Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 Jun;558(7709):313-317.
doi: 10.1038/s41586-018-0199-7. Epub 2018 Jun 6.

EMI1 switches from being a substrate to an inhibitor of APC/CCDH1 to start the cell cycle

Steven D Cappell  1   2 , Kevin G Mark  3   4 , Damien Garbett  5 , Lindsey R Pack  5 , Michael Rape  3   4 , Tobias Meyer  6
Affiliations

EMI1 switches from being a substrate to an inhibitor of APC/CCDH1 to start the cell cycle

Steven D Cappell et al. Nature. 2018 Jun.

Abstract

Mammalian cells integrate mitogen and stress signalling before the end of G1 phase to determine whether or not they enter the cell cycle1-4. Before cells can replicate their DNA in S phase, they have to activate cyclin-dependent kinases (CDKs), induce an E2F transcription program and inactivate the anaphase-promoting complex (APC/CCDH1, also known as the cyclosome), which is an E3 ubiquitin ligase that contains the co-activator CDH1 (also known as FZR, encoded by FZR1). It was recently shown that stress can return cells to quiescence after CDK2 activation and E2F induction but not after inactivation of APC/CCDH1, which suggests that APC/CCDH1 inactivation is the point of no return for cell-cycle entry 3 . Rapid inactivation of APC/CCDH1 requires early mitotic inhibitor 1 (EMI1)3,5, but the molecular mechanism that controls this cell-cycle commitment step is unknown. Here we show using human cell models that cell-cycle commitment is mediated by an EMI1-APC/CCDH1 dual-negative feedback switch, in which EMI1 is both a substrate and an inhibitor of APC/CCDH1. The inactivation switch triggers a transition between a state with low EMI1 levels and high APC/CCDH1 activity during G1 and a state with high EMI1 levels and low APC/CCDH1 activity during S and G2. Cell-based analysis, in vitro reconstitution and modelling data show that the underlying dual-negative feedback is bistable and represents a robust irreversible switch. Our study suggests that mammalian cells commit to the cell cycle by increasing CDK2 activity and EMI1 mRNA expression to trigger a one-way APC/CCDH1 inactivation switch that is mediated by EMI1 transitioning from acting as a substrate of APC/CCDH1 to being an inhibitor of APC/CCDH1.

PubMed Disclaimer

Conflict of interest statement

Competing financial interests

M.R. is a co-founder and consultant to Nurix, a biotech company in the ubiquitin space.

