Mammalian cell growth dynamics in mitosis
Abstract
The extent and dynamics of animal cell biomass accumulation during mitosis are unknown, primarily because growth has not been quantified with sufficient precision and temporal resolution. Using the suspended microchannel resonator and protein synthesis assays, we quantify mass accumulation and translation rates between mitotic stages on a single-cell level. For various animal cell types, growth rates in prophase are commensurate with or higher than interphase growth rates. Growth is only stopped as cells approach metaphase-to-anaphase transition and growth resumes in late cytokinesis. Mitotic arrests stop growth independently of arresting mechanism. For mouse lymphoblast cells, growth in prophase is promoted by CDK1 through increased phosphorylation of 4E-BP1 and cap-dependent protein synthesis. Inhibition of CDK1-driven mitotic translation reduces daughter cell growth. Overall, our measurements counter the traditional dogma that growth during mitosis is negligible and provide insight into antimitotic cancer chemotherapies.
https://doi.org/10.7554/eLife.44700.001Introduction
Animal cell growth, that is biomass accumulation (Lloyd, 2013), is classically viewed to take place during interphase. During mitosis and cytokinesis, when cells are assumed to prioritize their energy usage for executing cell division, growth is presumed to be minimal (reviewed in Kronja and Orr-Weaver, 2011; Pyronnet and Sonenberg, 2001; Salazar-Roa and Malumbres, 2017; Sivan and Elroy-Stein, 2008; White-Gilbertson et al., 2009). More specifically, mRNA synthesis is inhibited due to chromatin condensation and dissociation of transcription factors (Liang et al., 2015; Novais-Cruz et al., 2018; Parsons and Spencer, 1997; Prescott and Bender, 1962), and ribosomal RNA synthesis is blocked as the nucleolus disappears in prometaphase (Hernandez-Verdun, 2011). Protein synthesis has also been reported to be suppressed in cell populations enriched for mitosis (Bonneau and Sonenberg, 1987; Celis et al., 1990; Fan and Penman, 1970; Prescott and Bender, 1962; Pyronnet et al., 2001; Sivan et al., 2011; Sivan et al., 2007). Consistently, polysome and ribosome profiling studies have suggested that the translational efficiency of most mRNAs is reduced in mitosis (Park et al., 2016; Qin and Sarnow, 2004; Stumpf et al., 2013; Tanenbaum et al., 2015). Studies on individual components of the translational machinery, such as eukaryotic Translation Initiation Factor 4E (eIF4E), eIF4E Binding Protein 1 (4E-BP1), eukaryotic Translation Elongation Factor 2 (eEF2) and S6 Ribosomal Protein (S6RP) have also suggested reduced protein synthesis, especially cap-dependent translation initiation, in mitosis (Celis et al., 1990; Dobrikov et al., 2014; Pyronnet et al., 2001; Shah et al., 2003; Wilker et al., 2007). Furthermore, ribosomes disassociate from endoplasmic reticulum around metaphase, suggesting that translation may be limited in the middle of mitosis (Puhka et al., 2007).
However, this classical view that growth is inhibited during mitosis has recently been challenged. Parts of DNA remain accessible for transcription machinery and de novo transcription of genes involved in cell growth persists in mitosis (Chan et al., 2012; Chen et al., 2005; Liu et al., 2017; Palozola et al., 2017). Recent reports also suggest that protein synthesis may persist during mitosis (Coldwell et al., 2013; Shuda et al., 2015; Stonyte et al., 2018). Importantly, cyclin-dependent kinase 1 (CDK1), the key regulator of mitotic entry and progression (Diril et al., 2012; Gavet and Pines, 2010), phosphorylates and activates components of the protein synthesis machinery, including 4E-BP1 (Heesom et al., 2001; Jansova et al., 2017; Shuda et al., 2015), eEF2 kinase (Smith and Proud, 2008) and p70 S6 kinase (Papst et al., 1998), suggesting an activation of cap-dependent translation. In addition, cap-independent translation of many mRNAs remains active in mitosis (Cornelis et al., 2000; Marash et al., 2008; Pyronnet et al., 2000; Qin and Sarnow, 2004). It is therefore becoming evident that particular proteins, especially those required for completion of cell division and those critical for cell growth, are synthesized during mitosis (Aviner et al., 2013; Aviner et al., 2017; Cornelis et al., 2000; Marash et al., 2008; Park et al., 2016; Pyronnet et al., 2000; Stumpf et al., 2013; Tanenbaum et al., 2015). However, the extent and dynamics of protein synthesis in mitosis remain unclear.
Importantly, protein and RNA synthesis rates are only proxies of overall growth (biomass increase), which is determined by the balance between synthesis (anabolic) and degradation (catabolic) rates (Lloyd, 2013; Miettinen and Björklund, 2015; Miettinen et al., 2017). Overall growth behavior during mitosis has not been studied, primarily due to the lack of precise cell size measurement methods that are sensitive enough to quantify growth during the short mitotic stages. Here, we utilize suspended microchannel resonator (SMR), a high-precision microfluidic mass sensor, and protein synthesis assays in conjunction with cell cycle measurements to study the extent, dynamics, mechanisms and consequences of mitotic growth on a single-cell level.
Results
Animal cells grow during mitosis and cytokinesis
SMR is a microfluidic cantilever that is capable of measuring buoyant mass (a proxy of dry mass, referred to as mass from here on) of single cells with a precision of <0.1 pg (Figure 1a; Figure 1—figure supplement 1a–d) (Burg et al., 2007; Son et al., 2015a; Son et al., 2012). This resolution corresponds to <8 nm (<0.07%) change in a spherical lymphocyte cell diameter. We repeatedly measured mass of the same cell every ~1 min, resulting in a temporal resolution of approximately 2 min according to the Nyquist rate. We quantified growth, more specifically mass accumulation, throughout multiple cell cycles in L1210 mouse lymphocytes without perturbing normal growth rates (Figure 1b) (Son et al., 2015a; Son et al., 2012). We assigned approximate mitotic entry (i.e. G2/M transition), metaphase-to-anaphase transition (i.e. M/A transition) and cytokinetic abscission of the daughter cells for each cell using biophysical properties and FUCCI cell cycle reporter (Figure 1—figure supplement 2; Materials and methods) (Kang et al., 2019). This allowed quantification of mass accumulation during early mitosis (between G2/M transition and M/A transition) and cytokinesis (between M/A transition and daughter cell abscission) on a single-cell level (Figure 1c). In cytokinesis the elongated cells register smaller than round cells in our mass measurements, because of a change in mass distribution (Kang et al., 2019). Correcting for this cell elongation induced bias (correction is applied to all data shown, unless otherwise stated) (see Materials and methods), had little influence of the mass measurements during cytokinesis (Figure 1—figure supplement 1e).

Various animal cell types grow during mitosis and cytokinesis.
(a) Left, schematic of a suspended microchannel resonator (SMR). Single-cell buoyant mass is repeatedly measured as the cell flows back and forth through the vibrating cantilever. Right, at cell division, one of the daughter cells is randomly selected and monitored, while the other daughter cell is discarded from the SMR. (b) Buoyant mass trace of a single L1210 cell and its progeny over five full generations. The interdivision time (~9 hr) for cells growing in the SMR and in normal cell culture condition is equivalent. Blue arrows indicate the abscissions of daughter cells. (c) Overlay of 180 individual L1210 cell buoyant mass traces (transparent orange) and the average trace (black) around mitosis. Each mass trace has been normalized so that the typical cell abscission mass is 2. (d) Mass accumulation in mitosis (before metaphase/anaphase transition, red) and cytokinesis (blue) relative to the total mass accumulated during the cell cycle for various animal cell types Total relative mass accumulation in M-phase (sum of mitosis and cytokinesis) is indicated on top. Note that while the relative mass accumulation in cytokinesis varies between cell types, all cell types display similar mass accumulation % in early mitosis. n refers to the number of individual cells analyzed. Boxplot line: median, box: interquartile range, whiskers: ± 1.5 x interquartile range.
-
Figure 1—source data 1
L1210 buoyant mass measurement data.
- https://doi.org/10.7554/eLife.44700.005
In total, we analyzed 180 individual L1210 cells undergoing mitosis and observed that on average 12% of the total mass accumulated during the whole cell cycle was acquired during M-phase (i.e. during mitosis and cytokinesis) (Figure 1d; Figure 1—source data 1). 7% of total cell growth took place during early mitosis, while 5% took place during cytokinesis. During anaphase, duration of which was estimated based on cell elongation (Materials and methods), mass accumulation was negligible. Considering that in most cell lines M-phase lasts approximately 10% of the whole cell cycle, the 12% mass accumulation observed during M-phase makes M-phase growth comparable to interphase growth.
The extent of total cell growth during mitosis and cytokinesis was surprising. To determine how generalizable this finding is, we repeated our measurements in other animal cell types. Mouse FL5.12 and BaF3 pro-B lymphocytes, and chicken DT40 lymphoblasts grew 9–13% of their total mass in M-phase (Figure 1d). Suspension HeLa (S-HeLa) cells also grew 14% of their total mass during M-phase, validating that substantial growth in M-phase is not specific to lymphocytes. We also examined CD3 +and CD8+activated primary human T cells. Both T-cell subpopulations added approximately 7% of their total mass during M-phase. Thus, growth in mitosis and cytokinesis is an important contributor to the total cellular growth across a variety of cell types grown in suspension.
Cell mass accumulation persists through prophase, stops as cells approach metaphase-to-anaphase transition and recovers during late cytokinesis
To study the dynamics of cell growth during M-phase, we quantified the absolute mass accumulation rates (MAR) before and during M-phase (Figure 2—figure supplement 1; Materials and methods). To account for cell-size-dependent growth rates (Miettinen and Björklund, 2016; Miettinen et al., 2017; Son et al., 2012), we normalized MAR to the mass of the cell (MAR/mass). Surprisingly, after mitotic entry (during approximate prophase) L1210 cells exhibited on average 15.8 ± 3% (mean ± SEM, n = 180) increase in MAR when compared to late G2 phase (Figure 2a,b). As cells approached the metaphase-to-anaphase transition, MAR rapidly decreased and eventually reached zero at the end of metaphase. MAR remained near zero for the approximate duration of anaphase, after which MAR started to recover during late cytokinesis (Figure 2a,c,d). The recovery of MAR continued through the abscission of daughter cells (Figure 2d). These cell growth dynamics also persisted under different nutrient conditions and were reproducible with different SMR devices over multiple years of study (Figure 2—figure supplement 2a,c).

Cell mass accumulation persists through prophase, stops as cells approach metaphase-to-anaphase transition and recovers during late cytokinesis.
(a) Mass-normalized mass accumulation rate (MAR) of L1210 cells in late G2 and M-phase. G2/M and metaphase-to-anaphase transitions are indicated with dashed vertical lines. Typical durations of metaphase and cell elongation (singlet to doublet) are indicated in green and light brown areas, respectively. n refers to number of individual cells analyzed. (b) Quantification of L1210 cell maximal growth rate in late G2 and in mitosis (n = 180 cells). p-Values obtained using two-tailed Welch’s t-test. In boxplots, line: median, box: interquartile range, whiskers: 5–95%. (c) Representative L1210 cell phase contrast (grey) and mAG-hGeminin cell cycle reporter (green) images (n = 18 cells). Times correspond to (a) and (d). Note that the physical separation of daughter cells takes place when cells are not accumulating mass. (d) Examples of individual L1210 mass-normalized MAR traces in late G2, M-phase and early G1. Arrows indicate the final abscission of daughter cells, around which mass-normalized MAR is indicated with dashed lines. M/A denotes the metaphase-to-anaphase transition, G2/M denotes the approximate mitotic entry, both of which are indicated with dashed vertical lines. (e) Mass-normalized MAR of indicated cell types along with representative images displaying the duration of the physical separation of daughter cells. BaF3 and DT40 cells were imaged separately, whereas S-HeLa and CD3 +T cells were imaged on-chip simultaneously with MAR measurements. M/A denotes the metaphase-to-anaphase transition, G2/M denotes the approximate mitotic entry, both of which are indicated with dashed vertical lines. Solid dark blue lines indicate the mean and light blue areas represent ± SD. n refers to number of individual cells analyzed.
We next examined MAR in other cell types. Although increased MAR during early mitosis was only observed in some cell types, MAR remained high after mitotic entry (during approximate prophase) for all the cell types studied (Figure 2e; Figure 2—figure supplement 2b). All cell types displayed rapid reduction of MAR during metaphase (possibly starting in late prometaphase), near zero MAR during anaphase and a recovery of MAR during late cytokinesis. The temporary stop of cell growth in anaphase was consistently short (<15 min) and coincided with the physical separation of the daughter cells (Figure 2c–e). Notably, some cells displayed a negative MAR, indicating a small loss of cell mass during late metaphase and/or early anaphase (Figure 2e; Figure 2—figure supplement 2b). Together, these results indicate that the mitotic growth behavior is conserved across various animal cell types in suspension, suggesting a role for these specific growth dynamics during mitosis.
