The P53/P73 - P21cip1 Tumor Suppressor Axis Guards Against Chromosomal Instability by Restraining CDK1 in Human Cancer Cells

Home , CEP72, Chromosome instability, CKAP5, P73

Oncogene (2021) 40:436–451 https://doi.org/10.1038/s41388-020-01524-4

ARTICLE

The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal instability by restraining CDK1 in human cancer cells

1 1 2 1 2 Ann-Kathrin Schmidt ● Karoline Pudelko ● Jan-Eric Boekenkamp ● Katharina Berger ● Maik Kschischo ● Holger Bastians 1

Received: 2 July 2020 / Revised: 2 October 2020 / Accepted: 13 October 2020 / Published online: 9 November 2020 © The Author(s) 2020. This article is published with open access

Abstract Whole chromosome instability (W-CIN) is a hallmark of human cancer and contributes to the evolvement of aneuploidy. W-CIN can be induced by abnormally increased microtubule plus end assembly rates during mitosis leading to the generation of lagging chromosomes during anaphase as a major form of mitotic errors in human cancer cells. Here, we show that loss of the tumor suppressor genes TP53 and TP73 can trigger increased mitotic microtubule assembly rates, lagging chromosomes, and W-CIN. CDKN1A, encoding for the CDK inhibitor p21CIP1, represents a critical target gene of p53/p73. Loss of p21CIP1 unleashes CDK1 activity which causes W-CIN in otherwise chromosomally stable cancer cells. ﬁ Vice versa

1234567890();,: 1234567890();,: Consequently, induction of CDK1 is suf cient to induce abnormal microtubule assembly rates and W-CIN. , partial inhibition of CDK1 activity in chromosomally unstable cancer cells corrects abnormal microtubule behavior and suppresses W-CIN. Thus, our study shows that the p53/p73 - p21CIP1 tumor suppressor axis, whose loss is associated with W-CIN in human cancer, safeguards against chromosome missegregation and aneuploidy by preventing abnormally increased CDK1 activity.

Introduction perpetual missegregation of whole chromosomes or large parts thereof during mitosis [4]. An important type of Chromosomal instability (CIN) is a major form of genome mitotic errors contributing to whole chromosome mis- instability representing a hallmark of cancer [1, 2]. While segregation in human cancer cells are so-called lagging structural alterations are caused by structural chromosome chromosomes that are caused by unresolved merotelic instability (S-CIN), changes in the chromosome numbers, microtubule-kinetochore attachments [5]. Various referred to as aneuploidy, are a result of whole chromo- abnormalities in chromosomally unstable cancer cells can some instability (W-CIN). A majority of cancer cells contribute to the generation of lagging chromosomes shows aneuploidy that is strongly associated with tumor- including supernumerary centrosomes [6, 7] or hyper- igenesis, tumor progression, therapy resistance, and poor stable microtubule-kinetochore attachments [8, 9]. More prognosis [3]. W-CIN is deﬁnedbyanelevatedrateof recently, we have demonstrated that abnormally increased microtubule plus end assembly rates within mitotic spindles result in the induction of W-CIN in cancer cells Supplementary information The online version of this article (https:// [10–16]. In fact, increased microtubule assembly rates doi.org/10.1038/s41388-020-01524-4) contains supplementary cause merotelic microtubule-kinetochore attachments, material, which is available to authorized users. lagging chromosomes, and chromosome missegregation * Holger Bastians [10](Fig.1a). [email protected] Despite the strong link between increased mitotic microtubule assembly rates and W-CIN, the cancer-associated 1 Georg-August University Göttingen, Göttingen Center for genetic alterations leading to abnormal microtubule dynamics Molecular Biosciences (GZMB) and University Medical Center Göttingen (UMG), Institute of Molecular Oncology, Section for are not well understood. So far, only a few tumor suppressors Cellular Oncology, D-37077 Göttingen, Germany and oncogenes including BRCA1, CHK2, AURKA,and CEP72 2 University of Applied Sciences Koblenz, Department of were shown to be involved in triggering W-CIN by Mathematics and Technology, D-53424 Remagen, Germany increasing mitotic microtubule assembly rates [10–15]. The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal. . . 437

low-dose a Taxol

cancer-relevant triggers?

increased microtubule lagging chromosomes whole chromosome plus end growth during anaphase missegregation chromosomal instability (W-CIN)

HCT116 + DNp73α b clone 1 clone 2 clone 3 - - - Dox + + + p73α (endogenous) DNp73α TA DBD OD SAM TID p73α p73β (endogenous) p53 DBD OD SAM TID DNp73α β-actin

c 25 **** **** **** **** **** ****

15 (μm/min) 16.6 16.8 20.0 16.9 16.3 20.2 16.5 16.3 19.7 16.3 10 m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min n=30 n=30 n=30 n=30 n=30 n=30 n=30 n=30 n=30 n=30 + -++-++-++Dox microtubule polymerisation rate - --+--+--+0.2 nM Taxol HCT116 clone 1 clone 2 clone 3 HCT116 + DNp73 d DNA CENP-C α-tubulin merge 10 HCT116 + DNp73α cells (clone 1) 8 ** ** * ** ** * 6

HCT116 + DNp73α 4 (clone 1) + Dox 2

proportion of anaphase 0 with lagging chromosomes (%) + - ++ -++ -++ Dox HCT116 + DNp73α - --+ --+ --+ 0.2 nM Taxol (clone 1) HCT116 clone 1 clone 2 clone 3 + Dox + 0.2 nM Taxol HCT116 + DNp73

e 30 generations ± Dox ± 0.2 nMTaxol evolvement of chromosome number variability = W-CIN

single cells single cell clones metaphase spread

f HCT116 + DNp73α (clone 1) g 40 ber

- m 100 Dox Dox + 0.2 nM Taxol 30 80 e nu m lls (%) e c 60 20 omoso 40 r 10 20 proportion of of proportion deviating from (%) modal deviating 0 0 cells with ch 2 6 3 3 5 7 0 -++Dox 45 46 48 4 4 4 44 47 40 42 47 4 41 43 44 45 4 48 48 41 42 4 44 4 46 40 41 --+0.2 nM Taxol number of chromosomes per metaphase spread HCT116 + DNp73α (clone 1)

Interestingly, loss of the tumor suppressor gene TP53 is abnormal microtubule dynamics nor chromosome mis- strongly associated with the presence of W-CIN [17]. segregation and W-CIN [10, 18]. p53 belongs to a family of However, loss of p53 alone is neither sufﬁcient to trigger transcription factors that also includes p73 [19]. While TP53 438 A.-K. Schmidt et al.

