Delayed aneuploidy stress response of neural stem cells impairs adult lifespan in flies

Summary Studying aneuploidy during organism development has strong limitations, as chronic mitotic perturbations used to generate aneuploidy result in lethality. We developed a genetic tool to induce aneuploidy in an acute and time controlled manner during Drosophila development. This is achieved by reversible depletion of cohesin, a key molecule controlling mitotic fidelity. Larvae challenged with aneuploidy hatch into adults with severe motor defects shortening their lifespan. Neural stem cells, despite being aneuploid, display a delayed stress response and continue proliferating, resulting in the rapid appearance of chromosomal instability, complex array of karyotypes and cellular abnormalities. Notably, when other brain cell-lineages are forced to self-renew, aneuploidy-associated stress response is significantly delayed, indicating that stemness state confers resistance to aneuploidy. Sparing solely the developing brain from induced aneuploidy is sufficient to rescue motor defects and adult lifespan, suggesting that neural tissue is the most ill-equipped to deal with developmental aneuploidy. Highlights Reversible depletion of cohesin results in just a round or two of aberrant cell divisions, generating high levels of aneuploidy. Larvae challenged with aneuploidy during development hatch into impaired adults. Few cell cycles are sufficient for chromosomal instability to emerge from a previously stable aneuploid state. Neural stemness delays aneuploidy stress response. Protecting only the neural tissue from aneuploidy rescues adult abnormalities and lifespan.


19
Larvae challenged with aneuploidy hatch into adults with severe motor defects shortening their 20 lifespan. Neural stem cells, despite being aneuploid, display a delayed stress response and 21 continue proliferating, resulting in the rapid appearance of chromosomal instability, complex 22 array of karyotypes and cellular abnormalities. Notably, when other brain cell-lineages are 23 forced to self-renew, aneuploidy-associated stress response is significantly delayed, indicating 24 that stemness state confers resistance to aneuploidy. Sparing solely the developing brain from 25 induced aneuploidy is sufficient to rescue motor defects and adult lifespan, suggesting that 26 neural tissue is the most ill-equipped to deal with developmental aneuploidy. Aneuploidy, a state of chromosome imbalance, was observed over a century ago by Theodor 42 Boveri. Since then, numerous studies have shown that aneuploidy is largely detrimental both at 43 cellular and organism level. In multicellular organisms chromosome gain or loss results in 44 lethality or developmental defects (1, 2). At the cellular level, studies in yeast and cell culture 45 have demonstrated that aneuploidy has a high fitness cost for the cell, as unbalanced karyotypes 46 lead to activation of multiple stress response pathways, resulting in reduced proliferation, cell 47 cycle arrest, or cell death (Reviewed in (3). The aneuploidy stress response and consequential 48 drop in fitness seems at odds with the hypothesized role of aneuploidy in promoting 49 malignancy, which is usually marked by over-proliferation (4). Ninety percent of solid tumors 50 harbor whole chromosome gains and/or losses (5). Therefore, although usually detrimental to 51 cell fitness, aneuploidy and its effects on cell proliferation can be context dependent, which 52 emphasizes our need for a better understanding of the immediate and ultimate consequences of 53 this abnormal cellular condition in a tissue context and through development.

