CDK4/6 activity is required during G2 arrest to prevent stress-induced endoreplication

Abstract

Cell cycle events are coordinated by cyclin-dependent kinases (CDKs) to ensure robust cell division. CDK4/6 and CDK2 regulate the growth 1 (G₁) to synthesis (S) phase transition of the cell cycle by responding to mitogen signaling, promoting E2F transcription and inhibition of the anaphase-promoting complex. We found that this mechanism was still required in G₂-arrested cells to prevent cell cycle exit after the S phase. This mechanism revealed a role for CDK4/6 in maintaining the G₂ state, challenging the notion that the cell cycle is irreversible and that cells do not require mitogens after passing the restriction point. Exit from G₂ occurred during ribotoxic stress and was actively mediated by stress-activated protein kinases. Upon relief of stress, a significant fraction of cells underwent a second round of DNA replication that led to whole-genome doubling.

Figure 1.. Ribotoxic stress causes endoreduplication via ZAK-p38 signaling.

(A) Schematic of ZAK pathway. Various stresses cause ribosome collisions which are sensed by the MAP3K ZAK and lead to p38/JNK SAPK signaling. (Aa, amino acid). (B) Experimental timeline for measuring stress-induced WGD. (C) WT and ZAK KO MCF10A cells were treated for 24 hours with anisomycin (150 ng/ml), washed, and cultured for an extra 24 hours to allow cell cycle reentry. DNA content was analyzed as described in methods. Density plots of DNA content and representative images of DAPI stained nuclei are shown. (D) WT and ZAK KO MCF10A cells were treated with anisomycin with or without ZAK inhibitor (PLX4720, 5 μM) or p38i (SB203580, 25 μM) and quantified as in C. The proportion of cells with >4C DNA content is shown. N=3. (E) Western Blot of WT or ZAK KO MCF10A cells treated with anisomycin for 15 min with or without ZAK inhibitor or p38i. (F) MCF10A cells were treated with sorbitol (150 mM) or UV (500 J/m²) and quantified as in C. N=3. (G) MCF10A cells were transduced with a TRE3G:MKK3DD-geminin construct and incubated with or without doxycycline for 24 hours, washed, and cultured for an extra 24 hours. Cells were fixed, stained, and quantified as in C. (H) MCF10A cells expressing EGFP-PCNA and mCherry-Geminin were treated with indicated stresses and imaged for 24 hours. Cells were then washed and imaged for another 24 hours in fresh media, fixed, and stained with DAPI. Representative images are shown for cells undergoing endoreduplication. Red lines indicate DNA content of example cells shown. (I) Schematic of stress-induced endoreduplication compared to a normal cell cycle. Red indicates geminin expression. D Two-way ANOVA with interaction and Tukey-Kramer post-hoc test. F One-way ANOVA with Tukey-Kramer post-hoc test. Data are mean ± s.e.m. Scale bars are 10 μm.

Figure 2.. Ribotoxic stress causes premature APC/C activation in G2 via ZAK-p38 signaling.

(A) Experimental timeline for measuring stress-induced G2 cell cycle exit. (B) MCF10A cells were treated with (right) or without (left) anisomycin (150ng/ml) for 24 hours, fixed and immunostained against the APC/C substrate geminin and DAPI stained. Scatter plots of 1000 single cells are shown. 4C, geminin-negative cells are shown in red. (C) WT or ZAK KO MCF10A cells were treated with anisomycin in the presence or absence of ZAKi (PLX4720, 5 μM) or p38i (SB203580 25 μM) for 24 hours and stained as in B. Proportion of 4C, geminin-negative cells is shown. N=3. (D) MCF10A cells were treated with sorbitol (150 mM) or UV (500 J/m2) and fixed 24 hours later. Cells were stained as in B and quantified as in C. N=3. (E) Primary airway epithelial cells were treated with anisomycin in the presence or absence ZAKi or p38i and fixed 24 hours later. Cells were stained and quantified as in C. N≥3. C, E Two-way ANOVA with interaction and Tukey-Kramer post-hoc test. D One-way ANOVA with Tukey-Kramer post-hoc test. Data are mean ± s.e.m.

Figure 3.. A new KTR to simultaneously measure APC/C^Cdh1 and Cyclin A-CDK2/1 activities with a single fluorescent protein.