Fig. 1 Expression of oncogenic DNp73 induces whole chromosome Material and methods instability by increasing mitotic microtubule plus end assembly rates. a Model depicting the causal relationship between increased Cell culture microtubule plus end growth rates in mitosis and the induction of chromosome missegregation constituting W-CIN. b Left panel: Model for the structure of p73 and DNp73: transactivation domain (TA), HCT116, HT29, RKO, SW480, and SW620 cells were DNA-binding domain (DBD), oligomerization domain (OD), SAM purchased from ATCC (USA), DLD-1 cells were from domain (SAM), transactivation inhibitory domain (TID). Right panel: Sigma-Aldrich (Germany). DLD-1-CDKN1A−/− cells were Doxycycline-inducible expression of DNp73 in chromosomally stable TP53−/− HCT116 cells. Three independent stable cell lines were treated with obtained from Horizon Discovery (UK). HCT116- − − doxycycline for 24 h and expression of DNp73 was detected by [27] and HCT116-CDKN1A / cells [28] were kindly western blotting. A representative western blot is shown. Note that the provided by Bert Vogelstein (Johns Hopkins University, α β two bands for endogenous p73 represent the and isoforms of the Baltimore, USA). RKO cells expressing CDKN1A in a protein. c Mitotic microtubule plus end assembly rates upon expression of DNp73. HCT116-DNp73 cell lines were treated with or ponasterone A-inducible manner (RKO-p21-Pon) were without doxycycline and additionally with low-dose Taxol. As a kindly provided by Matthias Schmidt [29]. Details on cul- control HCT116 parental cells were treated with doxycycline. turing conditions are given in the Supplemental Methods. Microtubule growth rates were determined by tracking EB3-GFP comets in mitotic cells by live-cell microscopy. Scatter dot plots show average microtubule polymerization rates (20 microtubules/cell, n = Measurement of microtubule plus end assembly 30 mitotic cells from 3 independent experiments, mean ± SD, t-test). rates d Proportion of cells exhibiting lagging chromosomes in anaphase DNp73 after expression of and after restoration of proper microtubule To determine microtubule plus end assembly rates, live-cell growth rates. HCT116 parental cells were incubated with doxycycline as a control. Left: Representative examples of anaphase cells with or microscopy was performed on cells expressing GFP-tagged without lagging chromosomes (white arrows), Scale bar, 10 µm. Right: end-binding protein 3 (EB3) [10, 30] as described [11]. The graph shows the proportion of anaphase cells exhibiting lagging n = chromosomes for three independent stable cell lines ( 300 anaphase Detection of lagging chromosomes cells from three independent experiments, mean ± SD, t-test). e Scheme depicting the generation of single-cell clones in order to determine chromosome number variability as a measure of W-CIN. To quantify anaphase cells exhibiting lagging chromosomes, f Proportion of cells harboring the indicated chromosome numbers per cells were fixed, stained, and imaged as described [15]. cell (n = 50 metaphase spreads). g Proportion of cells with chromosome numbers deviating from the modal (45 chromosomes in HCT116 cells). The calculation is based on analyses of chromosome number W-CIN analysis variability shown in f. Single-cell clones were generated to determine the evolved chromosome number variability as a measure of W-CIN. is often mutated or deleted in cancer cells, TP73 is infre- Chromosome numbers were determined from metaphase quently mutated, and thus, it is still under debate whether spreads as described [10]. Details are given in the Supple- TP73 represents a tumor suppressor gene like TP53 [20]. mental Methods. However, a truncated oncogenic form of p73, DNp73,is frequently overexpressed in various tumor types and can act Tumor data and bioinformatic analysis as a dominant-negative regulator for both p73 and p53 [21, 22]. It is well established that both transcription factors Tumor data from the colon adenocarcinoma (COAD) and have overlapping target genes including CDKN1A, which breast invasive carcinoma (BRCA) projects of The Cancer encodes for the cyclin-dependent kinase (CDK) inhibitor Genome Atlas (TCGA) program were obtained from the protein p21CIP1.p21CIP1 was shown to inhibit CDK1 and harmonized NCI Genomic Data Commons (GDC) database. CDK2, thereby contributing to cell cycle arrest in G1 and G2 We used Single Nucleotide Variants (SNVs) from whole- phase of the cell cycle, e.g., upon DNA damage [23][24]. exome sequencing (Mutect2 Annotation), Copy Number Interestingly, loss of p21CIP1 in unperturbed cells was shown Variations (CNVs) from Affymetrix 6.0 Genotyping Arrays to cause problems in chromosome segregation, which might and gene expression quantifications from high-throughput reflect a requirement for p21CIP1 in fine-tuning mitotic pro- RNA sequencing. 389 COAD and 972 BRCA samples, for gression [25]. In cancer, CDKN1A expression is often which all three data modalities were available, were used for decreased, partly along with p53 inactivation, and associated further analyses. Additionally, RNA sequencing data of 113 with poor prognosis suggesting a possible tumor-suppressive BRCA and 41 COAD normal tissue samples were down- function of CDKN1A [26]. loaded from the database. Here, we provide a link between loss of the p53/p73 - For tumor transcriptome analysis, RNA sequencing p21CIP1 axis, mildly increased CDK1 activity and increased count data were transformed into log2-counts per million mitotic microtubule assembly rates and W-CIN. (logCPM) expression values utilizing mean-variance The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal. . . 439 modeling at the observational level (voom) [31] as imple- growth rates by low-dose Taxol treatment (Fig. 1c) efficiently mented and described in the R-package limma [32]. These suppressed the generation of lagging chromosomes (Fig. 1d) logCPM were centered around the mean logCPM of normal indicating a causal relationship between both defects in the tissue samples for each gene respectively. The resulting context of DNp73 expression. Next, we investigated whether differences were interpreted as the log2 fold change of gene DNp73 expression causes W-CIN in a microtubule growth expression in a tumor sample relative to normal tissue. To rate-dependent manner. We evolved single-cell clones from find differentially expressed genes, we used Wilcoxon’s uninduced (-Dox) or induced (+Dox) cells in the absence or rank sum test. To quantify chromosomal instability in the presence of Taxol treatment (Fig. 1e). Karyotypic analyses of tumor samples, the weighted genome instability index score the otherwise chromosomally stable HCT116 cells revealed an (wGII) was calculated using the SNP array-based copy- induction of high chromosome number variability in cells with number variation data as described previously [33]. DNp73 expression, which was suppressed upon correction of Classification of tumor samples: A sample is labeled as abnormal microtubule growth rates by Taxol (Fig. 1f, g). Thus, being mutated in a gene when a non-silent nucleotide var- expression of DNp73 acts as a bona fide inducer for abnormal iation can be observed in the SNV data. To divide the tumor mitotic microtubule assembly rates and W-CIN. samples into a lowCIN and highCIN group we used euclidean-distance average-linkage hierarchal clustering of Concomitant loss of p53 and p73 mimics expression the wGII. A sample was considered to exhibit low expression of DNp73 to induce an increase of microtubule of a gene if the difference between its logCPM in that sample growth rates and W-CIN and the mean logCPM in normal samples is lower than −1. This is equivalent to a ratio between CPM in the tumor and DNp73 represents an oncogenic N-terminally truncated (geometric) average CPM in normal tissue lower than half. form of p73 lacking its transactivation domain that can act as a dominant-negative regulator for both p53 and p73 [22] Statistical analysis (Fig. 1b, scheme). To test whether DNp73 expression influences mitosis through inactivation of p53 and p73, we For statistical analysis, the Graph Pad Prism 5.0 software generated stable HCT116 cell lines with either single loss of (Graph Pad Software, USA) was used. Mean values and p53, p73, or concomitant loss of both (Fig. 2a). As shown standard deviation (SD) were calculated. To analyze sta- previously [10], loss of p53 alone was not sufficient to tistical significance, unpaired two-tailed t-tests (SD ≠ 0) or induce increased microtubule assembly rates (Fig. 2b) or one-sample t-tests (SD = 0) were performed. P values were lagging chromosomes (Fig. 2c). The same was observed indicated as: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < upon single loss of p73. However, concomitant loss of p53 0.0001, ns (not significant) p ≥ 0.05. and p73 clearly increased microtubule growth rates to a level observed after DNp73 expression (Fig. 2b). This was also associated with the induction of lagging chromosomes Results in anaphase (Fig. 2c). Rescue experiments using Taxol treatment showed that increased microtubule growth rates Expression of oncogenic DNp73 induces increased triggered by p53/p73 loss are responsible for whole chro- mitotic microtubule growth rates, chromosome mosome missegregation (Fig. 2b, c). This was further missegregation, and W-CIN supported by additional experiments, in which proper microtubule assembly rates were restored in p53/p73 defi- We performed a mini-screen to identify cancer-relevant cient cells by partial downregulation of the microtubule plus alterations that might contribute to an increase in mitotic end polymerase ch-TOG (encoded by CKAP5; Fig. S1a, microtubule growth rates and to the development of W-CIN S1b) [10], which led to suppression of lagging chromo- and identified a truncated oncogenic form of p73 (DNp73) somes (Fig. S1c). [22](Fig.1b, scheme). We generated HCT116 cell lines that To investigate whether concomitant loss of p53 and p73 express DNp73 in a doxycycline (Dox) inducible manner induces W-CIN, we generated single-cell clones derived (Fig. 1b) and measured microtubule growth rates in mitotic from the different TP53, TP73,orTP53/TP73-deficient cell cells by tracking GFP-tagged microtubule end-binding protein lines and grew the cells in the presence or absence of Taxol 3 (EB3-GFP) by live-cell microscopy. Dox-induced expres- for 30 generations. Analysis of chromosome number sion of DNp73 was sufficient to increase mitotic microtubule variability as a measure of W-CIN demonstrated the growth rates to a level typically observed in chromosomally induction of aneuploidy over time only in cells with con- unstable cancer cells (Fig. 1c) [10]. Moreover, DNp73 comitant loss of p53 and p73, but not in cells with single expression caused the induction of lagging chromosomes loss of p53 or p73. W-CIN was efficiently suppressed upon during anaphase (Fig. 1d). Correction of abnormal microtubule correction of abnormal microtubule assembly rates by Taxol 440 A.-K. Schmidt et al. A a b **** 25 **** ****

shRNA * **** **** **** + control shRN

-/- ** HCT116-TP53-/- 20 * TP73 + TP73 shRNA -TP53 m/min)

( 15 HCT116 + control shRNA HCT116 HCT116 clone 1 clone 2 HCT116 + HCT116 clone 3 16.3 16.7 16.8 16.8 19.5 16.7 19.3 16.4 19.6 16.6 p73 tubule polymerisation rate m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min o 10 n=30 n=30 n=30n=30 n=30n=30 n=30 n=30 n=30 n=30

p53 micr -+- - -+-+-+0.2 nM Taxol

- / A - clone 1 clone 2 clone 3 β-actin RN T116 T116 HCT116 -TP53-/- + TP73 shRNA shRNA

-TP53 l shRNA HC HCT116 o r t con + TP73 HCT116 + + control sh

c DNA CENP-C α-tubulin merge ** 10 ** HCT116 lls

e ** + control shRNA ns ** ** ** 8 ns + DMSO (%) mes o ns s o 6 HCT116-TP53-/- + TP73 shRNA 4 (clone 2) + DMSO 2 oportion of anaphase c anaphase of oportion TP53-/- r HCT116- p 0 + TP73 shRNA with lagging chrom -+-- -+-+-+0.2 nM Taxol

- /

(clone 2) A - clone 1 clone 2 clone 3

+ 0.2 nM Taxol 116 RN -/-

T TP53 h hRNA hRNA HCT116- TP53 s + TP73 shRNA HC ol shRNA ol r 73 t TP on

c + HCT116- + d + control s 116

** T

r 60

*** HC 50 ns e numbe

m 40 o * 30 mos ns hro

c 20

with 10 deviating from (%) modal deviating lls

ce 0 2 2 3 3 3 3 clone 3 clone clone 1 clone clone 1 clone clone 1 clone 2 clone 1 clone 2 clone 2 clone 2 clone 2.3 clone 2.1 clone 2. clone 2. clone 2.1 clone 2.