55
Study of aneuploidy in vivo is challenging since somatic aneuploidy is a rare event and is 56 difficult to capture and trace in real time due to several constraints: i) Cells are equipped with 57 surveillance mechanisms that prevent chromosome mis-segregation (e.g Spindle Assembly 58 Checkpoint (SAC) (Reviewed in (6) making naturally occurring aneuploidy events virtually 59 impossible to evaluate; ii) experimentally-induced aneuploidy, by compromising mitotic 60 fidelity, if often of low prevalence, as it has been demonstrated for several mammalian (7) (8)   61 and Drosophila tissues (9, 10) and iii) induction of somatic or constitutional aneuploidy in 62 metazoans relies on chronic mitotic perturbation (Listed in (11) which usually causes embryonic 63 lethality (Reviewed in(12) as a result of progressive accumulation of damage in the developing 64 organism. Thus, from these studies, it is impossible to disentangle short term and long term 65 consequences of aneuploidy, or to examine kinetics of the response to aneuploid state during 66 development. To circumvent these limitations, we generated a genetic system with the power to 67 induce aneuploidy in an acute and time-controlled manner, in all the dividing tissues of the 68 developing Drosophila. The tool is based on reversible depletion of cohesin, a key molecule 69 regulating mitotic fidelity (13,14). Cohesin is a tripartite ring complex, composed by SMC1, 70 SMC3 and the bridging kleisin subunit RAD21 (15,16). The primary mitotic role of cohesin is 71 to mediate sister chromatid cohesion, by topologically entrapping DNA fibers from neighboring 72 chromatids (17,18). Cells entering mitosis with premature loss of cohesion and sister chromatid 73 separation activate the Spindle Assembly Checkpoint (SAC) resulting in prolonged mitosis (14, 74 19). During this SAC-dependent mitotic delay, chromosomes are shuffled from one cell pole to 75 the other by the mitotic spindle (19). Consequently, chromosome shuffling induces genome 76 randomization and aneuploidy upon mitotic exit with a theoretical rate of nearly 100%. Our 77 engineered system enables a quick restoration of this complex shortly after its inactivation, 78 thereby restricting mitotic abnormalities to a short time-frame, concomitantly with the 79 generation of high levels of aneuploidy. Using such tool, we dissect the kinetics of aneuploidy 80 response across various cell/tissue types and developmental timings. To induce aneuploidy in an acute and time-controlled manner, we developed a genetic system 87 based on rapid removal of cohesin complex, the molecular glue that holds sister chromatids 88 together. To prevent a chronic cohesion depletion state and restrict mitotic failure to a single 89 cell cycle, our genetic system is able to induce cohesin inactivation, followed by subsequent 90 cohesion rescue. The system relies on the artificial cleavage of a modified version of the 91 RAD21 cohesin subunit that contains TEV protease cleavage sites (RAD21-TEV). As 92 previously shown, this system is very efficient at inactivating cohesin upon expression of the 93 exogenous Tobacco Etch Virus (TEV) protease, induced by a heat-shock promoter, resulting in 94 long-term inactivation of this complex (>24h) (20). To restrict cohesin impairment, we modified 95 this system by promptly rescuing cohesin integrity through the expression of TEV-resistant 96 RAD21 protein (RAD21-WT) right after TEV-mediated inactivation. For this purpose, RAD21-97 WT expression is under the control of UAS promoter (UAS-Rad21-wt-myc) that is induced by a 98 Gal4 protein induced concomitantly with the TEV protease (also under a heat-shock promoter, 99 HSprom-GAL4) ( Figure 1A). Given that the TEV protease is under a direct control of heat-100 shock promoter, whereas RAD21-WT relies on a dual expression-system (Gal4-UAS), we 101 anticipated that the temporal delay in RAD21-WT expression relative to the induction of TEV 102 protease would lead to a short time window of cohesin inactivation (RAD21 cleavage) ( Figure   103 1A).

104
To test this, we probed for the kinetics of TEV-mediated cleavage of RAD21-TEV and 105 synthesis of RAD21-WT in different tissues of the developing larvae. After heat shock, both 106 Drosophila larvae brains and wing discs, showed similar kinetics of the TEV-sensitive RAD21 107 disappearance followed by the appearance of RAD21-WT ( Figures 1A´ and 1A´´

