E2F activity determines mitosis versus whole-genome duplication in G2-arrested cells

Abstract

While mitogenic signaling is known to regulate cell-cycle entry during the G1 phase, its function in the G2 phase remains elusive. Here we show that mitogenic signaling controls whether G2-arrested cells proceed through mitosis or undergo whole-genome duplication. Although mitogenic signaling is not required for the G2/M transition under normal conditions, it modulates E2F transcriptional activity via c-Myc. When G2 arrest occurs due to CDK4/6 and CDK2 suppression, E2F activity levels determine the status of APC/C inactivation and the CDK2-Rb feedback loop. Upon release from G2 arrest, cells maintaining APC/C inactivation promptly induce CDK2 activation and FoxM1 phosphorylation, driving mitotic entry. Conversely, APC/C reactivation degrades cyclin A and abolishes the CDK2-Rb loop, necessitating CDK4/6 activation for cell-cycle re-entry. This regulatory mechanism mirrors the G1-phase process, resulting in whole-genome duplication. In cancer cells, this process promotes genome instability and oncogene amplification, contributing to aggressive behavior. These findings reveal a previously unrecognized mitogen-dependent checkpoint that governs cell fate in the G2 phase.

Mitogenic signalling is known to regulate cell-cycle entry in G1, but its role in G2 is unclear. Here, the authors show that mitogenic signalling governs whether G2-arrested cells enter mitosis or undergo whole-genome duplication.

Introduction

Cells employ regulatory mechanisms, including checkpoints and decision points, to ensure precise and orderly transition between cell-cycle phases¹. In the G1 phase, the restriction point serves as a critical decision-making juncture where cells commit to continuing through the cell cycle^2,3. Mitogenic signaling is essential for regulating the restriction point by activating cyclin-dependent kinases 4 and 6 (CDK4/6)^4,5. The non-phosphorylated form of the retinoblastoma protein (Rb) binds and inhibits E2F transcription factors, which control the expression of cell-cycle genes⁶. CDK4/6 activation phosphorylates Rb and subsequently releases E2F transcription factors, promoting the transcription of target genes, including those that activate CDK2⁷. CDK2 activation further phosphorylates Rb, creating a positive feedback loop^8,9. This bistable switch enables cells to proceed through the cell cycle independently of mitogenic signaling and CDK4/6 activity after crossing the restriction point. Furthermore, this mechanism also safeguards genomic stability by preventing incomplete DNA replication. However, emerging evidence indicates that CDK4/6 functions extend beyond the restriction point, including roles in cellular senescence^10,11 and G2-phase regulation^12,13.

Following DNA replication, the G2 checkpoint serves as a critical surveillance mechanism to ensure the accuracy and integrity of DNA synthesis before transitioning into mitosis. The G2 checkpoint detects damage or abnormalities in newly synthesized DNA, preventing the propagation of defective genetic material to daughter cells¹⁴. The DNA damage response (DDR) signaling pathways, including p53/p21, ATM, and ATR pathways, are central regulators of the G2 checkpoint¹⁵. CDK1 is pivotal for the G2/M transition and plays a fundamental role in activating the G2 checkpoint^16,17. Cyclin B1 expression, induced by the transcription factor FoxM1 during the S and G2 phases^18–20, binds to and activates CDK1^16,17. CDK1-cyclin B complex is initially kept inactive by inhibitory phosphorylation mediated by Wee1 and Mty1 kinases²¹. The phosphatase CDC25C is responsible for dephosphorylating and activating this complex²². Activation of the DDR pathway inhibits FoxM1 and CDC25C activities while upregulating the CDK inhibitor protein p21, leading to CDK inactivation^19,22,23. Additionally, stress-activated protein kinases modulate the G2 checkpoint through CDK inhibition under various stress conditions^13,24. Consequently, CDK inhibition results in cell-cycle arrest in G2, delaying mitotic entry to allow time for DNA repair. Dysregulation of this process can lead to DNA re-replication and whole-genome duplication (WGD)^25–27, a process frequently associated with cancer development, chromosomal instability, and poor clinical outcomes^28,29. Despite these insights, the role of mitogenic signaling in this context and the mechanisms guiding the decision between WGD and mitotic entry during the G2 phase remain incompletely understood.

In this study, we investigated the role of mitogenic signaling and CDK activity in G2-phase progression and the decision-making process between WGD and mitosis. Our findings provide insights into the regulatory role of CDKs and underscore the importance of mitogenic signaling in maintaining genomic stability during the G2 phase.

Results

Release from G2 arrest requires CDK4/6 activation and leads to WGD

Using Hoechst and EdU staining, we classified cell-cycle phases in primary mouse embryonic fibroblasts (MEFs) and found that 13% of cells exhibited DNA content exceeding 4 N (>4 N cells) (Fig. 1a). To determine whether G2-arrested cells contribute to these >4 N cells, we analyzed Rb phosphorylation at S807/811, a marker of Rb hyperphosphorylation^9,30. While all S-phase cells induced Rb phosphorylation, we identified subpopulations lacking Rb phosphorylation in both G1 and G2 phases (Fig. 1b and Supplementary Fig. 1a). This loss of Rb phosphorylation in G2 was similarly observed in non-transformed MCF-10A, RPE1, and passage-limited HS68 cells (Supplementary Fig. 1b, c). Furthermore, mitogen removal gradually increased the percentage of G2-phase cells without Rb phosphorylation (Supplementary Fig. 1d, e). We subsequently focused on MCF-10A cells, a well-established non-transformed human model. To investigate cell-cycle re-entry in G2-arrested cells, we synchronized cells in quiescence through mitogen removal, resulting in G1-arrested (96%) and G2-arrested (4%) populations (Supplementary Fig. 1f). G2-arrested cells showed significantly elevated levels of DNA damage markers, phosphorylated H2AX at S139 (γH2AX) and 53BP1, indicating checkpoint activation (Fig. 1c, d). Additionally, G2-arrested cells exhibited larger nuclear areas compared to G1-arrested cells (Supplementary Fig. 1g). After mitogen stimulation of G2-arrested cells, we observed a tight temporal correlation between Rb phosphorylation and E2F transcriptional activity (E2F1 mRNA), measured by immunostaining and fluorescence in situ hybridization (FISH) (Fig. 1e, f). Notably, acute treatment with the CDK4/6 inhibitor palbociclib (CDK4/6i) for 15 min significantly reduced Rb phosphorylation in G2-arrested cells stimulated with mitogens for 12 h (Fig. 1g, h). These data suggest that CDK4/6 activity is essential for G2-arrested cells to reactivate the Rb/E2F pathway and resume cell-cycle progression, potentially leading to >4 N cells.