DMSO0.2 nM DMSO DMSO DMSO 0.2 nM Taxol Taxol HCT116 + HCT116- HCT116 HCT116-TP53-/- control shRNA TP53-/- + TP73 + TP73 shRNA + control shRNA shRNA Fig. 2 Concomitant loss of p53 and p73 induces chromosomal p73 or both, and after restoration of proper microtubule growth rates. instability by increasing mitotic microtubule plus end assembly Left panel: Representative examples of anaphase cells with or without rates. a Generation of stable cell lines with loss of p53, p73 or con- lagging chromosomes (white arrows), Scale bar, 10 µm. Right panel: comitant loss of both. A representative western blot detecting p53 and The graph shows the proportion of anaphase cells exhibiting lagging p73 protein levels is shown. b Mitotic microtubule plus end assembly chromosomes (n = 300 anaphase cells from 3 independent experi- rates upon loss of p53, p73 or both. The indicated stable cell lines were ments, mean ± SD, t-test). d Induction of aneuploidy upon loss of p53/ used to measure microtubule plus end growth rates in mitosis by p73 and rescue upon restoration of proper microtubule growth rates. tracking EB3-GFP comets by live-cell microscopy. To restore normal Single-cell clones derived from the indicated cell lines were grown in microtubule growth rates, cells were treated with 0.2 nM Taxol for the presence or absence of 0.2 nM Taxol for 30 generations, chro- 16 h before measurement. Scatter dot plots show average microtubule mosome numbers per cell were determined. The graph depicts the polymerization rates (20 microtubules/cell, n = 30 mitotic cells from proportion of cells harboring chromosome numbers deviating from the three independent experiments, mean ± SD, t-test). c Proportion of modal (45 chromosomes in HCT116 cells, n = 50 cells). cells exhibiting lagging chromosomes in anaphase after loss of p53, The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal. . . 441

(Fig. 2d, Fig. S2), emphasizing again the causality between suppressed lagging chromosomes (Fig. 3f). Hence, mild re- abnormal microtubule growth rates and W-CIN. Thus, expression of CDKN1A downstream of p53/p73 loss rescues concomitant loss of the transcription factors p53 and p73 the mitotic defects and chromosome missegregation indicating mimics the mitotic defects observed upon DNp73 expres- that p21CIP1 acts as an important mediator of the W-CIN sion and is sufﬁcient to induce W-CIN. driving defects.

Loss of CDKN1A expression in response to loss of Loss of the p53/p73 – p21CIP1 axis unleashes CDK1 p53/p73 is crucial for the induction of increased activity that mediates increased microtubule growth microtubule growth rates and W-CIN rates and W-CIN

It is well established that the transcription factors p53 and p21CIP1 acts as a CDK inhibitor inhibiting CDK1 and CDK2 p73 have overlapping gene targets [34]. One of the premier [24]. Since CDK1 is a key regulator of mitosis, we hypo- targets of p53 and p73 is CDKN1A, which encodes for the thesized that CDK1 activity might be unleashed upon loss cyclin-dependent kinase (CDK) inhibitor protein p21CIP1. of the p53/p73 - p21CIP1 axis and thereby trigger abnormal p21CIP1 levels are strongly induced e.g., after DNA damage, microtubule growth rates and chromosome missegregation inhibit CDK activities and thereby, mediate cell cycle arrest in mitosis. To test this, we performed rescue experiments [27]. In unstressed cells, the levels of p21CIP1 are low and the using the CDK1 specific inhibitor RO-3306 [37]. As role of p21CIP1 during an unperturbed cell cycle is less clear. expected, higher concentrations of RO-3306 resulted in cell We investigated if loss of p21CIP1 in response to loss of p53/ cycle arrest in G2 before cells enter mitosis (Fig. S5a). We p73 mediates the abnormal increase in microtubule growth titrated very low doses of RO-3306 on control cells, TP53/ rates and chromosome missegregation. We used chromoso- TP73-deficient and CDKN1A knockout cells and found mally stable DLD-1 cells with or without homozygous efficient restoration of proper microtubule growth rates and deletion of CDKN1A (DLD-1-CDKN1A−/−;Fig.3a). Indeed, a suppression of lagging chromosomes (Fig. 4a–d, loss of CDKN1A caused the same increase in microtubule Fig. S5b). Remarkably, already very low doses of RO-3306 growth rates during mitosis (Fig. 3b) and induced the gen- that did not affect cell cycle progression (Fig. S5a) were eration of lagging chromosomes in anaphase in a microtubule sufficient for these rescue effects suggesting that CDK1 growth rate-dependent manner (Fig. 3c). Experiments with activity is only mildly increased upon p53/p73 or p21CIP1 HCT116 cells harboring a homozygous knockout for loss. To investigate an involvement of mildly increased CDKN1A (Fig. S3a) verified these findings (Fig. S3b). CDK1 activity in the induction of W-CIN, we generated Moreover, acute siRNA-mediated depletion of TP53 and single-cell clones derived from control, TP53/TP73,or TP73 resulted in significant downregulation of p21CIP1 levels CDKN1A-deficient cells and grew them for 30 generations (Fig. S3c) and was associated with an increase in mitotic in the continuous absence or presence of the CDK1 inhi- microtubule growth rates (Fig. S3d). Karyotype analyses of bitor and determined the resulting chromosome number single-cell clones derived from DLD-1-CDKN1A−/− cells variability. In fact, weak inhibition of CDK1 was found to that were grown for 30 generations revealed the induction of be sufficient to restore proper microtubule growth rates in W-CIN upon loss of CDKN1A expression, which was the single-cell clones (Fig. 4e, f) and to suppress W-CIN in dependent on increased microtubule assembly rates (Fig. 3d, the absence of p53/p73 or p21CIP1 (Fig. 4g, h, Fig. S6a, b). Fig. S3e, S3f). Hence, the loss of CDKN1A expression fully It is of note that suppression of W-CIN and the evolvement mimics the loss of p53/p73. of aneuploidy in these long-term experiments required even To further support a causal relationship between loss of lower concentrations of RO-3306 (0.5 µM instead of 1.0 µM p53/p73-p21CIP1 and mitotic defects leading to W-CIN, we re- in the short-term experiments), which might be related to expressed CDKN1A in cells with p53/p73 depletion. To avoid intra-cellular accumulation of the inhibitor upon prolonged cell cycle arrest after high expression of CDKN1A,we treatment (Fig. 4a, b, Fig. S5b). Our results suggest that the employed a previously established chromosomally stable RKO inactivation of the p53/p73 - p21CIP1 axis mildly unleashes cell line [29], in which CDKN1A is expressed in a ponasterone CDK1 activity and this causes abnormal microtubule A-inducible manner (RKO-p21-Pon cells; Fig. S4a). As shown growth rates and chromosome missegregation in mitosis. before [24, 35, 36], high levels of CDKN1A expression caused cell cycle arrest in G1 and G2 (Fig. S4b). RKO cells exhibited Mild unscheduled activation of CDK1 activity normal microtubule assembly rates and low levels of lagging triggers increased microtubule growth rates and chromosomes, both of which were induced upon p53/p73 W-CIN depletion (Fig. 3e, f). Importantly, low-level expression of CDKN1A, which did not influence cell cycle progression (Fig. We considered that direct, mild activation of CDK1 inde- S4b), restored proper microtubule growth rates (Fig. 3e) and pendently of the p53/p73 - p21CIP1 axis might result in 442 A.-K. Schmidt et al.

a -/- b 25 ****

CDKN1A **** on rate i 20 ns DLD-1 DLD-1-

CIP1 p21 m/min) 15 ( p73 16.4 16.6 19.2 16.7 10 m/min m/min m/min m/min p53 n=30n=30 n=30 n=30 microtubule polymerisat microtubule --++0.2 nM Taxol β-actin DLD-1 DLD-1- CDKN1A-/-

** c 10 d 60 **

%) 50 8 ** ( ** 40

6 modal ns

m 30 hromosomes (%) hromosomes 4 20 ns 2 10 deviating fro deviating proportion of anaphase cells anaphase of proportion