Mitotic protein synthesis rates are consistent with mitotic MAR dynamics
Proteins constitute approximately 70% of cellular dry mass (Palm and Thompson, 2017), making it likely that the measured MAR dynamics reflect protein synthesis rates. Using L1210 cells as a model, we quantified the dynamics of mitotic protein synthesis using O-propargyl-puromycin (OPP)-based single-cell protein synthesis assays (Liu et al., 2012) together with a mitotic marker (phospho-Histone H3 (Ser10)) (Figure 3a; Figure 3—figure supplement 1a,b; Materials and methods). For unsynchronized cells, the average mitotic protein synthesis rates were 85 ± 6% (mean ± SD, n = 6) of the rate for G2 cells. We then synchronized cells using double thymidine block followed by CDK1 inhibitor (RO-3306)-mediated G2 arrest and a release, which was followed by 10 min OPP labelling at various timepoints. We normalized mitotic protein synthesis rates to G2 protein synthesis rates to avoid cell synchronization induced biases (Figure 3a). Immediately after the release from G2 arrest, when mitotic cells were in prophase, protein synthesis rates were higher in mitotic cells than in G2 cells (Figure 3b). Approximately 30 min later, when most mitotic cells proceeded to cytokinesis, the protein synthesis rates were reduced to below the normal G2 levels.

Mitotic protein synthesis dynamics are consistent with mass accumulation dynamics.
(a) Top, schematic of the protocol for quantifying mitotic protein synthesis rates using O-propargyl-puromycin (OPP). G2 synchronization was achieved by double thymidine block followed by RO-3306 mediated G2 arrest. Bottom, representative FACS scatter plots indicating L2110 cell cycle synchrony (n = 3 independent experiments). Phospho-Histone H3 (Ser10) was used as a mitotic marker. (b) Ratio of protein synthesis rate (blue) between mitotic and G2 L1210 cells after release from G2 arrest. Light green area displays the typical protein synthesis ratio between mitotic and G2 cells in the absence of cell cycle synchronization. The relative portion of mitotic cells is shown in orange. Each data point represents an individual replicate. (n = 3 separate cultures for each timepoint). Time of G2 release and the typical time to reach metaphase-to-anaphase transition are indicated with dashed vertical lines. (c) Representative FACS scatter plot indicating the separation of early (prophase) and late (metaphase to telophase) mitotic L1210 cells using Cyclin A antibody staining. (d) Protein synthesis rate of G2, early mitotic and late mitotic L1210 cells. (n = 6 separate cultures). Early and late mitotic cells were separated as shown in (g). p-Values obtained using ANOVA followed by Tukey’s posthoc test.
To validate that the observed protein synthesis dynamics are not an artefact of cell synchronization, we utilized a previously developed approach where Cyclin A, which is degraded during prometaphase (den Elzen and Pines, 2001), is used to separate early (prophase) and late (metaphase to telophase) mitotic cells (Ly et al., 2017) (Figure 3c). In the absence of cell cycle synchronization, OPP-based protein synthesis assay indicated that early mitotic cells have higher protein synthesis rates than G2 cells, whereas in late mitotic cells, protein synthesis is reduced (Figure 3d). We also separated G2, early mitosis and late mitosis based on cyclin B1, which is degraded at metaphase-to-anaphase checkpoint (Figure 3—figure supplement 1a). This approach also revealed that protein synthesis rates remain high in early mitosis but not in late mitosis (Figure 3—figure supplement 1c). Although protein synthesis assays do not have temporal resolution required for separation of all mitotic stages, the protein synthesis dynamics we observe correspond to those observed with MAR measurements. In conclusion, L1210 cells display increased growth in early mitosis and radically reduced growth in metaphase and anaphase.
Mitotic arrests, including antimitotic chemotherapies, inhibit cell growth
Our results show that growth is inhibited from metaphase (or late prometaphase) to the end of anaphase (Figure 2a). As many studies examine mitotic growth by arresting cells to mitosis, and this has been suggested to reduce cell growth (Coldwell et al., 2013; Sivan et al., 2011; Stonyte et al., 2018), we measured the effect of chemically induced mitotic arrest on cell growth. First, we monitored the MAR of L1210 cells treated with kinesin inhibitor S-trityl-l-cysteine (STLC), which arrests cells in prometaphase state. These cells displayed a growth burst in early mitosis, similarly to untreated cells, after which MAR approached zero over the course of 2–3 hr (Figure 4a). We also repeated our mitotic protein synthesis assay (Figure 3a) in the presence of STLC. Similarly to MAR, protein synthesis rates increased after mitotic entry, but then gradually decreased as cells were arrested in mitosis (Figure 4b).

Mitotic arrests result in growth inhibition independently of the mechanism of arrest.
(a) Mass-normalized MAR of 5 µM STLC or 1 mg/ml Nocodazole treated L1210 cells in late G2 and mitosis. Dashed vertical line indicates the approximate mitotic entry. Solid dark lines indicate the mean and light areas represent ± SD. n refers to number of individual cells analyzed. Drug treatments started 1–4 hr prior to mitotic entry and were maintained through the experiment. (b) Ratio of protein synthesis rate (blue) between mitotic and G2 L1210 cells after release from G2 arrest in to 5-µM STLC-mediated mitotic arrest. Light green area displays the typical protein synthesis ratio between mitotic and G2 cells in the absence of cell cycle synchronization. The relative portion of mitotic cells is shown in orange. Each data point represents an individual replicate (n = 3 separate cultures for each timepoint). Cells were synchronized to G2 as in (Figure 3a). (c) Protein synthesis rates of G2 (blue) and mitotic (red) L1210 cells after 4 hr treatment with indicated mitotic inhibitors. The proportion of mitotic cells relative to control is indicated below (mean ± SD). (n = 4 separate cultures). p-Values obtained using ANOVA followed by Tukey’s posthoc test. (d) Mass-normalized MAR of 10 nM Vinblastine or 10 nM Vincristine-treated L1210 cells in late G2 and mitosis. Dashed vertical line indicates the approximate mitotic entry. Solid dark lines indicate the mean and light areas represent ± SD. n refers to number of individual cells analyzed. Drug treatments started 1–4 hr prior to mitotic entry and were maintained through the experiment.
To separate drug-specific effects on growth from those that reflect mitotic arrest, we tested three additional chemical approaches for arresting cells in mitosis. These were microtubule inhibitor nocodazole, proteasome inhibitor MG-132 and Anaphase-Promoting Complex inhibitor proTAME (Zeng et al., 2010). All these chemicals resulted in similar reduction in the overall mitotic protein synthesis (Figure 4c). None of these chemicals caused protein synthesis to be significantly reduced in G2, except for MG-132. In addition, nocodazole treatment resulted in identical MAR behavior as STLC (Figure 4a).
Mitotic arrest is the mechanism of action for many chemotherapy drugs. We examined how clinically relevant concentrations of the chemotherapy drugs Vinblastine and Vincristine (Florian and Mitchison, 2016) affect cell MAR. Neither of the drugs affected cell growth in G2, but as cells were arrested in mitosis, their growth rate reduced to zero (Figure 4d). Thus, mitotic arrests, including antimitotic chemotherapies, stop cell growth independently of the arresting mechanism.
Cells in metaphase and anaphase display mitotic stage specific inhibition of mass accumulation
Next, we studied if metaphase and anaphase, where MAR was near zero, have a growth reducing mechanism(s) that is not active earlier in mitosis. We first considered the role of mitotic deswelling. During mitosis cells round up and increase their volume by approximately 10–20%, before shrinking (deswelling) back to their original volume during anaphase (Son et al., 2015a; Zlotek-Zlotkiewicz et al., 2015). While mitotic rounding has minimal influence on the cell types used in this study, as the suspension cells display a spherical morphology throughout the cell cycle, inhibition of mitotic cell swelling removed the MAR increase seen in early mitosis but did not affect MAR in metaphase and anaphase (Figure 5—figure supplement 1a–c). Furthermore, inhibition of mitotic swelling did not influence early mitotic protein synthesis rates (Figure 5—figure supplement 1d). Thus, while mitotic swelling influences the MAR observed in early mitosis, possibly by increasing cell mass due to the uptake of ions, this does not explain why cells suddenly stop growing around metaphase-to-anaphase transition.
Next, we examined if cytokinetic cell elongation is required for the near zero (or even negative) MAR around metaphase-to-anaphase transition. We treated L1210 cells with Tozasertib, an Aurora kinase inhibitor, which blocks cytokinesis but not mitosis as evident from the loss of Geminin (Figure 5a). Tozasertib-treated cells displayed low MAR around metaphase-to-anaphase transition, although MAR remained higher than in control cells (Figure 5b; Figure 5—figure supplement 2a). We then treated cells with blebbistatin, a myosin motor inhibitor which also blocks cytokinesis (Atilla-Gokcumen et al., 2010). Blebbistatin treatment resulted in MAR dynamics comparable to control cells (Figure 5—figure supplement 2b). In addition, both Tozasertib and blebbistatin prolonged the duration of early mitosis. Together, these data indicate that the physical separation of daughter cells does not explain the observed MAR dynamics.

Cells in metaphase and anaphase display stage-specific mass accumulation regulation independently of cell elongation.
(a) Representative L1210 cell phase contrast (grey) and mAG-hGeminin cell cycle reporter (green) images in control (n = 9 cells) and 200 nM Tozasertib (n = 6 cells) treated cells. The degradation of mAG-hGeminin indicates metaphase-to-anaphase transition. No cytokinesis takes place under Tozasertib treatment. (b) Mass-normalized MAR of control (blue) and 200 nM Tozasertib (orange) treated L1210 cells. Note that Tozasertib prolonged early mitosis, but most cells still displayed increased MAR after G2/M transition (see Figure 5—figure supplement 2a). Dashed vertical line indicates the metaphase-to-anaphase transition. Solid dark lines indicate the mean and light areas represent ± SD. n refers to number of individual cells analyzed. Arrows reflect typical time of G2/M transition for each sample. Drug treatment started 1–4 hr prior to mitotic entry and was maintained through the experiment. (c, d) Mass-normalized MAR of control (blue) and 5 µM STLC (red) treated L1210 cells (f). Dashed vertical lines indicate the approximate mitotic entry (for both samples) and metaphase-to-anaphase transition (only applies to control). Solid dark lines indicate the mean and light areas represent ± SD. Arrows indicate time points from which data was extracted to generate the boxplot in (g). n refers to number of individual cells analyzed. p-Values were obtained using ANOVA followed by Tukey’s posthoc test. In all boxplots, line: median, box: interquartile range, whiskers: 5–95%. See Materials and methods for details on MAR analysis resolution for this figure.
The radical reduction in MAR observed as cells approach metaphase-to-anaphase transition could be explained by two separate mechanisms: First, growth may be reduced as a function of time after mitotic entry, or possibly after an initial delay in growth reduction. This hypothesis is supported by the gradual decrease in growth rates following mitotic arrests (Figure 4), possibly because of the inhibition of transcription as DNA condenses. Second, there may be a separate growth inhibiting mechanism(s) specific to metaphase and anaphase. To separate these two options, we compared the MAR as a function of time from mitotic entry in control cells and cells arrested in prometaphase state using STLC (Figure 5c). 20 min after mitotic entry, when cells are in late prometaphase, both samples displayed similar growth rates (Figure 5d). However, 28 min and 32 min after mitotic entry, when control cells had proceeded to metaphase and anaphase, respectively, but STLC-treated cells remained arrested in prometaphase, the control cells displayed lower MAR. Similar results were obtained when prometaphase arrest was achieved using Nocodazole (Figure 5—figure supplement 2c). In conclusion, the mitotic morphological changes (Figure 5a,b; Figure 5—figure supplement 1; Figure 5—figure supplement 2a,b) and the time that cells have spent in mitosis cannot fully explain the observed MAR in metaphase and anaphase. Thus, additional MAR reducing mechanism(s) must exist around metaphase-to-anaphase transition.
Mitotic growth does not require mTOR activity
We then investigated what signaling promotes growth and protein synthesis in mitosis (Figure 6a). We measured the levels of phosphorylated 4E-BP1 (Thr37/46) and S6RP (Ser235/236) at the single-cell level in mitotic and G2 L1210 cells. Mechanistic Target Of Rapamycin (mTOR) regulates both of these proteins, which in turn control translation (Fingar et al., 2004). Importantly, 4E-BP1 is a negative regulator of translation and is inactivated by phosphorylation on several sites, including Thr37/46 (reviewed by Qin et al., 2016). Phosphorylated 4E-BP1 levels were approximately threefold higher in mitosis than in G2 (Figure 6b; Figure 6—figure supplement 1), whereas phosphorylated S6RP displayed only a minor increase in mitosis (Figure 6—figure supplement 2a). The mitotic phosphorylation of 4E-BP1 was validated with an independent antibody using microscopy and the mitotic increase was not observed when using antibody isotype controls or pretreating the sample using Lambda protein phosphatase (Figure 6—figure supplement 1). The mitotic phosphorylation of 4E-BP1 is also consistent with previous reports (Shuda et al., 2015) and the observation that the translational targets of mTOR are actively translated during mitosis (Park et al., 2016).

CDK1 drives mitotic growth through 4E-BP1 and cap-dependent protein synthesis.
