Disruption of augmin-mediated microtubule nucleation in neural stem cells causes p53-dependent apoptosis and aborts brain development

Microtubules that assemble the mitotic spindle are generated by three different mechanisms: centrosomal nucleation, chromatin-mediated nucleation, and nucleation from the surface of other microtubules mediated by the augmin complex. Impairment of centrosomal nucleation in apical progenitors of the developing mouse brain induces p53-dependent apoptosis and causes non-lethal microcephaly. Whether disruption of non-centrosomal nucleation has similar effects is unclear. Here we show, using mouse embryos, that conditional knockout of the augmin subunit Haus6 in apical progenitors led to spindle defects and mitotic delay. This triggered massive apoptosis and complete abortion of brain development. Co-deletion of p53 rescued cell death, but brain development was still aborted. This could be explained by exacerbated mitotic errors and resulting chromosomal defects including increased DNA damage. Surviving progenitors had lost apico-basal polarity and failed to organize a pseudostratified epithelium. Thus, in contrast to the centrosomal nucleation pathway, augmin is crucial for apical progenitor mitosis, and, even in the absence of p53, for progression of brain development.


Introduction 42
Spindle assembly crucially depends on microtubule nucleation by the -tubulin ring 43 complex ( TuRC). During mitosis TuRC generates microtubules through three 44 different pathways: centrosomal nucleation, chromatin-mediated nucleation, and 45 nucleation from the surface of other microtubules [1][2][3]. The latter mechanism is 46 mediated by the augmin complex and has been referred to as a microtubule 47 amplification mechanism [4][5][6][7]. Augmin binds to the lattice of microtubules generated 48 6 S1a) [40,41]. In contrast to the full knockout, Haus6 conditional knockout (Haus6 124 cKO) mice passed through all developmental stages and at E13.5 we observed 125 efficient deletion of Haus6 in the brain (Fig. S1b). Whereas mice with a heterozygous 126 Haus6 deletion developed normally and were fertile, homozygous Haus6 cKO mice 127 died around birth. Analysis of Haus6 cKO animals at E17.5 showed severe defects in 128 brain development, whereas overall body development appeared normal (Fig. 1b, c;  129 Fig. S1c). Histopathology analysis revealed a strong disruption or absence of 130 different forebrain structures (cortex, thalamus, hypothalamus) and of the cerebellum 131 ( Fig. 1c; Fig. S1c). To evaluate whether this was due to agenesis or tissue loss 132 during development, we analyzed embryos at E13.5. Even at this earlier stage brains 133 in Haus6 cKO embryos displayed severe defects compared to control embryos. 134 Lateral cortexes in Haus6 cKO embryos were almost completely absent and 135 thalamus structures, while partially formed, displayed a strong reduction in radial 136 thickness (Fig. 1d). Moreover, spaces between tissue structures were filled with 137 cellular debris. These data suggest that, in Haus6 cKO brains at early developmental 138 stages, formation of structures that would give rise to the cortex, thalamus and 139 hypothalamus is initiated but not completed, leading to tissue loss and abortion of 140 brain development at later stages. 141 142

