Sustained ERK Signaling Promotes G2 Cell Cycle Exit and primes cells for Whole Genome Duplication

Summary

Whole genome duplication (WGD) is a frequent event in cancer evolution that fuels chromosomal instability. WGD can result from mitotic errors or endoreduplication, yet the molecular mechanisms that drive WGD remain unclear. Here we use live single cell analysis to characterize cell cycle dynamics upon aberrant Ras-ERK signaling. We find that sustained ERK signaling in human cells leads to reactivation of the APC/C in G2 resulting in tetraploid G0-like cells that are primed for WGD. This process is independent of DNA damage or p53 but dependent on p21. Transcriptomics analysis and live cell imaging showed that constitutive ERK activity promotes p21 expression, which is necessary and sufficient to inhibit CDK activity and prematurely activate the Anaphase Promoting Complex (APC/C). Finally, either loss of p53 or reduced ERK signaling allowed endoreduplication completing a WGD event. Thus, sustained ERK signaling-induced G2 cell cycle exit represents an alternative path to whole genome duplication.

Graphical Abstract

eTOC Blurb

Guerrero and Aikin et al. use live cell imaging to characterize cell cycle dynamics upon aberrant Ras-ERK signaling to find that sustained ERK activity leads to G2 cell cycle exit. The resulting tetraploid G0-like cells are poised to refire the replication origins to complete a whole genome doubling event.

Introduction

Mutations in the Extracellular-signal Regulated Kinase (ERK) signaling pathway drive 46% of human cancers^1,2. However, therapeutic strategies targeting the ERK pathway are dominated by adaptation, drug-resistance, and relapse^3,4, highlighting an incomplete understanding of the influence of oncogenic mutations on malignant cell behaviors. Live single-cell analysis with signaling biosensors has revealed that the temporal patterns of ERK activity (i.e. signaling dynamics) can be altered by oncogenic mutations and play an important role in modulating cell behavior^5–7. Specifically, previous work showed that while ERK activity pulses increase proliferation, sustained ERK activation leads to increased migration, loss of cellcell contact, and cell cycle arrest⁵. Yet, the mechanisms by which cell behavior is differentially regulated by pulsatile or sustained ERK activity remain unknown.

Over the last decade, the use of live cell biosensors has enabled a deeper quantitative understanding of the molecular dynamics that govern the G1/S transition^8–14. While G2 dynamics have received less attention, several alternative cell cycle outcomes can occur including abortive mitoses (i.e. mitotic slippage and cytokinesis failure)^15,16 and G2 cell cycle exit (also known as mitotic bypass)^17–21. These alternative cell cycles can lead to whole genome duplication (WGD) in both normal and diseased cells. Recent work in zebrafish melanocytes revealed that oncogene-induced cytokinesis failure preceded WGD²². Other studies have shown that mitogenic signaling is integrated throughout the entire cell cycle and not only G1 to determine daughter cell fate^17,23,24. However, whether oncogenes can elicit WGD through altering G2 dynamics remains unclear.

Approximately one-third of human tumors have a near-tetraploid DNA content and are assumed to have undergone a WGD event in their evolutionary history^16,25–27. While the timing of WGD during oncogenesis is not known, inference from tumor sequencing suggests that both loss of p53 and acquisition of oncogenic drivers precedes WGD in most cases^25,28. WGD, in turn, leads to chromosomal instability, a hallmark of cancer that drives tumor evolution²⁷. Emerging evidence supports the association of WGD with metastasis, immune suppression, drug resistance, and decreased overall survival^{25,26,29–31}. These findings show the importance of WGD in cancer evolution, yet the precise timing and drivers of WGD during oncogenesis are unknown. Here, we use inducible oncogenes and live cell biosensors for ERK signaling, APC/C, Cyclin B, and CDK activities to study the effects of sustained ERK signaling dynamics on the cell cycle. We find that G2 cells with oncogenic ERK signaling can persistently inhibit CDKs and prematurely reactivate the APC/C without mitotic entry. This G2 cell cycle exit event results in tetraploid cells that resemble a G0/G1 biochemical state and are thus poised for WGD via endoreplication of the genome. Interestingly, we find that G2 cell cycle exit depends on p21 upregulation in a p53-independent manner. Altogether, our work shows that sustained ERK signaling leads to G2 cell cycle exit and primes cells for WGD.

Results

Our previous work showed that the temporal patterns of ERK activity elicited by common oncogenes can differentially regulate cell behavior⁵. Specifically, sustained ERK activity, associated with KRAS^G12V or BRAF^V600E expression, leads to a robust cell cycle arrest, while pulsatile ERK activity from overexpressed wild-type EGFR, KRAS, or BRAF leads to increased proliferation. However, the mechanisms and consequences of these dynamics-dependent cell fates remain unclear. To address this question, we first asked whether oncogene induced cell cycle arrest was dependent on serum levels or confluency. EdU incorporation in breast mammary epithelial cells (MCF10A) expressing doxycycline-inducible BRAF^V600E indicated that oncogene-induced arrest still occurred in high serum and at low confluency (Figure 1A). BRAF^V600E expression induced a switch from the rapid pulses of ERK activity observed under normal culturing conditions to sustained ERK activation (Figure 1B). DNA content analysis showed the expected decrease of S-phase cells but revealed cells arrested with both 2C and 4C DNA content (Figure 1C).

Figure 1: BRAF^V600E expression promotes APC/C reactivation in G2 independently of p53.

