meiosis progresses (Fig. S1 B and C, yellow rectangles, 'actin recovery' in 185 bottom panels). Additionally, actin cables are almost completely absent after the treatment (compare Fig. S1 186 A, -100' and -75' vs Fig. S1 B). Hence, we confirm that LatA is bioactive in our experimental conditions, 187 achieving a partial F-actin depolymerisation upon drug addition in meiosis. Noteworthy, the effect of the 188 depolymerizing activity of the drug extends throughout the whole meiosis as shown by the fact that actin rings, 189 taking place after actin patches recovery, are also affected ( Fig. S1 B 145' and C 180').

191
Next, we analysed the potential role of F-actin in self-assembled spindle formation and dynamics. In 192 the absence of LatA, the behaviour of F-actin patches (Fig. 4 A, -95' to 0') and cables (Fig. 4

SPB-associated γ-tubulin complex is not involved in self-assembled spindle nucleation 203
After completion of meiotic prophase, cytoplasmic astral oscillating microtubules dissolve just before 204 the formation of SPB-mediated MI and MII spindles. Nucleation of microtubules is carried out by a 205 macromolecular protein complex called the γ-tubulin complex, which serves as a structural template for 206 priming the de novo synthesis of microtubule filaments (Braunfeld et al., 2002). At the time of spindle 207 nucleation, this complex actively targets to the nuclear side of the already inserted SPBs, nucleating 208 microtubules that are then elongated via polymerization, projecting from the SPBs towards the nucleoplasm 209 (Vardy, 2000;Bestul et al., 2017). In the case of self-assembled spindles, the SPBs are not inserted into the 210 NE, thus, the possible role of γ-tubulin complex is unclear. To investigate whether self-assembled spindles 211 nucleation depends on this mechanism, a GFP-tagged version of Alp4 (Alp4-GFP), an essential component of the γ-tubulin complex (Vardy, 2000), was used as a proxy to monitor the localization of the complex relative to self-assembled spindles. In a control scenario of SPB-mediated spindles, one Alp4-GFP dot is observed 214 during meiotic prophase to localize at the leading edge of the astral microtubules structure, following its 215 oscillating movement (Fig. 5 A, -25'). After prophase, one dot of Alp4-GFP co-localizes with the microtubules 216 focus from which the spindle emerges at the time of its formation (Fig. 5 A, 0 chromosomes up to three to six masses and only a minority (32%) segregates into two masses i.e., normal 320 segregation ( Fig. 7 G, n=57). Strikingly, after klp6 deletion, the rate of chromosome missegregation decreases, from 32% to 0% for one mass and from 19% to 6% for four to six masses. Consequently, the percentage of normal segregation rises from 32% to 61% as well as that of mild missegregation, i.e. three masses, from 18% 323 to 32% (Fig. 7 G). Hence, klp6 deletion increases chromosome segregation fidelity of self-assembled spindles, 324 illustrating that spindle improvement occurs not only at the level of formation and structure, but also at the 325 functional level. 326 327 328

329
In this work, we carried out a molecular characterization of an unexpected type of self-assembled 330 spindle in fission yeast meiosis. SPB-dependent and independent spindles share similarities such as structural 331 dependence on microtubule crosslinker Ase1/PRC1, microtubule arrangement polarity and the recruitment of 332 the microtubule polymerase Alp14/XMAP215. In contrast, self-assembled spindle formation seems to be 333 independent of the LINC complex localization and γ-tubulin complex conventional nucleation. Moreover, we 334 increased self-assembled spindles robustness by deleting kinesin-8 klp6, which improved their structural 335 stability and chromosome segregation fidelity. Thus, we think that our system will be useful for future studies 336 to understand the molecular basis of the acentrosomal meiosis. 337 338

Self-assembly of SPB-independent microtubules behaves as a proper spindle 339
A major question in our work is whether the self-assembly of nuclear microtubules in the absence of

351
Among the wide range of configurations permitted by microtubules versatility, the structure and 352 polarity adopted by self-assembled microtubules repeats the scheme of SPB-dependent spindles, with 353 microtubules minus-end at the spindle poles as shown by the recapitulating localization pattern of Pkl1. We 354 hypothesize that self-assembled spindle polarity is the same than centrosomal spindles due to a set of 355 microtubule-associated factors that force microtubules to adopt the canonical spindle-type configuration instead of alternative configurations, such as that of interphase arrays. Microtubules configuration within the 357 spindle has fundamental implications for the mechanism of chromosome segregation. With minus-ends at the 358 poles facing outwards and plus-ends projecting inwards to chromosomes, we speculate that the mechanism is 359 likely to involve chromosome capturing by microtubules plus-ends and pulling to opposite poles driven by 360 their shrinkage towards opposite poles, as for SPB-mediated spindles (Tanaka et al., 2005). In addition, the poleward movement of chromosomes that we observe could be complemented by supporting mechanisms such Our observations suggest that conventional γ-tubulin complex-driven nucleation may not be involved 384 in self-assembled spindle formation since we did not observe accumulation of Alp4, one of the essential 385 members of the complex at the self-assembled spindle structure. However, we cannot totally rule out this 386 possibility. For instance, the accumulation of Alp4 molecules at the SPBs in a SPB-dependent spindle context 387 might be necessary for triggering SPB insertion into the NE instead for the nucleation of the SPB-dependent 388 spindles. Accordingly, Alp4-GFP dots show similar behaviour between inserted and uninserted SPBs (Fig. 5B,   389 75', yellow asterisk), which strongly suggests that these dots correspond to Alp4 molecules associated to the On the other hand, other alternative explanation to the absence of Alp4 signal in the vicinity of forming self-assembled spindle might derive from the fact that self-assembled spindles show a thinner structure than 394 SPB-dependent spindles, which indicates a lower number of microtubules. We hypothesize that self-assembled 395 spindles in our system would then need a smaller amount of nucleation factors, among them the γ-tubulin 396 complex, than the canonical spindle. Hence, these factors could not be easily detectable by live fluorescence 397 microscopy systems used in this study.  (Fig. 8 A and B). Longer length of klp6-deleted 437 self-assembled spindles could be explained by microtubule overgrowth derived from enhanced polymerization, 438 as well as by microtubule hyper-stabilization, which would confer the spindle resistance against microtubule-439 destabilizing mechanisms responsible for spindle disassembly, letting it reach a longer length (Fig. 8 A and 440 B). Enhanced polymerization would as well promote self-assembled spindle formation, explaining the increase 441 in the appearance of self-assembled spindles in MII (Fig. 8 B (Fig. 7). However, in our system, deletion of Klp6 444 significantly improves chromosome segregation carried out by self-assembled spindles, probably as a 445 consequence of reinforced spindle structure and kinetochore-microtubule interaction stability, which might 446 facilitate kinetochore attachment (Fig. 8 B). Hence, our results show a successful strategy aimed to improve 447 the acentrosomal spindle formation and function (Fig. 8 C), which we estimate not only relevant for future 448 studies in fission yeast meiosis but also to gain insight into mammalian female meiosis.

789
In the SPB-mediated spindles background, Alp4 localizes as one dot at the forefront of the oscillating 790 microtubules in the meiotic prophase (-25') and later localizes as one dot at each pole of MI (35' to 45') and arrowheads). After spindle disassembly, microtubules start to form around this Alp4 dot (45' to 75', orange asterisks), which appear to split into several dots (75', yellow asterisk). Alp4: Alp4-GFP.   ase1Δ SPB-mediated spindles ase1Δ Self-assembled spindles