Wnt10b-GSK3β–dependent Wnt/STOP signaling prevents aneuploidy in human somatic cells

Autocrine Wnt10b signaling ensures proper mitotic microtubule polymerization and faithful chromosome segregation.


Introduction
Wnt signaling pathways play critical roles in various developmental processes during embryogenesis and are important for the maintenance of adult tissue homeostasis (1). Wnt signaling has also a major role in cell proliferation, cell survival and differentiation and its misregulation is closely linked to human diseases including developmental pathologies and cancer (2). A variety of different Wnt ligands are expressed in human cells and secreted into the extracellular space to activate Wnt signaling in an auto-or paracrine manner (3). Secretion of Wnts requires their palmitoylation by the O-acyltransferase Porcupine and transport by the general Wnt cargo receptor Evi/Wntless (Wls) (4,5,6). Upon secretion, different Wnts bind to different Frizzled (FZD) receptors and activate intracellular signaling, which can be categorized into canonical/β-catenin-dependent and non-canonical/β-cateninindependent pathways (2).
The best characterized Wnt signaling pathway is the so-called canonical pathway that leads to the activation of the transcription factor β-catenin. In this pathway, Wnt ligands such as Wnt3a bind to different Frizzled (FZD) receptors and the co-receptors LRP5/6 (7,8). The ligand receptor complex clusters together with Dishevelled (Dvl) proteins in LRP6 signalosomes that recruit and inactivate a cytoplasmic destruction complex consisting of adenomatous polyposis coli, AXIN1 and the kinases CK1α and GSK3β (8,9,10). The inactivation of this complex prevents the ongoing phosphorylation of the transcription factor β-catenin by CK1α and GSK3β otherwise leading to its degradation by the ubiquitin-proteasome pathway (10,11,12). Hence, activation of the classical canonical Wnt signaling results in stabilization of β-catenin, which subsequently activates the expression of various target genes that are involved in the different cellular and developmental outcomes. It is known that different secreted Wnt ligands can function as canonical Wnts stabilizing β-catenin, albeit at different strength (3). As a counterpart, the secreted protein Dickkopf-1 (DKK1) inhibits canonical Wnt signaling by binding to and inducing the turnover of LRP5/6 coreceptors (13).
Interestingly, canonical Wnt signaling was shown to be involved in cell cycle regulation. A crucial target of β-catenin is c-myc, which drives the expression of CNND1 (encoding for cyclin D1) as a key driver of the G1/S transition of the cell cycle (14,15). Moreover, in mouse embryonic stem cells (mESCs), the loss of Evi/Wls, which is associated with a general lack of Wnt secretion (16), was found to impact mitotic progression and resulted in genome instability (17).
More recently, it became clear that Wnt/LRP6 signaling can also stabilize proteins other than β-catenin in a GSK3β-dependent 1 manner by sequestering the destruction complex in multivesicular bodies (10). This Wnt signaling pathway, now referred to as Wntmediated stabilization of proteins (Wnt/STOP) also involves LRP coreceptors and DVL (10,18,19,20,21) and promotes cell division and growth (18,22,23,24). Wnt/STOP activity peaks at the G2/M transition of the cell cycle because of the kinases CDK14-16, which phosphorylate and activate LRP6, and CDK1, which, in turn, phosphorylates and recruits BCL9 to the mitotic LRP6 signalosome (22,25).
Our previous work has demonstrated that loss of LRP5/6 or DVL, but not of β-catenin deregulates mitosis and causes chromosome missegregation suggesting that Wnt/STOP is required for faithful mitosis. In fact, we showed that loss of LRP6/DVL causes abnormally increased microtubule growth rates in mitotic spindles that act as a trigger for the generation of erroneous microtubulekinetochore attachments leading to whole chromosome missegregation (26,27). However, whether proper mitosis requires autocrine Wnt signaling and whether specific Wnt ligands are involved has not been addressed so far. In our work presented here, we show that basal canonical Wnt signaling involving LRP6 and leading to continuous suppression of GSK3β activity is required for the proper regulation of mitotic microtubule dynamics and chromosome segregation. We find that autocrine Wnt signaling requiring palmytoylation and secretion of Wnt ligands is essential for the proper execution of mitosis. Surprisingly, we identified Wnt10b as crucial Wnt ligand involved in the regulation of mitosis and required for preventing whole chromosome missegregation.