A. Bar plot showing % cells in S-phase. Inducible BRAF^V600E cells (rtTA, TRE3G::BRAF^V600E) were treated with media or doxycycline (2μg/ml) in full serum and low density for 48hrs before EdU incorporation to measure the fraction of S-phase cells. Bar plot of mean and standard deviation from n=20 observations. Statistics calculated by two-sided t-test, p=1.7e-27 B. Traces of ERK activity (black) from live-imaging. MCF10A cells expressing reporters for ERK localization and kinase activity (ERK1-mRuby2 and ERK-KTR-mCer3) and inducible BRAF^V600E were treated with media or doxycycline (2μg/ml) under full serum, low density conditions. Images were acquired every 5 minutes over 9 hours and analyzed as described in methods. 5 representative single-cell traces (top), population averages and 25^th-75^th percentiles (shaded) for n > 125 cells (bottom) are shown. C. Probability distribution of DNA content from inducible BRAF^V600E cells treated with 48 hours media or doxycycline (2μg/ml) as measured by DAPI staining. Data represents n > 2000 cells. D. Schematic representing DNA content and regulation of the cell cycle by the Anaphase Promoting Complex (APC/C). APC/C activity is on in G1, inactivated at the commitment point (“C”), and reactivated during mitosis (“M”). APC/C targets, such as geminin, accumulate through DNA synthesis (“S”) and G2, but are degraded with APC/C reactivation. E. Comparison of DNA content and Geminin from inducible BRAF^V600E cells. Cells were treated with media or doxycycline for 48 hours before fixation, geminin immunofluorescence, Hoechst staining, and imaging as described in methods. Representative scatters show 1000 randomly selected cells plotted from n > 2300 cells. Dotted lines separate threshold for Geminin and DNA content and cells in a 4C APC/C active state are highlighted in red. F. Comparison of DNA content and APC/C activity in WT and p53^KO inducible BRAF^V600E cells. Cells were treated with media or doxycycline for 48 hours before fixation, geminin immunofluorescence, Hoechst staining, and imaging as described in methods. Representative scatters show 1000 randomly selected cells plotted from n > 2600 cells. Dotted lines separate threshold for Geminin and DNA content and cells in a 4C APC/C active state are highlighted in red. G. Representative images of immunofluorescence staining for phospho-Histone2A.X (pH2A.X). Inducible BRAF^V600E and KRAS^G12V cells were treated with media (Med.), doxycycline (Dox. 2μg/ml), Nutlin-3A (N3A, 10μM), Etoposide (Etopo. 10μM) for 24 hours, fixed and stained for pH2A.X. Scale bar, 50 μM. H. Quantification of pH2A.X immunostaining from A. Violin plots represent distribution, mean, and 25^th-75^th percentiles for n > 2300 cells analyzed as in methods. p-value was calculated by 2 sample t-test using the mean values for three independent experiments. I. Representative images of immunofluorescence staining for p53. Inducible BRAF^V600E and KRAS^G12V cells were treated with media (Med.), doxycycline (Dox. 2μg/ml), Nutlin-3A (N3A, 10μM), Etoposide (Etopo. 10μM) for 24 hours, fixed and stained for p53. Scale bar, 50 μM . J. Quantification of p53 immunostaining from C. Violin plots represent distribution, mean, and 25^th-75^th percentiles for n > 2600 cells analyzed as in methods. p-value was calculated by 2 sample t-test using the mean values for three independent experiments.

Previous studies showed that DNA damage and HRAS^G12V expression leads to premature APC/C activation in G2^15,16,18. In agreement, geminin (APC/C substrate) and DNA content analysis showed a large population of 4C geminin-negative cells following BRAF^V600E induction that is absent in normally cycling cells (Figures 1D and 1E). Oncogene induced APC/C reactivation in G2 occurred also in other cell types including lung adenocarcinoma cells (HCC2935) and small cell lung cancer cells (HCC2286) but not in RPE cells suggesting that ERK driven reactivation of APC/C in G2 is context dependent (Figures S1A–S1H). Our synthetic system involves a non-physiological overexpression of oncogenes, however the levels of phosphorylated ERK achieved by this system were comparable to those observed in other cancer cell lines such as melanoma cells C32 (Figure S1I). Previous studies have shown that p53 is necessary and sufficient for premature APC/C reactivation upon DNA damage¹⁸ yet p53^KO cells still showed a large fraction of 4C geminin negative cells upon oncogene induction (Figure 1F). Furthermore, BRAF^V600E or KRAS^G12V overexpression did not show p53 stabilization or DNA damage as indicated by pH2A.X immunofluorescence (Figures 1G–1J). Taken together, these data suggest that sustained ERK signaling induced by oncogene overexpression can lead to G2 cell cycle exit independently of DNA damage or p53.

To interrogate the mechanism of oncogene induced G2 cell cycle exit we aimed to directly observe the effects of oncogene induction on cell cycle dynamics. Thus, we transduced MCF10A cells containing our oncogene expression system with the Anaphase Promoting Complex (APC/C) sensor mCherry-Geminin^{(1−110) 19} (mCh-Gem) (Figures 2A and 2B) and performed live imaging during induction. As expected, Geminin is rapidly degraded during normal mitosis and is low in G0/G1 daughter cells (Figures 2B and 2C). Interestingly, induction of BRAF^V600E or KRAS^G12V expression produced a subpopulation of Geminin positive cells that degraded the reporter without undergoing mitosis (Figures 2D, 2E, S2A and S2B; Video S1). Cells that were in G1 at the time of oncogene induction were more likely to prematurely reactivate the APC/C in S/G2 than cells that were in G2 (Figure S2C). This premature Geminin degradation was slower that the one observed during mitosis and occurred in ~30% of geminin positive BRAF^V600E cells (Figure 2F), a much more frequent event than cytokinesis failure (~2%). Taken together our data indicates that hyperactivation of the ERK pathway leads to premature APC/C reactivation in different cellular contexts.

Figure 2: BRAF^V600E expression promotes G2 cell cycle exit.

A. Schematic of the APC/C reporter cell line, containing a nuclear marker (H2B-iRFP) and a reporter of APC/C activity (mCherry-Geminin¹⁻¹¹⁰), which accumulates when APC/C is inactive. B. Representative images of APC/C reporter cells during mitosis. A geminin-positive cell enters metaphase, and mCherry-Geminin signal is rapidly degraded. Cell of interest marked by closed arrow and daughter cell by open arrow. Scale bar, 10 μM. C. Quantification of mCherry-Geminin nuclear fluorescence from live imaging. Cells were imaged every 12 minutes for 48 hours and analyzed as in methods. 9 single cell traces (grey) of normalized nuclear intensity were aligned in silico to the time of metaphase (t=0) and their average (red) are shown. A grey box obscures the time of nuclear envelope breakdown during mitosis, when nuclear mCherry-Geminin signal is unreliable. D. Representative images of a BRAF^V600E-expressing cell reactivating APC/C without mitosis. Images presented as in B. Scale bar, 10 μM. E. Quantification of mCherry-Geminin nuclear fluorescence from live imaging of BRAF^V600E cells reactivating APC/C without mitosis, as in C. Cells are imaged and analyzed as in methods. 8 single cell traces were aligned to the time of APC/C reactivation and presented as in C. F. Quantification of cell cycle fates. Cells from E-G were tracked from S/G2 until reactivation of the APC/C. The average fraction of cells undergoing mitosis (grey), APC/C reactivation without mitosis (red) or mitotic failure (cyan) from 6 observations of n = 30 cells each are presented in stacked bar plots with standard deviation (error bars). G. Comparison of DNA content, Geminin and cyclin A protein levels in inducible MCF10A BRAF^V600E cells. Cells were treated with media or doxycycline for indicated amounts of time hours before fixation. DNA content was measured by a DAPI stain, Geminin was measured by using the mCherry-Geminin¹⁻¹¹⁰ biosensor and cyclin A by quantitative immunofluorescence as described in methods. Violin plots show cyclin A expression at different cell cycle stages based on their DNA content and Geminin state. H. Comparison of DNA content, Geminin and cyclin B in inducible MCF10A BRAF^V600E cells. Cells were treated and analyzed as in G but using the cyclin B antibody. I. Comparison of DNA content, Geminin and cyclin D in inducible MCF10A BRAF^V600E cells. Cells were treated and analyzed as in G but using the cyclin D antibody. J. Comparison of DNA content, Geminin and phosphor Rb in inducible MCF10A BRAF^V600E cells. Cells were treated and analyzed as in G but using the cyclin D antibody.