0 number chromosome with cells 0 with lagging c with -- 1 + - + 0.2 nM Taxol 3 DLD-1DLD1 DLD-1- CDKN1A-/- clone 2 clone 3 clone 2 clone 3 clone 1 clone 2 clone clone 2 clone clone 1 clone 3 clone 1

DLD1-CDKN1A-/ DMSO0.2 nM DMSO 0.2 nM Taxol Taxol DLD-1 DLD-1-CDKN1A-/-

ef**** * **** * 25 **** 10 **** * **** 8 20 6 m/min) of anaphase cells anaphase of

( 15 4

16.4 19.5 18.4 17.0 16.7 16.9 2 10 m/min m/min m/min m/min m/min m/min

n=30 n=30 n=30 n=30 n=30 n=30 proportion 0 microtubule polymerisation rate polymerisation microtubule with lagging chromosomes (%) xol DMSO DMSO

0.5 μM PonA 0.2 nM Ta 0.2 nM Taxol 0.25 μM PonA 0.75 μM PonA 0.75 μM PonA siRNA + DMSO siRNA + DMSO + TP53 + TP73 siRNA + TP53 + TP73 siRNA LUC LUC RKO-p21-Pon RKO-p21-Pon increased microtubule growth rates and chromosome mis- already sufﬁcient to increase mitotic microtubule assembly segregation in mitosis. To directly activate CDK1, we rates and to induce lagging chromosomes (Fig. 5a, b). inhibited the wee1 kinase that is known to inhibit CDK1 Moreover, these effects were rescued upon mild inhibition activity by direct phosphorylation of CDK1 on tyrosine-15 of CDK1 by low-dose RO-3306 demonstrating that wee1 [38]. Titration experiments using the established wee1 inhi- inhibition acts on microtubule dynamics and lagging chro- bitor MK-1775 [39]veriﬁed that CDK1 is indeed depho- mosome induction through mildly increased CDK1 activity sphorylated and activated in an inhibitor concentration- (Fig. 5a, b). Further, we directly elevated CDK1 by dependent manner leading to premature entry into mitosis expression of a non-phosphorylatable, constitutively active (Fig. S7). Importantly, even very low concentrations of the mutant of CDK1 (CDK1-AF) and, as a control, of a kinase- wee1 inhibitor that cause only very little dephosphorylation dead mutant of CDK1 (CDK1-DN, Fig. 5c) [40]. of CDK1 and no premature entry into mitosis (Fig. S7) were Only expression of CDK1-AF triggered an increase in The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal. . . 443

Fig. 3 Increased microtubule growth rates and chromosomal cells we analyzed a panel of colorectal cancer (CRC) cell instability after concomitant loss of p53 and p73 is mediated by lines characterized by either MIN/MSI (chromosomally loss of p21CIP1.ap73, p53, and p21CIP1 protein levels in chromoso- − − stable, nearly diploid) or W-CIN (chromosomally unstable, mally stable DLD-1 and DLD-1-CDKN1A / cells. A representative western blot is shown. b Microtubule plus end assembly rates in aneuploid). As shown previously [10], only CRC cell lines mitotic DLD-1 and DLD-1-CDKN1A−/− cells in the absence or pre- characterized by W-CIN exhibit increased microtubule sence of low-dose Taxol. Average microtubule polymerization rates growth rates and lagging chromosomes (Fig. 6a, b). We n = are shown in the scatter dot plots. (20 microtubules/cell, 30 mitotic treated the different cell lines with low-doses of the CDK1 cells from three independent experiments, mean ± SD, t-test). c Pro- portion of cells exhibiting lagging chromosomes in anaphase after loss inhibitor RO-3306, and measured mitotic microtubule of p21CIP1, and after restoration of proper microtubule growth rates. polymerization rates. In fact, mild CDK1 inhibition was The proportion of anaphase cells exhibiting lagging chromosomes was sufﬁcient to restore proper microtubule assembly rates in = determined using the indicated cell lines (n 300 anaphase cells from cell lines exhibiting W-CIN and had no effect in MIN/MSI 3 independent experiments, mean ± SD, t-test). d Aneuploidy induction upon loss of CDKN1A expression. Single-cell clones derived from cells (Fig. 6a). Consequently, lagging chromosomes and parental DLD-1 and DLD-1-CDKN1A−/− cells were grown in the thus, chromosome missegregation were also signiﬁcantly presence or absence of 0.2 nM Taxol for 30 generations, chromosome suppressed in W-CIN cell lines upon mild CDK1 inhibition numbers per cell were determined. The graph shows the proportion of (Fig. 6b). It is of note that we were not able to consistently cells harboring chromosome numbers deviating from the modal (46 CIP1 chromosomes in DLD-1 cells; n = 50 cells). e Microtubule plus end associate alterations in p53, p73, p21 or CDK1 protein assembly rates in mitotic RKO-p21-Pon cells upon knockdown of levels with increased microtubule polymerization rates in TP53 and TP73 and ponasterone-A-induced re-expression of the different W-CIN cell lines (Fig. S9) suggesting that CDKN1A . Scatter dot plots show average microtubule polymerization various mechanisms and alterations might contribute to rates (20 microtubules/cell, n = 30 mitotic cells from three independent experiments, mean ± SD, t-test). f Proportion of cells exhibiting CDK1 deregulation in human cancer cells (see model in lagging chromosomes in anaphase after loss of TP53 and TP73 and re- Fig. 7). expression of CDKN1A. As control, cells were treated with low-dose Taxol to restore proper microtubule growth rates independently of Alterations of TP53/TP73 and CDKN1A are associated CDKN1A expression (n = 300 anaphase cells from 3 independent experiments, mean ± SD, t-test). with W-CIN in human cancer

Our experimental data presented here indicate that loss of microtubule growth rates in mitosis and induced lagging TP53 and TP73 can trigger W-CIN by mediating a loss of chromosomes (Fig. 5d, e). These effects of CDK1-AF CDKN1A. This prompted us to investigate whether this expression were found to be dependent on its kinase activity situation is reflected in human cancer samples. We since the mitotic defects were suppressed by CDK1 inhi- employed bioinformatics correlation analyses of 389 col- bition using low doses of RO-3306 (Fig 5d, e). Also, Taxol orectal cancer (CRC) samples and 972 breast carcinoma treatment corrected normal microtubule polymerization samples and were able to detect cases with normal rates and suppressed the generation of lagging chromo- p53 status, with loss of p53, and concomitant loss of p53 somes (Fig. 5d, e). To prove whether increased CDK1 and p73, the latter at rather low frequency. Importantly, we activity is sufficient to induce W-CIN, we generated single- found a significant correlation of low CDKN1A expression cell clones stably expressing either active CDK1-AF or with loss of TP53 and TP73 for CRC and breast cancer kinase-dead CDK1-DN (Fig. 5f). Even very low-level (Fig. 6c, Fig. S10a). We then categorized the tumor samples expression of CDK1-AF was sufficient to increase micro- based on the calculated weighted genome integrity index tubule growth rates (Fig. 5g), induced lagging chromo- (wGII) into low and high CIN groups and correlated the somes (Fig. 5h) and W-CIN (Fig. 5i, Fig. S8). Together, status of CIN with the expression level of CDKN1A. these results suggest that even mildly increased CDK1 Interestingly, there was a significant correlation between activity as a result of the loss of the p53/p73 - p21CIP1 axis low CDKN1A expression and the presence of chromosomal or upon inhibition of wee1 can act as a trigger for W-CIN instability (Fig. 6d, Fig. S10b). Thus, these bioinformatic by increasing microtubule assembly rates in mitosis. analyses support the hypothesis that loss of CDKN1A expression associated with unleashed CDK1 can be linked Mild inhibition of CDK1 is sufficient to suppress to CIN in human cancer. Given the observation that con- increased microtubule polymerization rates and comitant loss of TP53 and TP73 is rather infrequent in chromosome missegregation in cancer cells cancer (at least in CRC and breast cancer) this also suggests that routes other than p53/p73 loss can result in lowered To test whether unleashed CDK1 activity can act as a expression of CDKN1A and increased CDK1 activity and trigger for increased microtubule polymerization rates and contribute to abnormally increased microtubule poly- subsequent chromosome missegregation in human cancer merization rates and CIN in human cancer (Fig. 7). 444 A.-K. Schmidt et al.