Figures

Extended Data Figure 1
Extended Data Figure 1. Single-cell APC/CCdh1 data used to determine hysteresis curves
a, Experimental setup for Fig. 1b,d. and Extended Data Fig. 1b. Asynchronous MCF10A cells were tracked and imaged for approximately 18 hours and then treated with either DMSO or the indicated concentration of CDK1/2 inhibitor. Two subpopulations were analyzed: cells in G1 phase (drug added while APC/CCdh1 On) or cells in S/G2 phase (Drug added while APC/CCdh1 Off). b, Single-cell traces of APC/C activity in MCF10A cells treated with DMSO or 30 μM CDK1/2 either in G1 phase (top and middle) or S phase (bottom). Traces were colored red if APC/CCdh1 ended up inactive and traces were colored black if APC/CCdh1 ended up active 5 hours after addition of the drug. (n=200, 135, and 197 cells). Right: histogram of APC/C activity at the last frame of the movie for each cell. Inset: Pie chart of the percent of cells with APC/C still active at the end of the movie. c, Scatter plot of CDK2 activity at the time of drug spike plotted against the CDK2 activity from the same cell 1 hour later. Cells at all phases of the cell cycle were included. d, Top: representative APC/C activity traces for MCF10A cells in G1 phase at time of the indicated treatment. Bottom: representative APC/C activity traces for MCF10A cells in S/G2 phase at time of the indicated treatment. Traces were colored red if APC/CCdh1 was inactive and traces were colored black if APC/CCdh1 was still active 5 hours after addition of the drug. Pie chart shows the percent of cells with APC/CCdh1 still active, which was used to make the dose-response curve in Fig. 1b. e, Table of the EC50 for the dose response curves in Fig. 1b,d +/− SEM. n=3 independent experiments. Note: doses above 30 μM CDK1/2 inhibitor killed the cells making it impossible to calculate an EC50 above 30 μM. f, MCF10A Cells were treated with 10 μM CDK1/2 inhibitor and binned by the APC/CCdh1 activity at the time of drug addition. The percent of cells in each bin with APC/CCdh1 activity on 2.5 hours after drug addition was calculated. Data were analyzed by nonlinear regression (sigmoidal dose-response, variable slope). Error bars are standard deviation. from 2 independent experiments. g, Scatter plot of APC/C activity at the time of drug spike plotted against the APC/C activity from the same cell 2.5 hours later. Vertical dashed line represents the 70% inactive threshold. Horizontal line indicates when APC/C activity is below 0.1 and considered to be “Off”.
Extended Data Figure 2
Extended Data Figure 2. siRNA validation experiments
HeLa (a,b,e) or U2OS (c,d) cells were treated with the indicated siRNA for 24 hrs prior to fixation and immunostaining. Scale bar is 20 μm. Single-cell quantification of immunofluorescence data was used to generate the histograms. a, n=7,152 cells for siControl and n=4,051 cells for siCyclin A2. b, n=7,772 cells for siControl and n=3,916 cells for siSkp2. c, n=7,003 cells for siControl and n=8,589 cells for siCyclin F. d, n=9,393 cells for siControl and n=7,871 cells for siEmi1. e, n=64,574 cells for siControl and n=53,338 cells for siCyclin E1,2.
Extended Data Figure 3
Extended Data Figure 3. Emi1 is responsible for rapid APC/CCdh1 inactivation in multiple cell lines
a,b, Single-cell traces of APC/C activity computationally aligned to when APC/CCdh1 first starts to inactivate in cells treated with the indicated siRNA 6 hours prior to the start of imaging. Black line is median APC/C activity trace. Right: Median APC/C activity trace ± SD. HeLa cells, n=129/602, 126/384, 124/399, 120/228, 138/400 cells per condition respectively (representative traces plotted/all traces used to calculate median) (a), MCF10A cells, n=139/400, 130/400, 134/400 cells per condition respectively (b). c, Single-cell traces of APC/C activity computationally aligned to mitosis in U2OS cells treated with the indicated siRNA 6 hours prior to the start of imaging. Black line is median APC/C activity trace. Traces in U2OS cells had to be aligned by mitosis since APC/CCdh1 did not inactivate when cells were treated with siRNA targeting Emi1. n=200 representative cells per condition out of a total of 2068, 1531, 875 cells respectively. d, Single-cell traces of APC/C activity in MCF10A cells treated Emi1 siRNA. At the indicated time in S phase, 1 μM CDK1/2 inhibitor was spiked in. Traces were colored red if APC/CCdh1 stayed inactive and traces were colored black if APC/CCdh1 turned back on 5 hours after addition of the drug. n=80. Inset: Pie chart of the percent of cells that turned APC/C back on at the end of the movie. e, Schematic diagram of a proposed bistable switch involving Cyclin E1,2/CDK2, APC/C, and Emi1.