Next, we aimed to biochemically characterize these cells that activate APC/C in G2. To address this question we performed quantitative immunofluorescence against different cell cycle indicators in cells containing the geminin reporter upon induction of BRAF^V600E for 24, 48 and 72 hours. Results showed that the levels of cyclin A, cyclin B, cyclin D, and phosphorylated Rb in 4C geminin-negative cells closely resemble those found in 2C geminin-negative cells (G0/1) (Figures 2G–2J). Thus, oncogene induction promotes G2 cell cycle exit. To discriminate between mitotic slippage and G2 cell cycle exit, we expressed a reporter of mitotic entry consisting of full-length Cyclin-B1 and a tandem mNeonGreen (2xmNG-CycB) (Figure S2D). Upon mitotic entry, the reporter translocates to the nucleus (Figure S2D and S2E) before both Cyclin-B and geminin are rapidly degraded. In contrast, BRAF^V600E expressing cells showed geminin degradation without nuclear translocation of CycB (Figure S2F–S2H) followed by a slow but complete degradation of the Cyclin B reporter. This cyclin B persistence for a few hours suggests that APC/C-CDH1, which is the complex responsible for geminin degradation in G1, is reactivated during oncogene induced G2 cell cycle exit.

We then wanted to characterize CDK activity dynamics (Figure 3A) during G2 cell cycle exit. Thus, we used reporters for CDK4/6 and CDK2 activities (Figure 3B)^12,32. In a normal cell cycle, CDK4/6 and CDK2 activities remained high until mitosis (Figures 3C and 3D). Surprisingly, a subset of CDK4/6^HIGH CDK2^HIGH BRAF^V600E expressing cells inactivated CDK4/6 and CDK2 without mitosis (Figures 3E and 3F; Video S2). In most of these cells, the drop in CDK4/6 preceded the drop of CDK2 activities by 2–4 hours. Premature CDK inactivation occurred in 48% of S/G2 cells (Figure 3G), a similar fraction to what was observed for geminin-degradation (Figure 2F). CDK inactivation was also observed in 19% of S/G2 KRAS^G12V cells (Figure S3A–S3C), suggesting that CDK inhibition results from sustained ERK activity. DNA content analysis confirmed an increase in 4C CDK2-inactive cells (Figure 3H). Thus, both premature APC reactivation and CDK inhibition occur during oncogene-induced G2 cell cycle exit.

Figure 3: CDK4/6 and CDK2 are inactivated during oncogene induced G2 cell cycle exit.

A. Schematic showing Cyclin-CDK regulation of cell cycle progression. B. Schematic representation of CDK reporter cell line. MCF10As stably express a nuclear marker (H2B-iRFP), and kinase translocation reporters for G1 and S/G2 CDKs (CDK4/6-KTR-mCh and CDK2-KTR-mVen, respectively). C. Representative images of a CDK reporter cell undergoing mitosis. Cell of interest marked by closed arrow and daughter cell by open arrow. Scale bar, 10 μM. D. Quantification of CDK activities from live imaging of mitosis. Cells from C were imaged every 12 minutes for 48 hours and analyzed as in methods. 12 single cell cytoplasmic/nuclear activity ratios (grey) were aligned in silico to the time of metaphase. Average CDK4/6 activity (magenta, top), CDK2 activity (green, center) or overlayed averages and 25^th-75^th percentiles (bottom) are shown for the 12 hours before and after mitosis. A grey box obscures the time of nuclear envelope breakdown during mitosis, when reporter signal is unreliable. E. Representative images of BRAF^V600E expressing CDK reporter cell inactivating CDKs without undergoing mitosis. Cell of interest indicated by arrow. Scale bar, 10 μM. F. Quantification of CDK activities from BRAF^V600E expressing cells, showing inactivation of CDKs without mitosis. Inducible BRAF^V600E expressing cells were treated with doxycycline (2μg/ml) for 12 hours before live imaging and analysis as in D. 6 single cell traces are aligned in silico to the time of CDK2 inactivation presented as in D. G. Quantification of cell cycle fates after BRAF^V600E induction. Cells from C-F were tracked from S/G2 until inactivation of CDK2. The average fraction of cells undergoing mitosis (grey) or CDK2 inactivation without mitosis (red) from 6 observations of n = 30 cells each are presented in stacked bar plots with standard deviation (error bars). H. Comparison of DNA content and CDK2 activity. Inducible BRAF^V600E cells were treated with dox for 60 hours before fixation and staining with DAPI. Representative scatters show 1000 randomly selected cells plotted from n > 3150 cells. Dotted lines separate threshold for CDK activity and DNA content and cells in a 4C CDK2 inactive state are highlighted in red.

To dissect the mechanism of oncogene induced G2 cell cycle exit we performed transcriptomics analysis in MCF10A cells expressing BRAF^V600E for 4, 12, 24, and 96 hours. Interestingly, while most cell cycle related genes where repressed, CDKN1A was the only upregulated cell cycle inhibitor that consistently appeared in all timepoints (Figures 4A, S4A and S4B; Table S1). This data agrees with previous studies indicating that ERK-induced p21 upregulation occurs independently of p53^33–36. We therefore wondered whether oncogene-induced G2 cell cycle exit could be p21-dependent. To address this question, we used an endogenously tagged p21-mCitrine MCF10A cell line expressing the CDK2 reporter³⁷ (p21-mCit, hDHB-mCh) (Figure 4B) and derived a p53^KO subclone using CRISPR. In these p21 CDK2 reporter cells, p21 levels increased significantly in both WT and p53^KO cells upon BRAF^V600E expression (Figures 4B and 4C). Immunofluorescence against p21 in WT and p53^KO cells showed similar results (Figure S4C). In contrast to p21, other CDK inhibitor proteins of the same family, p27 and p57, were not upregulated following oncogene induction (Figure S4D). Addition of the TGFβ inhibitor A 83–01 did not show any effect on oncogene induced premature APC/C reactivation (Figure S4E). Collectively, these results support a mechanism by which oncogenic ERK signaling causes a p53-independent increase in p21.

Figure 4: p21 upregulation is necessary and sufficient for BRAF^V600E-induced G2 cell cycle exit.