Fig. 1. G2-arrested cells need CDK4/6 activity to re-enter the cell cycle, resulting in WGD.

a Density scatterplot showing DNA content versus EdU staining in primary MEFs (n = 3000 cells). b Histogram of phosphorylated Rb (p-Rb, S807/811) normalized to total Rb (t-Rb) levels in S and G2 phases in MEFs. Dotted lines represent the threshold used to classify cells into p-Rb negative and positive populations (n > 1000 cells/condition). c Representative images of Hoechst, γH2AX, and 53BP1 in MCF-10A cells 48 h after mitogen removal and sorted by DNA content. G1- and G2-arrested cells were classified based on DNA content. Scale bar is 20 μm. d Single-cell violin plots of γH2AX and 53BP1 puncta area in G1 and G2-arrested cells (n > 2000 cells/condition). Asterisks indicate significant differences in the two-tailed unpaired t-test (*** p ≤ 0.0001). e Representative images of Hoechst, p-Rb (S807/811), and E2F1 mRNA FISH in mitogen-starved MCF-10A cells stimulated with mitogens for 12 h. Scale bar is 20 μm. f Percentage of cells positive for p-Rb (S807/811) and averaged levels of E2F1 and Cdc25A mRNA as a function of time since mitogen stimulation. Solid lines represent sigmoidal best-fit curves. g, h Density scatterplot showing DNA content versus EdU staining (n > 1000 cells) (g) and percentage of p-Rb-positive cells in G2-arrested cells treated with DMSO or palbociclib (1 µM) for 15 min (h). Mitogen-starved cells were stimulated with mitogens for 12 h before fixation. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in a two-tailed unpaired t-test (*p ≤ 0.05). i Schematic illustrating live-cell sensors for APC/C activity (top) and cell-cycle phase (bottom). j, k Single-cell traces of nuclear area (left), Geminin-degron (middle), and Cdt1-degron (right) levels in G1- (j) and G2-arrested (k) cells. Cells were synchronized by mitogen removal for 48 h and classified as G1- or G2-arrested based on the nuclear area (n = 69 cells/condition). l Histogram showing DNA content in G1- and G2-arrested cells 25 hr after mitogen stimulation (G1-arrested: n = 642 cells; G2-arrested: n = 69 cells).

To directly assess whether G2-arrested cells transition into mitosis or undergo an additional round of DNA replication, we employed live-cell imaging using MCF-10A cells expressing biosensors for anaphase-promoting complex/cyclosome (APC/C) activity (Geminin degron)³¹ and cell-cycle phase (Cdt1 degron)^32,33, along with a nuclear marker (histone H2B) for individual cell tracking. APC/C is typically inactivated around the G1/S transition, leading to Geminin accumulation, whereas the Cdt1 degron is degraded during S phase, marking transitions at G1/S and S/G2 (Fig. 1i). After synchronizing cells in quiescence and stimulating with mitogens, live-cell imaging revealed that both G1- and G2-arrested cells entered DNA replication rather than directly transitioning into mitosis (Fig. 1j, k). Subsequent DNA content analysis revealed that 3% of cells had >4 N DNA content (Supplementary Fig. 1h). When specifically classifying cells based on live-cell tracking data, 59% of G2-arrested cells acquired >4 N DNA content, whereas G1-arrested cells predominantly displayed normal 2 N (9%) or 4 N (91%) DNA content (Fig. 1l). Together, these results demonstrate that G2-arrested cells re-enter the cell cycle similarly to G1-arrested cells, resulting in WGD.

p53-mediated CDK2/4/6 inhibition induces APC/C reactivation and WGD

To examine the contribution of CDK activity to G2 arrest, we employed live-cell reporters for CDK4/6³⁴ and CDK2³⁵ activity, in combination with measurements of Rb phosphorylation at S807/811 in MCF-10A cells (Supplementary Fig. 2a). Classification of G2-phase cells according to Rb phosphorylation status revealed a robust correlation between Rb phosphorylation and CDK activities (Supplementary Fig. 2b). In the presence of CDK4/6 inhibition, G2-phase cells retaining phosphorylated Rb exhibited elevated CDK2 activity (Supplementary Fig. 2c). Furthermore, we observed a significant correlation between low CDK activities and DDR pathway activation during G2 (Supplementary Fig. 2d, e). These findings indicate that loss of CDK2 and CDK4/6 activities is closely associated with DDR activation and G2 arrest.

We next investigated the cell-fate decision between mitotic progression and WGD following G2 arrest. To induce G2 arrest, we treated cells with the CDK2/4/6 inhibitor PF-06873600 (CDK2/4/6i) for 24 hr, followed by drug withdrawal to release cells from arrest. CDK4/6 and CDK2 activities were rapidly suppressed during treatment and promptly recovered upon drug removal (Supplementary Fig. 3a). Using Geminin and Cdt1 degron reporters, we selectively tracked cells that had entered G2 before treatment with either DMSO or CDK2/4/6i. Control cells progressed normally into mitosis, whereas CDK2/4/6 inhibition completely blocked mitotic entry, confirming effective G2 arrest (Fig. 2a, b, and Supplementary Fig. 3b). Notably, 75% of these G2-arrested cells exhibited premature APC/C reactivation (Fig. 2b). Following drug withdrawal, we found two distinct fates: cells maintaining APC/C inactivation proceeded directly to mitosis, while cells with premature APC/C reactivation underwent an additional round of DNA replication, indicative of WGD (Fig. 2b and Supplementary Fig. 3c). Knockdown of the APC/C activator Cdh1 significantly accelerated mitotic entry upon drug withdrawal, underscoring the functional importance of APC/C activity in controlling this fate decision (Supplementary Fig. 3d,e). Consistent with these findings, DNA content analysis showed an increased population of >4N cells 32 hr after CDK2/4/6 inhibition (Supplementary Fig. 3f). Further classification based on live-cell imaging indicated that 20% of cells experiencing APC/C reactivation acquired >4N DNA content, compared to only 1% without APC/C reactivation (Fig. 2c). These results demonstrate APC/C reactivation as a critical step determining cell fate toward mitotic progression and WGD.

Fig. 2. APC/C reactivation in G2-arrested cells is associated with low E2F activity and WGD.

a Histogram showing the time to mitosis in G2-phase cells treated with DMSO or PF-06873600 (500 nM) for 24 h (DMSO: n = 409 cells; CDK2/4/6i: n = 1984 cells). b Single-cell traces of Geminin-degron and Cdt1-degron levels. Circles mark the first mitosis after PF-06873600 (500 nM) treatment. Cells were classified based on APC/C reactivation status 24 h after treatment (n = 100 cells). c Histogram of DNA content 32 h after CDK2/4/6i withdrawal (APC/C reactivation: n = 847 cells; no APC/C reactivation: n = 284 cells). d Single-cell traces of Geminin-degron levels (left), CDK4/6 (middle), and CDK2 (right) activities. G2-phase cells were treated with NCS (1 µg/mL) and classified by APC/C status 24 h after treatment (n = 100 cells). e Percentage of cells with APC/C reactivation in wild-type (WT) and p53-knockout cells. G2-phase cells were treated with NCS (1 µg/mL), Etoposide (10 µM), or Zeocin (500 µg/mL) for 24 h and classified based on APC/C reactivation status. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the two-tailed unpaired t-test (*p ≤ 0.05; **p ≤ 0.001). f, g Single-cell traces of Geminin-degron levels (left), CDK4/6 (middle), and CDK2 (right) activities in WT (f) and p53-knockout (g) cells. G2-phase cells were treated with Zeocin (500 µg/mL) and classified by APC/C status 24 h after treatment (n = 100 cells).