Extended Data Figure 4
Extended Data Figure 4. Measuring protein and mRNA dynamics from fixed cells
a, HeLa cells were synchronized with thymidine followed by nocodazole and mitotic shake-off (see Fig. 2a). After shake-off, cells were immediately plated in a 96-well plate for live-cell imaging. Top: single-cell traces of APC/C-degron levels. Black line is the median APC/C-degron level. Bottom: single-cell APC/C activity traces. Dashed-line indicates time when APC/CCdh1 first starts to inactivate. n=200 representative cells out of a total of 3190 cells used to calculate the median. b, Asynchronous HeLa cells were exposed to 10 μM EdU for 15 minutes, fixed, and then stained for both Emi1 and EdU incorporation. Top: density scatter plot of DNA content vs EdU levels, used to gate cells into specific cell cycle phases. (e.g. G1 cells have 2n DNA and low EdU, S phase cells have greater than 2n DNA and high EdU, etc). Cells were designated as Early S phase if they had 2n DNA and sub-maximal EdU, indicating they were in S phase for less than 15 minutes at the time of fixation. Bottom: density scatter plot of DNA content vs Emi1 levels. Since we measured EdU and Emi1 in the exact same cells, cell cycle phase gates were applied to the Emi1 scatter plot. Each cell was colored based on the cell cycle phase that cell was in. Large dot with black boarder marks the median values for each gated population. The single-component histograms for Emi1 levels in each cell cycle phase are shown on the right. c, HeLa Cells were treated as in Fig. 2a, but 2 hours after release, cells were treated with either DMSO or 10 μM MG132. Lysates were collected 2 hours later for a total of 4 hours after release. Image is representative of n=3 independent experiments. d, Schematic diagram of the regulation of 3 key G1 and S phase genes, Geminin, Emi1, and Cyclin E1 and E2, by both transcription and ubiquitination-mediated degradation. e, HeLa cells were imaged and tracked for 20 hrs, and the time each cell went through anaphase relative to the end of the movie was recorded. Cells were then fixed and stained with single molecule mRNA FISH probes for the indicated genes (top), or with antibodies against the indicated proteins (bottom). Data are a scatter plot of the single-cell mRNA or protein levels as a function as the time relative to mitosis. Solid line represents the median levels from 1 hour bins. n=2 independent experiments. f, cells treated as in (e), but prior to fixation cells were incubated with 100 μM EdU for 5 min. Dashed line indicates the first bin when EdU incorporation goes above background, indicating cells have entered S phase. Data is from one representative experiment Right: Single-cell data was binned in 1 hour intervals and the median +/− S.E.M. was plotted. n=6 replicates.
Extended Data Figure 5
Extended Data Figure 5. Emi1 accumulates coincident with APCCdh1 inactivation
a-c, Normalized mRNA and protein levels for the indicated genes. Data from mRNA and protein are plotted on the same x-axis to allow comparison of the kinetics. S phase start determined by EdU incorporation (see Extended Data Fig. 4e). Data is median value for each bin. n=2 independent experiments. d, Density scatter plot of Emi1 protein levels and Geminin protein levels measured in the same single cells. Single-component histograms shown as insets. e, Density scatter plot of Cyclin E1 protein levels and Geminin protein levels measured in the same single cells. Single-component histograms shown as insets. f, HeLa cells were treated with either control siRNA (black line) or Cdh1 siRNA (colored line) and then imaged and tracked for 20 hrs. The time each cell went through anaphase relative to the end of the movie was recorded. Cells were then fixed and immunofluorescence was performed with antibodies to the indicated proteins. The time to reach half-maximal is indicated with a dashed line (t1/2). Data are median values +/− S.E.M and best fit line was calculated using nonlinear regression. Data was normalized to maximum value. n=8 independent experiments. g, Bar graph of the difference in t1/2 times in cells treated with Control siRNA and Cdh1 siRNA from (j). Individual data points are shown as dots. Error bars are S.E.M. p values from one-way ANOVA with Tukey multiple comparisons test. n=8 independent experiments. h,i, in vitro ubiquitination time-course using immuno-purified APC/C and either recombinant Emi1 (h) or Geminin (i). Lower panel is a lighter exposure showing the un-modified substrate. n=1 experiment. j, in vitro ubiquitination assay using immuno-purified APC/CCdh1, recombinant Emi1, and the indicated ubiquitin mutant. E2C/S, E2 enzymes Ube2C and Ube2S. n=1 experiment.