114
The cohesive function of cohesin is established in S-phase, concomitantly with DNA 115 replication. Once stabilized on the replicated genome, cohesive cohesin complexes do not turn 116 over (21). As such, loss of cohesin using our system will affect sister chromatid cohesion in all 117 cells that are in S/G2/M phase during the short period between TEV protease expression and 118 synthesis of RAD21-WT ( Figure 1A). In addition to its canonical cohesive function, cohesin 119 has also been recently implicated in other interphase functions, including regulation of gene 120 expression (22). In contrast to the cohesive pool, these cohesin molecules are known to be 121 highly dynamic (21, 23). Moreover, cohesin-mediated loops were recently reported 122 "memorable" and quickly reformed upon cohesin re-establishment (24). We therefore 123 anticipated that this function should not be severely affected by our system. In sharp contrast, 124 mitotic errors induced upon cohesin cleavage are irreversible as there is no way to restore 125 cellular ploidy after a compromised round of mitosis.

127
Whereas canonical chronic mitotic perturbations lead to several rounds of mitotic failures, our 128 novel genetic system should lead to cohesion defects only in the first mitosis following the heat-129 shock, as the expression of RAD21-WT should be able to rescue cohesion in the subsequent cell 130 cycle, if given enough time ( Figure 1A). To confirm that our genetic system works as 131 anticipated, we focused our analysis on two different cycling tissues from the larva: the 132 developing brain and the epithelial wing discs.

133
The developing brain of Drosophila is an excellent model to study the consequences of 134 developmental aneuploidy. The well characterized cell lineages of the tissue in combination 135 with our tractable system to induce mis-segregation of chromosomes offer a unique opportunity 136 to trace the fate of aneuploid cells in real time and analyze their effect on the nervous system 137 development. Through larval development ~100 large neural stem cells called Neuroblasts 138 (Nbs) (25) located in the central brain (CB) region divide asymmetrically to self-renew and 139 generate distinct neuronal lineages via differentiating progeny (26).

140
We evaluated, by live cell imaging, mitotic fidelity in these Nbs using two independent criteria 141 to estimate the state of sister chromatid cohesion: i) the presence of single sisters (a direct 142 consequence of cohesion loss), as opposed to metaphase chromosome alignment and ii) the time 143 cells spend in mitosis, given that premature loss of sister chromatid cohesion is known to 144 activate the SAC and delay mitotic exit (Mirkovic et al., 2015).

145
As expected, the first division after the heat shock results in full cohesin cleavage in Nbs, 146 followed by cohesin rescue in subsequent divisions (Movies S1 and S2). The fast cell cycle of 147 Nbs, coupled with continued proliferation of these cells despite their abnormal genome content 148 (further discussed below), enables analysis of mitotic fidelity throughout several consecutive 149 divisions in great detail. Consistently, in the first mitosis AHS, 95% of Nbs contain single 150 sisters, and exhibit mitotic delay and chromosome shuffling (Figures 2A and 2B Figure S1A to A´´).

156
In contrast to the Nbs, in the epithelial cells of the wing disc, we observe the presence of single 157 sisters and a mitotic delay even at 48hs AHS, despite the presence of high levels RAD21-WT 158 (Figures 1A´´;2D;2E and 2F).These findings are consistent with the long cell cycle of the wing 159 discs cells (27,28). The high incidence of cells affected by reversible-cohesin cleavage is also 160 consistent with a high frequency of cells in S/G2 in this tissue, estimated using the fly FUCCI 161 system (29) ( Figure S3B).To fully demonstrate the ability of our tool to induce aneuploidy in an 162 acute manner in epithelial tissues we tested their regeneration capacity. In Drosophila epithelial 163 cells, multiple cellular insults, including aneuploidy, can activate the Jun N-terminal kinase 164 (JNK) signaling pathway, thus inducing the expression of pro-apoptotic genes and triggering the 165 apoptotic cascade (9, 30). In agreement with these studies, 24hs AHS in the wing disc, Cleaved