A. Volcano plots of RNA sequencing counts comparing indicated doxycycline timepoints with no dox controls. Blue dots are cell cycle genes from the KEGG list indicated in Supp. Table 1. Red dots are upregulated genes (> that 2 fold induction and p value < 0.001) within the cell cycle genes in that list. Gene names correspond to the red dots. B. Schematic of CDK2, p21 reporter cell line. MCF10As express a nuclear marker (H2B-mTurq), a CDK2-KTR (CDK2-KTR-mCh), and endogenously tagged p21 (mCit-p21). C. Quantification of p21 expression following BRAF^V600E induction in WT and p53^KO CDK2, p21 reporter cells. Cells were imaged every 5 minutes for 18 hours following treatment with media (Med.), doxycycline (Dox. 2μg/ml), or Nutlin-3A (N3A, 10μM) at time 0 and analyzed as in methods. Means and 25^th-75^th percentiles shown for n > 400 cells. D. Representative images of p21 expression increase in BRAF^V600E-expressing CDK2, p21 reporter cell undergoing mitotic bypass. Cells were imaged every 10 minutes after treatment with doxycycline (Dox. 2μg/ml). Doxycycline addition occurred 13.5 hours before the observed bypass event, marked as t = 0. The same cell was continuously monitored for > 48 hours following bypass, without reactivation of CDK2 or a drop in mCit-p21 intensity. Scale bar, 10 μM. E. Single-cell analysis of p21 levels preceding mitosis or mitotic bypass from live imaging. Inducible BRAF^V600E CDK2, p21 reporter cells were pretreated with media or doxycycline for 8 hours and imaged every 10 minutes for another 48 hours. Single cell traces of mitosis or oncogenic mitotic bypass events were aligned in silico to the time of CDK2 inactivation and p21 levels were assessed for the 6 hours preceding either event. Boxplots show means and 25^th-75^th percentiles, along with standard deviations and outliers for n > 27 cells. Significance assessed by two-sided t-test, *** p=2.1e-8. F. Quantification of cell cycle fates from WT, p53^KO, p21^KO, and p53^KO/p21^KO inducible BRAF^V600E cells. CDK reporter inducible BRAF^V600E cells were treated with media or doxycycline for 12 hours before imaging every 12 minutes for 48 hours and analyzed as in methods. Cells in S/G2 were tracked from S/G2 until inactivation of CDK2. The average fraction of cells undergoing mitosis (grey), CDK2 inactivation without mitosis (red), or mitotic failure (cyan) from 6 observations of n = 30 cells each are presented in stacked bar plots with standard deviation (error bars). WT data is reproduced from Figure 3G. G. Comparison of DNA content and Geminin from WT, p53^KO, p21^KO, and p53^KO/p21^KO inducible BRAF^V600E cells. Cells were treated with media or doxycycline for 48 hours before fixation, geminin immunofluorescence, Hoechst staining, and imaging as described in methods. Fraction of polyploid cells was assessed from single-cell analysis of DNA content and Geminin as in methods. Bar plot gives mean and standard deviation (error bars) for four replicates of n > 1600 cells each, with statistics calculated by two-sided t-test, NS not significant, *** p<0.001. H. Quantification of CDK activities from inducible p21 cells, showing inactivation of CDKs without mitosis. Inducible p21 cells were treated with doxycycline (2μg/ml) for 12 hours before live imaging and analysis as in Figure 3D, F. 165 single cell traces are aligned in silico to the time of CDK2 inactivation with population averages and 25^th-75^th percentiles presented as in Figure 3D & F. I. Quantification of cell cycle fates after p21 induction. Cells from G were tracked from S/G2 until inactivation of CDK2. The average fraction of cells undergoing mitosis (grey) or CDK2 inactivation without mitosis (red) from 5 observations of n = 10–30 cells each are presented in stacked bar plots with standard error (error bars).

We hypothesized that oncogene-induced p21 increase may be responsible for the inhibition of CDKs that occurs during G2 cell cycle exit. In fact, after oncogene expression, p21 expression at the single cell level was predictive of whether cells would undergo normal mitosis or G2 cell cycle exit (Figure 4D and 4E; Video S3). To assess whether p21 was required for CDK inhibition, we generated p21^KO hDHB-Venus CDK4 KTR mCherry cells. Interestingly, p21 deletion almost completely abolished premature CDK inhibition in response to BRAF^V600E or KRAS^G12V expression regardless of p53 status (Figures 4F and S4F). This effect was still observed in p21^KO, p53^KO double knock-out cells, demonstrating a dominant role of p21 in CDK inhibition. Moreover, in p21-deficient cells, the 4C geminin-negative cells observed upon oncogene expression was completely abolished (Figure 4G). These results indicate that oncogene-mediated G2 cell cycle exit depends on p21, but not p53. To test whether p21 induction was sufficient to induce G2 cell cycle exit we established an inducible p21 cell line (PGK::rtTA, TRE3G::p21) with reporters for CDK4/6 and CDK2 activity. Following p21 overexpression, over 90% of S/G2 cells inactivated CDKs without mitosis in a pattern consistent with oncogene induction (Figure 4H–I). Taken together, this data is consistent with a model in which p21 upregulation by sustained ERK signaling inhibits CDK activities and stalls E2F transcription to the point where APC/C activity can no longer be inhibited. In agreement with this model transcriptomics and protein level analysis of the E2F target gene and APC/C inhibitor Emi1 showed a rapid transcriptional repression followed protein decay at around 24–48 hours post oncogene expression (Figure S4G and S4H) which is when most APC/C reactivation is observed (Figure 2).

While p53 is dispensable for G2 cell cycle exit, we observed that p53^KO cells frequently reactivated CDK2 (Figure 5A). Thus, we hypothesized that p53 could prevent cells from reentering the cell cycle after G2 cell cycle exit. EdU incorporation and Ki67 staining in p53^KO cells following BRAF^V600E induction showed that a constant fraction of cycling cells is maintained over extended periods (6 days) (Figures 5B and 5C). This data suggested that p53^KO cells stochastically exit quiescence even in the presence of sustained ERK activity. In fact, p53-competent cells had a marked increase in p21 levels immediately following CDK2 inhibition whereas in p53^KO cells p21 levels remained constant following CDK2 inactivation (Figure 5D). Accordingly, p53^KO cells were more likely to re-activate CDK2 and reenter S-phase after G2 cell cycle exit than WT cells (Figures 5E and 5F). However, neither WT nor p53^KO cells showed a stably cycling tetraploid population (4C-8C) suggesting that other mechanisms limit the growth of tetraploid cells independently of p53^26,28,38. In fact, multipolar divisions and other types of mitotic errors were observed in oncogene induced p53^KO cells (Video S4).

Figure 5: p53 prevents whole genome duplication after oncogene induced G2 Cell Cycle Exit.