Given the potential off-target effects of PF-06873600 on CDK1 activity³⁶, we further evaluated the impact of CDK inhibition on G2 progression using tagtociclib (CDK2i), a more selective CDK2 inhibitor. While CDK4/6 inhibition alone did not impede the G2/M transition, CDK2 inhibition caused a notable delay in mitotic entry (Supplementary Fig. 3g). Combined CDK4/6 and CDK2 inhibition more robustly induced G2 arrest accompanied by premature APC/C reactivation (Supplementary Fig. 3h). These results confirm the critical role of CDK2 in promoting timely G2/M progression and underscore APC/C reactivation as a consequence of CDK2/4/6 inhibition.

We examined whether diverse cellular stress conditions similarly trigger APC/C reactivation during G2 arrest. Using DNA-damaging agents (neocarzinostatin [NCS], etoposide, and zeocin) and ribotoxic stress (anisomycin), we monitored changes in CDK activities alongside APC/C dynamics using the Geminin degron. These stress conditions led to suppression of both CDK4/6 and CDK2 activities in a subset of G2-phase cells, resulting in premature APC/C reactivation during G2 (Fig. 2d and Supplementary Fig. 4a). Notably, knockout of p53 diminished stress-induced suppression of CDK2/4/6 activities, thereby significantly preventing APC/C reactivation (Fig. 2e–g and Supplementary Fig. 4b). Conversely, direct inhibition of CDK2/4/6 robustly activated APC/C independently of p53 expression (Supplementary Fig. 4c,d). Collectively, our data establish p53 as a pivotal mediator coupling stress-induced CDK2/4/6 inhibition to APC/C reactivation, leading ultimately to WGD induction.

E2F activity determines APC/C reactivation and the decision between mitosis and WGD in G2-arrested cells

Given that E2F transcriptional activity regulates the expression of the APC/C inhibitor Emi1^37,38, we next assessed how the Rb/E2F pathway contributes to APC/C reactivation during G2 arrest. Using live-cell imaging to monitor Geminin and Cdt1 degrons, we classified G2-phase cells based on their APC/C reactivation status after CDK2/4/6 inhibition (Fig. 3a and Supplementary Fig. 5a). Subsequently mRNA FISH analyses showed significantly higher expression of E2F target genes (E2F1 and Cdc25A) in cells maintaining APC/C inactivation compared to those exhibiting APC/C reactivation, despite comparable ACTB control mRNA levels (Fig. 3b). While CDK2/4/6 inhibition effectively blocked mitotic entry regardless of Rb expression, knockout of Rb significantly suppressed APC/C reactivation during G2 arrest (Fig. 3c–e and Supplementary Fig. 5b). Upon CDK2/4/6i withdrawal, Rb-knockout cells predominantly progressed to mitosis rather than undergoing WGD. Conversely, Emi1 knockdown significantly increased APC/C reactivation and promoted WGD, while Cdh1 knockdown decreased APC/C reactivation, promoting mitotic entry (Fig. 3f and Supplementary Fig. 5c). Additionally, expression of a non-degradable D-box mutant of Emi1^AxxA (R322A and L325A)³⁹ via a doxycycline-inducible system significantly suppressed APC/C reactivation, enabling mitotic entry upon drug removal (Fig. 3g and Supplementary Fig. 5d). These findings demonstrate that E2F activity critically determines APC/C reactivation and thus dictates the cell fate decision between mitosis and WGD in G2-arrested cells.

Fig. 3. Rb/E2F pathway determines APC/C reactivation in G2-arrested cells.

a Single-cell traces of Geminin degron in G2-phase cells treated with PF-06873600 (500 nM) and classified by APC/C reactivation status 24 h after treatment. Circles mark the time of fixation and staining (n = 50 cells). b Single-cell violin plot showing ACTB (left), E2F1 (middle), and Cdc25A (right) mRNA levels (n > 250 cells/condition). Asterisks indicate significant differences in the two-tailed unpaired t-test (***p ≤ 0.0001). c Histogram showing the time to mitosis in WT and Rb-knockout cells (WT DMSO, n = 352 cells; WT CDK2/4/6i, n = 939 cells; Rb-knockout DMSO, n = 410 cells; Rb-knockout CDK2/4/6i, n = 658 cells). d Single-cell traces of Geminin degron in WT and Rb-knockout cells treated with PF-06873600 (500 nM) and classified by APC/C reactivation status (n = 100 cells/condition). e–g Percentage of cells with APC/C reactivation in WT and Rb-knockout cells (e), after Emi1 or Cdh1 siRNA knockdown (f), and with/without Emi1^AxxA (R322A, L325A) induction (g). G2-phase cells were treated with PF-06873600 (500 nM) and −/+ doxycycline (1 µM) (g) and classified by APC/C reactivation status. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the two-tailed unpaired t-test (*p ≤ 0.05; **p ≤ 0.001).

APC/C reactivation regulates the CDK2-Rb feedback loop and CDK4/6 dependency in G2-arrested cells

While CDK2 phosphorylates Rb to reinforce E2F activation, APC/C reactivation targets the CDK2 activator cyclin A for degradation⁴⁰. To assess how APC/C reactivation influences CDK activity dynamics, we analyzed single-cell CDK4/6 and CDK2 activity traces following CDK2/4/6 inhibition, grouped by APC/C reactivation status. CDK4/6 activity was unaffected by APC/C status (Fig. 4a, b). In contrast, cells maintaining APC/C inactivation retained residual CDK2 activity that gradually increased over time, while cells undergoing APC/C reactivation showed further suppression of CDK2 activity. Upon drug withdrawal, cells with inactive APC/C rapidly activated CDK2 and proceeded to mitosis. Combined live-cell and fixed-cell analyses confirmed an inverse correlation between cyclin A levels and APC/C reactivation after 24 h of CDK2/4/6 inhibition (Fig. 4c). Additionally, cells exhibiting APC/C reactivation displayed elevated p21 expression, suggesting involvement of the p53/p21 axis in controlling CDK activities under these conditions (Fig. 4d). These results indicate that APC/C reactivation status influences basal CDK2 activity in G2-arrested cells.