219
Premature differentiation and apoptosis were suggested as the main mechanisms of aneuploid 220 Nb elimination, reported in two recent studies (10, 31). However, after acute aneuploidy 221 induction in the entire Nb population, we found a very low frequency of cells undergoing 222 premature differentiation or cell death ( Figure S4).As a proxy for premature differentiation events, we quantified Nb-like cells that had either lost the DPN marker or abnormally exhibit 224 the differentiation marker Prospero (Pros) with or without co-expression of DPN (Figures S4A   225 and S4A´, arrowheads and dashed circles).Pros is the key factor acting as a switch for the 226 transition from stem cell self-renewal to terminal differentiation (33); therefore, this marker 227 should not be present in Nbs. We observed that upon acute aneuploidy induction in the entire 228 Nb population, there is a very low frequency of cells indicative of premature differentiation 229 ( Figure S4). These findings suggest that premature differentiation, although still taking place, is 230 unlikely to be the major form of stem cell elimination. To estimate the levels of apoptosis, we 231 also counted cells positive for cell death markers like CC3 and DCP1. We found a significant

244
To dissect the kinetics of the aneuploid response, we took advantage of the temporal resolution 245 of our system allowing for the tracing of aneuploid fate in real time. We restricted our analysis 246 to 3 rd instar wandering larvae as at this stage no new Nbs are generated from the neuro-247 epithelium (26). Induction of aneuploidy at this developmental stage, therefore, affects the 248 entire Nbs population, which facilitates cell fate analysis. We observed a significant amount of 249 Nbs proliferating for several days and displaying a tendency for chromosome accumulation over Aneuploidy elicits a stress response in the brain tissue.

297
Our findings revealed that aneuploid cells are not promptly eliminated but instead continue to 298 proliferate within certain karyotype restrictions. This should lead not only to the maintenance of 299 aneuploid stem cells (due to Nb self-renewal) but also to the accumulation of differentiated 300 aneuploid progeny (note that each Nb divides every ~2 hours (32)). We therefore tested how 301 such increase in aneuploid cells within the tissue could affect cellular physiology and influence 302 normal tissue development.

303
Several aneuploidy-associated stresses that include oxidative, metabolic, and proteotoxic stress 304 are likely to alter cellular homeostasis (3), which ultimately lead to p53 activation and a p53-305 dependent cell-cycle arrest/senescence (34, 35). Interestingly, elevated levels of p53 have been 306 observed in the Central Nervous System of Down syndrome patients (36). We decided to take 307 advantage of our in vivo system to acutely induce aneuploidy to examine whether abnormal 308 karyotypes trigger a stress response in the developing Drosophila brain and if so, what is the 309 kinetics of such response. We assessed by immunohistochemistry the presence of P53 and the 310 senescence marker Dacapo (DAP, a p21/p27 homologue (37) Figure S5). We concluded that aneuploidy induction has a detectable 320 effect in the entire brain population, triggering stress responses across different cell types.

321
However, despite being promptly affected by aneuploidy induction, Nbs display a delay to this 322 cellular insult and only mount a detectable response ~2 days after becoming aneuploid.

325
The delayed stress response (i.e. ~48hs after induction of aneuploidy) in the neural stem cell 326 pool may imply selective stem cell tolerance for the aneuploid condition when compared to the 327 other cell types of the developing brain. To test this idea we took advantage of the brat mutant 328 condition (38). In brat mutant larvae brains, each Nb divides into two daughter cells grow that 329 retain Nbs properties, leading to the formation of a tumor-like neoplasm (39). We reasoned that 330 cellular stemness confers tolerance to aneuploidy, the complete occupancy of the developing 331 brain by Nbs-like cells observed in the brat mutant phenotype should be sufficient to prevent 332 the stress response observed at 24hs AHS. To test this idea we combined our system for acute 333 induction of aneuploidy with brat mutations to be able to induce aneuploidy in a brat mutant 334 background and analyze the presence of stress markers at 24 and 48hs AHS. As predicted, DAP 335 appearance was significantly delayed in aneuploid brat mutants when compared to aneuploid 336 brains alone ( Figures 6B and 6B´