Results
Inhibition of Wnt signaling at the receptor level causes abnormal microtubule dynamics and aneuploidy in human somatic cells To inhibit canonical Wnt signaling at the receptor level we treated chromosomally stable human HCT116 cells with DKK1, an inhibitor of the Wnt co-receptor LRP6 (13,28). HCT116 cells harbor a heterozygous stabilizing mutation in β-catenin, but these cells still respond to Wnt in reporter assays (16,29). Accordingly, we found that Wnt3a further increased and DKK1 treatment reduced basal TCF4/ β-catenin reporter activity. Co-treatment with the GSK3β inhibitor CHIR99021 reverted this effect (Fig S1A). Upon inspection of DKK1treated cells progressing through mitosis by live cell microscopy, we found evidence for chromosome missegregation as indicated by the appearance of lagging chromosomes during anaphase and by generation of micronuclei upon exit from mitosis ( Fig S1B). Rigorous quantification of the proportion of cells exhibiting lagging chromosomes indeed revealed a significant induction of lagging chromosomes upon 16-h DKK1 treatment, which was suppressed by co-treatment with a low concentration of CHIR99021 (Fig 1A), which did not affect cell viability (Fig S1C). Our previous work has linked the generation of lagging chromosomes to abnormally increased microtubule assembly rates in mitotic spindles in response to Wnt inhibition (26). Hence, we used live cell microscopy to track GFPtagged microtubule end binding protein 3 (EB3) at microtubule plus tips in DKK1 treated cells synchronized in prometaphase of mitosis (Fig S1D and Video 1). DKK1 treatment significantly increased mitotic microtubule growth rates in HCT116 cells (Fig 1B). In line with this, non-transformed human retina pigment epithelial cells (RPE1-hTert) also exhibited increased microtubule assembly rates upon DKK1 treatment ( Fig S2A). As demonstrated previously (26,30), increased microtubule growth rates can be efficiently suppressed by treatment with sub-nanomolar concentrations of Taxol (Fig 1B), a microtubule binding drug known to inhibit microtubule dynamics (31). Upon DKK1 treatment restoration of proper microtubule growth rates by low-dose Taxol prevented the generation of lagging chromosomes (Fig 1A), suggesting a direct link between increased microtubule growth rates and the induction of whole chromosome missegregation in human somatic cells after DKK1-mediated LRP6 inhibition.
Because perpetual chromosome missegregation leads to an evolvement of aneuploidy, we investigated whether continuous inhibition of canonical Wnt signaling causes a high chromosome number variability in a microtubule dynamics-dependent manner. For this, we generated single-cell clones grown for 30 generations in the presence or absence of inhibitors as illustrated in Fig 1C and determined the proportion of cells harboring chromosome numbers deviating from the modal (45 chromosomes in HCT116 cells and 46 chromosomes in RPE1 cells). Continuous DKK1 treatment resulted in an evolvement of aneuploidy, which was suppressed by concomitant inhibition of GSK3β, both, in HCT116 (Fig 1D and E) and in RPE1 cells (Fig S2B and C). Co-treatment with low doses of Taxol, which restores proper microtubule growth rates, efficiently suppressed the generation of aneuploidy over time, indicating that abnormally increased microtubule dynamics is responsible for perpetual chromosome missegregation after Wnt inhibition (Fig 1D  and E). Our results strongly corroborate the finding that inhibition of canonical Wnt signaling, which comprises LRP6 and increased GSK3β activity results in abnormally increased microtubule growth rates during mitosis leading to whole chromosome missegregation and aneuploidy. Thus, human somatic cells require basal Wnt signaling to maintain proper microtubule dynamics in mitosis, faithful chromosome segregation, and a stable karyotype.