A. Representative images of p21 expression increase in p53^KO, BRAF^V600E-expressing CDK2, p21 reporter cell undergoing mitotic bypass. Cells were imaged every 10 minutes after treatment with doxycycline (Dox. 2μg/ml). Doxycycline addition occurred 12.5 hours before the observed bypass event. The same cell was continuously monitored an additional 21 hours until CDK2 reactivation. Scale bar, 10 μM. B. Time course analysis of S-phase fractions from WT and p53^KO inducible BRAF^V600E cells. Cells were treated with media (black) or doxycycline (Dox., red, 2μg/ml) and grown for 2, 4, or 6 days before EdU incorporation as in methods. Mean and standard deviation are shown for n = 25 positions. C. Time course analysis of quiescence from WT and p53^KO inducible BRAF^V600E cells. Cells were treated as in B before fixation and immunofluorescence for Ki67. Data presented as in B. D. Comparison of CDK2 activity and p21 levels during BRAF^V600E-induced mitotic bypass in WT (black) and p53^KO cells (red) from live imaging. Inducible BRAF^V600E cells were treated with doxycycline for 8 hours before live imaging every 10 minutes for an additional 48 hours. n > 85 mitotic bypass cells were aligned in silico to the time of CDK2 inactivation. Average and 25^th-75^th percentile CDK2 activities (left) and p21 levels (right) are shown for the 6 hours before and after mitotic bypass. Statistics were calculated by comparing single time point p21 levels of WT and p53^KO cells by two-sided t-test (* p<0.05, ** p<0.01, *** p<0.001). E. Quantification of the rate of CDK2 reactivation following mitotic bypass in WT or p53^KO BRAF^V600E expressing cells. Cells from D were tracked from S/G2 until inactivation of CDK2, and then followed for at least 12 additional hours to observe the rate of CDK2 reactivation. Bar plot shows mean and standard error (error bars) for 6 observations of n > 10 cells each over 2 replicates. Statistics calculated by two-sided t-test, ** p=0.0045. F. Edu incorporation and DNA content analysis by flow cytometry. WT or p53KO MCF10A cells containing TRE3G::BRAF V600E were cultured with media or doxycycline for 44 hours plus 4 hours of Edu at 10μM before fixation. MEK inhibitor (PD0325901, 1.4 μM) was added 24hr after doxycycline. EdU and PI DNA staining was performed as described in methods. Data represents three independent experiments. G. Visual summary of model. Oncogenic ERK signaling causes p53-independent p21 induction, leading to arrest of cells in G0/G1 and mitotic bypass of cells in G2. Then, p21 further increases in a p53-dependent manner to ensure sustained arrest of bypassed, 4C cells. Loss of p53 or relief of ERK signaling can cause arrested 4C cells to reenter the cell cycle and become further polyploid.

Our experimental system relies on overproduction of oncogenes and continuous ERK signaling, but in other cell types and in physiological scenarios, ERK dynamics are varied and the strength of ERK signaling can be different over time. We hypothesized that since sustained ERK signaling is maintained after G2 cell cycle exit, S-phase re-entry (endocycling) may be limited. Thus we hypothesized that moderate suppression of ERK signaling could cause arrested cells to reenter the cell cycle and become polyploid. To test this model, we treated BRAF^V600E expressing cells with a low dose of the MEK or ERK inhibitor and analyzed DNA content by DAPI. Results showed a moderate increase in polyploid cells (Figures 5E, 5F and S5A–S5C). Cell fate analysis of 4C vs. 8C cells showed an increased rate of mitotic failure in 8C cells, especially in p53^KO cells treated with low dose MEK inhibitor (Figure S5D). Thus, a reduction in oncogenic ERK signal strength can cause arrested cells to endocycle after G2 cell cycle exit (Figure 5G and S5E).

Discussion

In this work, we show that sustained ERK signaling in cycling cells can violate the commitment point¹³ by causing p21 stabilization and premature inactivation of CDKs. In G1 cells this CDK inhibition will result in quiescence which is consistent with the role of sustained ERK signaling in cell differentiation^39,40. However, we show that in G2 cells sustained ERK activity can lead to premature reactivation of APC/C in G2 without mitosis, an event that primes cells for whole genome doubling (WGD). Notably, both p21 upregulation and G2 cell cycle exit occur independently of p53 which is often mutated in tumors that have undergone WGD events^15,28. In fact, previous work has shown that the first round of DNA synthesis after WGD leads to widespread DNA damage³⁸ which activates p53 and drives senescence. Previous studies have shown that the transcription factor SP1 is necessary to induce p21 in response to oncogene expression³⁴. However, how the temporal patterns of ERK activity are decoded by transcription factors to define promoter activity will require further investigation.

Our data suggests that sustained ERK signaling triggered by different oncogenes leads to G2 cell cycle exit and primes cells for WGD. Our observations that both KRAS or BRAF mutations lead to G2 cell cycle exit confirms previous findings using transient transfection of oncogenic HRAS¹⁸ and highlights a general role for ERK pathway oncogenes, and specifically, sustained ERK signaling, in promoting G2 cell cycle exit. This data provides a potential route to WGD in addition to replicative stress and cytokinesis failure^21,22. While we did not observe extensive cytokinesis failure or binucleation in our study, differences in p21, p27, or p57 expression in different cell types could result in different degrees of CDK inhibition changing the path to WGD from endocycling to endomitosis or cytokinesis failure. The CDK threshold model posits that different levels of CDK activity may be required for mitotic entry, chromosome segregation, and mitotic exit⁴¹. Thus, complete CDK inhibition in G2 would cause cell cycle exit, while partial CDK inhibition could instead cause mitotic slippage or cytokinesis failure. A similar principle may underly differences in developmental tetraploidy, as giant trophoblast cells rely on p57 for endocycling^42,43, megakaryocytes use p21 to cause mitotic slippage^44,45 and cardiomyocytes and hepatocytes become binucleated through cytokinesis failure¹⁶. Future studies will address whether the CDK threshold model for alternate G2 fates can explain the different mechanisms of WGD observed in different cancer types or mutational burdens.

Our study also points to paradoxical roles for p21 in tumorigenesis. p21 is involved in the DNA damage response along with p53 and is generally considered a tumor suppressor^46,47. However, compared to p53 and other tumor suppressors, p21 mutations are relatively rare in tumors^48,49, despite the observation that p21 loss contributes to the occurrence of mammary Ras-driven tumors^50,51. One explanation is the redundancy with the CDK inhibitor family members p27 and p57. Another possibility is that p21 functions as both a tumor suppressor and an oncogene^52–54. Our study supports this notion by showing that while p21 is a cell cycle inhibitor, it also mediates WGD. In vivo models should be used to test whether p21 suppresses or promotes WGD during oncogenesis.

Additionally, it is not known whether G2 cell cycle exit occurs in diseased tissues, nor whether those events will lead to cycling polyploid cells in vivo. In our study, most cells arrest after G2 cell cycle exit. Thus, this event may not represent an immediate advantage for cancer cell fitness. However, the observation that WGD tumors often settle with pseudotriploid karyotype suggests that a tetraploid intermediate is followed by additional mitotic errors¹⁶. Accordingly, our data shows that 8C cells that reach mitosis are likely to undergo mitotic catastrophe. Thus, while there is a low likelihood of cell cycle re-entry after G2 cell cycle exit and a high potential for catastrophic mitosis, a genetic or epigenetic state that facilitates G2 cell cycle exit could, over long periods of time, elevate the chances of producing malignant karyotypes. We have shown that either preexisting p53 mutations or relief of oncogenic ERK signaling increase the chances of refiring of the replication origins after G2 cell cycle exit. Additional studies will determine whether therapeutic ERK-pathway inhibition or other forms of transient relief of ERK signaling can lead to reentry of WGD cells, and how this event affects cancer evolution.

Limitations of the study

In this study we artificially overexpress mutant BRAF and/or KRAS proteins to induce sustained ERK signaling. While our results show that cell cycle arrest is mediated by ERK signaling and that similar levels of active ERK are observed in cancer cell lines, the degree of ERK induced G2 cell cycle exit and WGD that occurs during tumor evolution will require further investigation. Similarly the genetic or environmental conditions that promote refiring of the origins of replication after G2 cell cycle exit remain to be explored. Finally we note that G2 cell cycle exit upon sustained ERK signaling is context dependent since RPE cells did not show this phenotype.