354
To test this hypothesis, we devised a system to selectively protect only the brain from cohesin 355 removal and consequent aneuploidy. To achieve this, we complemented our reversible cohesin 356 cleavage system with brain-specific expression of RAD21-WT throughout the course of the 357 experiment ( Figure 7A). In this way, TEV expression should lead to cohesion loss in all larval 358 tissues that survive solely on RAD21-TEV at the time of heat shock. In contrast, neural stem 359 cells should be resistant to this challenge, as they express both RAD21-TEV and RAD21-WT 360 ( Figure 7B). Neuroblast-specific expression of RAD21-WT was achieved by the use of 361 inscutable-Gal4 (insc-Gal4) or worniu-Gal4 (wor-Gal4) drivers, to constitutively express UAS-362 Rad21-wt-myc in the developing brain ( Figure 7B). As expected, constitutive presence of TEV-363 resistant RAD21 in the brain prevents any cohesion defects in 3 rd instar larvae Nbs ( Figure 7C).

364
To confirm that the rescue of sister chromatid cohesion occurs exclusively in the brain, we 365 performed parallel characterization of the first mitotic division after the heat shock in the wing, 366 derived from the same larvae. As anticipated, full cohesin cleavage was observed in all the 367 dividing epithelial cells from the wing discs ( Figure 7D). Notably, protecting only the brain 368 from developmental aneuploidy fully rescued the severe motor defects of the ecloded flies from 369 the 72hs AEL heat-shock, as demonstrated by mobility essays (Figures 7E´ and Movie S6).
Even more surprisingly, the brain protection was enough to rescue the lifespan of ~70% of the 371 adult flies affected by organism-wide aneuploidy during development, demonstrating that the 372 brain is indeed the most sensitive tissue when challenged with aneuploidy ( Figure 7E).

403
Neuroblasts have been used as a system to study aneuploidy response in previous studies (10, 404 31). These studies postulate two different but not mutually exclusive mechanisms of response to 405 induced aneuploidy: premature differentiation (31) and cell death by apoptosis (10). We 406 reasoned that if these are the major mechanisms of response to aneuploidy in neural stem cells, 407 they should be detectable in high frequency after the aneuploidy induction by our acute 408 approach. Contrary to that notion, after examined in detail the kinetics of the response, both 409 premature differentiation and cell death were detected at low frequency even days after cells 410 became aneuploid. It is important to note that the degree of aneuploidy in the Nbs upon cohesin 411 loss should be around 98% due to the extensive genome shuffling prior to mitotic exit.

412
Therefore, the finding that aneuploidy does not eliminate the entire Nb population, strongly 413 argues against the existence of specific, active mechanisms controlling the integrity of the 414 neural stem cell genome. The more plausible explanation is that the Nb elimination due to 415 aneuploidy stems from a wide spectrum of abnormalities due to a randomized genome.  Interestingly, our results highlight that cell identity determines the kinetics of this stress 445 response. Aneuploidy response is specifically delayed in the neural stem cell pool (displayed 446 mainly at ~48hs AHS) compared to the rest of the tissue, which exhibits it considerably earlier.

447
Forcing self-renewal is sufficient to delay stress response in the entire tissue, suggesting that 448 cellular stemness alone makes cells less sensitive to aneuploidy-induced stresses. Accordingly, 449 unusual resistance to altered ploidy was observed in human and mouse embryonic stem cells 450 (ESCs), mostly achieved by relaxing the cell cycle control and uncoupling the spindle 451 checkpoint from apoptosis (49). The ability of neural stem cells to continue dividing despite the 452 aneuploid karyotype dubbed them as aneuploidy "tolerant" (10). Yet, based on our findings it is 453 clear that keeping these aneuploid cells is catastrophic for normal tissue architecture and 454 development. Thus, aneuploidy may be "tolerated" better in Nbs, but the tissue as a whole is 455 unable to be functional. In contrast, the "sensitivity" of epithelial cells enables the tissue to 456 clean up and regrow properly.