Loss of augmin impairs mitotic progression in cortical and thalamic neural 143 progenitors 144
To analyze defective brain development in Haus6 cKO animals at E13.5 at the 145 cellular level we focused on the thalamus, which was at least partially preserved. We 146 co-stained brain sections with antibodies against PAX6 and βIII-tubulin to label apical 147 progenitors and neurons, respectively. In Haus6 cKO embryos we observed that the 148 reduced radial thickness in the thalamus was due to a striking thinning of the 149 neuronal layer by ~90% when compared to controls (Fig. 1e, f), indicating severely 150 impaired neurogenesis. In some parts, where tissue organization appeared to be 151 disrupted, we also observed neurons that were misplaced in apical regions (Fig. 1e). 152 To directly test if augmin deficiency impaired mitoses, we identified and quantified 153 mitotic cells in the thalamus using Ser10-phospho-Histone H3 staining. In Haus6 154 cKO embryos we observed a ~4-fold increase in the number of mitotic cells in the 155 region closest to the ventricular surface compared to controls, whereas there were 156 no significant differences in more basal regions (Fig. 2a, b). The percentage of 157 Haus6 cKO mitotic cells in prometaphase was strongly increased, whereas 158 metaphases and ana/telophases were reduced relative to controls (Fig. 2c). Together 159 these observations suggest that augmin deficiency in progenitors of the thalamus 160 leads to a defect in progression to metaphase, causing mitotic delay. 161 To analyze cortical progenitors and since there were no intact cortical 162 structures in Haus6 cKO brains at E13.5, we analyzed embryos at E11.5. At this 163 stage cortical structures were present suggesting that, as for the thalamus, cortical 164 tissue is originally formed but lost at later stages. Similar to the situation in the 165 thalamus at E13.5, in Haus6 cKO cortexes at E11.5 the percentage of mitotic 166 progenitors was increased when compared to controls and this occurred specifically 167 in the apical region and not in more basal layers. Again this increase in mitotic cells 168 was due to accumulation in prometaphase (Fig. S2a, b, c, d). Together the data 169 show that augmin plays an important role in allowing timely mitotic progression of 170 apical progenitors in different regions of the developing mouse brain. 171 To test if augmin-deficient progenitors displayed spindle defects we analyzed 172 brain sections with antibodies against -tubulin and -tubulin (Fig. 2d, e, f, g). Mitotic 173 8 apical progenitors in the thalamus of control animals displayed strong, centrosomal 174 staining of -tubulin at spindle poles and more diffuse -tubulin signals along spindle 175 microtubules. In Haus6 cKO embryos -tubulin could not be detected on spindle 176 microtubules. Moreover, in ~56% of cells the staining of -tubulin at spindle poles 177 was dispersed into multiple smaller foci, consistent with pole fragmentation (Fig. 2d, 178 e). Indeed, labeling of microtubules by -tubulin antibodies revealed abnormal 179 spindle configurations, frequently with multiple poles, in about half of the mitotic 180 progenitors in Haus6 cKO animals (Fig. 2f, g). However, bipolar configurations 181 including at ana/telophase were also observed and cell divisions occurred in Haus6 182 cKO progenitors, suggesting that in some cases multipolarity may be avoided 183 through pole clustering and that mitosis was not completely blocked. 184 Considering that augmin-deficiency caused pole fragmentation, we wondered 185 whether this affected spindle positioning. We measured spindle angles relative to the 186 ventricular surface in dividing apical progenitors in the thalamus and in the cortex of 187 E13.5 and E11.5 Haus6 cKO embryos, respectively. We found that in both cases the 188 majority of spindles axes were oriented horizontally similar to spindles in control cells 189