*STAR*METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Sergi Regot (sregot@jhmi.edu)

Materials Availability

Raw data, cell lines, and materials are available from the lead contact upon reasonable request.

Data and Code Availability

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell Lines and Reagents

MCF10A human mammary epithelial cells (ATCC) were grown at 37° and 5% CO2 in DMEM/F12 (Gibco) with 5% horse serum (HS) (Sigma), 10 μg/ml Insulin (Sigma), 20 ng/ml EGF (Peprotech), 1x Penicillin-Streptomycin (P/S) (Gibco), 0.5 mg/ml Hydrocortisone (Sigma), 100 ng/ml Cholera Toxin (Sigma). Cells were passaged every 3 days with 0.25% Trypsin-EDTA (Gibco), are mycoplasma free, and were verified by STR-profiling (ATCC).

ERK-reporter cells express H2B-iRFP, ERK1-mRuby2, and ERK-KTR-Cer3 as previously described⁵. APC/C reporter cells express H2B-iRFP and mCherry-Geminin¹⁻¹¹⁰ (mCh-Gem). The mitotic entry reporter is a tandem codon scrambled mNeonGreen-CyclinB (2 x mNG-CycB) from a plasmid provided by Andrew Holland (Johns Hopkins University) which was subcloned into a pENTR backbone by Gibson cloning⁵⁵ and recombined by Gateway cloning into a pLenti-PGK-Hygro destination vector⁵⁶. Mitotic entry reporter cells are APC/C reporter cells lentivirally transduced with 2 x mNG-CycB and selected (hygromycin 10 μg/ml Corning). CDK2/4/6 reporter cells express H2B-iRFP, ERK-KTR-Cer3, and hDHB-mVenus-p2a-mCherry-CDK4/6 KTR¹² (Addgene #126679) introduced via lentivirus followed by selection (blasticidin 3 μg/ml Corning). CDK2 p21 reporter cells are MCF10A expressing H2B-mTurq and hDHB-mCh (hDHB), with an mCitrine-p21 endogenous knock-in³⁷ provided by Sabrina Spencer (University of Colorado, Boulder). Reporter cell lines were lentivirally transduced with doxycycline inducible BRAF^V600E, KRAS^G12V, or p21 (PGK::rtTA, TRE3G::[gene]) and selected (puromycin 1 μg/ml Sigma, pHC125 and pHC136). All lentiviral production, infections, and selections occurred as previously described⁵.

To create CRISPR knock out cell lines, we used the CRISPR V2 system⁵⁷. A CRISPR V2 Puro plasmid (Addgene #52961) targeting A159 of exon 5 from p53 was used to generate lentivirus, and infected cells were selected for 3 passages in Nutlin-3A. p53^KO cells are polyclonal and verified by IF for p53 following treatment with Nutlin. A CRISPR V2 Neo backbone and gRNA oligos targeting L59 of exon 2 from the p21 encoding gene CDKN1A was used to generate lentivirus, and infected cells were selected for 3 passages in Neomycin (500 μg/ml, Sigma). Clonal cell lines were expanded from p21^KO clonal cell lines before validation by IF for p21 after Nutlin in full serum, low density. p53^KO/p21^KO cells were generated by infecting p53^KO cells with the p21^KO construct, selection with Neomycin, clonal isolation, and validation by immunofluorescence.

METHOD DETAILS

Live Imaging

Reporter cells were seeded on fibronectin-coated (EMD Millipore) glass-bottom 96 well plates (CellVis). Before imaging, cells were switched to imaging media composed of 5% horse serum, 1% penicillin-streptomycin, 1% GlutaMax in phenol-red-free DMEM/F12. Plates were imaged using with a 20x air objective on a Nikon Eclipse Ti-E epifluorescence microscope equipped with a multi-LED SpectraX light source (Lumencor), multiband dichroic mirrors (Chroma) and Hamamatsu sCMOS camera (photmetrics). Imaging control performed with Metamorph. OKO Labs control units were used to ensure temperature (37°C), humidity, and CO₂ control throughout imaging.

DNA Content, EdU Incorporation, and Immunofluorescence Assays

For DNA content analysis, nuclei were stained with either DAPI or Hoescht dyes, as indicated. DAPI (Invitrogen) was diluted 1:1000 in PBS, and incubated with cells 10 minutes (room temperature, dark), and washed before imaging. Hoescht 33342 (Invitrogen) dye was diluted 1:1000 in PBS and incubated with cells for 30 minutes immediately before imaging.

S-phase fraction analysis in Figure 1 A was performed by EdU incorporation as previously described. Briefly, following indicated growth and treatments, cells were incubated with EdU (10 μM, Thermo Fischer Scientific) for four hours, fixed with cold methanol, washed, and stained with Alexa-Fluor Azide 488 click reagent (Thermo Fisher Scientific) before DAPI staining as above.

For immunofluorescence, cells were fixed in 4% paraformaldehyde fixation (10 minutes, on ice), blocked in 5% bovine serum albumin, 0.3% Triton-X in PBS for 1–2 hours, and stained overnight at 4°C with primary antibody diluted in blocking buffer. The following day, plates were washed before a 1-hour incubation in secondary antibody diluted in blocking buffer (room temperature, dark). Cells were then washed before DAPI/Hoechst staining or imaging. The following primary and secondary antibody combinations were used in this work:

Immunoblotting

For immunoblotting, cells were grown for 2 days until plates reached ~80% confluency unless otherwise specified. Protein was extracted in Lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% Triton, 0.5% Na-Deoxycholate, 1% SDS, 2 mM EDTA, 1x fresh Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific), and 1 mM DTT). Protein was heated in Laemmli sample buffer at 70°C for 15 min (Bio-Rad), resolved by SDS-PAGE, and transferred to a PVDF membrane (Milipore-Sigma). Blocking and antibody dilutions were done in Odyssey Blocking Buffer (Li-COR) and washes in PBS/Tween-20 (0.1%). Antibodies used were pERK (CST 4370), tERK (CST 4696), EMI1 (SC 365212), p21 (CST 2947), HSC70 (Santa Cruz 7298). Membranes were read with a Li-COR scanner using ImageStudioLite software.

Edu Incorporation and Flow Cytometry

For Edu Incorporation in Figure 5 F, cells were incubated with EdU (10 μM, Thermo Fischer Scientific) for four hours, fixed with cold methanol, washed, and stained with Alexa-Fluor Azide 488 click reagent (Thermo Fisher Scientific) before PI staining. EdU stained cells were then washed in 1x PBS and stained with PI (10μg/mL, Thermo Fischer Scientific) in the presence of RNase A (Qiagen 19101) and incubated for 15 minutes (room temperature, dark), and then washed. The DNA Content was measured on a BD FACSCalibur flow cytometer (BD Biosystems).