(a) Schematic of growth regulation pathways. Chemical and genetic inhibitors (red), kinases (yellow) and the measured downstream consequences (green) are shown. 1NM-PP1-mediated inhibition of CDK1 is dependent on kinase mutation. (b) L1210 cell levels of phosphorylated 4E-BP1 (Thr37/46) in G2 (blue) and mitosis (red) after 2 hr treatment with 250 nM TORIN-1, 1 µM RO-3306 or 50 nM OTSSP167. (n = 5–6 separate cultures). (c) Protein synthesis rates of G2 (blue) and mitotic (red) L1210 cells after 2 hr treatment with 1 µM RO-3306 or 50 nM OTSSP167. (n = 6 separate cultures). (d) Mass-normalized MAR of control, 1 µM RO-3306 or 30 nM OTSSP167-treated L1210 cells. Solid dark lines indicate the mean and light areas represent ± SEM. Arrows reflect typical time of G2/M transition for each sample. n refers to number of individual cells analyzed. Drug treatments started 1–4 hr prior to mitotic entry and were maintained through the experiment. (e) DT40 CDK1as cell protein synthesis rates in G2 (blue) and mitosis (red) after 5 hr treatment with 1NM-PP1. (n = 5–8 separate cultures). (f) Mass-normalized MAR of control or 200 nM 1NM-PP1-treated DT40 CDK1as cells. Solid dark lines indicate the mean and light areas represent ± SEM. Arrows reflect typical time of G2/M transition for each sample. n refers to number of individual cells analyzed. Drug treatments started 1–4 hr prior to mitotic entry and were maintained through the experiment. (g) Protein synthesis rates of G2 (blue) and mitotic (red) L1210 cells expressing scrambled or 4E-BP1 targeting shRNAs. The cells were treated for 2 hr with 1 µM RO-3306 before sample preparation. (n = 6 separate cultures). (h) Protein synthesis rates of G2 (blue) and mitotic (red) L1210 cells after 2 hr treatment with 1 µM RO-3306, 5 hr treatment with 50 µM 4EGI-1 or combined treatment with RO-3306 and 4EGI-1. (n = 6–8 separate cultures). All p-Values were obtained using ANOVA followed by Tukey’s posthoc test.
To examine the role of mTOR, we treated cells for 2 hr with 250 nM mTOR inhibitor TORIN-1. In G2 cells, the levels of phosphorylated 4E-BP1 (Thr37/46) and S6RP (Ser235/236) were near zero (Figure 6b; Figure 6—figure supplement 2b). In mitosis, phosphorylation of S6RP was reduced by TORIN-1 (Figure 6—figure supplement 2b), indicating that mTOR remains active in mitosis. However, TORIN-1 did not change the levels of phosphorylated 4E-BP1 in mitosis (Figure 6b), suggesting that an mTOR-independent mechanism activates 4E-BP1 in mitosis. Next, we measured mitotic protein synthesis rates in the presence of TORIN-1. Although G2 protein synthesis rates were reduced, mitotic protein synthesis rates were not affected by TORIN-1 (Figure 6—figure supplement 2c). Thus, mTOR is not a major contributor to mitotic growth.
CDK1 drives phosphorylation of 4E-BP1, protein synthesis and mass accumulation in mitosis
We examined how the levels of phosphorylated 4E-BP1 (Thr37/46) and S6RP (Ser235/236) change when CDK1 is inhibited with 1 µM RO-3306 (Vassilev et al., 2006). At this RO-3306 concentration, L1210 cells can still progress through mitosis, as CDK1 is only partially inhibited (Son et al., 2015a). Only in mitosis did RO-3306 reduce phosphorylated 4E-BP1 but not phosphorylated S6RP (Figure 6b; Figure 6—figure supplement 2b). Although the role of CDK1 in controlling 4E-BP1 phosphorylation has been reported before (Heesom et al., 2001; Shuda et al., 2015), the consequences of this on protein synthesis and cell growth remain unknown. We observed that 1 µM RO-3306 treatment also reduced mitotic protein synthesis rates without affecting G2 protein synthesis rates (Figure 6c). In addition, RO-3306 treatment prolonged early mitosis and resulted in a clear reduction of MAR in mitosis but not in G2 (Figure 6d).
To further validate the role of CDK1 in promoting mitotic growth, we utilized chemical genetics in the DT40 CDK1as cell line. In these cells, the wild-type CDK1 has been replaced with Xenopus laevis CDK1 that has a F80G mutation (Gibcus et al., 2018), which sensitizes CDK1 to inhibition by the ATP analog 1NM-PP1 (Hochegger et al., 2007). We observed that protein synthesis was reduced by 1NM-PP1 in a dose-dependent manner in mitosis but not in G2 (Figure 6e). Consistently, 1NM-PP1 also reduced MAR in mitosis (Figure 6f).
We also investigated the role of other mitotic kinases in promoting protein synthesis and cell growth. Inhibitors for CDK2, Aurora kinases and DYRK kinases did not affect mitotic protein synthesis (Figure 6—figure supplement 3a). OTSSP167, a drug designated as a MELK inhibitor (Chung et al., 2012), reduced phosphorylation of 4E-BP1, protein synthesis and mass accumulation rates in mitosis (Figure 6b–d). Consistently, MELK has been suggested to promote translation in mitosis (Wang et al., 2016). However, alternative MELK inhibitors (Klaeger et al., 2017) did not display mitosis-specific inhibition of protein synthesis (Figure 6—figure supplement 3b,c), suggesting that the mitotic growth effects observed in OTSSP167-treated cells were not mediated by MELK but by OTSSP167 off-targets.
We then moved to validate that 4E-BP1 mediates the CDK1 driven protein synthesis. We generated two L1210 cell lines expressing 4E-BP1 targeting shRNAs, which reduced 4E-BP1 levels by approximately 85% (shRNA #1) and by 50% (shRNA #2) (Figure 6—figure supplement 4a,b). The 4E-BP1 knockdowns had little effect on proliferation rate (Figure 6—figure supplement 4c) or G2 cell protein synthesis (Figure 6g), possibly reflecting that 4E-BP1 is kept mostly inactive under our experimental conditions. However, when we reduced mitotic protein synthesis by inhibiting CDK1 with RO-3306, the knockdowns of 4E-BP1 partially rescued the mitotic protein synthesis inhibition (Figure 6g). These data indicate that CDK1 promotes mitotic growth at least partly through 4E-BP1.
4E-BP1-driven cap-dependent protein synthesis is classically considered to be inhibited in mitosis (Bonneau and Sonenberg, 1987; Pyronnet et al., 2001), although recently this view has been challenged (Coldwell et al., 2013; Shuda et al., 2015). We therefore tested if i) cap-dependent translation remains active in L1210 cell mitosis, and if ii) cap-dependent translation is required for CDK1-mediated mitotic growth. To test these, we first inhibited cap-dependent translation using 4EGI-1, an inhibitor of eIF4F complex assembly (Cencic et al., 2011). Both G2 and mitotic protein synthesis rates were reduced by a similar amount, approximately 50%, following 4EGI-1 treatment (Figure 6h). However, treatment with both 4EGI-1 and RO-3306 did not significantly change mitotic protein synthesis rates when compared to treatment with 4EGI-1 alone, suggesting that cap-dependent protein synthesis is involved in CDK1-driven mitotic translation. In conclusion, our results are consistent with a previous report (Shuda et al., 2015) that CDK1 substitutes for mTOR in mitosis to promote phosphorylation of 4E-BP1, to maintain cap-dependent translation and to promote mass accumulation.
CDK1-driven mitotic protein synthesis supports daughter cell growth
Mitotic transcription and translation have been suggested to be geared toward ribosomal proteins and other components that promote growth (Aviner et al., 2017; Palozola et al., 2017). Therefore, inhibiting mitotic protein synthesis could also impact growth in cytokinesis and in daughter cells. We compared the MAR of RO-3306 or OTSSP167-treated L1210 cells to control cells in late G2 (last 30 min before G2/M transition), early mitosis (before metaphase-to-anaphase transition), cytokinesis (after metaphase-to-anaphase transition) and newborn G1 (first 30 min after abscission of daughter cells). In the presence of the mitotic growth inhibitors, G2 MAR was not affected, but MAR remained low in cytokinesis and in newborn G1 (Figure 7a). In contrast, cells treated with 100 nM cycloheximide, a translation inhibitor which at this concentration reduces total protein synthesis by approximately 50% (Figure 3—figure supplement 1d), did not display similar cell cycle specificity in growth inhibition (Figure 7a). In addition, mother cell MAR in early mitosis and daughter cell MAR in early G1 correlated in both control (R2 = 0.42) and RO-3306 (R2 = 0.33) treated cells (Figure 7—figure supplement 1). 1NM-PP1-mediated inhibition of CDK1 in the DT40 CDK1as cell line also resulted in reduced growth during cytokinesis and in daughter cells (Figure 7b).

CDK1-driven mitotic protein synthesis supports daughter cell growth.
(a) MAR during indicated stages of cell cycle in control, 1 µM RO-3306, 30 nM OTSSP167 or 100 nM cycloheximide (CHX) treated L1210 cells. The MAR values were normalized to control mean at each stage. n refers to number of individual cells analyzed. Drug treatments started 1–4 hr prior to mitotic entry and were maintained through the experiment. (b) MAR during indicated stages of cell cycle in control or 200 nM 1NM-PP1-treated DT40 CDK1as cells. The MAR values were normalized to control mean at each stage. n refers to number of individual cells analyzed. Drug treatment started 1–4 hr prior to mitotic entry and was maintained through the experiment. (c and d) MAR of newborn L1210 G1 cells from control and from cells that have undergone mitosis in the presence of 1 µM RO-3306 (c) or from cells that have been arrested to mitosis for 4 hr with STLC before releasing to undergo cytokinesis (d). Data acquired using serial SMR (see Figure 7—figure supplement 2 for details). n refers to number of individual cells analyzed. (e) Protein synthesis rates of G1 L1210 cells expressing a scrambled or 4E-BP1 targeting shRNA after the cells progressed through mitosis in the presence or absence of 1 µM RO-3306. Two timepoints (3 hr and 8 hr) after G2 release are shown. (n = 4 separate cultures). (f) Long-term growth, as measured by cell confluency, in L1210 cells that have been arrested to mitosis for 4 hr with STLC or MG132 before releasing to undergo cytokinesis (n = 5 separate cultures, top), or have undergone mitosis in the presence of 1 µM RO-3306 (n = 6 separate cultures, bottom). Dark colors indicate mean and light areas indicate ± SEM. In (a) and (e), p-values obtained using ANOVA followed by Tukey’s posthoc test. In (b–d), p-values obtained using two-tailed Welch’s t-test. In boxplots, line: median, box: interquartile range, whiskers: 5–95%.
We speculated that the daughter cell growth inhibition is a consequence of growth reduction in mother cell mitosis. To validate that mitotic growth is needed to promote daughter cell growth, we synchronized L1210 cells to G2 phase, released them back to cell cycle in the presence or absence of mitotic growth inhibitions, and carried out daughter cell growth measurements (see Figure 7—figure supplement 2a for workflow). First, we measured single-cell MAR using a serial SMR, which measures MAR over a 15–30 min window (Calistri et al., 2018; Cermak et al., 2016) (Figure 7—figure supplement 2b). We identified the newborn G1 cells based on their smaller mass and quantified their MAR (Figure 7—figure supplement 2c). In addition, we quantified protein synthesis rates of G1 daughter cells, as identified through DNA staining, and also measured long-term proliferation of the cells. Following mitotic growth inhibitions, either by temporary mitotic arrest (4 hr STLC treatment) or by partial CDK1 inhibition, newborn cells displayed significantly reduced MAR and protein synthesis rates (Figure 7c,d; Figure 7—figure supplement 3). We then repeated the daughter cell protein synthesis assays in 4E-BP1 knockdown cells after the mother cells had progressed through mitosis in the presence or absence of CDK1 inhibition. 4E-BP1 knockdown partly rescued the daughter cell protein synthesis rates that were reduced by CDK1 inhibition (Figure 7e). Thus, 4E-BP1 mediates mitotic protein synthesis of CDK1 (Figure 6), and this may also affect daughter cell protein synthesis. However, we cannot fully exclude the possibility that daughter cell growth is affected by some other effects of partial CDK1 inhibition, such as chromosome missegregation, which could consequently reduce daughter cell growth independently of mitotic growth. The protein synthesis rate and long-term proliferation rate of the daughter cells recovered over time (Figure 7e,f), suggesting that mitotic growth inhibitions do not permanently affect cell viability.
Discussion
We show that animal cells grow approximately 10% of their total mass during M-phase (Figure 1), indicating that the growth during M-phase is comparable to interphase when the short duration of M-phase is taken into consideration. This contradicts the classical dogma that growth takes place primarily during interphase and not during M-phase. Our single-cell measurements show that mass accumulation behavior during mitosis is dynamic (Figure 2), dependent on mitotic stage (Figure 5), conserved across a variety of animal cell types grown in suspension (Figure 2) and reflected in protein synthesis rates (Figure 3). Importantly, growth is only stopped for a short duration in mitosis, as cells approach the metaphase-to-anaphase transition, and growth recovers in late cytokinesis after anaphase. Most studies on translation and growth signaling during mitosis use cell-population-based experimental approaches that require population enrichment for mitotic cells. This is commonly achieved by mitotic blockades, which result in growth inhibition (Figure 4). Alternatively, mitotic cells can be collected with mitotic shake-off. However, this only enriches for the small subset of mitotic cells that are temporarily not growing (i.e. metaphase and anaphase cells). Thus, many of the controversial reports regarding translational control during mitosis can be explained by experimental approaches.
Conceptually, the tight and conserved coordination between growth rates and mitotic stages suggests that mitotic growth control is important for cell division. It has been thought that prioritization of energy away from the ATP consuming macromolecule synthesis and towards mechanical reorganization of the cell could explain the reduction in mitotic protein synthesis rates (Kronja and Orr-Weaver, 2011; Sivan and Elroy-Stein, 2008; White-Gilbertson et al., 2009). Consistent with this hypothesis, we show that growth is stopped when the physical separation of daughter cells takes place (Figure 2). Yet, there is no direct evidence that cell division would increase the energetic needs of a cell to a point where growth could not persist, and the inhibition of growth may therefore be due to other reasons. For example, the inhibition of growth may protect cells from harmfully excessive cell growth during prolonged mitosis (Miettinen and Björklund, 2016; Miettinen et al., 2017; Neurohr et al., 2019).