Loss of augmin in neural progenitors induces p53 expression and apoptosis 198
We sought to determine the fate of progenitors undergoing abnormal mitoses after 199 loss of augmin. We probed thalamus and cortex of E13.5 and E11.5 Haus6 cKO 200 embryos, respectively, for p53 induction and the presence of the apoptotic marker 201 cleaved caspase-3. Indeed, p53 and cleaved caspase-3 were strongly upregulated in 202 both brain regions (Fig. 2h, i; Fig. S2h), whereas cells positive for these markers 203 were barely found in the corresponding tissues of control embryos. To reveal the 204 identity of cells overexpressing p53, we performed a triple staining with antibodies 205 against p53, the neuronal marker III-tubulin and the apical progenitor marker PAX6 206 ( Fig. 2j). This experiment showed that in the Haus6 cKO thalamus ~87% of the p53-207 positive cells were also positive for PAX6 and only a minor fraction (~5%) for III-208 tubulin (Fig. 2k). Moreover, we observed that PAX6-positive progenitors displaying 209 p53 induction were exclusively interphase cells, based on the presence of intact 210 nuclei. We concluded that p53 induction occurred specifically in augmin-deficient 211 progenitors, after exit from abnormal mitoses. 212 Together the data suggests that mitotic spindle defects in Since massive apoptosis in Haus6 cKO brains was correlated with p53 induction, we 219 wondered whether cell death was p53-dependent and the cause of aborted brain 220 development. To address this, we crossed Haus6 cKO mice with p53 KO mice (Fig.  221   3a). Strikingly, at E13.5, a stage at which Haus6 cKO brains displayed massive 222 apoptosis, lacked cortical structures and had a poorly developed thalamus, Haus6 223 cKO p53 KO brains showed only minimal signs of apoptosis and there was some 224 growth in the regions where cortex and thalamus would be expected to form (Fig. 3b,  225 c, d). This growth was enhanced when compared to the single Haus6 cKO brains, 226 but seemed to lack the layered organization observed in control brains at this stage 227 ( Fig. 3b). At E17.5, however, when thalamus and cortex were well formed in controls, 228 in Haus6 cKO p53 KO embryos cortex and thalamus structures appeared thin and 229 undeveloped (Fig. 3e). Moreover, as observed for Haus6 cKO embryos, Haus6 cKO 230 p53 KO animals were not viable and died around birth. 231 In summary, massive apoptosis in Haus6 cKO brains is rescued in Haus6 232 cKO p53 KO brains, promoting growth in the affected brain regions, but this growth is 233 not productive for proper brain development. throughout the tissue including more basal regions (Fig. 4a, b, c). 274 The presence of large numbers of basally positioned mitotic figures in the 275 cortex and thalamus of Haus6 cKO p53 KO embryos could indicate that apical 276 progenitors had delaminated, that their nuclei did not migrate to the apical region 277 prior to division, or that the cells displaying mitotic defects in basal layers were not 278 apical progenitors. The latter possibility was tested by PAX6 staining (Fig. 6a). progenitors. In the cortex of control embryos, nestin-stained progenitors displayed a 296 highly polarized, apico-basal morphology and a laterally aligned arrangement within 297 13 the tissue (Fig. S3a). In contrast, polarized morphology and lateral alignment were 298 completely disrupted in progenitors of Haus6 cKO p53 KO embryos (Fig. S3a). 299 Consistent with these observations, staining with -tubulin antibodies revealed that 300 microtubules displayed apico-basal organization in control cells, running along the 301 length of the highly polarized cell bodies (Fig. S3b, c). In contrast, microtubules in 302 Haus6 cKO p53 KO progenitors lacked apico-basal orientation and appeared 303 disorganized (Fig. S3b, c). 304 Together these data suggest that in Haus6 cKO p53 KO embryos apical 305 progenitors had lost their polarized organization and divided ectopically. As a result 306 neuroepithelium integrity was severely disrupted. 307 308 309

Discussion 310
The mitotic spindle serves to segregate the replicated chromosomes faithfully into 311 two daughter cells. This task is carried out by spindle microtubules and a multitude of 312 proteins that nucleate, organize and remodel these microtubules during mitotic 313 progression. Here we have analyzed the contribution of one of three different 314 microtubule nucleation pathways, augmin-mediated microtubule amplification, to 315 mitotic spindle assembly in proliferating neural progenitor cells during mouse brain 316 development. Previous work found that impairment of centrosomal microtubule 317 nucleation in apical progenitors slowed mitotic spindle assembly and progression, 318 leading to p53-dependent apoptosis and causing microcephaly [24][25][26][27][28]. Similarly, 319 augmin-deficiency also impaired spindle assembly, delayed mitosis, and induced 320 p53-dependent apoptosis. In agreement with previous functional studies in cell lines, 321 augmin-deficient progenitors displayed fragmented spindle poles, but this did not 322 14 significantly impair spindle positioning. The most important outcome of these defects 323 was cell death. Our finding that the large majority of cells positive for expression of 324 p53 and the apoptotic marker cleaved caspase-3 were PAX6-positive interphase 325 cells, suggests that cell death occurred after completion of abnormal mitoses. 326 Despite the similarities with centrosome defects, the Haus6 conditional knockout 327 phenotype is much more severe. Rather than leading to microcephaly, augmin 328 deficiency completely aborted brain development. To our knowledge this has not 329 been reported for any other microtubule regulator affecting mitotic spindle assembly It should be noted that a more recent Cenpj conditional knockout mouse model 336 displayed more severe disruption of forebrain structures, causing lethality a few 337 weeks after birth [27]. Still, these defects seem less severe than what we observed 338 after augmin knockout. One may expect that preventing cell death in augmin-339 deficient progenitors would, at least to some degree, rescue brain development. 340 Codeletion of p53 in Haus6 cKO mice largely rescued apoptosis, revealing that cell 341 death was p53-dependent, but did not rescue brain development and lethality. In the 342 absence of apoptosis, augmin-deficient progenitors likely underwent repeated cycles 343 of abnormal mitoses, leading to increasingly severe mitotic abnormalities. This 344 behavior has recently been described after induced knockout of the augmin subunit 345 . In these animals co-deletion of p53 also exacerbated 359 mitotic defects and aneuploidy, but the outcome was still a microcephalic brain [42]. 360 In contrast, in the case of Haus6 cKO p53 KO progenitors in our study, continued 361 proliferation was not productive for brain development. While some cortical structures 362 were present at E13.5, they lacked a pseudostratified epithelial organization. Haus6 363 cKO progenitors had lost their characteristic, highly polarized morphology and 364 formed a disorganized cell mass that was intermingled with III-tubulin-positive 365 differentiated neurons, in both apical and basal regions. Considering that the 366 polarized apical progenitor morphology is integral to the organization of the 367 neuroepithelium, providing scaffold function and guidance for translocating basal 368 progenitors and migrating neurons, it is not surprising that these defects lead to 369 abortion of brain development. In summary, our work shows that, in contrast to centrosomal nucleation, 387 augmin-mediated microtubule amplification in neural apical progenitors is essential 388 for brain development and cannot be compensated for by the chromatin-and 389 centrosome-dependent nucleation pathways. As in the case of progenitors lacking 390 centrosomal nucleation, mitotic delay caused by augmin deficiency triggers p53-391 dependent apoptosis. While cell death can be prevented by co-deletion of p53, the 392 specific defects that result from the loss of augmin are sufficient to completely abort 393 brain development, independent of p53 status.