**** ab**** te **** 25 **** 25 **** **** *** *** on ra **** **** ati * ns ns 20 ns is 20 in) mer m / m m/min) 15 15 ( ( le poly 16.4 16.3 16.3 19.2 18.4 17.2 16.1 16.4 16.2 16.5 16.7 19.3 18.3 17.6 16.5 16.5 m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min

10 tubu 10 n=30 n=30 n=30 n=30n=30 n=30 n=30 n=30 n=30 n=30n=30 n=30 n=30 n=30 n=30 n=30 ro

l 6 6 ic 6 6 6 0 0 m microtubule polymerisation rate polymerisation microtubule -3306 DMSO Taxol DMSO -330 -330 DMSO DMSO O O O R RO-33 R R M RO-3306 M RO-33M M RO-330 M μ 0.2 nM Taxo μ μM 0.2 nM Taxol μM RO-33060.2 nM μM 0.2 nM Taxol 0 0 μ 1.0 0.5 1.0 1. 0.5 1. 0.25 0.25 HCT116 + HCT116-TP53-/- + DLD-1 DLD-1-CDKN1A-/- control shRNA TP73 shRNA (clone 2)

**** ** 10 (%) * ** 10 * c **** d es 8 8 osom 6 6 om ns hr ns

ns c 4 4 ns ng 2 i 2 agg

0 h l 0 --+ --+ 1.0 μM RO-3306 t --+ --+ 1.0 μM RO-3306 proportion of anaphase cells anaphase of proportion cells anaphase of proportion

with lagging chromosomes (%) chromosomes lagging with --+ --+ 0.2 nM Taxol wi --+ --+ 0.2 nM Taxol HCT116 HCT116-TP53-/- DLD-1 DLD-1- + control + TP73 shRNA CDKN1A-/- shRNA (clone 2)

e 25 **** f 25 **** ation rate ation ation rate ation **** ****

s ns ns i ris 20 r 20 yme yme l l m/min) m/min) 15 15 ( ( le po 16.6 16.6 16.4 16.6 16.1 16.5 19.5 19.0 19.4 16.7 16.7 16.5 16.6 16.5 16.7 16.6 16.5 16.2 19.3 19.0 19.1 16.4 16.5 16.4 m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min 10 10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 tubu n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 otubule po otubule r 3 2 3 2 1 1 3 micro mic clone clone 2 clone clone clone 1 clone 1 clone clone clone clone clone 3 clone 2 clone 1 clone 2 clone 3 clone 1 clone 2 clone 3 clone 2.3 clone 2.1 clone 2.1 clone 2.2 clone 2.2 clone 2.3 DMSO0.5 μM DMSO 0.5 μM DMSO0.5 μM DMSO 0.5 μM RO-3306 RO-3306 RO-3306 RO-3306 DLD-1 DLD-1-CDKN1A-/- HCT116 + HCT116-TP53-/- + control shRNA TP73 shRNA

*** g 60 60 ) **** h

%) *** (

50 *** l 50 e number

40 m 40 o ns s

30 m moda 30 o ns 20 fr 20 ng i chromo ing from modal (% modal from ing t h t 10 ia

v 10 wi de deviat 0 0 1 1 2 2 cells cells cells with chromosome number clone clone 3 clone clone 2 clone 3 clone clone 3 clone 3 clone 2 clone 2 clone 3 clone 1 clone 1 clone 1 clone 3 clone 1 clone 2 clone clone 2.3 clone 2.2 clone 2.3 clone 2.1 clone 2.2 clone 2.1 DMSO0.5 μM DMSO 0.5 μM DMSO0.5 μM DMSO 0.5 μM RO-3306 RO-3306 RO-3306 RO-3306 DLD-1 DLD-1-CDKN1A-/- HCT116 + HCT116-TP53-/- control shRNA + TP73 shRNA The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal. . . 445

Fig. 4 Inhibition of CDK1 suppresses abnormal microtubule anaphase as a pre-stage of chromosome missegregation polymerization rates and chromosomal instability after loss of p53/ [10]. Hence, transient spindle positioning alterations might p73 or p21CIP1.aRescue of abnormal microtubule growth rates upon TP53/TP73 fi act similarly to transient multipolar spindle intermediates partial CDK1 inhibition in -de cient cells. HCT116 and fi TP53/TP73-deficient cells were treated with the CDK1 inhibitor RO- upon centrosome ampli cation or abnormal spindle geo- 3306 or with low-dose Taxol for 16 h and mitotic microtubule plus end metry to increase the probability for the generation of assembly rates were measured. Scatter dot plots show average erroneous merotelic microtubule-kinetochore attachments n = microtubule polymerization rates (20 microtubules/cell, 30 mitotic as the basis for the formation of lagging chromosomes cells from three independent experiments, mean ± SD, t-test). b Rescue of abnormal microtubule growth rates upon partial CDK1 inhibition in [6, 7, 41, 42]. CDKN1A-deficient cells. DLD-1 and DLD-1-CDKN1A−/− cells were In our work presented here, we focused on the treated as in a and mitotic microtubule polymerization rates were upstream events responsible for triggering abnormal n = determined (20 microtubules/cell, 30 mitotic cells from three microtubule polymerization in cancer cells. Specifically, independent experiments, mean ± SD, t-test). c Suppression of lagging chromosomes upon partial CDK1 inhibition in TP53/TP73-deficient we asked which cancer-relevant alterations can contribute cells. HCT116 and TP53/TP73-deficient cells were treated with the to W-CIN in cancer cells. By screening for known CDK1 inhibitor RO-3306 or with low-dose Taxol. The graph shows oncogenes and tumor suppressor genes, we found that the proportion of cells exhibiting lagging chromosomes during ana- loss of the tumor suppressor gene TP53 is not sufficient to phase (n = 300 anaphase cells from three independent experiments, mean ± SD, t-test). d Suppression of lagging chromosomes upon trigger W-CIN, which is in line with former results partial CDK1 inhibition in CDKN1A-deficient cells. Parental and [10, 18]. On the other hand, it is well known that loss of CDKN1A-deficient DLD-1 cells were treated as in c. The graph shows TP53 is among the most significant cancer alterations the proportion of cells exhibiting lagging chromosomes during ana- n = associated with chromosomal instability [17]. It was phase ( 300 anaphase cells from three independent experiments, TP53 mean ± SD, t-test). e, f Mitotic microtubule growth rates in single-cell suggested that the association between loss of and clones derived from control, TP53/TP73-deficient HCT116 cells or CIN might be explained by the fact that loss of TP53 from parental and CDKN1A-deficient DLD-1 cells and treated with or creates a permissive situation for the proliferation of without CDK1 inhibitor (RO-3306). Scatter dot plots show average aneuploid cells and is therefore required for long-term microtubule growth rates (20 microtubules/cell, n = 10 mitotic cells for each clone, mean ± SD, t-test). g, h Chromosome number varia- proliferation of chromosomally unstable cancer cells [43]. bility in single-cell clones derived from control, TP53/TP73-deficient However, this view has been challenged by the obser- HCT116 cells or from parental and CDKN1A-deficient DLD-1 cells vation that chromosome missegregation at a level com- and treated with or without CDK1 inhibitor. The proportion of cells parable to cancer cells does not obligatory result in a with a chromosome number deviating from modal (45 chromosomes for HCT116, 46 chromosomes for DLD-1) was calculated for three strong p53 response leading to cell cycle arrest [44]. independent clones (n = 50 cells per condition). Also, it has been well documented that chromosomally unstable, but p53 proficient cells are still able to pro- liferate. This applies for instance to MAD2 or CHK2 Discussion deficient HCT116 cells that are chromosomally unstable, but harbor functional p53 [45, 46]. Nevertheless, loss of Our work demonstrates that mildly increased CDK1 p53 might still be crucial to improve the proliferation of activity is sufficient to trigger whole chromosome mis- aneuploid cancer cells, which are otherwise strongly segregation in mitosis and thus, whole chromosome growth impaired [10, 47, 48]. instability (W-CIN) in otherwise chromosomally stable Our work now revealed that loss of p53 can contribute to human cells. In accordance with our previous work the induction of W-CIN, but only if its cousin p73 is also [10–12, 16], W-CIN in response to unleashed CDK1 is impaired indicating that p53 and p73 collaborate to prevent mediated by an abnormal increase in microtubule plus end aneuploidy. Indeed, concomitant loss of p53 and p73 or polymerization rates within mitotic spindles. This parti- expression of DNp73, which is known to act as a trans- cular mitotic defect is a characteristic of chromosomally repressor of p53- and p73-dependent transcription [49], unstable cancer cells and is defined by an approx. 20-30% results in increased microtubule polymerization rates, increase in microtubule plus end polymerization rates chromosome missegregation, and W-CIN. without grossly affecting other microtubule dynamics In contrast to TP53, TP73 is rarely mutated in cancer, parameters and without affecting the overall spindle size but loss of its expression can be detected in various and structure [10]. It is currently not understood in detail human cancers. This might be related to promotor how increased microtubule polymerization rates cause methylation or a loss of the genomic locus of TP73 on chromosome missegregation, but existing data indicate that chromosome 1p.36, which is frequently deleted e.g., in abnormal microtubule polymerization rates trigger tran- neuroblastomas [49, 50]. We were able to detect con- sient spindle mispositioning, thereby facilitating the gen- comitant loss of p53 and p73 in colorectal and breast eration of erroneous merotelic microtubule-kinetochore cancer samples, albeit at low frequency. However, a loss attachments and inducing lagging chromosomes during of the common p53/p73 downstream target CDKN1A 446 A.-K. Schmidt et al.