Extended Data Figure 6
Extended Data Figure 6. Emi1 and Geminin protein levels are tightly correlated during the cell cycle
a, Model diagram of APC/CCdh1 architecture. For simplification, the substrate binding region comprised primarily of Cdh1 and APC10 is designated as site 1, and the E2 ubiquitin conjugating enzymes Ubch10 and Ube2s are designated as site 2. c-d, Regulation of APC/C by two Emi1 mutants. Substrates are recruited to APC/C through a substrate binding region (site 1). Ubiquitin is transferred to the substrate by UBCH10 and UBE2S (site 2). Sub, substrate; Ub, ubiquitin. e, Signaling diagram of the G1 cell-cycle program. Shows inhibiting CDK4/6 activity with a small molecule inhibitor will prevent APC/CCdh1 inactivation. f, MCF10A cells expressing either empty vector or various Emi1 variants were treated with DMSO. Single-cell APC/C activity traces were computationally aligned to the time of mitosis. DMSO treatment administered before mitosis as indicated by arrow. N=120 cells for each condition. g, Median levels of APC/C degron (red) or wild-type Emi1 (Emi1WT; green) stably expressed in the same MCF10A cells. n=330 cells. h, Median expression levels of ectopically expressed Emi1 mutants tagged with mCitrine. Measurments for each single-cell were the maximum fluorescence detected during each cell cycle. Single-cell data was then averaged for each replicate. Individual data points are shown as dots. Data are median values +/− S.D. n=3 independent experiments. i, Left: median APC/C degron levels measured over 24 hours in MCF10A cells. Arrows indicate mitosis (mit). n=88,125 cells. Right: median levels of APC/C degron (red) or Emi1C401S (green) stably expressed in the same cells. n=3,618 cells. Inset diagram shows that Emi1C401S is a substrate of APC/CCdh1 and not an inhibitor. j, Single-cell traces of APC/C degron (red) or Emi1C401S (green) levels in MCF10A cells. Six representative cells from (i) are shown. k, Density scatter plot of APC/C degron levels versus Emi1C401S levels measured in the same cells. l, Percent of cells from Fig. 3b-e that inactivated their APC/CCdh1 at least 5 hours after mitosis. n=3 biological repeats. Individual data points are shown as dots. Error bars are S.D.
Extended Data Figure 7
Extended Data Figure 7. Mathematical model for an APC/C-Emi1 dual negative feedback loop
a, Differential equation of the regulation of Emi1 protein concentration. Constant synthesis (S) is assumed for the steady state model. The model includes reversible low affinity binding of Emi1 to the substrate binding site (γ+/−) and inhibitory binding that blocks ubiquitin chain elongation, but not mono-ubiquitination. Emi1 ubiquitination is assumed to be distributive, requiring 5 ubiquitin elongations to become fully ubiquitinated for effective degradation. APC/CCdh1 activity is inhibited by CDK2 activity or by Emi1 binding to the inhibitory site, allowing Emi1 to be subsequently de-ubiquitinated and stabilized. b, A single-cell APC/C activity and CDK2 activity trace. Each steady-state from Fig. 3g is marked. At the indicated time 3 μM CDK1/2 inhibitor was added, but APC/C activity remains off, as predicted in Fig. 3g.
Extended Data Figure 8
Extended Data Figure 8. Emi1 is an APC/C substrate at low concentrations but an inhibitor of ubiquitin chain elongation at high concentrations