Fig. 4. APC/C reactivation disrupts the CDK2-Rb feedback loop.

a Single-cell traces of Geminin-degron levels (left) and CDK4/6 (middle) and CDK2 (right) activities in G2-phase cells treated with PF-06873600 (500 nM), classified by their APC/C reactivation status 24 h after treatment (n = 100 cells). b Phase plot of CDK activities (y-axis) versus Geminin-degron levels (x-axis), with cell trajectories classified by APC/C reactivation status and color-coded to represent the time since CDK2/4/6i treatment. Distinct color schemes are applied to highlight differences based on APC/C reactivation (n > 300 cells/condition). Single-cell violin plot showing levels of cyclin A (c) and p21 (d) in G2-phase cells treated with PF-06873600 (500 nM) for 24 h (n = 1200 cells/condition). Asterisks indicate significant differences in the two-tailed unpaired t-test (***p ≤ 0.0001). e, f Histogram of p-Rb (S807/811) normalized to t-Rb. MCF-10A (top) and RPE1 (bottom) cells were treated with DMSO, palbociclib (1 µM), or PF-06873600 (500 nM) for 1 h (n > 1000 cells/condition) (e). G2-phase MCF-10A cells were treated with either DMSO or PF-06873600 (500 nM) for 24 h, followed by fixation and staining (DMSO: n = 630 cells; CDK2/4/6i without APC/C reactivation: n = 1066 cells; CDK2/4/6i with APC/C reactivation: n = 2888 cells) (f). g Single-cell traces of Geminin-degron levels in G2-phase cells that reactivated APC/C during PF-06873600 (500 nM) treatment. Cells were treated with PF-06873600 (500 nM) for 24 h, followed by a drug switch to either DMSO or palbociclib (1 µM). Cells were classified into proliferation (red) and quiescence (blue) based on Geminin degron after drug switch (n = 50 cells/condition). h Percentage of proliferating cells. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the two-tailed unpaired t-test (*p ≤ 0.05).

To investigate how basal CDK2 activity affects the CDK2-Rb feedback loop in G2-arrested cells, we measured Rb phosphorylation at S807/811 in MCF-10A and RPE1 cells. CDK4/6 inhibition abolished Rb phosphorylation in G1 but not in S/G2 phases (Fig. 4e). Inhibition of CDK2 or CDK1 alone did not affect Rb phosphorylation in any cell-cycle phase (Supplementary Fig. 6a). CDK2/4/6 inhibition markedly reduced Rb phosphorylation in G1/S phases but only partially in G2. Increasing CDK2i concentration alongside CDK4/6i further diminished—but did not fully eliminate—Rb phosphorylation in G2 cells (Supplementary Fig. 6b). Addition of CDK1 inhibitor to CDK2/4/6 inhibition had no further effect (Supplementary Fig. 6c). Live-cell followed by fixed-cell analyses revealed that Rb phosphorylation correlated inversely with APC/C reactivation status under CDK2/4/6 inhibition (Fig. 4f). These data suggest the association of Rb phosphorylation maintenance and APC/C reactivation status in G2-arrested cells.

To evaluate the requirement for CDK4/6 activity upon release from G2 arrest, we treated cells with CDK4/6i after withdrawing CDK2/4/6i. Cells maintaining APC/C inactivation proceeded to mitosis regardless of CDK4/6 inhibition (Supplementary Fig. 7a). However, CDK4/6 inhibition significantly blocked cell-cycle re-entry in cells exhibiting APC/C reactivation (Fig. 4g, h and Supplementary Fig. 7b). Collectively, these findings reveal that APC/C reactivation status critically modulates basal CDK2 activity and the CDK2-Rb feedback loop, determining the requirement for CDK4/6 activity upon exit from G2 arrest.

Mitogenic signaling regulates E2F activity and the cell-fate decision between mitosis and WGD in G2-arrested cells

To evaluate the role of mitogenic signaling in the G2/M transition and APC/C reactivation, we monitored Geminin and Cdt1 degrons. In line with the restriction model, mitogen removal alone did not affect the G2/M transition in G2-phase cells (Fig. 5a and Supplementary Fig. 8a). Moreover, mitogen withdrawal did not influence CDK1 activity, as assessed by phosphorylation of its substrate and a live-cell sensor⁴¹ (Supplementary Fig. 8b–d). However, when G2 arrest was induced by CDK2/4/6 inhibition, increasing mitogen levels significantly and progressively reduced APC/C reactivation (Fig. 5b, c and Supplementary Fig. 8e). Although mitogen withdrawal suppressed CDK4/6 activity, it did not alter the kinetics of CDK2 inhibition in APC/C-reactivating cells (Supplementary Fig. 8f, g). These observations indicate that mitogenic signaling directly controls APC/C reactivation rather than influencing residual CDK2 activity. We alternatively induced G2 arrest using a 20-min pulse of NCS and further examined the effect of mitogen withdrawal on G2 progression. By titrating NCS concentrations in G2-phase cells, we confirmed G2-checkpoint activation, as evidenced by a progressive delay in the G2/M transition (Fig. 5d). While mitogen withdrawal alone did not impact the G2/M transition, it suppressed this transition in the presence of DNA damage (Fig. 5e). These findings indicate the significance of mitogenic signaling in determining the cell-fate decision between mitosis and WGD in G2-arrested cells.

Fig. 5. Reduction in mitogenic signaling results in APC/C reactivation in G2-arrested cells.

a Cumulative distribution function showing the time to mitosis in G2-phase cells, showing mitogen removal (Control: n = 308 cells; Mitogen removal: n = 331 cells). b Single-cell traces of Geminin-degron levels after mitogen removal or exposure to indicated mitogen concentrations 4 h before treatment with PF-06873600 (500 nM) for 24 h. Cells were classified based on APC/C reactivation status at 24 h after drug treatment (n = 100 cells/condition). c Percentage of cells exhibiting APC/C reactivation. Data are shown as mean ± SD (n = 4 biological replicates). Asterisks indicate significant differences in the one-way ANOVA test (*p ≤ 0.05; **p ≤ 0.001; ***p ≤ 0.0001). d, e Cumulative distribution function showing the time to mitosis as a function of 20-min NCS pulse at varying concentrations (d) and at 200 ng/ml combined with mitogen removal (e) (d, NCS 0 ng/ml: n = 157 cells; 50 ng/ml: n = 224 cells; 100 ng/ml: n = 208 cells; 200 ng/ml: n = 223 cells; e, NCS 0 ng/ml: n = 185 cells; NCS 0 ng/ml + Mitogen removal: n = 234 cells; NCS 200 ng/ml: n = 278 cells; NCS 200 ng/ml + Mitogen removal: n = 119 cells). f Histogram of p-Rb (S807/811) normalized to t-Rb in S- and G2-phase MCF-10A cells. Cells were treated with DMSO, palbociclib (1 µM), or PF-06873600 (500 nM) for 1 h after 8 h of mitogen removal (n > 1000 cells/condition). g Relative mRNA levels of E2F1 (left) and Cdc25A (right) in G2 phase cells exposed to the indicated mitogen concentration for 8 hr. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the one-way ANOVA test (*p ≤ 0.05; **p ≤ 0.001). h, i Relative mRNA levels of E2F and Cdc25A in G2 phase MCF-10A cells 8 h after control or c-Myc siRNA knockdown (h) or with and without doxycycline (1 µM) (i) for 8 h. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the two-tailed unpaired t-test (*p ≤ 0.05; ***p ≤ 0.0001). j Percentage of cells with APC/C reactivation. MCF-10A cells expressing a doxycycline-inducible c-Myc were treated with PF-06873600 (500 nM) ±doxycycline (1 µM) for 24 h, followed by a drug withdrawal. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the two-tailed unpaired t-test (*p ≤ 0.05).