RNA sequencing

MCF10A cell containing TRE3G:BRAFV600E were cultured as described in Cell Lines and Reagents. Cell pellets (~2 million cells) from each experiment were flash frozen in liquid nitrogen and stored at −80C until use. Sample information can be found on Supplemental Table 1. RNA extraction was performed in three batches with each batch containing one sample from each timepoint to reduce batch effects. Samples (2 million cell pellets) were removed from −80C and placed on ice to slightly thaw. Before completely thawing, 1 mL of TRI reagent (AM9738, Thermo Fisher Scientific) was added to all samples. Samples were lysed by repeated pipetting and incubated for 5 minutes at room temperature. Chloroform (200 uL) was added to each and samples were shaken vigorously for 15 seconds. Samples were then incubated at room temperature for 3 minutes followed by centrifugation at 12000xg for 15 minutes at 4C. The aqueous layer was retained and RNA was purified using Direct-zol RNA Miniprep kits (R2050, Zymo Research) with DNAse treatment. RNA was quantified with Qubit RNA Broad Range assays (Q10210, Thermo Fisher Scientific). RNA quality was determined by using RNA integrity numbers (RINs) calculated from extracted RNA run on RNA ScreenTape (5067–5576, Agilent) in a TapeStation 4200 (G2991BA, Agilent). Extracted RNA was stored at −80C until use. Sample information can be found on Supplemental Table 1.

RNA-seq sequencing libraries were prepared using the NEBNext Ultra II RNA Library Prep Kit for Illumina (E7770S, NEB) in combination with the NEBNext Poly(A) mRNA Magnetic Isolation Module (E7490S, NEB). PolyA tail selection and library preparation were performed in the same batches as extraction to reduce batch effects. Briefly, approximately 1 ug of total RNA was used as input to polyA tail selection for each sample. After polyA selection, samples were fragmented for 10 minutes at 94C to achieve RNA fragmentation between 200–500 bp, followed by first and second strand cDNA synthesis. DNA end repair and adapter ligation were then performed and followed by size selection according to manufacturer’s recommendations to achieve a final insertion size of ~300 base pairs (bp) and final library size of ~420 bp. Samples were amplified following manufacturer’s recommendations for 1 ug of input using unique combinations of i5 and i7 primers from NEBNext Multiplex Oligos for Illumina Dual Index Primers Set 1 (E7600S, NEB). Combinations of i5 and i7 primers were chosen by using the checkMyIndex tool⁵⁹ with settings for NovaSeq chemistry. Libraries were quantified using Qubit dsDNA Broad Range assays (Q32850, Thermo Fisher Scientific). Library size was determined using a Genomic DNA ScreenTape in a TapeStation 4200 (G2991BA, Agilent). Completed libraries were stored at −20C until sequencing. Libraries were diluted to 4 nM, pooled at equal ratios, and 2×150 bp paired-end sequencing was performed on two lanes of a SP flow cell (300 cycle) on an Illumina NovaSeq 6000 instrument.

FASTQs were aligned to a masked version of the GRCh38 human genome⁶⁰ using HISAT2 (v2.2.1)⁶¹ with the key parameters “--seed 24 --no-mixed --no-discordant --no-unal” included. Samtools (v1.7)⁶² was then used to convert SAM files to sorted BAM files. Gene-level expression counts were generated with featureCounts (v2.0.1)⁶³ using the GENCODE v39 human transcriptome reference with the key parameters “-p -B --primary -t exon -g gene_id” included.

The count matrix produced by featureCounts was then processed in R with custom scripts. First, filtering was performed to retain “expressed” genes. “Expressed” was defined as having a counts per million (cpm) > 0.202 in at least three samples. Counts were then transformed to log2(cpm) using the “cpm” function from the edgeR package (v3.32.1)⁶⁴ with the parameter, “log = TRUE.” Principal component analysis was then performed using the “prcomp” R function with “retx = TRUE” setting included. Genes were then annotated with gene symbols using biomaRt (v2.46.3)⁶⁵ and the Ensembl 105 database⁶⁶. Code for processing RNA-seq data can be found on GitHub (https://github.com/timplab/pwh_projects/tree/master/regot_rna) in a private repository that will be made public upon publication. Raw RNA-seq sequencing data will be deposited in a public database prior to publication. Processed RNA-seq count matrices can be found on GitHub (https://github.com/timplab/pwh_projects/tree/master/regot_rna) in a private repository that will be made public upon publication.

QUANTIFICATION AND STATISTICAL ANALYSIS

Image Analysis and Quantification

For fixed cell analysis, images were flat-fielded, segmented, and analyzed by Cell Profiler as previously described (Aikin et al., 2020 and code available at https://github.com/tjaikin/Regot-Lab). Nuclear shape parameters were used to exclude artifacts (i.e. solidity > 0.95). Integrated intensity measurements from DAPI or Hoechst staining were plotted, and DNA content distributions were normalized to the center of the G1 density before using control (+ media vs N3A/CDKi arrest) conditions to set thresholds for 2C, 4C, and >4C populations. Polyploid fractions in Figure 4G include 4C Gemininnegative and all >4C cells, with mean and standard deviation calculated from four independent replicates.

Time-lapse images were flat-fielded and registered before Cell Profiler segmentation and analysis and custom tracking. See previous works for full description of live-image processing, segmentation, tracking, and analysis^5,12,58. mCit-p21 intensities were dim compared to background, so local background subtraction was necessary to distinguish signal (median nuclear YFP intensity – median cytoring YFP intensity).

Assessment of the rate of APC/C reactivation or CDK2 inactivation was performed by blinding positions, selecting 30 cells from each position in G2 12–24 hours after doxycycline induction of oncogenes (BRAF^V600E or KRAS^G12V), and manually tracking until mitosis, APC/C reactivation, CDK inhibition, or mitotic failure. Following analysis, positions were unblinded and the rate of each event is plotted in the figures. Cyclin B1 degradation times in Supplementary Figure 4 required observation of 2 x mNG-CycB in cells after mitosis or APC/C reactivation. Similarly, the rate of CDK2 reactivation in Figure 5E required observation of CDK2 activity for at least 12 hours after initial CDK2 inactivation.

In silico alignment of mitosis, APC/C reactivation, or CDK inhibition required plotting individual traces of nuclear mCh-Gem intensity or CDK2 sensor ratios along with nuclear size and H2B-iRFP intensity. APC/C reactivation or CDK2 inactivation events that coincided with nuclear area decrease and a spike in H2B intensity usually corresponded to mitosis, whereas APC/C reactivation or CDK2 inactivation without changes in nuclear area or H2B intensity were counted as G2 cell cycle exit events. After identifying examples of each, videos of individual cells were visually inspected. These traces were then in silico aligned for figures.

Supplementary Material

1

Supplementary Video 1: Premature reactivation of APC/C in oncogene expressing cells, related to Figure 1 APC/C reporter BRAF^V600E-expressing cells reactivating APC/C without mitosis. Nulcear marker (red, left, H2B-iRFP) and APC/C activity reporter (mCh-Gem, middle, cyan), and merged images (right) are shown for three example cells which are aligned to mitotic bypass at time 0.

2

Supplementary Video 2: Premature inactivation of CDKs in oncogene expressing cells, related to Figure 3 CDK KTR BRAF^V600E-expressing cells inactivating CDKs without mitosis. CDK4/6 (left, mCherry-CDK4/6 KTR) and CDK2 activities (right, DHB-mVenus) are shown, with CDK inactivation occurring at t=0.