458
The developing brain restricts organism recovery after induced aneuploidy.

459
Chromosomal aberrations have been long associated with neurological disorders (50). However, 460 their impact on brain development and function remains complex and poorly understood, 461 partially due to limitations of available experimental approaches. In almost all animal model 462 systems used to study aneuploidy and its consequences until now, the organisms die 463 prematurely due to the chronic disruption of mitotic fidelity to generate chromosome imbalance. 464 Therefore, it is only possible to address the short term effect of aneuploidy in nervous system 465 development, but not to understand the ultimate consequences for brain function. Our acute 466 system reversibly affects chromosome segregation to induce just a pulse of aneuploidy, 467 enabling the organism to recover from the insult and complete its development. The most 468 noticeable phenotype observed in the adult was the severe motor and behavioral defects that 469 clearly affect the lifespan of the flies, evidencing the sensitiveness of the nervous system to 470 aneuploidy. Previous studies in Drosophila have shown that the mitotic disruption in larvae Nbs 471 generates a reduction of their brain size (10, 31) reinforcing the idea about a link between 472 aneuploidy and microcephaly. However, our results showed that induced acute aneuploidy has 473 no significant impact in the size of the adult brain. These findings suggest that the continued 474 proliferation of neural stem cells, caused by incomplete cell elimination and delayed 475 aneuploidy-stress response, is sufficient to support the development of an apparently normal-476 sized organ. It is conceivable that the observed normal size reflects a sample selection, as this 477 analysis was restricted to flies that survived the aneuploid challenge (~70%). Supporting this 478 possibility, a screening performed to isolate anatomical brain mutants of Drosophila have 479 shown that mutant strains showing altered brain shape and particularly small brains are very 480 weak being mostly lethal at pupa stage (51). Despite unaltered shape and size of the adult brains, we reasoned that the neural circuits are likely impaired in those brains giving rise to the 482 adult phenotype observed in all the surviving flies.

483
In accordance with the notion of the brain as the tissue most sensitive to aneuploidy, we show 484 that preventing aneuploidy exclusively in the brain is sufficient to rescue all the behavioral 485 defects previously observed. This brain protection not only rescued motor defects but also the 486 lifespan of the flies ecloded upon 72hs AEL heat-shock, suggesting that neural tissue is the most

508
RFP, respectively. The progeny was then heat shocked once at 37°C for 45min at the desired 509 developmental stage. The correct genotype larvae were selected based on the absence of the 510 "tubby" phenotype; the heat shocked "tubby" larvae were used as negative controls (control 511 HS). As genetic control we used the same genotypes for the induction of aneuploidy but without 512 performing the heat-shock.

513
To determine the proportion of adult eclosion, the crosses mentioned were raised in cages to 514 monitor the time of egg collection. After 6hs collection, the plates were removed from the 515 cages, the number of eggs counted and the plates were kept until larvae hatched. The plates 516 were then heat-shocked at 37°C for 45min at different larvae developmental time (~48hs AEL, 517 ~72hs AEL, ~96hs AEL and ~120hs AEL (± 6hs)) and placed in a new clean plastic cage. Once 518 they reached pupae stage ("yellow body") the pupae were gently removed with a wet brush and 519 separated in "tubby" (control HS) and "no tubby" phenotype (condition). The different batches 520 of pupae were placed over agar plates covered with two layers of absorbent paper to maintain 521 the humidity and counted. The plates with the pupae were kept at room temperature until flies 522 ecloded and the proportion of eclosion calculated.

547
Lifespan was measured at room temperature according to standard protocols. In brief, newly 548 ecloded animals (0 to 3 days) were collected (50 per genotype: "control", "Aneuploidy" and 549 "Aneuploidy + brain rescue"), and then placed in vials (up to 10 per vial), and transferred to

932
C-Quantification of Cohesive states of 3 rd instar larvae Nbs and epithelial cells from wing discs 933 following heat-shock. Insc-Gal4 protects the brain from cohesin loss, but has no effect in the 934 wing disc.