a, in vitro ubiquitination reaction of recombinant Geminin using immuno-purified APC/C and a range of concentrations of recombinant Emi1. Non-specific bands are marked with (*). n=1 experiment. b, in vitro ubiquitination time-course using immuno-purified APC/C and 2944 nM recombinant Emi1. Only mono-ubiquitinated Emi1 was observed after 30 minutes. n=1 experiment. c, in vitro ubiquitination assay using immuno-purified APC/CCdc20 from cells arrested in mitosis and either 23 nM or 2944 nM recombinant Emi1. After the in vitro reaction, equal amounts of Emi1 were loaded onto the gel for western blotting. n=1 experiment. d, Quantification of fully-ubiquitinated Emi1 from Fig. 4b using densitometry. Data is plotted as the percent of total Emi1. Best-fit line was determined using the Hill equation. Dashed lines indicate an EC50 of ~175 nM. e, Steady-state analysis of Emi1 levels as a function of the degradation and synthesis rate. Assuming constant Emi1 synthesis (dashed lines), there are two stable steady states (red and blue dots). The system can transition to a mono-stable state with APC/C “Off” and high Emi1 levels by increasing the Emi1 synthesis rate above a threshold level (red arrow). Conversely, the system can transition to a different mono-stable state with APC/C “On” and low Emi1 levels by decreasing the Emi1 synthesis rate (blue arrow). This system also exhibits hysteresis, since the occupied steady-state depends on the previously occupied state (red dots vs blue dots).
Extended Data Figure 9
Extended Data Figure 9. Cyclin E is dispensable for APC/CCdh1 inactivation in cells with constitutive active E2F
a,b, Single-cell traces of APC/C activity computationally aligned to mitosis in either HeLa (n=200 cells per condition out of a total of 1284, 434, 462 cells respectively used to calculate the median and standard deviation) (a) or U2OS (n=200 cells per condition out of a total of 551, 300, 271 cells respectively used to calculate the median and standard deviation) (b) cells treated with the indicated siRNA 6 hours prior to the start of imaging. Black line is median APC/C activity trace. Bottom: Median APC/C activity trace ± SD. c, U2OS cells were treated with either Control, Cyclin E1 and E2, or Emi1 siRNA for 4 hours prior to the start of time-lapse imaging. APC/C and CDK2 activity were measured in the same cells, and traces were aligned to mitosis. Data is median activity trace +/− SEM. n=200 cells. d, Time between anaphase and when relative APC/C activity reaches below 0.2 was calculated for each cell from Fig. 4f. Times from all cells were averaged for each replicate. Individual data points are shown as dots. Data are median times +/− S.E.M. p values from one-way ANOVA with Tukey multiple comparisons test. n=3 independent experiments. e, Single-cell APC/CCdh1 inactivation slope at the G1/S transition for each cell from Fig. 4f. Slopes from all cells were averaged for each replicate. Data are median APC/CCdh1 inactivation slopes +/− S.E.M. p values from one-way ANOVA with Tukey multiple comparisons test. Individual data points are shown as dots. n=3 independent experiments. f, Histogram of the time between CDK2 activation and APC/CCdh1 inactivation in HeLa cells treated with the indicated siRNA. Values above zero indicate CDK2 turns On before APC/CCdh1 turns Off, and values below zero indicate APC/CCdh1 turns Off before CDK2 turns On.
Extended Data Figure 10
Extended Data Figure 10. APC/CCdh1 inactivates prior to CDK2 activation in cells with Cyclin E1,2 knocked down
a,b, Single-cell traces of APC/C (red) and CDK2 (green) activity in HeLa cells treated with either control siRNA (a) or cyclin E1 and E2 siRNA (b). Ten representative cells are shown for each condition. Black arrow, time of mitosis; green dashed line, time CDK2 activity first starts to rise; red dashed line, time APC/C activity first starts to inactivate. c, Phase-plane diagram of median CDK2 and APC/C activity measured simultaneously in the same U2OS cells over time, starting in Mitosis or G0 phase and progressing to S phase. n=270 cells per condition. d, Scheme of the regulation of APC/CCdh1 inactivation in the presence (Wild-type) or absence of Cyclin E. Cells can bypass the requirement for Cyclin E1,2/CDK2 in G1 phase by upregulating Emi1 mRNA. HeLa and USO2 cells bypass Cyclin E1,2/CDK2 by having constitutively active E2F1-3.
Figure 1
Figure 1. Emi1 conveys hysteresis to APC/CCdh1 inactivation
a, Requirements for a bistable switch. b, Dose response curve for the two subpopulations of cells treated with CDK1/2 inhibitor. Data were analyzed by nonlinear regression (sigmoidal dose-response, variable slope). n=3 independent experiments, errobars are S.E.M. c, APC/C activity traces aligned to when APC/CCdh1 inactivates in HeLa cells. Top: Median and single-cell traces of APC/C activity in control cells. Bottom: Median APC/C activity traces. Error bars are SD (n=602, 384, 399, 228, 400 cells respectively). d, Same experimental setup as (b) but MCF10A cells were first treated with Emi1 siRNA. Data were analyzed by nonlinear regression (sigmoidal dose-response, variable slope). n=3 independent experiments, errobars are S.E.M.