It has also been proposed that growth, and especially protein synthesis, are required during mitosis. A complete growth inhibition for the duration of M-phase would result in loss of short-lived proteins needed for mitosis and cytokinesis (White-Gilbertson et al., 2009). Consistently, we observe that growth is not inhibited during prophase or late cytokinesis, suggesting that short-lived proteins, such as survivin (White-Gilbertson et al., 2009), can be produced immediately prior to and after anaphase. Future studies examining cellular energy production and mapping the levels of short-lived proteins in different mitotic stages will help elucidate why cells display such a dynamic growth behavior in mitosis.
Mechanistically, we found that CDK1 promotes mitotic mass accumulation and protein synthesis, and this is at least partly mediated by 4E-BP1 (Figure 6). Our results are consistent with previously reported interactions between CDK1 and the translational machinery (Heesom et al., 2001; Shuda et al., 2015; Smith and Proud, 2008). Importantly, our results do not exclude the existence of other CDK1 dependent or independent mechanisms regulating mitotic growth. Indeed, CDK1 remains active until metaphase-to-anaphase transition (Gavet and Pines, 2010), whereas mass accumulation rates begin to decay in late prometaphase and metaphase, indicating that CDK1 alone cannot fully explain the observed mass accumulation dynamics.
Mitotic growth may be dependent on the time that cells have spent in mitosis, as growth can be limited by the reduced chromatin accessibility and consequently reduced transcription. This is consistent with the gradual growth reduction we observe during mitotic arrests (Figure 4). However, considering the long half-life of most mRNAs (Schwanhäusser et al., 2011), and the observations that mRNA levels do not decline during mitosis (Novais-Cruz et al., 2018; Tanenbaum et al., 2015), limited chromatin accessibility does not explain why protein synthesis would be reduced in the absence of mitotic arrests. Furthermore, our results also show that normal mitotic progression results in more radical reduction in mass accumulation than what is observed in prometaphase arrested cells (Figure 5), indicating that there are also other mechanism(s) that radically reduce mass accumulation around metaphase-to-anaphase transition.
Cell mass accumulation reflects the flux of components, such as nutrients, in and out of the cell. Because we observe that mass accumulation fully stops during anaphase, or even becomes negative in some cell types, the mechanism(s) controlling mass accumulation around metaphase-to-anaphase transition must either block the cells from taking up nutrients or even expel cellular components. Yet, this does not necessarily mean that protein synthesis comes to a complete stop, as cells can maintain translation even in the absence of nutrient uptake by degrading proteins and recycling the components (Son et al., 2015b). In fact, prolonged mitotic arrests resulted in near zero mass accumulation although ~25% of the protein synthesis rate persisted (Figure 4). We therefore hypothesize that around metaphase-to-anaphase transition, where the anaphase-promoting complex drives protein degradation, cells largely stop nutrient uptake but maintain low level of protein synthesis by recycling amino acids.
We also observed that reducing CDK1 activity results in reduced growth in daughter cells (Figure 7). This may be explained by CDK1-driven translation of growth components during mitosis, such as ribosomal proteins, that are required to ‘jump-start’ growth in the newborn cells. Consistently, several studies have suggested that transcription and translation of growth-related genes are prioritized during mitosis (Aviner et al., 2017; Palozola et al., 2017; Park et al., 2016). These results imply that CDK1 does not only coordinate cell division, but also optimizes mitotic translation to promote immediate daughter cell growth. However, it should be noted that the active mitotic translation of growth promoting components remains controversial (Stumpf et al., 2013; Tanenbaum et al., 2015), CDK1 inhibition may reduce daughter cell growth independently of mitotic growth, and the physiological significance of mitotic translational control in vivo remains unexplored.
Finally, our observations that mitotic arrests block cell growth have implications for antimitotic cancer chemotherapy. We observed that the vinca alkaloids Vincristine and Vinblastine, both commonly used to treat lymphomas, stop cell growth when used in concentrations lower than those measured in patient plasma (Florian and Mitchison, 2016). This mitotic growth arrest may contribute to the induction of mitotic catastrophe and the efficacy of antimitotic chemotherapies.
Materials and methods
Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
---|---|---|---|---|
Cell line (M. musculus) | L1210 | ATCC | Cat#CCL-219; RRID:CVCL_0382 | |
Cell line (M. musculus) | L1210 - FUCCI | Other | Generated in a previous study (Son et al., 2012), Nature Methods), cells originate from ATCC (Cat#CCL-219). | |
Cell line (M. musculus) | L1210 - ECACC | ECACC | Cat#87092804 | |
Cell line (M. musculus) | Fl5.12 | Other | Kindly provided by laboratory of Prof. Matthew Vander Heiden from MIT | |
Cell line (M. musculus) | BaF3 | RIKEN BioResource Center | Cat#RCB4476 | |
Cell line (G. gallus) | DT40-CDK1as | Other | Kindly provided by laboratory of Prof. Bill Earnshaw from University of Edinburgh | |
Cell line (H. sapiens) | S-HeLa | Other | Kindly provided by laboratory of Kevin Elias from Brigham And Women's Hospital | |
Transfected construct (M. musculus) | Scrambled shRNA | VectorBuilder | Cat#LVS (VB151023-10034) | Refers to a lentiviral construct used to transfect and express the indicated shRNA. |
Transfected construct (M. musculus) | 4E-BP1 shRNA #1 | VectorBuilder | Cat#LVS (VB181217-1124dqm)-C | Refers to a lentiviral construct used to transfect and express the indicated shRNA. |
Transfected construct (M. musculus) | 4E-BP1 shRNA #2 | VectorBuilder | Cat#LVS (VB181217-1125ypy)-C | Refers to a lentiviral construct used to transfect and express the indicated shRNA. |
Biological sample (H. sapiens) | Unpurified buffy coat for isolation of T-cells | Research Blood Components | NA | |
Antibody | Phospho-Histone H3 (Ser10) (D2C8) XP Rabbit monoclonal Ab (Alexa Fluor 647 Conjugate) | Cell Signaling Technology | Cat#3458; RRID:AB_10694086 | Dilution 1/100 in PBS supplemented with 5% BSA |
Antibody | Phospho-Histone H3 (Ser10) (D2C8) XP Rabbit monoclonal Ab (Alexa Fluor 488 Conjugate) | Cell Signaling Technology | Cat#3465; RRID:AB_10695860 | Dilution 1/100 in PBS supplemented with 5% BSA |
Antibody | Cyclin B1 (V152) Mouse monoclonal Ab (Alexa Fluor 488 Conjugate) | Cell Signaling Technology | Cat#4112; RRID:AB_491024 | Dilution 1/50 in PBS supplemented with 5% BSA |
Antibody | Phospho-4E-BP1 (Thr37/46) (236B4) Rabbit monoclonal Ab (PE Conjugate) | Cell Signaling Technology | Cat#7547; RRID:AB_10949897 | Dilution 1/100 in PBS supplemented with 5% BSA |
Antibody | Phospho-S6 Ribosomal Protein (Ser235/236) (D57.2.2E) XP Rabbit monoclonal Ab (Alexa Fluor 647 Conjugate) | Cell Signaling Technology | Cat#4851; RRID:AB_10695457 | Dilution 1/100 in PBS supplemented with 5% BSA |
Antibody | Rabbit (DA1E) monoclonal Ab IgG XP Isotype Control (PE Conjugate) | Cell Signaling Technology | Cat#5742; RRID:AB_10694219 | Dilution 1/100 in PBS supplemented with 5% BSA |
Antibody | α-Tubulin (11H10) Rabbit monoclonal Ab (Alexa Fluor 488 Conjugate) | Cell Signaling Technology | Cat#5063; RRID:AB_10694858 | Dilution 1/200 in PBS supplemented with 5% BSA |
Antibody | Cyclin A Mouse monoclonal Ab (H-3) (FITC Conjugate) | Santa Cruz Biotechnology | Cat#sc-271645; RRID:AB_10707658 | Dilution 1/25 in PBS supplemented with 5% BSA |
Antibody | 4E-BP1 (53H11) Rabbit monoclonal Ab (PE Conjugate) | Cell Signaling Technology | Cat#34470 | Dilution 1/100 in PBS supplemented with 5% BSA |
Antibody | Phospho-4EBP1 (Thr37, Thr46) Rabbit monoclonal Ab (4EB1T37T46-A5) | Thermo Fisher Scientific | Cat#MA5-27999; RRID:AB_2745012 | Dilution 1/100 in PBS supplemented with 5% BSA |
Antibody | Goat polyclonal anti-Rabbit IgG (H + L) Cross-Adsorbed Secondary Antibody (Alexa Fluor 568 Conjugate) | Thermo Fisher Scientific | Cat#A-11011; RRID:AB_143157 | Dilution 1/1000 in PBS supplemented with 5% BSA |
Commercial assay or kit | Click-iT Plus OPP Alexa Fluor 594 Protein Synthesis Assay Kit | Thermo Fisher Scientific | Cat#C10457 | |
Commercial assay or kit | T-cell isolation kit | Miltenyi Biotec | Cat#130-096-495 | |
Commercial assay or kit | Lambda Protein Phosphatase | New England Biolabs | Cat#P0753S | |
Chemical compound, drug | O-propargyl-puromycin (OPP) | Jena Bioscience | Cat#NU-931–5; CAS:1416561-90-4 | |
Chemical compound, drug | S-trityl-l-cysteine (STLC) | Sigma-Aldrich | Cat#164739; CAS:2799-07-7 | |
Chemical compound, drug | Nocodazole | Sigma-Aldrich | Cat#M1404; CAS:31430-18-9 | |
Chemical compound, drug | MG132 | Sigma-Aldrich | Cat#474787; CAS:133407-82-6 | |
Chemical compound, drug | proTAME | Thermo Fisher Scientific | Cat#I44001M; CAS:1362911-19-0 | |
Chemical compound, drug | Vinblastine | Cayman Chemical | Cat#11762; CAS:143-67-9 | |
Chemical compound, drug | Vincristine | Cayman Chemical | Cat#11764; CAS:2068-78-2 | |
Chemical compound, drug | TORIN-1 | Tocris Bioscience | Cat#4247; CAS: 1222998-36-8 | |
Chemical compound, drug | RO-3306 | Cayman Chemical | Cat#15149; CAS:872573-93-8 | We observed the chemical loses its activity within a month when stored in −20°C. All experiments were done using a stock under 2 weeks old. |
Chemical compound, drug | OTSSP167 (OTS) | Cayman Chemical | Cat#16873; CAS:1431698-10-0 | |
Chemical compound, drug | 1NM-PP1 | Sigma-Aldrich | Cat#529581; CAS:221244-14-0 | |
Chemical compound, drug | 4EGI-1 | Cayman Chemical | Cat#15362; CAS:315706-13-9 | |
Chemical compound, drug | Cycloheximide (CHX) | Cayman Chemical | Cat#14126; CAS:66-81-9 | |
Chemical compound, drug | Defactinib | Cayman Chemical | Cat#17737; CAS:1073154-85-4 | |
Chemical compound, drug | PF-3758309 | Cayman Chemical | Cat#19186; CAS:898044-15-0 | |
Chemical compound, drug | Nintedanib | Cayman Chemical | Cat#11022; CAS:656247-17-5 | |
Chemical compound, drug | SNS-032 | Cayman Chemical | Cat#17904; CAS:345627-80-7 | |
Chemical compound, drug | Tozasertib | Cayman Chemical | Cat#13600; CAS:639089-54-6 | Alternative Names : MK 0457, VX 680 |
Chemical compound, drug | GSK 626616 | R and D Systems | Cat#6638; CAS:1025821-33-3 | |
Chemical compound, drug | EIPA | Sigma-Aldrich | Cat#A3085; CAS:1154-25-2 | |
Chemical compound, drug | (-)-Blebbistatin | Sigma-Aldrich | Cat#B0560; CAS:856925-71-8 | |
Software, algorithm | MATLAB R2014b | MathWorks | Used to analyze the SMR raw data and generate data plots. | |
Software, algorithm | OriginPro 2019 | OriginLab | Used to perform statistical analyses and generate data plots. | |
Software, algorithm | Mass accumulation rate analysis code | This paper | Used to analyze the SMR data. Code can be found attached to this manuscript. |
Cell lines, primary cells and culture conditions
Request a detailed protocolL1210 cells were obtained directly from ATCC, with the exception of L1210 cells shown in Figure 2—figure supplement 2b, which were obtained from ECACC. L1210 cells expressing the FUCCI cell cycle sensor were generated in a previous study (Son et al., 2012) using ATCC originating cells. The BaF3 cells were originally obtained from RIKEN BioResource Center and engineered to express BCR-ABL in a previous study (Stevens et al., 2016). S-HeLa cell line was a gracious gift from Dr Elias. FL5.12 cell line was a gracious gift from Dr Vander Heiden. All DT40 cell experiments were carried out using DT40 CDK1as cell line, which was a gracious gift from Dr Samejima and Dr Earnshaw. The DT40 CDK1as cells have had their CDK1 replaced with Xenopus laevis CDK1 with a F80G mutation, as detailed in Gibcus et al. (2018), to sensitize the CDK1 to inhibition by 1NM-PP1. Note that the CDK1as cell line is protected by MTA and the rights to this cell line belong to Prof Earnshaw and the University of Edinburgh. All cell lines tested negative for mycoplasma. Cell line identity was validated by vendors and the identity of L1210 cells from ATCC was further validated using RNA-seq data.