Abnormal microtubule dynamics and chromosome missegregation is triggered by increased GSK3β activity
The GSK3β kinase is the critical downstream component of the canonical Wnt-regulated destruction complex (1). Because abnormally increased microtubule growth rates and chromosome missegregation upon DKK1 treatment were suppressed by GSK3β inhibition (Fig 1), we asked whether increased GSK3β activity is sufficient to trigger the mitotic errors. To this end, we generated two independent stable cell lines derived from HCT116 cells showing low-to-moderate overexpression of GSK3β (Fig 2A). These cell lines were subjected to EB3 tracking experiments to determine microtubule growth rates in living mitotic cells. Clearly, increasing GSK3β levels were sufficient to increase microtubule assembly rates to the same levels seen upon DKK1 treatment. The abnormal microtubule growth rates were restored to control levels by treatment with lowdoses of Taxol and by small molecule-mediated GSK3β inhibition (CHIR99021) indicating that the kinase activity is responsible for the observed phenotypes ( Fig 2B). Accordingly, increased GSK3β levels are sufficient to induce lagging chromosomes during anaphase, which were also suppressed by Taxol or by GSK3β inhibitor treatment ( Fig 2C). To test whether an increase in GSK3β can trigger the development of aneuploidy, we generated single-cell clones from control and GSK3β overexpressing cells and determined chromosome number variability that evolved within 30 d of cultivation. In fact, overexpression of GSK3β in otherwise chromosomally stable cells was found to be sufficient to trigger the evolvement of aneuploidy ( Fig 2D). Thus, increased GSK3β activity upon inhibition of canonical Wnt signaling represents a relevant trigger for abnormal mitotic microtubule dynamics, chromosome missegregation, and aneuploidy. Because GSK3β represents the kinase in the Wnt signaling pathway that phosphorylates and destabilizes not only β-catenin, but also additional proteins (10), it seems likely that Wnt/STOP is required for the normal regulation of microtubule dynamics.

Wnt secretion ensures normal microtubule dynamics during mitosis and prevents aneuploidy
Next, we focused on the upstream events of Wnt signaling involved in mitotic regulation and asked whether continuous autocrine Wnt signaling is required to suppress aneuploidy in human somatic cells. To address this question, we generated Evi/Wls-knockout cells derived from chromosomally stable HCT116 cells using CRISPR/cas9 (HCT116 Evi_KO ). HCT116 cells were transfected with gRNA targeting Evi/Wls along with a Cas9 expression plasmid as described (32). Three independent knockout cell clones were isolated. Further details on the characterization of the HCT116 Evi_KO cells will be provided elsewhere (Voloshanenko et al, manuscript in preparation). Evi/Wls deficiency in the three knockout cell clones was verified by Western blotting (Fig 3A). The HCT116 Evi_KO cells were used to determine microtubule growth rates in mitosis by live cell analyses. All three cell clones exhibited increased microtubule assembly rates as seen for DKK1 treated cells, which were restored to normal levels by low-dose Taxol treatment and also upon GSK3β inhibition indicating that abnormal microtubule dynamics induced by Evi/Wls knockout is due to global loss of Wnt signaling ( Fig 3B). Consequently, loss of Evi/Wls causes a significant increase in cells with lagging chromosomes and this defect was also suppressed by Taxol or GSK3β inhibitor treatment ( Fig 3C). In line with the abnormal microtubule dynamics and lagging chromosome induction, karyotype analyses revealed that single cell clones derived from the three independent HCT116 Evi_KO cell lines developed aneuploidy within 30 generations ( Fig 3D). Thus, loss of Wnt secretion causes the same mitotic defects leading to aneuploidy as seen upon LRP6 inhibition. To further support this finding, we inhibited Porcupine, which is the O-acyltransferase required for palmitoylation of Wnts and, thus, for their secretion (33). Porcupine inhibition by the small molecule inhibitor IWP12 (34) increased mitotic microtubule assembly rates ( Fig 3E) and induced lagging chromosomes during anaphase ( Fig 3F) as seen in HCT116 Evi_KO cells. These data indicate that modification and secretion of Wnts is a prerequisite for normal mitotic microtubule dynamics, proper mitosis, and faithful chromosome segregation in human somatic cells.