ab 25 10 ** s(%) ***

**** e

**** ecells **** **** s 8 20 som

mo 6 anapha hro m/min) 15

of of 4 (

16.4 17.6 18.3 18.8 16.6 ing c 2 10 m/min m/min m/min m/min m/min n=30 n=30 n=30 n=30 n=30

hlagg 0 proportion wit microtubule polymerisation rate

DMSO DMSO

+ 1

25 nM 50MK-1775 nM 75MK-1775 nM 177575MK-1775 nM MK-1775 75 nM MK-177575 nM MK-1775

M MK- + 1.0 μM RO-3306 + 1.0 μM RO-3306 75HCT116 n HCT116

c d e )

HCT116 s (% 25 **** ll 10 ** **** ** **** ce ** ns e 8 ns ns ns 20 ns 6 ns CDK1-AF vector CDK1-DN mosomes o m/min) FLAG hr 4 15 c ( CDK1-FLAG 16.4 16.5 16.5 16.6 19.0 16.0 16.5 ion anaphas of 2 CDK1 m/min m/min m/min m/min m/min m/min m/min t 10 n=30 n=30 n=30 n=30 n=30 n=30 n=30 0 α-tubulin -+-- -+-1.0 M RO-3306 -+-- -+-1.0 M RO-3306 propor microtubule polymerisation rate --+- --+0.2 nM Taxol with lagging --+- --+0.2 nM Taxol vector CDK1 CDK1-AF vector CDK1 CDK1 -DN -DN -AF HCT116 HCT116 e HCT116 t f g a 25 **** vector CDK1-DN CDK1-AF on r **** ti

a ns s i 20 n) i er clone 3 clone 1 clone 2 clone 3 clone 1 clone 2 clone 3 clone 1 clone 2

FLAG m/m ( 15 epolym l CDK1-FLAG CDK1 16.5 16.7 16.8 17.0 16.3 16.5 18.6 19.0 19.6 10 m/min m/min m/min m/min m/min m/min m/min m/min m/min crotubu i n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10 n=10

α-tubulin m 2 3 1 2 1 1 3 2 clone clone clone 3 clone clone clone clone clone clone vector CDK1-DN CDK1-AF HCT116 h ** *** r

i be 60 )

lls 10 ns ns m (% ce

ns l 50 e 8 **** 40 *** 6 ns ns moda ns ns 30 anaphas rom

f * f 4 romosome nu h 20 ng on o

2 viati 10 de 0

proporti 0 cells with c with lagging chromosomes (%) 3 3 2 1 2 clone 3 clone 1 clone 2 clone clone 2 clone clone 1 clone clone 1 clone clone 2 clone 1 clone 3 clone 2 clone 3 clone 1 clone clone 3 vector CDK1 CDK1 vector CDK1 CDK1 -DN -AF -DN -AF HCT116 HCT116 seems to be more frequent in human cancer [51]. Along not with MIN/MSI [52], which is in agreement with our this line, it was shown that 50% of CRC cases show a loss results from colorectal and breast cancer showing that of CDKN1A expression and this correlated with CIN, but loss of CDKN1A correlates with the CIN status of the The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal. . . 447

Fig. 5 Increased CDK1 induces chromosomal instability. a Mitotic whichwasshowntobephosphorylatedbyCDK1in microtubule polymerization rates upon induction of CDK1 activity mitosis and this might be involved in stabilization of mediated by inhibition of wee1. HCT116 cells were treated with microtubule-kinetochore attachments [54]. Interestingly, increasing concentrations of the wee1 inhibitor MK-1775 in the absence or presence of the CDK1 inhibitor RO-3306 and mitotic hyper-stable microtubule-kinetochore attachments, which microtubule plus end assembly rates were determined. Average might result from hyper-phosphorylation of CLASP2, microtubule polymerization rates are depicted in scatter dot plots (20 have been associated with the induction of lagging n = microtubules/cell, 30 mitotic cells from three independent chromosomes and whole chromosome missegregation in experiments, mean ± SD, t-test). b Induction of lagging chromosomes upon CDK1 activation mediated by wee1 inhibition. HCT116 cells mitosis [9]. were treated with MK-1775 in the absence or presence of RO-3306. Our previous work has identified the tumor suppressors The graph shows the proportion of anaphase cells exhibiting lagging BRCA1 and CHK2 and the oncogenic kinase AURKA as n = chromosomes ( 300 anaphase cells from three independent regulators of mitotic microtubule growth rates and for W- experiments, mean ± SD, t-test). c Transient expression of FLAG- tagged CDK1-AF (constitutive active) or CDK1-DN (inactive) in CIN [10, 11, 13]. Whether the BRCA1-CHK2-AURKA chromosomally stable HCT116 cells. A representative western blot axis is influenced by increased CDK1 activity needs to be detecting CDK1 protein levels and exogenously expressed CDK1 further investigated. (FLAG) is shown. d Mitotic microtubule polymerization rates after Given the fact that unleashed CDK1 is central to the transient expression of CDK1-AF or CDK1-DN in HCT116 cells. Average microtubule polymerization rates are depicted in scatter dot induction of CIN, one has to consider various routes that plots (20 microtubules/cell, n = 30 mitotic cells from three indepen- can lead to increased CDK1 activity in cancer cells (Fig. 7). dent experiments, mean ± SD, t-test). e Quantification of cells exhi- As demonstrated here, functional inactivation of p53/p73 or biting lagging chromosomes after transient expression of CDK1-AF or CIP1 CDK1-DN n = loss of p21 can drive CIN in a CDK1-dependent man- in the absence or presence of RO-3306 or Taxol ( 300 CDK1 anaphase cells from three independent experiments, mean ± SD, t-test). ner. Direct overexpression of or of its activating f Generation of HCT116 cell lines stably expressing CDK1-AF or subunit cyclin B (CNNB1/2) are obvious additional possi- CDK1-DN. The expression level of the CDK1 variants was detected in bilities to achieve higher CDK1 activity in cancer. In fact, three independent cell clones. A representative western blot is shown. overexpression of CDK1 or of CNNB1/2 is frequently found g Mitotic microtubule polymerization rates in single-cell clones of HCT116 cells stably expressing CDK1-AF or CDK1-DN. Average in various human cancers and correlates with poor prog- microtubule polymerization rates are depicted in scatter dot plots (20 nosis [41, 55]. Moreover, CDK1 and CNNB1/2 have been microtubules/cell, n = 10 mitotic cells for each independent single-cell identified as CIN signature genes whose overexpression clone, mean ± SD, t-test). h Quantification of cells exhibiting lagging CDK1- correlates with CIN in cancer [56]. In this regard it is also chromosomes in HCT116 single-cell clones stably expressing CNNB2 AF or CDK1-DN (n = 300 anaphase cells for each independent single- interesting to note that overexpression of , which cell clone, mean ± SD, t-test). i Chromosome number variability in could also promote CDK1, was shown to result in spindle single-cell clones derived from HCT116 cells stably expressing positioning defects, chromosome missegregation, and the CDK1-AF CDK1-DN or . The proportion of cells with a chromosome induction of aneuploidy [41]. number deviating from modal (45 chromosomes for HCT116 cells) was calculated for three independent clones (n = 50 cells per single- In addition to direct alterations in CDK1-cyclin B itself, cell clone). also upstream regulators of CDK1 can potentially be altered in cancer cells and can be involved in the induction of CIN. In fact, a network of kinases and phosphatases are involved cancer samples. However, reduced levels of CDKN1A are in the precise regulation of CDK1 activity at the G2/M not only a consequence of loss of p53/p73. Also increased transition (for review see: [57]). Among these upstream promotor repression, which can be mediated e.g., by high regulators, cdc25 phosphatases are known to act as activa- levels of the myc oncogene [53], can account for reduced tors of CDK activity and are frequently overexpressed in CDKN1A expression. Hence, different cancer-relevant cancer [58]. Also, changes in the subcellular localization or alterations might contribute to downregulation of a spectrum of mutations might account for functional CDKN1A and thereby, possibly to CIN. alterations in upstream pathways leading to hyper-activation Our data implicate increased CDK1 activity as a med- of CDK1. iator of CIN. This low level of additional CDK1 activity is It is intriguing that W-CIN associated with whole chro- not sufficient to induce premature entry into mitosis sug- mosome missegregation is not only leading to whole gesting that a severe deregulation of cell cycle timing is chromosome missegregation and aneuploidy, but also to the unlikely to account for the observed mitotic defects. induction of structural chromosome aberrations. In fact, Instead, it seems more plausible that even minor changes recent work has demonstrated that missegregation of a in CDK1-mediated phosphorylation on specifictarget single chromosome can lead to its encapsulation into a proteins can contribute to increased microtubule poly- micronucleus, which, in turn, is prone for chromotripsis merization. However, these targets are not yet known. One leading to complex genomic rearrangements [59, 60]. Thus, example of such a candidate protein might be CLASP2, W-CIN is considered a driver for the rapid development of 448 A.-K. Schmidt et al.