We investigated the impact of mitogen removal on Rb phosphorylation at S807/811. Under control and CDK4/6 inhibition conditions, mitogen removal primarily suppressed Rb phosphorylation in the G1 phase, but not in the S/G2 phases (Fig. 5f and Supplementary Fig. 8h). Conversely, mitogen removal significantly decreased the proportion of G2-arrested cells maintaining Rb phosphorylation under CDK2/4/6 inhibition, highlighting the critical role of mitogenic signaling in regulating APC/C and Rb phosphorylation specifically during G2 arrest.

Mitogenic signaling regulates c-Myc, thereby amplifying the expression of actively transcribed genes^42–44. We hypothesized that mitogenic signaling enhances E2F activity through c-Myc expression, controlling APC/C reactivation in G2-arrested cells. Indeed, reducing mitogen levels progressively lowered mRNA levels of the E2F target genes, E2F1 and Cdc25A, in G2-phase cells (Fig. 5g). Furthermore, modulating c-Myc expression via siRNA knockdown or doxycycline-inducible overexpression reduced or increased E2F1 and Cdc25A mRNA levels, respectively, in G2-phase cells (Fig. 5h, i). Consistent with this, c-Myc depletion increased APC/C reactivation, whereas c-Myc induction suppressed APC/C reactivation in G2-arrested cells (Fig. 5j). Thus, mitogenic signaling modulates APC/C reactivation by regulating c-Myc-dependent E2F activity, critically influencing the decision between mitosis and WGD.

CDK2 phosphorylates FoxM1 to induce cyclin B1 expression and mitotic entry

We explored how CDK2/4/6 inhibition delays mitosis independently of the Rb/E2F pathway. Since CDK1 activity, regulated by cyclin B and Wee1/Myt1, drives mitosis^16,17, we tested whether CDK2/4/6 inhibition delays mitotic entry via Wee1/Myt1 activity or cyclin B expression. Although Wee1/Myt1 inhibition with PD166285 accelerated mitosis, CDK2/4/6 inhibition blocked mitotic entry even in the presence of Wee1/Myt1 inhibitor, indicating a Wee1/Myt1-independent mechanism (Supplementary Fig. 9a–c). To examine cyclin B dynamics, we employed MCF-10A and RPE1 cells expressing fluorescently tagged endogenous cyclin B1⁴⁵ along with Geminin and Cdt1 degrons. In control cells, cyclin B1 levels progressively rose, preceding mitotic entry (Supplementary Fig. 9d). After CDK2/4/6 inhibition, cells with APC/C reactivation promoted cyclin B1 degradation (Supplementary Fig. 9e). Upon drug withdrawal, these cells sequentially initiated APC/C inactivation, DNA re-replication, cyclin B1 re-accumulation, and mitotic entry. In contrast, cells maintaining APC/C inactivation halted cyclin B1 accumulation upon CDK2/4/6 inhibition but rapidly resumed cyclin B1 increase and entered mitosis after drug removal (Fig. 6a and Supplementary Fig. 9f). CDK4/6 inhibition alone did not impair cyclin B1 accumulation or mitotic entry, indicating a specific requirement for CDK2 in cyclin B1-mediated mitotic progression (Supplementary Fig. 10a, b).

Fig. 6. CDK2 phosphorylates FoxM1 to induce cyclin B expression.

a Single-cell traces of endogenous cyclin B1 (top) and Geminin-degron (bottom) levels in MCF-10A (left) and RPE1 (right) cells without APC/C reactivation following 24 h treatment with PF-06873600 (500 nM) and subsequent drug withdrawal (n = 50 cells). b Scatterplot showing DNA content versus EdU, color-coded by p-FoxM1 (T600) levels in MCF-10A cells (n = 3000 cells/condition). c Single-cell violin plot of p-FoxM1 (T600) levels in MCF-10A cells (n > 2000 cells/condition). d Relative p-FoxM1 (T600) levels normalized to G1-phase levels in DMSO-treated MCF-10A (left) and RPE1 (right) cells. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the one-way ANOVA test (*p ≤ 0.05; **p ≤ 0.001). b–d Cells were treated with DMSO, palbociclib (1 µM), tagtociclib (1 µM), or PF-06873600 (500 nM) for 1 h, followed by fixation and staining.

FoxM1 phosphorylation is essential for initiating a mitotic transcription program, including cyclin B expression^18,19,46. We examined whether CDK2 activation is necessary for inducing FoxM1 phosphorylation at T600, a CDK-dependent activating modification⁴⁷. FoxM1 phosphorylation was gradually increased during cell-cycle progression (Fig. 6b). To test the impact of CDK4/6 or CDK2 inhibition on FoxM1 phosphorylation, we used CDK4/6i and CDK2i and confirmed that treatment with CDK4/6i and CDK2i suppressed CDK4/6 and CDK2 activities, respectively, in G2-phase cells (Supplementary Fig. 11a). We found that CDK2 inhibition, but not CDK4/6 inhibition, significantly blocked FoxM1 phosphorylation in the S and G2 phases (Fig. 6b–d). Furthermore, premature phosphorylation of FoxM1 by ATR inhibition¹⁹ can also be suppressed by CDK2 inhibition (Supplementary Fig. 11b, c). By performing live-cell imaging followed by fixed-cell experiments, we classified G2-arrested cells based on APC/C reactivation status and measured FoxM1 phosphorylation an hour after withdrawing CDK2/4/6i. We found an inverse correlation between FoxM1 phosphorylation and APC/C reactivation (Supplementary Fig. 11d). Our data suggest that CDK2 phosphorylates FoxM1 to promote cyclin B expression and facilitate the G2/M transition. Therefore, when APC/C inactivation is maintained, cells released from G2 arrest can rapidly induce high CDK2 activation and FoxM1 phosphorylation, facilitating cyclin B accumulation and the G2/M transition.