Identification of Wnt ligands crucial for proper execution of mitosis
To test whether proper mitotic regulation can be restored in HCT116 Evi_KO cells by paracrine Wnt signaling, the knockout cells were co-cultured with parental HCT116 cells, and microtubule growth rates were determined specifically in (EB3-GFP-transfected) HCT116 Evi_KO cells. In fact, abnormally increased microtubule assembly rates in Evi/Wls-deficient cells were corrected by co-culture with parental cells in a cell number-dependent manner indicating restoration of mitosis-relevant Wnt signaling by Wnt ligands derived from parental HCT116 cells (Fig 4A). Based on this, we asked whether specific Wnt ligands are able to suppress abnormally increased microtubule polymerization rates in HCT116 Evi_KO cells. For this, we generated conditioned media for 14 different Wnts, treated Evi/Wls-deficient cells with control or Wnt-conditioned media and determined microtubule assembly rates in living mitotic cells. This screening approach revealed that conditioned media for Wnt1, Wnt2, Wnt3, Wnt3a, Wnt5a, Wnt6, Wnt7a, Wnt8a, Wnt9b, Wnt10a, and Wnt16 had no effect on mitotic microtubule regulation ( Fig 4B). However, addition of Wnt10b-conditioned medium fully restored proper mitotic microtubule polymerization rates in HCT116 Evi_KO cells, whereas addition of Wnt7b and Wnt9aconditioned media resulted in partial rescues ( Fig 4B). Thus, particularly Wnt10b, but possibly also Wnt7b and Wnt9a, represents candidate Wnt ligands involved in the regulation of mitotic microtubule dynamics and chromosome segregation.
Basal Wnt10b signaling is required for proper mitotic regulation Next, we wished to test whether basal endogenous levels of Wnt7a, Wnt9b, or Wnt10b are required for normal mitotic microtubule dynamics in HCT116 cells. We down-regulated a subset of endogenous Wnts, including Wnt7b, Wnt9a, and Wnt10b, which showed some rescue activity HCT116 Evi_KO cells, and determined mitotic microtubule dynamics. These five Wnts show highest expression levels in HCT116 cells (35), and siRNA-mediated down-regulation was achieved at similar levels for all of them ( Fig S3A). Significantly, we found that only down-regulation of endogenous Wnt10b, but not of Wnt3a, 7b, 9a, and 16 was sufficient to increase microtubule polymerization rates, suggesting that Wnt10b is indeed the most relevant Wnt ligand involved in the regulation of mitosis in HCT116 cells (Fig 5A). Thus, although conditioned media containing high levels of Wnt7a and Wnt9b showed some rescue effect in HCT116 Evi_KO cells, basal levels of these Wnts seem not to be required for normal regulation of microtubule dynamics in HCT116 cells. In addition to HCT116 cells, down-regulation of Wnt10b was also sufficient to increase mitotic microtubule polymerization rates in non-transformed RPE1-hTert cells (Fig S3B). After Wnt10b downregulation, both the increase in mitotic microtubule polymerization as well as the resulting induction of lagging chromosomes were not only rescued by low-dose Taxol treatment, as expected (see e.g., Fig  1A and B), but also by addition of purified recombinant Wnt10b protein and also by partial inhibition of GSK3β (Fig 5A and B) indicating specificity for Wnt10b and dependence on GSK3β activity. Moreover, addition of purified recombinant Wnt10b, but not of Wnt3a fully rescued the abnormally increased mitotic microtubule polymerization rates and also chromosome missegregation in HCT116 Evi_KO cells (Fig 5C and D). Thus, Wnt10b, but not other canonical Wnts including Wnt3a, which activates β-catenin-dependent transcription stronger than Wnt10b in reporter assays in HCT116 cells (Fig S3C), is crucial for the regulation of mitotic chromosome segregation.