The L1210 cells basic experimental and culture conditions were in 2 mM L-glutamine and 11 mM glucose containing RPMI (Thermo Fisher Scientific, #11835030) supplemented with 10% FBS (Sigma-Aldrich), 1 mM sodium pyruvate (Thermo Fisher Scientific), 20 mM HEPES (Thermo Fisher Scientific) and antibiotic/antimycotic (Thermo Fisher Scientific). In Figure 2—figure supplement 2a, where L1210 cells were grown under different nutrient conditions, the media was kept otherwise identical except for 4% FBS and indicated concentrations of glucose (Sigma-Aldrich) or galactose (Sigma-Aldrich). The L1210 experimental conditions also included 10 nM TMRE in the media, which did not affect cell growth.
BaF3, FL5.12 and DT40 CDK1as cells were grown in media identical to L1210 cells, with the exceptions that FL5.12 cell media was supplemented with 10 ng/ml IL-3 (R and D Systems) and DT40 cell media was supplemented with 3% chicken serum (Sigma-Aldrich). S-HeLa cells were grown in DMEM (Thermo Fisher Scientific) containing 10% FBS (Sigma-Aldrich), 1 mM sodium pyruvate (Thermo Fisher Scientific), 20 mM HEPES (Thermo Fisher Scientific) and antibiotic/antimycotic (Thermo Fisher Scientific).
Human peripheral blood mononuclear cells (PBMCs), including T-cells, were isolated from unpurified buffy coat (Research Blood Components) and activated as described previously (Cermak et al., 2016). Briefly, PBMCs isolation was carried out using Ficoll-Paque Plus density gradient (GE) according to the manufacturer’s recommendations. After isolation of the PBMC layer, the cells were subjected to red blood cell lysis using ACK lysis buffer (Thermo Fisher Scientific). The PBMCs were then washed three times and either frozen or used for isolation of specific T-cell subsets. The T-cell data shown in this paper is derived from both frozen and non-frozen samples obtained from two independent blood samples. Primary T-cells were isolated using Naive CD8 +or CD3+T Cell Isolation Kits (Miltenyi Biotec) according to supplier’s instructions. The primary T-cells were activated on an anti-human CD3 (BioLegend) coated cell culture plate in identical to L1210 culture media, with the exception that the media was supplemented with 10 µM 2-mercaptoethanol, 100 U/ml IL-2 (R and D Systems) and 2 µg/ml anti-human CD28 (BioLegend). The same media was used for SMR experiments. After activation, cells were left undisturbed for 24 hr before SMR experiments, and the activation was validated by monitoring cell number and volume changes using Coulter Counter (Beckman Coulter).
The shRNA expressing L1210 cells were generated as in Kang et al. (2019). L1210 cells (obtained from ATCC) were transfected using mammalian shRNA knockdown lentiviral vectors obtained from VectorBuilder Inc Each construct contained an shRNA sequence under a U6 promoter, as well as EGFP and Puro linked by T2A for selection. Full construct details can be found online at VectorBuilder.com (Vector IDs: VB190401-1106ebw, VB190401-1107bty and VB190401-1108sfm)
Control shRNA target sequence: CCTAAGGTTAAGTCGCCCTCG
4E-BP1 shRNA #1 target sequence: ATTATCTATGACCGGAAATTT (location: 233–253, CDS)
4E-BP1 shRNA #2 target sequence: CCAGTGTTTATGGTGTGATTT (location: 912–932, 3’UTR)
Transfection was carried out using 4 rounds of spinoculation. In each round, 1.5 × 105 L1210 cells were mixed with 10 µg/ml Polybrene (EMD Millipore) and approximately 1 × 106 transducing units of lentivirus. The mixture was centrifuged at 800 g for 60 min at 25°C, and the cells were moved to normal culture media. This procedure was repeated every 12 hr for a total of 4 times. 24 hr after the last round of transfection selection was started using 10 µg/ml puromycin (Sigma-Aldrich). After 5 days of selection the shRNA and EGFP expressing subpopulation was sorted out using BD FACS Aria. shRNA knockdown efficiency was validated by immunostaining 4E-BP1 and quantifying the staining levels using FACS (antibody staining details are below, Figure 6—figure supplement 4a,b).
SMR device setup and experimental details
Request a detailed protocolThe SMR chips were fabricated as previously described (Cermak et al., 2016; Son et al., 2015a; Son et al., 2012) by CEA-LETI, Grenoble, France. The device setup and experimental details were similar to those described previously (Cermak et al., 2016; Son et al., 2015a; Son et al., 2012). Briefly, a piezo-ceramic placed under the device vibrates the cantilever in its second flexural bending mode resonant frequency, which is typically ~1.1 MHz. The resonant frequency was monitored using piezo-resisters embedded at the base of the cantilever. A digital control platform was used to actuate the cantilever in a direct feedback mode, where an actuating signal is generated by amplifying and delaying the detected motion signal from the cantilever. Utilizing the feedback mode with a data rate of ~3000 Hz, our SMR measurement bandwidth was fixed to ~1500 Hz, which was adequate to capture fast modulating frequency signal resulting from a cell transit through the cantilever.
SMR device was operated at a fixed temperature of 37°C, by mounting the SMR chip on a copper clamp that was connected to a circulating, temperature-controlled water bath (Julabo). Fluid flow was controlled by pressure difference across the SMR input ports. Each input port was connected to a reservoir of normal cell culture media that was pressurized with 5% CO2, 21% O2 (Airgas) to maintain stable pH. The amount of pressure applied to each vial were controlled using electronic pressure regulators (Proportion Air QPV1) and the applied pressure difference across the channel was set to achieve a typical flow rate of ~2 nL/s to minimize the shear stress of cells growing within the SMR. This resulted in a typical ~300 ms transit time through the cantilever. Both the absolute pressure and flow direction were controlled using a custom software (LabVIEW 2012 and LabVIEW 2016). The software controls the pressure levels, and consequently the flow direction and speed, in real-time, and is automated to quickly respond to a change in resonant frequency signal. For example, a set of pressures are applied to flow a cell through the cantilever (Figure 1—figure supplement 1a, left #1). The resonant frequency change caused by cell transit through the cantilever automatically stops the flow (Figure 1—figure supplement 1a, left #2). Flow is maintained at zero for desired amount of time (~50 s), after which the pressures are changed to reverse the flow direction (Figure 1—figure supplement 1a, left #3). See Figure 1—figure supplement 1 for the detailed steps of the fluid control and consequent cell movement.
For monitoring morphology changes during mitosis and measuring FUCCI cell cycle reporter intensity, we utilized an on-chip imaging system described previously (Son et al., 2012). Briefly, a modular Nikon microscope equipped with a Nikon LU plan ELWD 50x/0.55 objective, a Lumencor Spectra X light engine and an 8 mm Voltage Output Type photomultiplier tube (Edmund Optics) or a monochrome camera (BFS-U3-13Y3M-C, FLIR). As a cell passed through the SMR cantilever, the change in the resonance frequency was used as a trigger to turn on illumination and measure the FUCCI reporter fluorescence. On-chip cell imaging was done as in Calistri et al. (2018).
Data acquisition and processing
Request a detailed protocolThe motion of the cantilever and thus the resonant frequency of the SMR was measured by a digital control platform described previously (Cermak et al., 2016). The measured signal in digital platform was fed into custom LabVIEW code that records the signal while cell is in transit. Recorded frequency was then post-processed by custom MATLAB code, as described previously (Cermak et al., 2016). Briefly, the code locates two local minima in frequency peaks, fits a fourth order polynomial to the raw data, and the minimum resonance frequency values are extracted from the fittings. The average of these two resonance frequency minima measured in Hz was then transformed in to picograms by calibrating the measurements using monodisperse polystyrene beads (Thermo Fisher Scientific, Duke Standards) with a known buoyant mass. No frequency peak exclusion was performed, except in an extremely rare event where two daughter cells separate inside the cantilever during transit.
Assigning cell cycle transitions to buoyant mass traces
Request a detailed protocolWe identified three distinct cell cycle transition points: mitotic entry (i.e. G2/M transition), metaphase-to-anaphase transition (i.e. M/A transition) and daughter cell abscission (Figure 1—figure supplement 2). We defined M-phase as the sum of mitosis and cytokinesis, starting at mitotic entry and ending at daughter cell abscission. We identified metaphase-to-anaphase transition using the mAG-Geminin signal of the FUCCI cell cycle reporter, which is degraded at metaphase-to-anaphase transition (Figure 1—figure supplement 2c), and using on-chip brightfield imaging, where we identified the metaphase-to-anaphase transition as the last timepoint of cell being round. These signals always coincided with two biophysical signals measured by SMR, a drop in node deviation signal (an acoustic signal corresponding to cell shape and stiffness) and a momentary reduction in buoyant mass trace (Figure 1—figure supplement 2b), both partly due to cell elongation in anaphase (Kang et al., 2019). These elongation dependent signals are applicable to all cell types studied here and this was validated by on-chip imaging, allowing us to designate metaphase-to-anaphase transition for all cells.
The daughter abscission was assigned for all cells based on the approximately 50% loss of buoyant mass within 2 min. Cytokinesis was defined as the time between metaphase-to-anaphase transition and daughter cell abscission. G1 was defined to start immediately following cell abscission.
Detection of mitotic entry (G2/M transition) for L1210 cells was carried out using biophysical parameters. First, we have previously shown that node deviation starts to decrease following mitotic entry (Kang et al., 2019) (Figure 1—figure supplement 2b), allowing us to locate mitotic entry for L1210 traces. The timing of the assigned mitotic entry (30 min prior to metaphase-to-anaphase transition) also matched with previously analyzed mitotic entry point based on single-cell volume measurements (Son et al., 2015a; Zlotek-Zlotkiewicz et al., 2015). Similar node deviation based assignment of approximate mitotic entry was also done for other cell types, whenever node deviation changes were observable. We also estimated the approximate mitotic entry by comparing the whole cell cycle duration to that of L1210 cells. When cell cycle durations were similar, we utilized the same timing of mitotic entry (30 min prior to metaphase-to-anaphase transition) for the other cell types, which was also consistent with the timing of mitotic entry observed by node deviation measurements. While this is an approximation, it is consistent with the notion that the duration of mitosis does not vary drastically cell-to-cell (Araujo et al., 2016). For chemical perturbations that prolonged early mitosis, the approximate mitotic entry is separately indicated in the figures (for example, arrows in Figure 6d).
Approximate duration of L1210 cell metaphase was assigned using the G2/M and metaphase-to-anaphase transitions together with previous characterization of the relative durations of different mitotic stages in L1210 cells (Son et al., 2015a). The duration of cell elongation (i.e. anaphase) was quantified for L1210 cells previously (12 min) (Kang et al., 2019) and the duration was approximated for other cell types using on-chip imaging.
Analyzing mass accumulation rate (MAR) and MAR/mass
Request a detailed protocolTo quantify average MAR/mass of for late G2, early mitosis, cytokinesis and newborn G1, as seen for example in Figure 7a,a single linear fit was made to the buoyant mass traces for each indicated cell cycle stage and the slope of the fit represents the MAR (Figure 2—figure supplement 1a). To minimize the error in the length of fitted segment, data points were linearly interpolated from the buoyant mass trace to accurately pinpoint the beginning and the end of the fitted segments. Then, each slope of the linear fits (MAR) was divided by the average mass during the fitting period to obtain MAR/mass.
To quantify MAR/mass dynamics within mitotic stages, as seen for example in Figure 2a, we quantified the slope and average mass during each 10 min segments in the buoyant mass traces (Figure 2—figure supplement 1b). The 10 min segments were separated by 5 min. The 10 min segments were separately fitted a linear line, and each slope of the fitted line represents MAR (Figure 2—figure supplement 1c). Then MAR of each 10 min segment was divided by the average mass within that 10 min segment to calculate MAR/mass (Figure 2—figure supplement 1d). To minimize the error in the length of fitted segment, data points were linearly interpolated from the buoyant mass trace to increase number of data points. The MAR/mass dynamics shown in Figure 5; Figure 5—figure supplement 1; Figure 5—figure supplement 2 were processed similarly but the length of each segment and separation between each segment were reduced to 4 min and 2 min, respectively. When showing MAR/mass dynamics of individual cells (Figure 2d; Figure 2—figure supplement 1c,d; Figure 5—figure supplement 2a), buoyant mass traces were filtered using a moving average covering a 10 min window of data.
To correct for the cell elongation induced bias in buoyant mass measurement during cytokinesis, we performed a data correction as shown before (Kang et al., 2019). Briefly, the change in cell mass distribution (cell elongation) reduces the resonance frequency shift of the SMR, in a manner that is dependent on cell geometry. Using on-chip imaging, cell mass and volume information, we estimated the cell geometry to be i) spherical before elongation, ii) overlapping spheres during anaphase, and iii) spherical doublets after elongation is over (Figure 1—figure supplement 1d,e). With this information we can calculate the estimated extent of the measurement bias. The details can be found in the (Kang et al., 2019) and in the MATLAB code attached to the supplements. The calculated extent of the measurement bias during L1210 cell cytokinesis can be seen in Figure 1—figure supplement 1d and in Figure 2—figure supplement 1d.
Chemical perturbations
Request a detailed protocolAll SMR experiments with chemical perturbations of cell growth or cell cycle, apart from serial SMR experiments, were carried out by diluting the chemicals directly in to the media within SMR. Untreated cells were then loaded to the SMR for growth monitoring, so that in a typical experiment the cell was exposed to the chemical for at least 2 hr before entering mitosis. Experiments with chemical inhibitors only lasted through one cell division, so that exposure to the chemicals was always started in interphase, approximately 1–4 hr prior to mitosis. Thus, the n indicated for chemical treatments always reflects separate experiments. Control experiments were always carried out between experiments with chemical perturbations to assure that control cell growth rates were reproducible. Note that the control L1210 cell data is accumulated over several years and the growth behavior was reproducible also between different SMR devices (Figure 2—figure supplement 2c).