(A) Schematic of new kinase translocation reporter (KTR). Briefly, Cdc25B 174-227 domain was cloned upstream of the KTR peptide and the fluorescent protein mClover. KEN box sequence within this domain is highlighted. (B) MCF10A cells expressing H2B-iRFP and Cdc25B^174-227 KTR-mClover were live imaged for 24 hours every 5 minutes. Representative images of a single tracked cell from mitosis to mitosis and corresponding quantifications of cytoplasmic over nuclear (C/N) intensity and expression levels (nuclear median intensity + 6*cytoring median intensity) are shown. C/N ratio is only shown when KTR expression is above half maximal (see methods for details). (C and D) Representative single cells (C) or population averages (D) in silico synchronized at mitosis from data obtained in B. Population median (bold line) and interquartile range (shaded area) are shown for both expression and C/N ratio. Data represents >100 individual cells. C/N ratio is only shown when KTR expression is above half maximal. (E and F) MCF10A cells expressing Cdc25B^174-227 KTR-mClover were transfected with indicated siRNAs and live imaged as in B 24 hours later. Representative images (E) and natural log expression levels of individual cells 100 min after mitosis (F) are shown. (G) MCF10A cells expressing Cdc25B^174-227 KTR-mClover were imaged in response to indicated inhibitors (CDK4/6i, Palbociclib 1μM; CDK2, CVT-313 10μM; CDK1i, RO3306 10μM). C/N ratio was calculated as described in methods. Data represents the average of >100 single cells (bold lines) and the 40th to 60th percentile regions (shaded area). C/N ratio is only shown when KTR expression is above half maximal. (H) MCF10A cells expressing Cdc25B^174-227 KTR-mClover were transfected with indicated siRNAs and imaged 24 hours after. Quantification of C/N ratios of >1000 individual cells is shown. Cells with cytoplasmic translocation of the KTR are shown in red. C/N ratio is only shown when KTR expression is above half maximal. (I) MCF10A cells expressing Cdc25B^174-227 KTR-mClover were fixed and immunostained for Cyclin A as described in methods. A scatter plot of C/N ratio and Cyclin A levels in the same single cells is shown. C/N ratio is only shown when KTR expression is above half maximal. Pearson correlation coefficient is shown. (J) MCF10A cells expressing KEN box mutant Cdc25B^174-227 KTR-mClover and PCNA-mRuby3 were transfected with indicated siRNAs and live imaged 24 hours later. KTR C/N was quantified and in silico synchronized at the end of S phase (dissolution of PCNA puncta). Scale bars are 10 μm.

Figure 4.. SAPK signaling promotes G2 cell cycle exit through sequential CDK inhibition.

(A) MCF10A cells expressing H2B-iRFP, mCherry-CDK4/6 KTR, CDK2 (DHB)-mCerulean3, and ACdC-mClover were live imaged for 24 hours, quantified as described in methods and in silico synchronized at mitosis. Representative images are shown. ACdC C/N ratio is only shown when expression is above half maximal. (B) Experimental timeline for live imaging of stress-induced G2 cell cycle exit dynamics. (C) Cells in A were incubated with media or indicated stress agents (anisomycin 150ng/ml; sorbitol 150 mM; UV 500 J/m²) and live imaged for 24 hours. Fates of cells in S/G2 at the time of stress were manually counted as described in methods. Data represents the average relative frequency of each indicated cell fate (N=3). (D) MCF10A cell cycle reporter cells were treated with anisomycin during live imaging. Representative images of the same individual cell undergoing G2 cell cycle exit are shown at indicated times. Arrows indicate key reporter transitions during the time course and correspond with arrows in E. (E-G) MCF10A cell cycle reporter cells were treated with anisomycin (E), sorbitol (F), or UV (G) and live imaged for 24 hours. Single cell traces of the four indicated cell cycle parameters where obtained, in silico synchronized at the point of APC/C^Cdh1 reactivation (blue arrow in E) and averaged (>100 cells). Data represents average (bold line) and interquartile range (shaded area). ACdC C/N ratio is only shown when expression is above half maximal. (H-J) WT or ZAK KO MCF10A cell cycle reporter cells were treated with anisomycin (dashed line) in the presence or absence of ZAKi and live imaged. Data represents average (bold line) and 40^th-60^th percentile of cells in S/G2 at the time of treatment. ACdC C/N ratio is only shown when expression is above half maximal. (K) MCF10A cells expressing endogenously-tagged mCitrine-p21 were treated with anisomycin in the presence or absence of ZAKi (PLX4720, 5 μM) or p38i (SB203580, 25 μM) and imaged 24 hours later. Fold-change expression level is shown. N=3. (L) WT, p21 KO, or p53 KO MCF10A reporter cells were treated with anisomycin and live imaged to quantify the fraction of S/G2 cells that prematurely exited the cell cycle. N≥3. K Two-way ANOVA with interaction and Tukey-Kramer post-hoc test. L One-way ANOVA with Tukey-Kramer post-hoc test. Data are mean ± s.e.m. Scale bars are 10 μm.