Figure 2
Figure 2. The APC/CCdh1 inhibitor Emi1 is also an APC/CCdh1 substrate degraded during G1 and building up at G1/S
a, HeLa cells were synchronized with nocodazole and mitotic shake-off. Image is representative of 3 independent experiments. b, HeLa Cells were treated as in (a), but 1.5hrs after release, cells were treated with either DMSO or 30 μM proTAME, a small-molecule APC/C inhibitor. Lysates were collected 4 hours after release. Image is representative of 2 independent experiments. c,d, Live-cell imaging was combined with fixed cell analysis to measure mRNA (c) or protein and EdU (d) levels over time since mitosis. Median levels from 1 hour bins were plotted. n=2 independent experiments. e, in vitro ubiquitination assay using APC/C from cells in G1 phase and recombinant Emi1 (amino acids 299-447). E2C/S, E2 enzymes Ube2C and Ube2S. Image is representative of 2 independent experiments.
Figure 3
Figure 3. APC/CCdh1 inactivation is a bistable switch driven by dual-negative feedback
a, Emi1 structural elements. D-Box, APC/C degron sequence (RxxL); L, linker; ZBR, zinc-binding region; T, tail. b-e, MCF10A cells expressing either empty vector or various Emi1 mutants were treated with DMSO (black lines; b) or 1 μM CDK4/6 inhibitor (red lines; b-e). Drug treatment administered before mitosis as indicated by arrow. n=120 cells for each condition. Data representative of n=3 independent experiments. f, Two-binding site model for regulation of the APC/C by Emi1. Emi1 can bind either the substrate binding region (1) via its APC/C degron motif or the E2 binding site (2) via its ZBR and C-terminal tail. g, Three steady-state solutions for the mathematical model (see Extended Data Fig. 7a).
Figure 4
Figure 4. Cyclin E/CDK2 activity and Emi1 mRNA synthesis synergistically trigger the Emi1-APC/CCdh1 inactivation switch and cell cycle commitment
a, in vitro ubiquitination assay using immuno-purified APC/CCdh1 and recombinant Emi1. After the in vitro reaction, equal amounts of Emi1 were loaded onto the gel for western blotting. Image representative of n=2 independent experiments. b, same as (a) but using a range of concentrations of recombinant Emi1. Image representative of n=2 independent experiments. c, Schematic showing at low Emi1 concentrations, Emi1 can be poly-ubiquitinated but at high Emi1 concentrations, Emi1 inhibits E2 activity and chain elongation. d, Schematic diagram of a bistable switch involving Cyclin E1,2/CDK2 and a dual-negative feedback loop between APC/CCdh1 and Emi1. e, Simulated changes in APC/CCdh1 activity in G1 and S phase. f, HeLa cells treated with the indicated siRNA. APC/C and CDK2 activity were measured in the same cells, and traces were aligned to mitosis. Data is median activity trace +/− SEM. g, Phase-plane diagram of median CDK2 and APC/C activity measured simultaneously in the same HeLa cells. n=270 cells per condition. h, APC/CCdh1 inactivation at the G1/S transition is a one way-switch, triggered by CDK2 activity and made irreversible by a dual-negative feedback mechanism with Emi1.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Arora M, Moser J, Phadke H, Basha AA, Spencer SL. Endogenous Replication Stress in Mother Cells Leads to Quiescence of Daughter Cells. Cell reports. 2017;19:1351–1364. doi: 10.1016/j.celrep.2017.04.055. - DOI - PMC - PubMed
    1. Barr AR, et al. DNA damage during S-phase mediates the proliferation-quiescence decision in the subsequent G1 via p21 expression. Nature communications. 2017;8:14728. doi: 10.1038/ncomms14728. - DOI - PMC - PubMed
    1. Cappell SD, Chung M, Jaimovich A, Spencer SL, Meyer T. Irreversible APC (Cdh1) Inactivation Underlies the Point of No Return for Cell-Cycle Entry. Cell. 2016;166:167–180. doi: 10.1016/j.cell.2016.05.077. - DOI - PMC - PubMed
    1. Yang HW, Chung M, Kudo T, Meyer T. Competing memories of mitogen and p53 signalling control cell-cycle entry. Nature. 2017;549:404–408. doi: 10.1038/nature23880. - DOI - PMC - PubMed
    1. Hsu JY, Reimann JD, Sorensen CS, Lukas J, Jackson PK. E2F-dependent accumulation of hEmi1 regulates S phase entry by inhibiting APC(Cdh1) Nat Cell Biol. 2002;4:358–366. doi: 10.1038/ncb785. - DOI - PubMed

Publication types