Discussion
Our work indicates that Wnt signaling is pivotal for the proper progression mitosis and for the maintenance of genome stability. In particular, whole chromosome missegregation in mitosis resulting   aneuploidy is triggered by loss of Wnt signaling and is mediated by an abnormal increase in microtubule dynamics in mitotic spindles. This causes the generation of erroneous, merotelic microtubulekinetochore attachments leading to the formation of lagging chromosomes in anaphase as a pre-stage of chromosome missegregation. It is currently not clear on a mechanistic level how abnormal microtubule dynamics is translated into faulty microtubule-kinetochore attachments, but previous work from our laboratory has shown that increased microtubule growth rates cause transient spindle misorientation in the early phases of mitosis (30). This might generate a condition where transient spindle geometry changes facilitate the generation of erroneous, merotelic microtubulekinetochore attachments that overwhelm the cellular correction machinery that otherwise resolves such erroneous kinetochore attachments   (27). Interestingly, mitotic spindle misorientation has also been described after depletion of DVL2 in HeLa cells (36), but whether this was also associated with increased microtubule dynamics was not addressed. Our own previous data, however, showed that siRNA-mediated depletion of DVL2 was sufficient to increase microtubule dynamics in mitosis (26).
Here, we further defined the mitosis and chromosome segregation relevant Wnt signaling pathway. It involves the components of the "canonical pathway" comprising Frizzled receptors, LRP5/6 co-receptors, DVLs and the destruction complex, but not β-catenin.
Thus, Wnt-mediated stabilization of proteins other than β-catenin (Wnt/STOP) seems to be required for proper mitosis (10,18,19,20,21). In fact, previous work has established that Wnt signaling downregulates GSK3β activity, thereby preventing phosphorylation of proteins that are otherwise targeted for degradation (10,37). This is fully in line with our results showing that increased mitotic microtubule polymerization rates and chromosome missegregation after inhibition of Wnt signaling fully depends on GSK3β activity. Moreover, a moderate increase in GSK3β activity is sufficient to trigger the same mitotic defects as seen after inhibition of Wnt signaling. Hence, an important function of Wnt signaling in this context seems to be a constant down-regulation of a sub-pool of Wnt-controlled basal GSK3β activity. It is interesting that fine-tuned regulation of Wnt associated GSK3β activity seems not only be important for mitotic regulation, but also, for example, for macropinocytosis and metabolite changes as demonstrated recently (38). Importantly, down-regulation of basal GSK3β required for proper mitosis is mediated by basal Wnt signaling that requires ongoing secretion of Wnt ligands. In fact, our work showed that PORCN inhibition or knockout of the Wnt cargo receptor Evi/Wls is sufficient to trigger the mitotic defects leading to chromosome missegregation in a GSK3β-dependent manner. Thus, basal Wnt signaling represents a self-maintaining machinery required for proper mitosis and genome stability. This is in full agreement with a previous study in mESCs with Evi/Wls knockout where loss of Wnt secretion is also associated with mitotic abnormalities and genome instability (17). Because of technical limitations, we were not able to test whether increased microtubule polymerization is also involved in genome instability in these mESCs. Interestingly, this previous work showed that mitotic abnormalities in Evi/Wls knockout mESCs could be rescued by re-addition of Wnt3a and Wnt5 (17). In contrast, our data in somatic cells clearly indicate that Wnt3a or Wnt5 are not able to rescue the abnormally increased microtubule growth rates and chromosome missegregation, which might indicate that different Wnts might contribute to mitotic regulation, depending on the cell type. In fact, we performed a comprehensive screen of 14 different human Wnts for their involvement in the regulation of microtubule dynamics in mitosis. We identified Wnt10b as a major Wnt ligand regulating mitotic microtubule dynamics, at least in HCT116 and in non-transformed RPE1 cells. In addition to that, also Wnt7b and Wnt9a might be candidate ligands with functions for mitosis, possibly in other cell types and/or cell lines. It is intriguing that only a small subset or even one specific Wnt ligand regulates mitotic microtubule dynamics through the "canonical" pathway and by restraining GSK3β activity. It has been reported that Wnt10b acts as a "canonical" Wnt, but with lower activity in β-catenin reporter assays when compared with, for example, Wnt3a ((3) and our data). This might reflect that strong canonical Wnts (e.g., Wnt3a), which induced signaling through a broad spectrum of FZD receptors (7), might preferentially regulate β-catenin stability. Other Wnts, including Wnt10b, which act through a more restricted set of FZDs (7), might use the same pathway to mediate Wnt/STOP and thus, stability of proteins other than β-catenin. It will be important for future work to dissect separable physiological function of different Wnts using the same signaling pathway. To our knowledge, the finding that Wnt10b fulfills an essential function for mitosis is the first example of a specific Wnt ligand required for normal cell cycle regulation in human somatic cells. Wnt10b is expressed during embryogenesis and also in adult tissues, but its role has not been well studied in somatic cells. Studies using transgenic and knockout mice, however, demonstrated a role for Wnt10b in the immune system, in the development of the mammary gland and the regulation of bone mass (39). It is interesting to note that Wnt10b expression was found to be down-regulated in the aging brain and in Alzheimer disease brains (40), conditions that are known to be associated with aneuploidy. Intriguingly, down-regulation of the Wnt components DVL and LRP6 and up-regulation of the Wnt/LRP6 antagonist DKK1 was also found in aging brains (40). Based on our work, all these conditions are sufficient to induce aneuploidy by preventing the faithful execution of mitosis, possibly providing a link between loss of Wnt10b-LRP6-GSK3β signaling and the induction of aneuploidy in human age-related disease conditions.
The link between Wnt signaling and aneuploidy might also have clinical implications. It is well established that hyper-active Wnt signaling has a tumor-promoting function (37). Therefore, inhibition of Wnt signaling has been considered as an anti-cancer treatment strategy. In fact, several PORCN inhibitors, FZD antagonists, secreted frizzled-related proteins, or tankyrase inhibitors are currently tested in clinical phase1/2 trials for the treatment of various cancers (37). Based on our results, these treatments would induce chromosomal instability and aneuploidy, conditions known to contribute to tumor progression by supporting tumor evolution and by facilitating the development of resistance towards chemotherapy (41). This would put Wnt inhibitors at risk for the treatment of cancer and might even promote tumor progression. On the other hand, therapies aiming to activate Wnt signaling, including DKKneutralizing antibodies or GSK3β kinase inhibitors, have been considered for treatment of diseases with reduced Wnt signaling such as Alzheimer's disease (40,42). Here, evolving aneuploidy, which is associated with such neurodegenerative diseases (43), might be indeed suppressed by Wnt activators as we have demonstrated in our cell-based studies.