CDK2/4/6 inhibition promotes WGD, genomic instability, and aggressive cancer cell growth

Recent studies have highlighted dual CDK2 and CDK4/6 inhibition as a promising strategy to overcome drug resistance in breast cancer^48–53. We investigated the impact of intermittent CDK2/4/6 inhibition on WGD using non-transformed cells (MCF-10A and HS68), as well as breast cancer cells (MDA-MB-231 and MCF-7). Cells underwent repeated cycles of 24-hr CDK inhibitor treatment followed by 24-h drug holiday intervals, mirroring clinical schedules designed to minimize toxicity⁵⁴. While repeated CDK4/6 inhibition alone rarely produced >4 N cells, repeated CDK2/4/6 inhibition markedly increased the percentage of >4 N cells, indicating effective induction of WGD (Fig. 7a, b and Supplementary Fig. 12a, b). Additionally, repeated CDK2/4/6 inhibition frequently led to multinucleation (Fig. 7c). These data show that CDK2/4/6i treatment promotes WGD and genomic instability in both non-transformed and cancer cells.

Fig. 7. CDK2/4/6 inhibition induces genomic instability.

a Percentage of cells with >4 N DNA content after 24 h treatment with either palbociclib (1 µM) or PF-06873600 (500 nM), followed by a 24 h drug holiday across 0–3 cycles. Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the one-way ANOVA test (*p ≤ 0.05; **p ≤ 0.001; ***p ≤ 0.0001). b Density scatterplot showing DNA content versus EdU staining in MDA-MB-231 cells (n = 2000 cells/condition). c Representative images of Hoechst and EdU staining in MDA-MB-231 cells after 3 cycles of intermittent PF-06873600 (500 nM) treatment. The scale bar represents 20 µm. d Growth curves following 3 cycles of intermittent treatment with either DMSO or PF-06873600 (500 nM). Data are shown as mean ± SD (n = 3 biological replicates). Asterisks indicate significant differences in the two-tailed unpaired t-test (*p ≤ 0.05; **p ≤ 0.001). e The proposed model illustrating the selection between mitosis and WGD in G2-arrested cells.

Oncogenic mutations and tumor-suppressor gene deletions are hallmark cancer features⁵⁵. Non-transformed MCF-10A and HS68 cells lack active mutations in oncogenes despite the absence of p16 expression in MCF-10A cells⁵⁶. In contrast, breast cancer cell lines MDA-MB-231 and MCF-7 harbor active mutations in oncogenes such as PIK3CA and KRAS, along with deletions in the tumor suppressor p53⁵⁷. WGD in cancer cells can potentiate oncogene amplification, enhancing tumor aggressiveness and therapeutic resistance. To test this possibility, we monitored cell proliferation for 28 days after three cycles of intermittent CDK2/4/6 inhibition. While WGD had minimal impact on the proliferation of non-transformed cells, cancer cells that underwent WGD displayed significantly accelerated growth relative to controls (Fig. 7d). These findings suggest that genomic instability induced by CDK2/4/6 inhibition promotes oncogene amplification, driving more aggressive tumor cell growth.

Discussion

Our study underscores the critical role of mitogenic signaling in regulating G2-phase progression and the cell-fate decision between mitosis and WGD. While mitogenic signals are well-known regulators of the G1 phase, our results demonstrate a broader function extending into G2, particularly during G2 arrest induced by CDK2/4/6 inhibition. In this scenario, mitogenic signaling, mediated by c-Myc-driven E2F activation, becomes essential for APC/C reactivation (Fig. 7e). APC/C status subsequently modulates basal CDK2 activity and regulates the CDK2-Rb feedback loop, dictating the requirement for CDK4/6 activity and determining whether cells enter mitosis or undergo WGD. These findings align with reports indicating that elevated E2F activity in cancer cells with activated mitogenic signaling pathways confers resistance to APC/C reactivation and DNA damage-induced G2 arrest²⁷.

Previous studies demonstrated that p53 activation following DNA damage can trigger APC/C reactivation, cyclin B degradation, and senescence in G2-phase cells^58,59. Consistent with this, our results indicate that stress-induced CDK2/4/6 inhibition reduces Rb phosphorylation and E2F activity, suppresses expression of the APC/C inhibitor Emi1, and thereby promotes APC/C reactivation and cell-cycle exit during the G2 phase. Although CDK2/4/6 inhibition can trigger APC/C reactivation independently of p53, the loss of p53 significantly attenuates CDK2/4/6 inhibition in response to stress, thereby preventing APC/C reactivation.

Upon release from G2 arrest, cells maintaining APC/C inactivation rapidly induce CDK2 activity and FoxM1 phosphorylation, facilitating cyclin B accumulation and mitotic entry. Conversely, cells undergoing APC/C reactivation require CDK4/6 activity to restart Rb phosphorylation and CDK2 activation, steering cells toward WGD. These observations reinforce emerging roles for CDK4/6 activity beyond the traditional G1 restriction point^12,13. Furthermore, our findings highlight the importance of APC/C inactivation in protecting cyclin A protein, thereby sustaining the CDK2-Rb positive feedback loop.

WGD requires uncoupling mitosis from DNA replication licensing, typically mediated by APC/C activation during mitosis^60,61. Premature APC/C reactivation drives cyclin B degradation (suppressing CDK1 activity) and geminin degradation (releasing the licensing factor Cdt1), thereby facilitating DNA replication re-licensing^62,63. Our findings are consistent with studies linking altered APC/C activity to DNA re-replication events.

Cancer cells undergoing WGD exhibit increased genomic instability and oncogene amplification, enhancing tumor aggressiveness. Clinical data indicate that WGD occurs in approximately 30% of advanced cancers, correlating with poor patient prognosis^28,64. However, WGD-associated genomic instability may also introduce unique vulnerabilities. Increased dependency on amplified oncogenic pathways presents a therapeutic opportunity to specifically target these cancer cells. Conversely, therapies that induce DNA damage or inhibit CDK2/4/6 activity might inadvertently promote premature APC/C reactivation, facilitating WGD and exacerbating tumor aggressiveness. These insights underscore the importance of designing combination therapies that minimize WGD risk while effectively targeting vulnerabilities of genomically unstable cancer cells.

In this study, we utilized pharmacological inhibitors to rapidly suppress CDK2/4/6 activities specifically in G2-phase cells. Although we employed selective CDK2 inhibitors, we acknowledge that off-target inhibition, particularly affecting CDK1, cannot be entirely excluded. Given the critical role of CDK1 in orchestrating the G2/M transition, unintended inhibition of CDK1 could complicate interpretations related specifically to CDK2. To rigorously distinguish the individual contributions of CDK1 and CDK2 to G2-phase arrest and downstream cellular outcomes, future studies incorporating genetic tools or inhibitors with greater specificity will be essential.

In conclusion, our study reveals the complex interplay between mitogenic signaling, APC/C regulation, and genomic integrity during G2-phase progression. These results provide critical insights into the mechanisms underlying genomic amplification and instability, offering potential avenues for therapeutic intervention.

Methods

This study adheres to all applicable ethical regulations.