For serial SMR experiments where cells were arrested to mitosis using STLC, cells were loaded in to the serial SMR in normal culture media immediately after STLC wash off. For serial SMR experiments where cells were treated with RO-3306, cells were loaded in to the serial SMR in 1 µM RO-3306 containing culture media immediately after release from G2 arrest. In Figure 4c, mitotic arrests were obtained by treating unsynchronized L1210 cells with 5 µM STLC, 1 µg/ml Nocodazole, 2 µM MG132 or 20 µM proTAME.
Importantly, we observed that RO-3306 stock stored in −20°C in DMSO was not stable over several months and we therefore carried out all experiments using RO-3306 that was prepared within 1 week of the experiment.
Cell cycle synchronizations
Request a detailed protocolL1210 cells were synchronized to G2 using a double thymidine block followed by a RO3306 mediated G2 arrest. L1210 cells in confluency of 4*105 cells were first treated with 2 mM Thymidine for 15 hr, then washed with PBS and moved to normal culture conditions to release from G1/S arrest. After 6 hr, 2 mM Thymidine was added for 6 hr. Cells were again washed with PBS and moved to normal culture conditions for 3 hr, after which 5 µM RO-3306 was added. 7 hr later cells were arrested in G2 (Figure 3a). Cells were then washed with PBS and moved to normal culture conditions (unless otherwise stated) to allow cells to uniformly progress through mitosis. Note while most cells enter mitosis soon after release from G2 arrest, some cells fail to exit G2. The release from G2 arrest was considered as time zero in protein synthesis, serial SMR and proliferation experiments.
When cells were temporarily arrested in mitosis (Figure 7c,d,f), cells were first synchronized to G2, then released in to 5 µM STLC. 4 hr later the cells were washed twice with media and replaced in to normal cell culture conditions. The release from STLC mediated mitotic arrest was considered as time zero in protein synthesis, serial SMR and proliferation experiments.
Protein synthesis rate sample preparation
Request a detailed protocolProtein synthesis rates were quantified using the Click-iT Plus OPP Alexa Fluor 594 Protein Synthesis Assay Kit (Thermo Fisher Scientific) together with antibody staining specific for different mitotic stages. For each replicate approximately 3*106 cells were treated with 20 µM OPP for 10 min under normal culture conditions. The OPP accumulation was stopped by mixing the cells with ice cold PBS, after which the cells were quickly washed with ice cold PBS and then fixed with formaldehyde for 10 min at room temperature. To reduce the number of cell doublets fixed together, the fixation was carried out by shaking the cells on vortex in PBS and slowly adding a corresponding volume of 8% paraformaldehyde to reach a final paraformaldehyde concentration of 4%. Fixative was washed away with PBS and the cells were permeabilized using 0.5% Triton X-1000 in PBS for 10 min at room temperature. The permeabilization solution was washed away with PBS and cells were incubated in 5% BSA in PBS for 30 min to block non-specific antibody binding.
Mitotic cells were separated from interphase cells using p-Histone H3 monoclonal antibody (S10, D2C8, conjugated to Alexa 647, Cell Signaling Technology, #3458S; or S10, D2C8, conjugated to Alexa 488, Cell Signaling Technology, #3465S). For separation of mitosis in to early and late mitosis, Cyclin B1 (V152, conjugated to Alexa 488, Cell Signaling Technology, #4112S) and Cyclin A (H-3, conjugated to FITC, Santa Cruz Biotechnology, #sc-271645 FITC) monoclonal antibodies were used. All antibody labeling steps were carried out o/n in 4°C in 5% BSA containing PBS. All antibodies were used at the concentration recommended by supplier. Antibodies were washed away using 5% BSA in PBS.
The OPP Click-IT reaction was carried out according to manufacturer’s (Thermo Fisher Scientific) instructions. After OPP conjugation with Alexa Fluor 594 fluorophore, the cells were washed twice with PBS and DNA was stained for 30 min in RT with 1:2000 dilution of NuclearMask Blue (#H10325, Thermo Fisher Scientific). Finally, the cells were washed three times with PBS, mixed in to PBS supplemented with 1% BSA and put on ice until FACS analysis.
Protein content and phosphorylation level sample preparation
Request a detailed protocolTo analyze the levels of total 4E-BP1 or phosphorylated S6RP and 4E-BP1, unperturbed cells were prepared using same fixation, permeabilization and blocking protocol as for protein synthesis assays. The cells were then incubated o/n in 4°C with 4E-BP1 monoclonal antibody (53H11, Cell Signaling Technology, #34470), p-4E-BP1 monoclonal antibody (Thr37/46, 236B4, conjugated to PE, Cell Signaling Technology, #7547S), p-S6RP monoclonal antibody (S235/236, D57.2.2E, conjugated to Alexa 647, Cell Signaling Technology, #4851S) or isotype specific controls (Rabbit mAb IgG, conjugated to PE or Alexa 647, Cell Signaling Technology, #5742S) in PBS solution containing 5% BSA. All antibodies were used at the concentration recommended by supplier. For analysis of total 4E-BP1 levels, cells were washed and treated with 2 µg/ml secondary antibody (Goat anti-Rabbit IgG secondary antibody conjugated to Alexa Fluor 568, #A-11011, Thermo Fisher Scientific) for 2 hr in RT. Antibodies were washed away using 5% BSA in PBS and the cells were stained for p-Histone H3 (S10) and DNA, as in protein synthesis assays. Finally, the cells were washed three times with PBS, mixed in to PBS supplemented with 1% BSA and put on ice until FACS analysis.
Flow cytometry
Request a detailed protocolFACS-based quantifications were done using BD Biosciences flow cytometer LSR II HTS with excitation lasers at 355 nm, 488 nm, 561 nm and 640 nm, and emission filters at 450/50, 530/30, 585/15 and 660/20. See (Figure 3a,c; Figure 3—figure supplement 1a) for DNA and antibody labeling that were used to separate cell cycle stages. At least 20,000 cells were analyzed for each replicate so that each analyzed subpopulation contained at least 250 cells, and typically over 1000 cells.
Microscopy
Request a detailed protocolFor validation of OPP staining, L1210 cells were plated on coverslips coated with 0.1% poly-L-lysine, and prepared using same fixation, permeabilization, blocking, antibody labeling and DNA staining protocol as for protein synthesis assays. After all staining procedures were done, the cells were mounted on to microscopy slides in Vectashield mounting medium (Vector Laboratories).
For examination of phosphorylated 4E-BP1, L1210 cells were plated on coverslips coated with 0.1% poly-L-lysine, and prepared using same fixation, permeabilization and blocking protocol as for protein synthesis assays. The cells were then labeled o/n in 4°C with p-4E-BP1 monoclonal antibody (Thr37/46, 4EB1T37T46-A5, #MA5-27999, Thermo Fisher Scientific). The following day the cells were washed three times with PBS and then treated with 2 µg/ml secondary antibody (Goat anti-Rabbit IgG secondary antibody conjugated to Alexa Fluor 568, #A-11011, Thermo Fisher Scientific) for 2 hr, before an o/n labelling in 4°C with α-Tubulin monoclonal antibody (11H10, conjugated to Alexa Fluor 488, #5063, Cell Signaling Technology). The following day the cells were washed three times with PBS and DNA was stained for 30 min in RT with 1:2000 dilution of NuclearMask Blue (#H10325, Thermo Fisher Scientific). The cells were washed three times with PBS and mounted on to microscopy slides in Vectashield mounting medium (Vector Laboratories). OPP staining and 4E-BP1 phosphorylation levels were imaged using DeltaVision wide-field deconvolution microscope using standard filters (DAPI, FITC, TRITC) and 100 × objective. After deconvolution using SoftWoRx 7.0.0 software, approximately 3 µm thick section from the middle of z-slices was merged into a maximum intensity image for visualization.
Cell proliferation rate was analyzed by imaging the cells every 1 hr using IncuCyte live cell analysis imaging system by Sartorius. The relative cell count was then assessed from the phase images by analyzing the relative confluency in each sample. These values were normalized to the value at start and hourly average values were plotted together with the standard error of mean (SEM). Representative images displaying the duration of daughter cell separation (Figure 2c,e) were obtained in a parallel experiment using IncuCyte live cell analysis imaging system by Sartorius with 20X objective or using on-chip imaging in the SMR with the imaging setup detailed in the section ‘SMR device setup and experimental details’.
Lambda protein phosphatase treatment
Request a detailed protocolTo verify the phospho-specificity of the p-4E-BP1 antibodies, cells were prepared using same fixation, permeabilization and blocking protocol as detailed above. The cells were first treated with p-Histone H3 antibody to label mitotic cells and to protect p-Histone H3 sites from Lambda phosphatase, after which the phosphatase treatment and DNA staining were carried out. The cells were then treated with 10,000 units/ml of Lambda protein phosphatase in 1X NEBuffer for protein MetalloPhosphatases that contained 1 mM MnCl2 for 12 hr in 30°C, after which cells were washed twice with PBS. The Lambda phosphatase and the treatment buffer were obtained from New England BioLabs (#P0753S). Finally, DNA was stained as detailed above and the cells were analyzed using flow cytometer.
Combining and normalizing data
Request a detailed protocolTotal mass accumulation for each cell was calculated from the mass accumulation traces, by assuming that the birth size of the cell was exactly half of the abscission size. Mass accumulation during mitosis (between G2/M transition and metaphase-to-anaphase transition) and during cytokinesis (between metaphase-to-anaphase transition and daughter cell abscission) was then normalized to the calculated total mass accumulation for each cell. MAR/mass traces (i.e. MAR/mass vs time) from each single cell were aligned to metaphase-to-anaphase transition. We then linearly interpolated data points from each MAR/mass trace, consequently making the total number of timepoints for each cell to be 100. Mean and SD were calculated for each timepoint and plotted as a function of approximate mitotic entry (30 min before metaphase-to-anaphase transition for most cell types). In Figure 4, where the drug-induced mitotic arrests inhibit us from aligning the data to metaphase-to-anaphase transition, all data was aligned to approximate mitotic entry using node deviation and MAR/mass signals. In addition, in Figure 4d, the MAR/mass traces were smoothed with a moving average filter of length 3. Data smoothing was not done for any other datasets. In Figure 7a,b, average MAR/mass for each indicated cell cycle stage (late G2, early mitosis, cytokinesis and early G1) was normalized to control values within that cell cycle stage.
Statistical information
Request a detailed protocolAll statistical tests carried out, as well as descriptions of error bars and numbers of replicates, are detailed in the figure legends. All t-tests were two tailed. No replicates were excluded, except for image analysis, when cells could not be analyzed, as detailed above. No power analysis was used. Sample size was kept at or above three independent replicates, with exact sample size depending on the experimental setup. Many of the experimental approaches required time-sensitive sample processing, which limited the maximum sample size. In all FACS-based assays, the replicate number refers to independent cell cultures. In all SMR-based assays, the replicate number refers to independent cells measured through mitosis. In SMR-based assays, control samples were grown for multiple generations yielding several replicates in one experiment, whereas drug treated samples were only grown through one division so that each drug treated replicate represents a separate experiment. Experiments were repeated at least three times on separate days, unless otherwise stated in the figure legends.
The increase in L1210 cell MAR/mass from G2 to early mitosis was quantified by comparing the highest MAR/mass values observed in G2 and in early mitosis for each cell separately. The statistical comparison between these two groups was carried out by two-tailed Student’s t-test.
Statistical tests were carried out using OriginPro 2019 software. The analysis of buoyant mass traces and the analysis of images was carried out using MATLAB R2014b. Visualization of microscopy images was carried out using ImageJ. All figures were compiled using Adobe Illustrator CC 2018.
Data and data analysis code availability
Request a detailed protocolAll L1210 control buoyant mass measurement shown in Figure 1 and used for quantification of MAR/mass in Figure 2a can be found in Figure 1—source data 1. This data has not been corrected for the cell elongation. Data analysis code for correcting the cell elongation bias and obtaining MAR from the buoyant mass traces can be found attached to this manuscript.
Data availability
All L1210 control buoyant mass measurement around M-phase, which were used for quantification of mitotic growth (Figure 1), MAR/mass dynamics (Figure 2), can be found in Figure 1-source data 1.
References
-
Making the cut: the chemical biology of cytokinesisACS Chemical Biology 5:79–90.https://doi.org/10.1021/cb900256m
-
Involvement of the 24-kDa cap-binding protein in regulation of protein synthesis in mitosisThe Journal of Biological Chemistry 262:11134–11139.