Generation of Evi/Wls knockout cells
HCT116 Evi_KO cells were generated using CRISPR/Cas9 following a protocol described in reference 44. In short, the HCT116 cells were transfected with px459 vector (a gift from Feng Zhang, Addgene plasmid # 48139; http://n2t.net/addgene:48139; RRID: Addgene_48139) expressing Cas9 and sgRNA targeting human Evi/Wls (59-GTAAGC-CAGGGAAACGTCCA-39). After 48 h, the cells were selected with 2 μg/ml of puromycin (Invitrogen) for 72 h. Because Evi silencing in HCT116 cells resulted in growth inhibition (16), either recombinant Wnt3a (50 ng/ml; R&D Systems) and/or medium from parental cell lines was added to the growth medium. Single-cell clones were selected and were used for further analysis. The generation of HCT116 Evi_KO cells will be described in detail in Voloshanenko et al (manuscript in preparation).

Generation of Wnt-conditioned media
Wnt-conditioned media were generated according to standard procedures. Briefly, HEK293T cells were transfected using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer protocol. 2.25 μg of untagged pcDNA3-Wnt plasmids (Addgene, Open Source Wnt Project Plasmids, Kit #1000000022) (3) and 0.25 μg pEGFP (to test for equal transfection efficiency) were used for transfection of cells in six-well format. 48 h post transfection, fresh medium was added, and conditioned medium was collected after 72 h.

FACS
FACS analysis was performed using a FACS Canto II flow cytometer (Beckton Dickinson) as described previously (30). To detect apoptotic cells, entire cell populations including floating cells were harvested and stained with 5 μg/ml propidium iodide. Cells with a sub-G1 DNA content were quantified as apoptotic cells using BD FACSDIVA software.

Live cell microscopy
HCT116 stably expressing H2B-GFP cells were grown in RPMI1640 medium without phenol red supplemented with 10% FBS and seeded into eight-well glass-bottom dishes (Ibidi). Live-imaging was performed using a Deltavision ELITE microscope (GE Healthcare) equipped with an Olympus x60 1.42 NA objective and a PCO Edge sCMOS camera (PCO). Images were recorded every 5 min for up to 5 h, whereas cells were incubated at 37°C and 5% of CO 2 . Images were deconvolved using SoftWorx 5.0/6.0 software (Applied Precision).