Cell lines

MCF-10A (#CRL-10317), RPE1 (#CRL-4000), HS68 (#CRL-1635), MDA-MB-231 (#CRM-HTB-26), and MCF-7 (#HTB-22) cells were obtained from ATCC. MCF-10A cells were cultured in phenol red-free DMEM/F12 media (Gibco, #11039047) supplemented with 5% horse serum (Gibco, #16050122), 20 ng/ml EGF (Proteintech, #HZ-1326), 10 µg/ml insulin (Sigma, #I1882), 100 ng/ml cholera toxin (Sigma, #C8052), and 500 µg/ml hydrocortisone (Sigma, #H0888). Rb-knockout MCF-10A cells were obtained from Dr. Sabrina Spencer’s laboratory⁴⁹. p53-knockout MCF-10A cells were described previously⁶⁵. MCF-10A and RPE1 cells expressing endogenously tagged cyclin B1 with a fluorescent protein were provided by Dr. Rene Medema’s laboratory⁴⁵. RPE1 cells were cultured in phenol red-free DMEM/F12 media supplemented with 10% fetal bovine serum (FBS) (Gibco, #10-437-028) and 0.01 mg/ml hygromycin B (InvivoGen, #ant-hg). HS68, MDA-MB-231, and MCF-7 cells were cultured in DMEM media (Genesee Scientific, #25-500) supplemented with 10% FBS. For mitogen removal, MCF-10A cells were washed three times with PBS and cultured in GM-GFS media [phenol red-free DMEM/F12 media supplemented with 0.3% Bovine Serum Albumin (BSA) (Millipore, #A3311), 100 ng/ml cholera toxin, 500 µg/ml hydrocortisone]. MEFs were isolated from C57BL/6 mouse embryos, which were extracted from the uteri of pregnant mice on embryonic days 11–14. The embryonic bodies were minced in a 0.25% trypsin-EDTA solution (Gibco, #25-510) and incubated at 37 °C for 15 min. After incubation, the suspension was centrifuged, and the supernatant was removed. The cell pellet was resuspended and cultured in DMEM media (Genesee Scientific, #25–500) supplemented with 10% FBS. All cell lines and MEFs were maintained at 37 °C with a 5% CO₂ atmosphere and tested negative for mycoplasma.

Plasmid generation

Previous described constructs include pLenti-DHB (a.a. 995–1087)-mVenus-p2a-mCherry-Rb (a.a. 886–928)-IRES-blasticidin (CDK2 and CDK4/6 sensors³⁴, Addgene, #126679), pLenti- H2B-iRFP670-p2a-mCerulean-Cdt1 (a.a. 1-100)-IRES-neomycin⁶⁶, pLenti-H2B-iRFP670-p2a-mCerulean-Geminin (a.a. 1-110)-IRES-neomycin⁶⁷, and pCW57.1-c-Myc (doxycycline-inducible construct)^66,67. Using Gibson cloning, Geminin (a.a. 1-110) was cloned into a pLenti-IRES-puromycin construct. The D-box mutant Emi1^AXXA (R322A, L325A) was created using overlap extension PCR and inserted into the pCW57.1 vector (Addgene, #50661) following Nhe1 and BamH1 restriction digestion.

Cell line generation

Stable cell lines were established through lentiviral transduction. Lentiviruses were produced in HEK-293T cells by co-transfecting them with the cDNA of interest, pCMV-VSV-G (Addgene, #8454), pRSV-rev (Addgene, #12253), and pMDLg/pRRE (Addgene, #12251) using the PEI transfection reagent. Viral supernatants were collected at 72 and 96 h post-transfection, filtered through a 0.45 µm filtration unit (Millipore, #SLHA033SB), and concentrated using 100 kDa ultra-centrifugal filters (Millipore, #UFC910024). Concentrated viruses were stored at −80 °C until use. For generating cells with multiple constructs, plasmids with different antibiotic selection markers were sequentially introduced. Post-transduction, cells were selected using appropriate antibiotics: 10 µg/ml puromycin (InvivoGen, #ant-pr-1) for RPE1 and 1 µg/ml for other cell lines, 10 µg/ml blasticidin (InvivoGen, #ant-bl-1), or 200 µg/ml neomycin (Thermo Scientific, #BP673-5).

Drugs and chemicals

Stock solutions of the following drugs and chemicals were dissolved in DMSO (Sigma, #D2438): palbociclib (Selleck Chemicals, #S1116), PF-06873600 (Cayman Chemicals, #35502), Etoposide (Sigma, #E1383), Anisomycin (Sigma, #A9789), tagtociclib (ChemieTek, #CT-PF0710), PD166285 (Tocris Bioscience, #3785), AZ20 (MedChemExpress, #HY-15557), and doxycycline (Sigma-Aldrich, #D9891). 5-ethynyl-2′-deoxyuridine (EdU) was obtained from Sigma (#900584), and Hoechst 33342 dye from Thermo Scientific (#62249). Neocarzinostatin (Sigma, #N9162) and Zeocin (Invitrogen, #46-0509) were procured as a ready-to-use solution.

Antibodies

The following antibodies were purchased from Cell Signaling Technology: Rabbit anti-phospho-Rb (S807/811) (#8516, 1:2000 for IF), Mouse anti-Rb (4H1) (#9309, 1:2000 for IF), Rabbit anti-53BP1 (#4937, 1:500 for IF), Rabbit anti-p21 (#2947, 1:2500 for IF), Rabbit anti-phospho-Chk1 (S317) (#12302, 1:1000 for IF), Rabbit anti-phospho-Chk2 (T68) (#2661, 1:500 for IF), Rabbit anti-phospho-FoxM1 (T600) (D9M6G) (#14655, 1:200 for IF) and Mouse anti-phospho-Histone H3 (Se10) (G63) (#9706, 1:500 for IF). Mouse anti-phospho-H2AX (S139) (γH2AX) was obtained from Millipore (#05-636, 1:500 for IF). Mouse anti-Cyclin A (B-8) (#sc-271682, 1:500 for IF) was acquired from Santa Cruz Biotechnology. Secondary antibodies from Invitrogen were diluted 1:2000 in blocking buffer: Goat anti-rabbit IgG (H + L) conjugated with Alexa Fluor 488 (#A32731), Goat anti-rabbit IgG (H + L) conjugated with Alexa Fluor 568 (#A11036), Goat anti-mouse IgG (H + L) conjugated with Alexa Fluor 488 (#A32723), and Goat anti-mouse IgG (H + L) conjugated with Alexa Fluor 568 (#A11031).