a 25

ns ns ns **** **** **** 20

m/min) 15 (

16.7 16.7 16.6 16.4 16.6 16.3 19.0 16.7 19.1 16.7 19.9 17.0 m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min m/min 10 n=30 n=30 n=30 n=30 n=30 n=30 n=30 n=30 n=30 n=30 n=30 n=30 microtubule polymerisation rate polymerisation microtubule - +- + - + - + - + - + 1.0 M RO-3306 HCT116 DLD-1 RKO SW480 SW620 HT29 MIN/MSI W-CIN

b 10 ** ** **

es (%) es 8 m

6 ns ns ns 4 ng chromoso ng

2 proportion of anaphase cells anaphase of proportion with laggi 0 - + - + - + - + - + - + 1.0 M RO-3306 HCT116DLD-1 RKO SW480 SW620 HT29 MIN/MSI W-CIN

c TCGA: colorectal adenocarcinoma d TCGA: colorectal adenocarcinoma

0.6 5.05.0 Wilcoxon rank sum p−value: 1.783e−57 Kruskal−Wallis, p < 2.2e−16

p= 4.6e−05 CIN status

p= 0.0084 expression low CIN (n=73) 2.52.5 p=1.2e−15 high CIN (n=316) 0.4 CDKN1A

0.00.0 density

-2.5−2.5 0.2 log2 fold change of TP53: WT TP53: mutant TP53: mutant TP73: normal TP73: normal TP73: low 0.0 −4 −2 0 2

log2 fold change of CDKN1A expression Fig. 6 Relevance of the TP53, TP73, CDKN1A, and CDK1 status anaphase cells from three independent experiments, mean ± SD, t-test). for chromosomal instability in colorectal cancer. a Rescue of c Relationship between TP53 and TP73 status to CDKN1A expression increased mitotic microtubule polymerization rates in various color- levels in colorectal cancer. 389 human tumor samples from The ectal cancer cell lines exhibiting W-CIN by mild inhibition of CDK1. Cancer Genome Atlas (TCGA) were analyzed for their TP53 muta- The indicated cell lines exhibiting either MIN/MSI or W-CIN were tional status and TP73 expression level status, and its association with used to measure mitotic microtubule polymerization rates in the CDKN1A expression. The pair-wise p values were obtained from a absence or presence of low concentrations of the CDK1 inhibitor RO- Wilcoxon rank sum test. d The relationship between CDKN1A 3306 (20 microtubules/cell, n = 30 mitotic cells from three indepen- expression and chromosomal instability (weighted Genome Integrity dent experiments, mean ± SD, t-test). b Suppression of lagging chro- Index, wGII). 389 colorectal adenocarcinoma samples from The mosomes upon partial CDK1 inhibition in colorectal cancer cell lines Cancer Genome Atlas (TCGA) were categorized into low and high exhibiting W-CIN. The indicated cell lines were treated with or CIN groups and aligned to their expression level of CDKN1A. The without low concentrations of RO-3306 and the proportion of cells combined density plot and histogram of CDKN1A gene expression showing lagging chromosomes in anaphase was determined (n = 300 indicates the shift of the distribution between high and low CIN scores. The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal. . . 449

expression Compliance with ethical standards of DNp73 expression or of myc loss of p53/p73 Conﬂict of interest The authors declare that they have no conﬂict of interest.

high Publisher’s note Springer Nature remains neutral with regard to CDK1 loss of p21 low high wee1 high jurisdictional claims in published maps and institutional afﬁliations. cyclin cdc25 ...... Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, increased CDK1 activity adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not ’ increased microtubule included in the article s Creative Commons license and your intended plus end growth use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/. References lagging chromosomes during anaphase 1. Bakhoum SF, Landau DA. Chromosomal instability as a driver of tumor heterogeneity and evolution. Cold Spring Harb Perspect Med. 2017;7(6):a029611. 2. Turajlic S, Sottoriva A, Graham T, Swanton C. Resolving genetic heterogeneity in cancer. Nat Rev Genet. 2019;20:404–16. 3. Sansregret L, Swanton C. The role of aneuploidy in cancer evolution. Cold Spring Harb Perspect Med. 2017;7:a028373. whole chromosome 4. Lengauer C, Kinzler KW, Vogelstein B. Genetic instabilities in missegregation human cancers. Nature. 1998;396:643–9. chromosomal instability (W-CIN) 5. Gregan J, Polakova S, Zhang L, Tolic-Norrelykke IM, Cimini D. Fig. 7 Model: increased CDK1 activity as a trigger for whole Merotelic kinetochore attachment: causes and effects. Trends Cell – chromosome instability in human cancer. Several possible ways that Biol. 2011;21:374 81. lead to an increase in CDK1 activity in human cancer cells are illu- 6. Ganem NJ, Godinho SA, Pellman D. A mechanism linking extra strated. Examples are loss of the p53/p73 - p21CIP1 axis, loss of wee1 centrosomes to chromosomal instability. Nature. 2009;460: – or overexpression of CDK1 or associated cyclins. In turn, CDK1 278 82. activity increases microtubule polymerization rates in mitosis leading 7. Silkworth WT, Nardi IK, Scholl LM, Cimini D. Multipolar to the generation of lagging chromosomes and causing whole chro- spindle pole coalescence is a major source of kinetochore mis- mosome missegregation and aneuploidy. attachment and chromosome mis-segregation in cancer cells. PloS One. 2009;4:e6564. genome alterations on both, a numerical and structural level 8. Bakhoum SF, Thompson SL, Manning AL, Compton DA. Gen- ome stability is ensured by temporal control of kinetochore- and increased CDK1 activity might contribute to these microtubule dynamics. Nat Cell Biol. 2009;11:27–35. chromosomal aberrations. 9. Bakhoum SF, Genovese G, Compton DA. Deviant kinetochore microtubule dynamics underlie chromosomal instability. Curr – Acknowledgements We thank Magdalena Hennecke for technical Biol. 2009;19:1937 42. assistance. We are grateful to Thorsten Stiewe, Linda Wordeman, 10. Ertych N, Stolz A, Stenzinger A, Weichert W, Kaulfuss S, Bur- ﬂ Matthias Schmidt, Makato Iimori, and Lienhard Schmitz for providing feind P, et al. Increased microtubule assembly rates in uence plasmids and cell lines. This work was supported by grants from the chromosomal instability in colorectal cancer cells. Nat Cell Biol. – Wilhelm Sander-Stiftung, from the Deutsche Krebshilfe, and from the 2014;16:779 91. Deutsche Forschungsgemeinschaft to HB and by the FOR2800 funded 11. Ertych N, Stolz A, Valerius O, Braus GH, Bastians H. CHK2- by the Deutsche Forschungsgemeinschaft (MK and HB). Open Access BRCA1 tumor-suppressor axis restrains oncogenic Aurora-A funding enabled and organized by Projekt DEAL. kinase to ensure proper mitotic microtubule assembly. Proc Natl Acad Sci USA. 2016;113:1817–22. 12. Luddecke S, Ertych N, Stenzinger A, Weichert W, Beissbarth T, Author contributions A-KS designed and performed experiments, Dyczkowski J, et al. The putative oncogene CEP72 inhibits the analyzed data and contributed to writing of the manuscript. KP per- mitotic function of BRCA1 and induces chromosomal instability. formed experiments and analyzed data. KB generated cell lines and Oncogene. 2016;35:2398–406. performed initial experiments. J-EB and MK performed analyses of 13. Stolz A, Ertych N, Bastians H. Loss of the tumour-suppressor tumor data and contributed to writing of the manuscript. HB designed genes CHK2 and BRCA1 results in chromosomal instability. the study, analyzed data, and wrote the manuscript. Biochem Soc Trans. 2010;38:1704–8. 450 A.-K. Schmidt et al.