Figure 5.. CDK4/6 inhibition during G2 arrest promotes premature cell cycle exit.

(A) MCF10A reporter cells were incubated with indicated inhibitors (CDK4/6i, Palbociclib 1μM; CDK2i, CVT-313 10μM; CDK1i, RO3306 10μM) individually or in combination and live imaged for 40 hours. S/G2 cell fates were quantified as in figure 4C. N=4. (B) MCF10A, BJ5Ta, and RPE1 cells were transfected with non-targeting or Cyclin A siRNA for 24 hours and then treated with or without CDK4/6i for 24 hours. Cells were fixed, immunostained for geminin, and DAPI stained, and the proportion of 4C, geminin-negative cells is shown. N=9 technical replicates. Data are mean ± s.e.m. (C) WT (top) and p53 KO (bottom) MCF10A cells were transfected with non-targeting or Cyclin A siRNA for 24 hours and then treated with or without CDK4/6i for 24 hours. Cells were then washed, fixed 24 hours later, and DAPI stained. DNA content was analyzed as described in methods. (D) Schematic of experimental hypothesis. Briefly, mitogen signaling and CDK4/6 activity are required for pocket protein phosphorylation in G1. During S/G2, CDK2/1 are reported to maintain pocket protein phosphorylation. During G2 arrest, CDK2/1 activity is inhibited and cells may once again require mitogen signaling and CDK4/6 to maintain pocket protein phosphorylation and prevent APC/C^Cdh1 reactivation. (E) MCF10A cells were treated as in B and fixed, immunostained for phospho-Rb, and DAPI stained. Scatterplots of 1000 single cells are shown. 4C, phospho-Rb-negative cells are shown in red. (F) MCF10A cells were treated with or without Cyclin A siRNA (24 hours prior) or CDK1i and with or without CDK4/6i for 24 hours. Cells were fixed, immunostained for phospho-Rb, and DAPI stained. 4C, phospho-Rb-negative cells are shown in red. (G) MCF10A cells were treated as in F and immunostained for geminin. 4C, geminin-negative cells are shown in red.

Figure 6.. Ribotoxic stress induces G2 cell cycle exit via loss of Cyclin A.

(A) MCF10A cells were treated with (bottom) or without (top) anisomycin (150 ng/ml) for 24 hours, fixed and immunostained against phospho-Rb (left), Cyclin A (middle), or geminin (right) and DAPI stained. (B) MCF10A cells expressing H2B-iRFP, mCherry-geminin, and endogenously-tagged mNeonGreen-Cyclin A were live imaged for 24 hours. Single cell traces were obtained, in silico synchronized at mitosis, and averaged. Data represents average (bold line) and interquartile range. (C) Cells from B were treated with anisomycin and live imaged for 24 hours. Single cell traces were obtained, in silico synchronized at the point of geminin degradation, and averaged. Data represents average (bold line) and interquartile range. (D) Indicated derivatives of our MCF10A reporter cells (WT, Cyclin D1 overexpression, SV40 LTag, and Cyclin A2 overexpression) were treated with anisomycin and live imaged to quantify the fraction of S/G2 cells that undergo premature cell cycle exit. N≥3. (E) Schematic of our working model. During prolonged G2 arrest, mitogen signaling and CDK4/6 activity are required to maintain pocket protein phosphorylation and APC^Cdh1 inhibition (left). If CDK4/6 activity is inhibited, cells will exit from G2 to a G0/G1-like state (right) D One-way ANOVA with Tukey-Kramer post-hoc test. Data are mean ± s.e.m.