siRNA transfection

siRNA was purchased from Integrated DNA Technologies, pooled, and transfected into MCF-10A cells using DharmaFECT1 (Horizon Discovery, #T-2001) according to the manufacturer’s instructions. The siRNAs used include Non-targeting negative control siRNA, Emi1 (FBXO5) targeting siRNAs (1: 5′-CUCUUACUCAAUUGAUACCAACAGA-3′; 2: 5′-CAAAAAGAAUUUACGAAGAUUGUGA-3′; 3: AGAAUUUCGGUGACAGUCUACAATC-3′), Cdh1 (FZR1) targeting siRNAs (1: 5′-GGUCUUACUGUUUCAAGGUUUUUAA-3′; 2: 5′-GCGUGAACUUCCACAGGAUUAACGA-3′; 3: 5′-GGUCUGGAAUCACUCGAGCCUGAGC-3′), and c-Myc targeting siRNAs (1: 5′-AUCAUUGAGCCAAAUCUUAAAAAAA-3′; 2: 5′-CGACGAGACCUUCAUCAAAAACATC-3′). For knockdown experiments, MCF-10A cells were seeded into 96-well plates (Cellvis, #P96-1.5 P) at least 12 h prior to transfection. DharmaFECT1 was diluted at a ratio of 1:50, and siRNA stock was diluted at a ratio of 1:10 (final 20 nM) in Opti-MEM media (Gibco, #31985070) and incubated at room temperature for 5 min. The diluted DharmaFECT1 and siRNA were mixed and incubated at room temperature for 15 min before adding to the growth media, resulting in a final composition of 20% mixture to 80% growth media.

Immunofluorescence

Cells were cultured, stained, and imaged in 96-well plates (Cellvis, #P96-1.5 P). For EdU staining, cells were incubated with EdU (10 µM) for 15 min at 37 °C prior to fixation. Fixation was performed with 4% paraformaldehyde (PFA) (Thermo Scientific, #28906) containing 10 mM HEPES (Sigma, #H3537) in PBS, diluted 1:1 ratio with media (final 2% PFA), for 15 min at room temperature. Cells were then permeabilized using 0.2% Triton X-100 (Sigma, #T8787) in PBS for 15 min at room temperature. For photobleaching, cells expressing fluorescent proteins were treated with 3% H₂O₂ + 20 mM HCl in PBS for 2 h at room temperature. Incorporated EdU was labeled using a click reaction in 2 mM CuSO₄ (Sigma, #C1297), 20 mg/ml sodium ascorbate (Sigma, #A4034), and AFDye 647 picolyl azide (Click Chemistry Tools, #1300) in TBS for 15 min, followed by PBS washes. Cells were blocked in PBS containing 0.1% Triton X-100, 10% FBS, 1% BSA, and 0.01% NaN₃ (Sigma, #S2002) for 1 h at room temperature before staining with primary antibodies diluted in blocking buffer overnight at 4 °C. Secondary antibodies conjugated to Alexa Fluor dyes were diluted 1:2000 in blocking buffer and incubated with cells for 1 h at room temperature. Hoechst 33342 dye (Thermo Scientific, #62249) was used to stain nuclei at a 1:1000 dilution in PBS for 15 min at room temperature. Washing steps were performed using an automated plate washer (aspirate to 50 µl, dispense 250 µl, repeated three times) (BioTek 405 TS).

RNA fluorescence in situ hybridization (FISH)

The RNA FISH was carried out using the ViewRNA ISH Cell Assay Kit (Thermo Scientific, #QVC00001), with probes targeting ACTB (Thermo Scientific, #VA1-10351-VC), E2F1 (Thermo Scientific, #VA1-12108-VC), and Cdc25A (Thermo Scientific, #VA1-15814-VC). Following cell fixation, photobleaching and permeabilization were performed as described in the immunofluorescence section. Probe hybridization, amplification, and labeling with Alexa Fluor 555 were carried out according to the manufacturer’s instructions.

Fixed- and live-cell imaging

Fixed- and live-cell imaging was conducted using a Nikon microscope with an inverted Eclipse Ti-2 body (Nikon)⁶⁸. Images were captured using either a 10× objective (Nikon CFI Plan Apo Lambda, 0.45 NA, no binning) or a 20× objective (Nikon CFI Plan Apo Lambda, 0.75 NA, 2-by-2-pixel binning). For fixed-cell imaging, thirty-two nonoverlapping sites per well were imaged. mRNA FISH images were captured with a z-stack interval of 1.5 µm for enhanced clarity. A 96-well plate was placed in a humidified chamber (Tokai Hit) maintained at 37 °C with 5% CO₂ for live-cell imaging. Images were captured every 12 min at three to six non-overlapping sites per well. The total light exposure time was kept below 500 ms per time point to minimize photodamage.

Image analysis

Automated image analysis was performed using a custom MATLAB pipeline (MathWorks, R2021a). Raw images were flat-field corrected for light bias. Nuclei were segmented using Hoechst staining based on a histogram-defined threshold for fixed-cell imaging. In live-cell imaging, nuclei were segmented using H2B-iRFP670 with a Laplacian of Gaussian blob detection method. The deflection-bridging algorithm was employed to track cell movement across live-cell frames and fixed-cell images. Mitosis detection involved identifying H2B signals in proximate daughter cells, which should be between 45% and 55% of the mother cell’s signal. To evaluate KTR-based CDK2 and CDK4/6 reporter activities, the cytoplasmic region was defined by a ring with an inner radius of 2 µm and an outer radius of 10 µm from the nuclear mask, excluding intersecting cytoplasmic rings of neighboring cells. Background signal was determined as the median pixel intensity of non-masked pixels, excluding those within a nuclear mask dilated by 50 µm. Each fluorescent signal was measured after global background subtraction for all immunofluorescence and fluorescent protein intensities. The nuclear signal was based on nuclear segmentation, while the cytoplasmic signal was calculated from the median signal within a ring outside the nucleus, extending from 0.65 to 3.25 µm beyond the nucleus edge. For mRNA FISH analysis, cells were segmented to include the entire cell regions, enveloping the nucleus and extending up to 50 µm beyond the nuclear mask, avoiding overlap with adjacent cells. Top hat filtering with a 4 µm radius circular kernel was used to create the FISH puncta mask. Intensity thresholds were manually adjusted for each probe and cell line to quantify regions and minimize false detections accurately.

Statistics and reproducibility

Statistical analyses were conducted using GraphPad Prism version 10.2.1. For parametric data, unpaired two-tailed Student’s t-tests and one-way ANOVA with Tukey’s post hoc test were used as appropriate. Statistical significance was defined as a p-value of 0.05 or lower. The results of these statistical tests, along with the number of biological replicates and sample sizes, are reported in the figure legends. Further details are provided in the Supplementary Table 1. All experiments were conducted in at least two independent experiments. No statistical methods were used to predetermine sample size. The experiments were not blinded to allocation during experiments and outcome assessment.

Author contributions

Experimental design, investigation, and visualization: K.K. and H.Y.; Investigation: J.A. and S.K.; writing, supervision, and funding acquisition: H.Y.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

Correspondence and material requests should be addressed to the corresponding authors. Source data are provided with this paper.

Code availability

The imaging analysis code is available at https://github.com/Kim-Yang-Lab/Kim-et-al.−2025-Nature-Comm.-.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-62061-w.