-
High-throughput measurement of single-cell growth rates using serial microfluidic mass sensor arraysNature Biotechnology 34:1052–1059.https://doi.org/10.1038/nbt.3666
-
Condensed mitotic chromatin is accessible to transcription factors and chromatin structural proteinsThe Journal of Cell Biology 168:41–54.https://doi.org/10.1083/jcb.200407182
-
Cyclin A is destroyed in prometaphase and can delay chromosome alignment and anaphaseThe Journal of Cell Biology 153:121–136.https://doi.org/10.1083/jcb.153.1.121
-
Anti-Microtubule drugsMethods in Molecular Biology 1413:403–421.https://doi.org/10.1007/978-1-4939-3542-0_25
-
Progressive activation of CyclinB1-Cdk1 coordinates entry to mitosisDevelopmental Cell 18:533–543.https://doi.org/10.1016/j.devcel.2010.02.013
-
An essential role for Cdk1 in S phase control is revealed via chemical genetics in vertebrate cellsThe Journal of Cell Biology 178:257–268.https://doi.org/10.1083/jcb.200702034
-
Translational regulation of the cell cycle: when, where, how and why?Philosophical Transactions of the Royal Society B: Biological Sciences 366:3638–3652.https://doi.org/10.1098/rstb.2011.0084
-
Cdc2-cyclin B phosphorylates p70 S6 kinase on Ser411 at mitosisJournal of Biological Chemistry 273:15077–15084.https://doi.org/10.1074/jbc.273.24.15077
-
Mitotic repression of RNA polymerase II transcription is accompanied by release of transcription elongation complexesMolecular and Cellular Biology 17:5791–5802.https://doi.org/10.1128/MCB.17.10.5791
-
Synthesis of RNA and protein during mitosis in mammalian tissue culture cellsExperimental Cell Research 26:260–268.https://doi.org/10.1016/0014-4827(62)90176-3
-
A cell cycle-dependent internal ribosome entry siteMolecular Cell 5:607–616.https://doi.org/10.1016/S1097-2765(00)80240-3
-
Suppression of cap-dependent translation in mitosisGenes & Development 15:2083–2093.https://doi.org/10.1101/gad.889201
-
Cell-cycle-dependent translational controlCurrent Opinion in Genetics & Development 11:13–18.https://doi.org/10.1016/S0959-437X(00)00150-7
-
Preferential translation of internal ribosome entry site-containing mRNAs during the mitotic cycle in mammalian cellsJournal of Biological Chemistry 279:13721–13728.https://doi.org/10.1074/jbc.M312854200
-
Fueling the cell division cycleTrends in Cell Biology 27:69–81.https://doi.org/10.1016/j.tcb.2016.08.009
-
Mitotic regulation of ribosomal S6 kinase 1 involves Ser/Thr, Pro phosphorylation of consensus and non-consensus sites by Cdc2Journal of Biological Chemistry 278:16433–16442.https://doi.org/10.1074/jbc.M300435200
-
Ribosomal slowdown mediates translational arrest during cellular divisionMolecular and Cellular Biology 27:6639–6646.https://doi.org/10.1128/MCB.00798-07
-
Mitotic modulation of translation elongation factor 1 leads to hindered tRNA delivery to ribosomesJournal of Biological Chemistry 286:27927–27935.https://doi.org/10.1074/jbc.M111.255810
-
Resonant microchannel volume and mass measurements show that suspended cells swell during mitosisThe Journal of Cell Biology 211:757–763.https://doi.org/10.1083/jcb.201505058
-
Cooperative nutrient accumulation sustains growth of mammalian cellsScientific Reports 5:17401.https://doi.org/10.1038/srep17401
-
Drug sensitivity of single cancer cells is predicted by changes in mass accumulation rateNature Biotechnology 34:1161–1167.https://doi.org/10.1038/nbt.3697
-
Regulation of global translation during the cell cycleJournal of Cell Science 131:jcs220327.https://doi.org/10.1242/jcs.220327
-
The translational landscape of the mammalian cell cycleMolecular Cell 52:574–582.https://doi.org/10.1016/j.molcel.2013.09.018
-
The role of protein synthesis in cell cycling and cancerMolecular Oncology 3:402–408.https://doi.org/10.1016/j.molonc.2009.05.003
-
Optical volume and mass measurements show that mammalian cells swell during mitosisThe Journal of Cell Biology 211:765–774.https://doi.org/10.1083/jcb.201505056
Decision letter
-
Jon PinesReviewing Editor; Institute of Cancer Research Research, United Kingdom
-
Jonathan A CooperSenior Editor; Fred Hutchinson Cancer Research Center, United States
-
Marvin TanenbaumReviewer; Hubrecht Institute,, Netherlands
-
Matthieu PielReviewer; Institut Curie, France
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your article "Animal cells grow during mitosis to maintain cytokinetic fidelity" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Jonathan Cooper as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Marvin Tanenbaum (Reviewer #2); Matthieu Piel (Reviewer #3).
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
Summary:
In this study, the authors make very sensitive single cell mass measurements to determine cellular growth during different phases of the cell cycle and find that cellular growth can also occur during (some stages of) mitosis. They go on to measure translation rates in different stages of mitosis and use a number of pharmacological inhibitors in an attempt to identify the signaling pathways that drive mitotic growth, and to study the functional consequences of inhibiting mitotic growth.
Essential revisions:
All the reviewers found your approach to be innovative and the potential of your assay to be exciting. They were all concerned, however, that several of your key conclusions (that cell growth is inhibited in metaphase, but not in prometaphase; that CDK1 acts through cap-dependent translation; that CDK acts through 4EBP1; that the cell growth promoting activity of CDK1 is needed for cell growth in G1 phase; and that the cell growth promoting activity of CDK1 is needed for symmetric cell division) are not directly supported by the data, either because of the possibility of off-target effects, or because the text and the figures appear to contradict each other. The majority of the reviewers' concerns can be addressed by re-writing the manuscript to make clear what can be directly concluded from the data, by removing the last sections of the paper concerning asymmetric cell division and 4EBP1 localisation, and, as a consequence, by changing the title.
Additional data will be required to support your conclusion on the effects on growth in G1 phase: the reviewers suggest using your single cell data to correlate, at the single cell level, the growth in mitosis and the growth in early G1, in non perturbed and in perturbed cells.
We are attaching the three reviewers concerns to guide you in amending your study.
Reviewer #1
1) Throughout the paper, what was being measured, what time-points were acquired and how the different sub-periods of mitosis should be made more explicit. A more consistent nomenclature of periods and timepoints should be used.
For example, the authors mention mass accumulation during anaphase but it is unclear how this period was defined, considering the timepoints identified on Figure 1C (it is surprising that growth can be measured during such a short event). The periods defined in Figure 1C do not have the same names as the periods on Figure 1D or Figure 2A. It was also unclear how the total cell cycle duration was measured (from the end of abscission? Or from cytokinesis completion?). Perhaps a clearer schematic could be presented that incorporates some elements of Supplementary Figure 1 to the main Figure.
Treatments applied to cells were also difficult to follow. Figure 4E,H: is the treatment done throughout the whole measurement? Figure 5C-E: Is this 1) a block to synchronize in G2, 2) a treatment with STLC, 3) a release, 4) growth measurement? (the text describing this in subsection “CDK1-driven mitotic protein synthesis supports daughter cell growth” could be made simpler). Again, schematics could be added to the figures to aid the reader.
2) Figure 2: In Figure 2D, the authors mention that MAR reaches 0 for all cell types. It looks like for S-HeLa and CD3+ T cells, MAR reaches negative values, which would mean that cells loose mass at M/A transition. Could the authors comment, and make sure that this is not a problem from measurement of adherent cells with the SMR? Figure 2A is used to document a 15.8% increase in MAR in prophase vs. late G2 but it would be better to show a box plot with <MAR> for the 2 windows of time.
3) Figure 4 is very interesting but requires clarification:
Interpretation of the effects of OTS treatment: It is hypothesized to act only via its off-target effect on CDK1, so why does it have a stronger effect on protein synthesis rate than direct inhibitors of CDK1?
Interpretation of the authors for the negative MAR in Figure 4E and H: Does it suggest that the ratio of protein degradation rate/protein synthesis rate becomes positive in the absence of CDK1-driven translation?
Some of the statements in the text could be clearer: For example, in the third paragraph of subsection “CDK1 drives phosphorylation of 4E-BP1, cap-dependent protein synthesis and mass accumulation in mitosis”, it might be more precise to say: 'cap-dependent protein synthesis persists in mitosis and partially involves CDK1-mediated regulation of cap-dependent translation'. There remains a delta in protein synthesis rate between the inhibition of translation via the inhibition of CDK1 activity using RO-3306 and the inhibition of translation via a direct inhibition of 4EBP1 (Figure 4F), thus suggesting that part of the growth in mitosis involves cap-dependent translation but via another regulatory pathway than through CDK1.
4) Figure 5, testing the recovery of growth after the release of CDK1 inhibition during mitosis is an important result that would benefit from a bit more work.
5 c and d: why are the experiments treated separately? It is surprising that the control from exp2 in 5d is similar to the treated condition in exp1. A third experimental replicate for 5c and 5d is recommended to make sure that the results are reproducible. In addition, the number of events compared in the control and treated conditions should be comparable.
The plot in 5f is supposed to show recovery of cell growth, but by examining confluency, proliferation, not growth is being measured. The size distribution of the different conditions at different time-points could prove this point (e.g. 1/2 cell cycle duration, 1 and 2 cell cycle durations after the treatment release). It would be even more interesting to add a few time points to the measurement on Figure 5E if feasible.
Some single-cell experiments datasets were relatively small compared to the control condition. Moreover, the number of replicates in Figure 5A and B was unclear and whether measurements were made on the same day (with the same batch of cells) in the control and the treated conditions. If not, a justification should be provided (e.g. at least showing reproducible MAR measurement distributions on control populations in different experiments).
5) Figure 6: is there only one biological replicate for Figure C-D?
6) In the Discussion, the authors suggest that the growth during mitosis is comparable to that of prophase. This could be more directly tested with the data the authors already have in Figure 1. The comparison of the different cell lines is interesting but could be improved by comparing the relative mass accumulation to the relative duration of mitosis of each of these cell types. Perhaps by plotting DM_mitosis/DM_total cell cycle vs. DT_mitosis/DT_total cell cycle, or measuring <DM_mitosis/DT_mitosis> and comparing with <(DM_total_cell_cycle – DM_mitosis)/(DT_total_cell_cycle – DT_mitosis)>, or doing the same boxplot as Figure 1D but sorting the cells by increasing DT_mitosis/DT_total cell cycle. Mitosis duration has been shown to function as an insulated period of the cell cycle, with very constant duration across cell types, regardless of the variety of duration of the rest of the cell cycle (see Aurajo et al., 2016). Perhaps the authors could use their data to see if the same homogeneous behavior holds true for growth. This would raise interesting hypotheses about whether general constraints on optimal energy resource allocation/protein synthesis exist during mitosis that lead to some reproducible growth behavior across cell types.
7) The statement at the start of the second paragraph of the Discussion that growth is required for cell division has not been proven. The authors do not test directly the requirement for growth to complete cell division. Doing so would require inhibiting growth (independently of the cell cycle proteins) and assessing the rate of mitotic progression (and the rate of asymmetric divisions). The authors could perform this experiment (for example using 4EBP1-I) to make this point more rigorously.
Reviewer #2
– Can the authors exclude that the morphological changes of the cell that occur during cell rounding in prometaphase affect their measurements, and perhaps cause the small bump in the graph describing cell mass in early prometaphase? Similar for the morphological changes in anaphase / telophase?
– Is the reduction in cell growth in metaphase due to metaphase itself, of simply a consequence of being in mitosis for 20-40 min? When the authors delay metaphase onset, for example with a small molecule inhibitor of Eg5, growth progressively slows or halts after 20 min of entering prometaphase. Therefore, it appears that growth does slow down after cells enter prometaphase, but the slow down is somewhat slow and therefore not complete until metaphase.
– If I'm correct, HeLa cells are the only non-lymphoid cells they use, and HeLa cells show a fairly strong reduction in growth during mitosis in their data (Figure 2D) (they don't distinguish between prometaphase and metaphase, but if looks like the growth is already reduced in PM). Their Abstract states: "Growth is only stopped in metaphase", but the HeLa cell data (As well as the DT40 cell data) appear to show a strong growth reduction already in prometaphase. Thus, there conclusions should only be limited to lymphoid cells, or the conclusion should better match the data, i.e. the observed slow down of growth in prometaphase.
– The authors are not always very accurate in their description of the different mitotic phases. For example, cells treated with STLC are in prometaphase, not in metaphase. In contrast, MG132 and proTAME to arrest cells in metaphase. Since the main focus of this study is the precise timing of cell growth during different stages of mitosis, such inaccuracies are fairly confusing.
– Based on the STLC and Vinblastine etc experiments, I belief it should be concluded that the growth inhibition does not occur specifically in metaphase, but rather occurs after entering mitosis, with some delay. Consistent with this, HeLa cells are slower in reaching anaphase, and show a greater reduction in growth before anaphase.
– Throughout the manuscript the authors claim that CDK1 stimulates translation. Yet, in late prometaphase and early metaphase there is plenty of CDK1 activity, but low translation rates. Similarly, in G2 there is low CDK1 activity, but high translation rates. So there is a fairly poor correlation between CDK1 activity and translation rates. This is poorly discussed throughout the manuscript.
– "However, treatment with both 4EGI-1 and RO-3306 did not significantly change mitotic protein synthesis rates when compared to treatment with 4EGI-1alone. Thus, cap-dependent protein synthesis persists in mitosis, and CDK1 regulates growth at least partly through cap-dependent translation."
I don't understand how the authors make this conclusion (and similar ones elsewhere in the manuscript) based on these experiments. Is the reasoning that 4EGI-1 inhibits cap-dependent translation, and in the absence of cap-dependent translation CDK1 inhibition does not reduce protein synthesis, thus CDK1 must be acting through cap-dependent translation? If so, I don't belief that the current data are solid enough to support this conclusion. 4EGI-1 treatment only reduces protein synthesis rates by 50%, both in G2 and M, so clearly there is still substantial cap-dependent translation under 4EGI-1 inhibitor treatment.
– Did the authors examine the time from mitotic entry to the metaphase-to-anaphase transition in control cells vs cells treated with a low dose of CDK1 inhibitor? It is certainly possible that chromosome alignment is perturbed under these conditions, so the time from NEB to the M-A transition is longer under partial CDK1 inhibition. All the data in this study shows that translation rates steadily decrease either from the moment of mitotic entry, or with a ~20 min delay (apparently depending on the cell line), including STLC treatment.