14. Stolz A, Neufeld K, Ertych N, Bastians H. Wnt-mediated protein 35. Bates S, Ryan KM, Phillips AC, Vousden KH. Cell cycle arrest stabilization ensures proper mitotic microtubule assembly and and DNA endoreduplication following p21Waf1/Cip1 expression. chromosome segregation. EMBO Rep. 2015;16:490–9. Oncogene. 1998;17:1691–703. 15. Bohly N, Kistner M, Bastians H. Mild replication stress causes 36. Niculescu AB, Chen X, Smeets M, Hengst L, Prives C, Reed S. Cip1/Waf1 aneuploidy by deregulating microtubule dynamics in mitosis. Cell Effects of p21 at both the G1/S and the G2/M cell cycle Cycle. 2019;18:2770–83. transitions: pRb is a critical determinant in blocking DNA repli- 16. Stolz A, Ertych N, Bastians H. Microtubule plus tips: a dynamic cation and in preventing endoreduplication. Mol Cell Biol. route to chromosomal instability. Mol Cell Oncol. 2015;2: 1998;18:629–43. e960768. 37. Vassilev LT, Tovar C, Chen S, Knezevic D, Zhao X, Sun H, et al. 17. Cancer Genome Atlas N. Comprehensive molecular character- Selective small-molecule inhibitor reveals critical mitotic functions ization of human colon and rectal cancer. Nature. 2012;487: of human CDK1. Proc Natl Acad Sci USA. 2006;103:10660–5. 330–7. 38. Schmidt M, Rohe A, Platzer C, Najjar A, Erdmann F, Sippl W. 18. Bunz F, Fauth C, Speicher MR, Dutriaux A, Sedivy JM, Kinzler Regulation of G2/M transition by inhibition of WEE1 and KW, et al. Targeted inactivation of p53 in human cells does not PKMYT1 kinases. Molecules. 2017;22:2045. result in aneuploidy. Cancer Res. 2002;62:1129–33. 39. Hirai H, Iwasawa Y, Okada M, Arai T, Nishibata T, Kobayashi M, 19. Maas AM, Bretz AC, Mack E, Stiewe T. Targeting p73 in cancer. et al. Small-molecule inhibition of Wee1 kinase by MK-1775 Cancer Lett. 2013;332:229–36. selectively sensitizes p53-deficient tumor cells to DNA-damaging 20. Flores ER, Sengupta S, Miller JB, Newman JJ, Bronson R, agents. Mol Cancer Ther. 2009;8:2992–3000. Crowley D, et al. Tumor predisposition in mice mutant for p63 40. Krek W, Nigg EA. Mutations of p34cdc2 phosphorylation sites and p73: evidence for broader tumor suppressor functions for the induce premature mitotic events in HeLa cells: evidence for a p53 family. Cancer Cell. 2005;7:363–73. double block to p34cdc2 kinase activation in vertebrates. EMBO 21. Engelmann D, Meier C, Alla V, Putzer BM. A balancing act: J. 1991;10:3331–41. orchestrating amino-truncated and full-length p73 variants as 41. Nam HJ, van Deursen JM. Cyclin B2 and p53 control proper decisive factors in cancer progression. Oncogene. 2015;34: timing of centrosome separation. Nat Cell Biol. 2014;16:538–49. 4287–99. 42. Silkworth WT, Cimini D. Transient defects of mitotic spindle 22. Di C, Yang L, Zhang H, Ma X, Zhang X, Sun C, et al. geometry and chromosome segregation errors. Cell Div. Mechanisms, function and clinical applications of DNp73. Cell 2012;7:19. Cycle. 2013;12:1861–7. 43. Thompson SL, Compton DA. Proliferation of aneuploid human 23. Warfel NA, El-Deiry WS. p21WAF1 and tumourigenesis: 20 cells is limited by a p53-dependent mechanism. J Cell Biol. years after. Curr Opin Oncol. 2013;25:52–8. 2010;188:369–81. 24. Medema RH, Klompmaker R, Smits VA, Rijksen G. p21waf1 can 44. Santaguida S, Richardson A, Iyer DR, M’Saad O, Zasadil L, block cells at two points in the cell cycle, but does not interfere Knouse KA, et al. Chromosome Mis-segregation generates cell- with processive DNA-replication or stress-activated kinases. cycle-arrested cells with complex karyotypes that are eliminated Oncogene. 1998;16:431–41. by the immune system. Dev Cell. 2017;41:638–51 e635. 25. Kreis NN, Sanhaji M, Rieger MA, Louwen F, Yuan J. p21Waf1/ 45. Stolz A, Ertych N, Kienitz A, Vogel C, Schneider V, Fritz B, et al. Cip1 deficiency causes multiple mitotic defects in tumor cells. The CHK2-BRCA1 tumour suppressor pathway ensures chro- Oncogene. 2014;33:5716–28. mosomal stability in human somatic cells. Nat Cell Biol. 26. Kreis NN, Louwen F, Yuan J. The multifaceted p21 (Cip1/Waf1/ 2010;12:492–9. CDKN1A) in cell differentiation, migration and cancer therapy. 46. Michel LS, Liberal V, Chatterjee A, Kirchwegger R, Pasche B, Cancers. 2019;11:1220. Gerald W, et al. MAD2 haplo-insufficiency causes premature 27. Bunz F, Dutriaux A, Lengauer C, Waldman T, Zhou S, Brown JP, anaphase and chromosome instability in mammalian cells. Nature. et al. Requirement for p53 and p21 to sustain G2 arrest after DNA 2001;409:355–9. damage. Science. 1998;282:1497–501. 47. Williams BR, Prabhu VR, Hunter KE, Glazier CM, Whittaker 28. Waldman T, Kinzler KW, Vogelstein B. p21 is necessary for the CA, Housman DE, et al. Aneuploidy affects proliferation and p53-mediated G1 arrest in human cancer cells. Cancer Res. spontaneous immortalization in mammalian cells. Science. 1995;55:5187–90. 2008;322:703–9. 29. Schmidt M, Lu Y, Liu B, Fang M, Mendelsohn J, Fan Z. Dif- 48. Sheltzer JM, Amon A. The aneuploidy paradox: costs and benefits ferential modulation of paclitaxel-mediated apoptosis by p21Waf1 of an incorrect karyotype. Trends Genet. 2011;27:446–53. and p27Kip1. Oncogene. 2000;19:2423–9. 49. Melino G, De Laurenzi V, Vousden KH. p73: Friend or foe in 30. Stepanova T, Slemmer J, Hoogenraad CC, Lansbergen G, Dort- tumorigenesis. Nat Rev Cancer. 2002;2:605–15. land B, De Zeeuw CI, et al. Visualization of microtubule growth 50. Dong S, Pang JC, Hu J, Zhou LF, Ng HK. Transcriptional inac- in cultured neurons via the use of EB3-GFP (end-binding protein tivation of TP73 expression in oligodendroglial tumors. Int J 3-green fluorescent protein). J Neurosci. 2003;23:2655–64. Cancer. 2002;98:370–5. 31. Law CW, Chen Y, Shi W, Smyth GK. voom: precision weights 51. Abbas T, Dutta A. p21 in cancer: intricate networks and multiple unlock linear model analysis tools for RNA-seq read counts. activities. Nat Rev Cancer. 2009;9:400–14. Genome Biol. 2014;15:R29. 52. Ogino S, Kawasaki T, Kirkner GJ, Ogawa A, Dorfman I, Loda M, 32. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. et al. Down-regulation of p21 (CDKN1A/CIP1) is inversely limma powers differential expression analyses for RNA- associated with microsatellite instability and CpG island methy- sequencing and microarray studies. Nucleic Acids Res. 2015;43: lator phenotype (CIMP) in colorectal cancer. J Pathol. e47. 2006;210:147–54. 33. Burrell RA, McClelland SE, Endesfelder D, Groth P, Weller MC, 53. Garcia-Gutierrez L, Delgado MD, Leon J. MYC oncogene con- Shaikh N, et al. Replication stress links structural and numerical tributions to release of cell cycle brakes. Genes. 2019;10:244. cancer chromosomal instability. Nature. 2013;494:492–6. 54. Maia ARR, Garcia Z, Kabeche L, Barisic M, Maffini S, Macedo- 34. Harms K, Nozell S, Chen X. The common and distinct target Ribeiro S, et al. Cdk1 and Plk1 mediate a CLASP2 phospho- genes of the p53 family transcription factors. Cell Mol Life Sci. switch that stabilizes kinetochore-microtubule attachments. J Cell 2004;61:822–42. Biol. 2012;199:285–301. The p53/p73 - p21CIP1 tumor suppressor axis guards against chromosomal. . . 451

55. Li J, Wang Y, Wang X, Yang Q. CDK1 and CDC20 over- 58. Sur S, Agrawal DK. Phosphatases and kinases regulating CDC25 expression in patients with colorectal cancer are associated with activity in the cell cycle: clinical implications of CDC25 over- poor prognosis: evidence from integrated bioinformatics analysis. expression and potential treatment strategies. Mol Cell Biochem. World J Surg Oncol. 2020;18:50. 2016;416:33–46. 56. Carter SL, Eklund AC, Kohane IS, Harris LN, Szallasi Z. A 59. Zhang C, Spektor A, Cornils H, Francis JM, Jackson EK, Liu S, signature of chromosomal instability inferred from gene expres- et al. Chromothripsis from DNA damage in micronuclei. Nature. sion proﬁles predicts clinical outcome in multiple human cancers. 2015;522:179–84. Nat Genet. 2006;38:1043–8. 60. Umbreit NT, Zhang C, Lynch LD, Blaine LJ, Cheng AM, Tourdot 57. Crncec A, Hochegger H. Triggering mitosis. FEBS Lett. 2019; R, et al. Mechanisms generating cancer genome complexity from 593:2868–88. a single cell division error. Science. 2020;368:eaba0712.