3

Supplementary Video 3: Increase in p21 levels before and after oncogene-induced G2 cell cycle exit, related to Figure 4 BRAF^V600E-expressing CDK2, p21 reporter cells inactivate CDKs after an increase in mCitrine-p21 intensity. Nuclear marker (left), CDK2 activity (CDK2-KTR-mCh, middle), and p21 levels (mCitrine-p21, right) are shown aligned to the time of CDK2 inactivation (t=0). Note p21 increase before CDK2 inactivation, inflection following inactivation, and long duration of high p21 levels and CDCK2 inactivity.

4

Supplementary Video 4: Multipolar division following reactivation of CDK2 after G2 cell cycle exit in p53^KO BRAF^V600E cells, related to Figure 5 P53^KO BRAF^V600E-expressing CDK2, p21 reporter cells reactivate CDK2 after G2 exit and undergo a multipolar division. Nuclear marker (left), CDK2 activity (CDK2-KTR-mCh, middle), and p21 levels (mCitrine-p21, right) are shown aligned to the time of CDK2 inactivation (t=0). Note p21 decreases after CDK2 inactivation, CDK2 reactivates (t=1160), and the cell undergoes a multipolar division (t=2230).

Key Resource Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-Geminin	Cell Signaling Technology	Cat# 52508
Rabbi anti-pH2A.X	Cell Signaling Technology	Cat# 9718
Rabbit anti-p53	Cell Signaling Technology	Cat# 2527
Rabbit anti-Cyclin A2	Cell Signaling Technology	Cat# 67955
Rabbit anti-Cyclin B1	Cell Signaling Technology	Cat# 12231
Rabbit anti-Cyclin D1	Cell Signaling Technology	Cat# 55506
Rabbit anti-phospho Rb	Cell Signaling Technology	Cat# 8516
Rabbit anti-Ki67	Cell Signaling Technology	Cat# 9129
Rabbit anti-phospho ERK 1/2	Cell Signaling Technology	Cat# 4370
Mouse anti-ERK 1/2	Cell Signaling Technology	Cat# 4696
Mouse anti-HSC70	Santa Cruz	Cat# sc-7298
Mouse anti-p21Waf1	EMD Millipore	Cat# OP64
Rabbit anti-p27 Kip1	Cell Signaling Technology	Cat# 3686
Rabbit anti-p57 Kip2	Cell Signaling Technology	Cat# 2557
Mouse anti-EMI1	Santa Cruz	Cat# sc-365212
Rabbit anti-p21Waf1/Cip1	Cell Signaling Technology	Cat# 2947
Donkey anti-Rabbit Cy3	Jackson Labs	Cat# 711-165-152
Goat anti-Rabbit AlexaFluor488	Invitrogen	Cat# A-11034
Donkey anti-mouse AlexaFluor488	Invitrogen	Cat# A-21202
IRDye 680RD_Goat anti-rabbit_lgG (H+L)	LI-COR	Cat# 926-68071
IRDye 800CW_Donkey anti-mouse_lgG (H+L)	LI-COR	Cat# 926-32212
IRDye 680RD_Goat anti-mouse_lgG (H+L)	Li-COR	Cat# 926-68070
Chemicals, peptides, and recombinant proteins
Doxycycline	Sigma Aldrich	Cat# D9891
Nutlin-3A (N3A)	Sigma Aldrich	Cat# N2513
Etoposide	Sigma Aldrich	Cat# E1383
MEKi (PD0325901)	SelleckChem	Cat#S1490
CDK1i (RO3306)	SelleckChem	Cat#S7747
TGFB (A 83-01)	Tocris	Cat#2939
EdU	Thermo FIsher	Cat# A10044
CuSO4	Biorbyt	Cat# orb532418
Critical commercial assays
Western Blotting Kit VII RD	LI-COR	Cat# 926-35014
Deposited data
RNA-seq	This paper	BioProject ID: PRJNA1082333
Experimental models: Cell lines
MCF10A	ATCC	Cat# CRL-10317
RPE-hTERT	ATCC	Cat# CRL-4000
NCI-HCC2286	ATCC	Cat# CRL-5938
HCC-2935	ATCC	Cat# CRL-2869
C32	ATCC	Cat# CRL-1585
MCF10A (H2B-iRFP, ERK1-mRuby2, ERK-KTR-Cer3, and TRE3G∷BRAFV600E or TRE3G∷KRASG12V)	This paper	Available upon request
MCF10A (H2B-iRFP, mCherry-Geminin and TRE3G∷BRAFV600E)	This paper	Available upon request
MCF10A (H2B-iRFP, ERK-KTR-Cer3, hDHB -mVenus-p2a-mCherry-CDK4/6 KTR and TRE3G∷BRAFV600E)	This paper	Available upon request
MCF10A (H2B-mTurq and hDHB-mCh (hDHB), mCitrine-p21 endogenous knock-in and TRE3G∷BRAFV600E)	This paper	Available upon request
MCF10A (p21KO hDHB-Venus, CDK4 KTR mCherry and TRE3G∷BRAFV600E)	This paper	Available upon request
MCF10A (p53KO hDHB-Venus, CDK4 KTR mCherry and TRE3G∷BRAFV600E)	This paper	Available upon request
NCI-HCC2286 (TRE3G∷BRAFV600E)	This paper	Available upon request
HCC-2935 (TRE3G∷BRAFV600E)	This paper	Available upon request
RPE-hTERT (TRE3G∷BRAFV600E)	This paper	Available upon request
Recombinant DNA
pLenti PGK-ERK1-mRuby 2 (pSR1214)	Sergi Regot Lab (Johns Hopkins)	Available upon request
pLenti PGK-ERKKTR-mCerulean3	Sergi Regot Lab (Johns Hopkins)	Addgene Cat# 90229
pLenti PGK∷rtTA, TRE3G∷BRAFV600E (pHC125)	Sergi Regot Lab (Johns Hopkins)	Available upon request
pLenti PGK∷rtTA, TRE3G∷KRASG12V (pHC142)	Sergi Regot Lab (Johns Hopkins)	Available upon request
pLenti-DHB-mVenus-p2a-mCherry-CDK4KTR	Addgene	Addgene Cat# 126679
pLenti H2B-iRFP	Markus Covert Lab (Stanford); Sergi Regot Lab (Johns Hopkins)	Addgene Cat# 90237
Software and algorithms
Fixed Image Analysis	Regot Lab	Zenodo DOI: 10829903
Code for processing RNA-seq	Sergi Regot Lab (Johns Hopkins), Winston Timp Lab (Johns Hopkins)	Zenodo DOI: 10835498

Highlights.

• Sustained ERK signaling leads to reactivation of the APC/C in G2.

• CDK inhibition precedes oncogene induced APC/C reactivation.

• ERK induced G2 cell cycle exit is p21 dependent but p53 independent.

• Loss of p53 or reduced ERK activity after G2 cell cycle exit led to endocycling.