Determination of microtubule plus-end assembly rates
Microtubule plus-end growth rates were determined by tracking Eb3-GFP in living cells as described (30). Cells were transfected 48 h before the measurement with pEGFP-EB3 (kindly provided by L Wordeman, University of Washington), seeded onto glass-bottom dishes (Ibidi) and treated with Dimethylenastron (DME, 2 μM, Calbiochem) for 2 h before measurements. Live-imaging was performed using a Deltavision ELITE microscope (GE Healthcare) equipped with an Olympus x60 1.42 NA objective and a PCO Edge sCMOS camera (PCO) and images were recorded every 2 s, whereas cells were incubated at 37°C and 5% of CO 2 . Images were deconvolved using SoftWorx 5.0/6.0 software (Applied Precision) and average assembly rates were calculated for 20 individual microtubules per cell. 30 cells were analyzed in three independent experiments. To restore proper microtubule plus-end growth rates, 0.2 nM Taxol (Sigma-Aldrich) or 0.6 μM CHIR99021 (Sigma-Aldrich) was used.

Analysis of lagging chromosomes in anaphase
To enrich cells in anaphase, the cells were treated with 2 mM thymidine for 20 h followed by release into fresh medium for 9 h. The cells were fixed in methanol at −20°C and analyzed by immunofluorescence microscopy. Mitotic spindles, chromosomes, and centromeres were detected using anti-α-tubulin antibodies (1:100; Santa Cruz), Hoechst 33342 (1:20,000; Biomol), and anti-CENP-C antibodies (1:1,000; MBL), respectively. Only CENP-C-positive chromosomes clearly separated from two masses of chromosomes were defined as lagging chromosomes.

Reporter assays
HCT116 cells were seeded into 96-well plates and transfected with 10 ng Topflash luciferase and 5 ng Renilla luciferase reporter plasmid using XtremeGene9 DNA transfection reagent (Roche). 24 h later, cells were treated with PBS, 400 ng/ml Wnt10b or 400 ng/ml Wnt3a. After 16 h, the cells were lysed with commercial Passive Lysis Buffer (Promega) for 15 min at 4°C. 30 μl of lysate was transferred to a white 96-well plate and analyzed using a Dual-Luciferase Reporter Assay System (Promega) using a Tecan Microplate Reader (Tecan Infinite M1000). Topflash signals were normalized to the Renilla reporter and Wnt activity was calculated relative to the control condition.

qRT-PCR analysis
Total cellular RNAs were isolated using peqGold TriFast reagent (Peqlab) according to the manufacturer's instructions. The concentrations of RNA were determined using a NanoDrop ND-2000 (NanoDrop). The first-strand cDNA synthesis using total cellular RNAs as the template was performed with Invitrogen RT Reagent Kit. qRT-PCR was performed with iTaq Universal SYBR Green Supermix (Bio-Rad Laboratories GmbH) as the dsDNA fluorescence dye using CFX96 Real-Time PCR Detection Systems (Bio-Rad Laboratories, Inc.). The reactions were performed under the following conditions: 95°C for 3 min; 40 cycles of 95°C for 10 s, 60°C for 30 s, and 72°C for 30 s; and a melting curve from 65°C to 95°C. The results of qRT-PCR were defined from the threshold cycle (Ct), and relative mRNA levels were determined using the 2 −44Ct method with beta-actin (β-actin) as an internal control. The following primer pairs used for qRT-PCR:

Karyotype analysis
To determine chromosome number variability evolved over time, single-cell clones were generated and grown for 30 generations. To enrich for mitotic cells, the cells were treated with 2 µM Dimethylenastron (Sigma-Aldrich) for 4 h. The cells were collected and incubated in hypotonic medium (40% RPMI-1640, 60% H 2 O) at RT for 15 min. The cells were fixed in Carnoy's solution (methanol:acetic acid = 3:1). Chromosomes were spread onto glass slides and stained with Giemsa solution (Sigma-Aldrich). The proportion of cells with chromosome numbers deviating from the modal was determined by chromosome counting.

Data Availability
This study includes no data deposited in external repositories.