Image acquisition and analysis 492
Histology sections stained with hematoxilin/eosin (Fig. 1c, 1d, 3b, 3e; Fig. S1c) or 493 used for immunohistochemistry (Fig. 5a, 5c) were imaged with the digital slide 494 scanner Nanozoomer 2.0 HT from Hamamatsu and processed with NDP.view 2 495 software from Hamamatsu. Immunofluorescence labeled histology sections in Fig.  496  1e, 2a, 2d, 2h, 2j, 3c, 4a, 4b, 6 and Fig. S2, S3 were imaged with a Leica TCS SP5 497 laser scanning spectral confocal microscope. Confocal Z-stacks were acquired with 498 0.5 µm or 1 µm of step size depending on the experiment and using laser 499 parameters that avoided the presence of saturated pixels. Immunofluorescence-500 labeled histology sections shown in Fig. 2f and Fig. 4e were imaged with a Zeiss 880 501 confocal microscope equipped with an Airyscan. In the images shown in Fig. 2f, for  502 the Superresolution Airyscan mode a 63x magnification, 1.4NA oil-immersion lens 503 with a digital zoom of 1.8x was used. The z-step between the stacks was set at 0.211 504 µm. In the images shown in Fig. 4e, for the Fast Airyscan mode a 40x magnification 505 1.2 NA multi-immersion lens with a digital zoom of 1.8x was used. The Z-step 506 between the stacks was set at 0.5 µm. XY resolution was set at 1588x1588. Airyscan 507 raw data were preprocessed with the automatic setting of Zen Black. Additional 508 image processing and maximum intensity z-projections were done in ImageJ 509 software. In each experiment, serial brain sections from multiple animals per 510 genotype were analyzed (details in figure legends). 511 Radial thickness of the thalamus was measured with ImageJ as the distance 512 between the ventricular surface and the basal surface of this brain region in E13.5 513 embryos. In the same regions, radial thickness of the area occupied by PAX6 and 514 βIII-tubulin cell populations was measured. 515  In all experiments, for each brain at least two coronal tissue sections were quantified. 540 To measure interphase nucleus size in cortical neural progenitors (Fig. 4h), 541 tissue sections were immunostained with PAX6 antibodies and DAPI to label DNA. 542 The area of nuclei in PAX6-positive cells in the cortex was measured in z-stack 543 images using the "Positive cell detection" tool of QuPath software. Mitotic cells were 544 excluded from this analysis. 545

23
To quantify the distribution of neural progenitors within the cortex, 546 cryosections providing lateral views of the cortex were immunostained against PAX6 547 or TBR2. In both cases, the "Plot Analysis" tool of ImageJ was used to measure 548 signal intensity along the apico-basal axis of the cortex. Measurements were 549 grouped into 9.8 µm-wide bins and the average value for each bin was plotted as