– Therefore, a delay in prometaphase induced by the CDK1 inhibitor would be sufficient to explain the observed decrease in the protein synthesis rates. This is especially true, because the authors align their traces at the M-A transition; if CDK1 inhibition slows down chromosome alignment, then cells that are, say 10 min before the M-A transition in control cells have been in mitosis shorter than cells 10 min before the M-A transition in CDK1 inhibited cells, thus potentially explaining why CDK1 inhibitor treatment apparently reduced translation.
– Subsection “CDK1-driven mitotic protein synthesis supports daughter cell growth”: "…the growth reduction in daughter cells must reflect a consequence of mitotic growth inhibition".
This is an overstatement, CDK1 inhibition does all kinds of things besides inhibition of growth. For example, it may delay mitosis, which activates p53, among other things and has consequences in G1 (Uetake and Sluder, 2010, Curr Biol), it could induce chromosome missegregation, or change the levels or localization of different proteins. Thus, attributing the effects of growth in daughter cells specifically to the role of CDK1 in stimulating mitotic translation or growth seems premature.
– Figure 6. Just because 4EBP1 localizes a bit more to the cleavage plane, does not prove that translation is preferentially occurring there. An OPP assay could be performed to support this claim.
– "Control cell divisions were mostly symmetrical (Figure 6C). However, when we analyzed cells treated with RO-3306, OTSSP167 or cycloheximide, we observed that all these treatments resulted in more asymmetric cell divisions"
CDK1 phosphorylates hundreds of proteins during mitosis, including many proteins that regulate microtubule and spindle function. Thus, the spindle is probably also not functioning completely normally after CDK1 inhibition, which is far more likely to be the cause of the asymmetric cell division. Therefore, I don't think that linking the observed affects on cell growth after CDK1 inhibition to the asymmetric divisions is warranted.
– "Our single-cell measurements show that mass accumulation behavior during mitosis is dynamic, dependent on mitotic stage"
I disagree that the data proves that mass accumaltion depends on the mitotic stage per se. The rate of mass accumulation appears to depend mainly on the time since entering mitosis, with a gradual shutdown of protein synthesis and growth occurring after mitotic entry. This conclusion is supported by the strong growth inhibition before the M-A transition in HeLa cells, as well as the strong growth inhibition in STLC treated cells, which never reach metaphase.
– "Indeed, we show that growth is only stopped during anaphase"
Not correct, in all cell lines, growth is already stopped in metaphase, in many cell lines, it is already strongly reduced in prometaphase.
– "Mechanistically, we found that CDK1 drives M-phase growth through 4E-BP1 and cap-dependent protein synthesis"
As stated above, I don't think this is shown. In G2 there is very low CDK1 activity, yet strong growth and translation. Similarly, in late prometaphase/metaphase there is also strong CDK1 activity, yet growth gets downregulated.
– The authors should be careful not to be selective in their choice of the different studies examining translation in mitosis to find support for their conclusions, for example:
"This is consistent with the observations that transcription and translation of
growth related genes are prioritized during mitosis". They do not discuss Stumpf et al. and Tanenbaum et al. here, who performed genome-wide analysis of translation in mitosis, and which are discussed in other parts of the manuscript.
– Title: I'm not sure that the title is sufficiently supported by the data. While the authors provide one experiment that translation is needed for cell division symmetry; cells treated with CHX (As discussed above, I don't think their inhibition of CDK1 experiments prove that cell growth in M is needed for division symmetry), they do not show that "cell growth" is needed for cell division symmetry. The authors themselves state the translation and cell growth are not the same: "Importantly, protein and RNA synthesis rates are only proxies of overall growth (biomass increase), which is determined by the balance between synthesis (anabolic) and degradation (catabolic) rates"
Reviewer #3
1) The main concern is the specificity of the effect of CDK1 inhibition on early mitotic growth. The authors show by multiple approaches that CDK1 activity is required for the maintenance of growth during early mitosis – this is fine. They then propose that it is because CDK1 phosphorylates 4E-BP1. Nevertheless, they do not formally prove this point, and CDK1 could act via other pathways. The authors should amend their text to make it clear that they did not prove that the early mitosis growth is due to CDK1 acting via 4E-BP1. They should also discuss the potential role of mitotic swelling, which they showed is also CDK1 dependent, and could affect early mitotic growth rate – when cells enter mitosis, they increase volume in a CDK1 dependent manner. One would thus expect that inhibiting CDK1 activity, which would prevent this swelling, could have an effect on overall cell density and affect the growth rate in this early mitotic phase.
2) One consequence of the previous concern is that the last piece of data provided is not really easy to interpret. It is in fact unlikely that the effect of CDK1 inhibition on the asymmetry of division is due to its effect on mitotic growth. This effect of CDK1 inhibition on mitotic spindle positioning, or in fact of almost any drug inducing a mitotic delay, as already been reported and rather depends on the importance of CDK activity for other processes that are needed to orient the spindle properly (see for example Beamish JBC 2009). I would remove this part, which does not bring much to the article and is unlikely to be correct – it is probably not because of a lack of early mitotic growth that the division ends up being asymmetrical upon CDK1 inhibition. In my opinion, the conclusion drawn “Together, these results show that CDK1 and the maintenance of active translation in M-phase are required for the fidelity of cytokinesis” is not supported by the data. But it is really not important for the rest of the article.
3) Similarly, I would remove the very short chapter on the localization of 4E-BP1. It is not clear how it helps the main message of the article.
4) To help the reader estimate the global effect of inhibiting early mitotic growth, the authors should provide, just like in Figure 1, the effective growth in mitosis (in% and not just the MAR), cytokinesis and early G1 comparing control and cells in which early mitotic growth has been inhibited.
[Editors' note: further revisions were requested prior to acceptance, as described below.]
Thank you for submitting your article "Mammalian cell growth dynamics in mitosis" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Jonathan Cooper as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Marvin Tanenbaum (Reviewer #2); Matthieu Piel (Reviewer #3).
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. This should only require a little rewording and we look forward to your rapid revision.
Essential revisions:
The 4E-BP1 knockdown experiment does not really prove that translation in mitosis is needed for growth in G1. One could still envision a scenario in which CDK1 inhibition causes some sort of stress in G1 phase, either through an increased mitotic duration or for example through induction of chromosome missegregation (or DNA damage associated with chromosome missegregation). Such stress could trigger a translational downregulation in G1, which is (partially) restored by depletion of 4E-BP1. In this scenario it is not the role of CDK1 in mitotic translation / growth that is responsible for slower G1, but another function of CDK1. Also, the depletion of 4E-BP1 does not increase G1 translation through its activity in M phase, but through its activity in G1 phase in this scenario.
The authors should discuss this alternative possibility if they cannot fully exclude it.
https://doi.org/10.7554/eLife.44700.027Author response
Essential revisions:
All the reviewers found your approach to be innovative and the potential of your assay to be exciting. They were all concerned, however, that several of your key conclusions (that cell growth is inhibited in metaphase, but not in prometaphase;
We have now amended our manuscript text (both Results and Discussion) to more clearly state that growth rates start to decrease in late prometaphase.
CDK1 acts through cap-dependent translation; that CDK acts through 4EBP1
4E-BP1 is a negative regulator of cap-dependent translation, and the phosphorylation of 4E-BP1 by mTOR or CDK1 inactivates 4E-BP1. We have now knocked down 4E-BP1 and monitored how this affects mitotic protein synthesis rates. Our new data show that in the absence of 4E-BP1, CDK1 inhibition no longer reduces mitotic protein synthesis as radically. Thus, this rescue experiment validates that at least part of the CDK1-driven mitotic growth is mediated by 4E-BP1. Please see our new Figure 6G, Figure 6—figure supplement 4, and paragraph four of subsection “CDK1 drives phosphorylation of 4E-BP1, protein synthesis and mass accumulation in mitosis”.
We have clarified in our Discussion that other CDK1 dependent or independent growth regulations may also exist, as our data does not rule out this possibility.
The cell growth promoting activity of CDK1 is needed for cell growth in G1 phase
Using the 4E-BP1 knockdown to rescue mitotic growth effects of CDK1 inhibition, as described above, we now show that daughter cell protein synthesis rates are less impaired in 4E-BP1 knockdown cells than in control knockdown cells. Thus, rescuing the mitotic growth inhibition also rescues the daughter cell growth inhibition. However, these rescues were not complete, and we have now clarified our text by pointing out that other mechanisms may also be involved. Please see our new Figure 7E, subsection “CDK1-driven mitotic protein synthesis supports daughter cell growth” and also the discussion of our manuscript.
In addition, as suggested by the reviewers, we have used our single cell data to correlate, at the single cell level, the growth in mitosis and the growth in early G1, in nonperturbed and in perturbed cells. This revealed a modest correlation between mitotic MAR and G1 cell MAR, and drug treatments which reduced mitotic MAR maintained a similar level of correlation. These new data can be found in our new Figure 7—figure supplement 1, and in subsection “CDK1-driven mitotic protein synthesis supports daughter cell growth”.
The cell growth promoting activity of CDK1 is needed for symmetric cell division) are not directly supported by the data, either because of the possibility of off-target effects, or because the text and the figures appear to contradict each other. The majority of the reviewers' concerns can be addressed by re-writing the manuscript to make clear what can be directly concluded from the data, by removing the last sections of the paper concerning asymmetric cell division and 4EBP1 localisation, and, as a consequence, by changing the title.
As suggested by the editor, we have removed the section regarding the asymmetrical cell divisions and subcellular localization of p4EBP1.
We have changed our manuscript title to “Mammalian cell growth dynamics in mitosis”.
Additional data will be required to support your conclusion on the effects on growth in G1 phase: the reviewers suggest using your single cell data to correlate, at the single cell level, the growth in mitosis and the growth in early G1, in non perturbed and in perturbed cells.
As suggested by the reviewers, we have now correlated our single-cell growth data non-perturbed and in perturbed cells to show a modest correlation between mitotic growth and daughter cell growth. When mitotic growth was perturbed, this correlation persisted, as also the daughter cells reduced their growth. These new data can be found in our new Figure 7—figure supplement 1, and in subsection “CDK1-driven mitotic protein synthesis supports daughter cell growth”.
As suggested by reviewer #1, we have repeated our experiments on inhibiting mitotic CDK1 activity and monitoring daughter cell growth rates using serial SMR. These new data are consistent with our previous data and our conclusion that CDK1-driven mitotic growth supports daughter cell growth. Please see Figure 7C, third panel (experiment #3).
As stated above, by knocking down 4E-BP1 we can now partly rescue the daughter cell growth impairment caused by CDK1 inhibition (Figure 7E), indicating that the mechanism regulating mitotic growth is also affecting daughter cell growth.
Other major changes:
We have taken advantage of the free eLife manuscript format by reorganizing some of our data. In the current organization, each figure has a more specific message and content, which will make our manuscript more accessible to readers.
As requested by reviewers #2 and #3, we have now addressed how morphological changes (mitotic swelling, rounding and cytokinesis) affect our data. We have now added several completely new figures (Figure 5A,B; Figure 5—figure supplement 1; Figure 5—figure supplement 2) to our manuscript. Experiments in these figures show that when we inhibit mitotic swelling or cytokinetic cell elongation the mitotic MAR dynamics are not radically altered. Thus, the morphological changes in mitosis do not explain the mitotic growth dynamics. Please see subsection “Cells in metaphase and anaphase display mitotic stage specific inhibition of mass accumulation” for full details.
As requested by reviewer #2, we have now studied if mitotic growth rate is purely dependent on the time that cells have spent in mitosis or if there are additional growth reducing mechanism(s) in later mitotic stages (metaphase and anaphase) that inhibit growth. In our new figures (Figure 5C,D; Figure 5—figure supplement 2C) we show that time after mitotic entry does not alone explain the cell growth rate decrease observed during control cell mitosis. Please see subsection “Cells in metaphase and anaphase display mitotic stage specific inhibition of mass accumulation”, as well as our revised Discussion section, for full details.
We are attaching the three reviewers' concerns to guide you in amending your study.
Reviewer #1
1) Throughout the paper, what was being measured, what time-points were acquired and how the different sub-periods of mitosis should be made more explicit. A more consistent nomenclature of periods and timepoints should be used.
Thank you for pointing out these important shortcomings. We have now added clarity to the text in the beginning of our Results section to better define early mitosis (from mitotic entry to metaphase-to-anaphase transition) and cytokinesis (from metaphase-to-anaphase transition to the abscission of daughter cells). We also more clearly state that G2/M transition is the same as mitotic entry. In addition, our Material and methods section has been amended to clarify the points raised by the reviewer.
For example, the authors mention mass accumulation during anaphase but it is unclear how this period was defined, considering the timepoints identified on Figure 1C (it is surprising that growth can be measured during such a short event).
We have now clarified in our main text and Materials and methods section that anaphase refers to the approximate duration of cell elongation (anaphase and telophase overlap in most cells studied).
The periods defined in Figure 1C do not have the same names as the periods on Figure 1D or Figure 2A.
We have changed the naming in Figure 1C to correspond to Figures 1D and 2A. We have also clarified in the beginning of our Results section the terminology used.
It was also unclear how the total cell cycle duration was measured (from the end of abscission? Or from cytokinesis completion?).
We measure the total cell cycle duration by quantifying the duration between successive abscissions (see Figure 1B for an example where the blue arrows indicate the abscissions). We also use the doubling time in cell culture to validate the cell cycle time for the cell types for which we have limited single-cell data.
Perhaps a clearer schematic could be presented that incorporates some elements of Supplementary Figure 1 to the main Figure.
We have incorporated color bars to Figure 1C which correspond to the durations quantified in Figure 1D (i.e. early mitosis and cytokinesis).
Treatments applied to cells were also difficult to follow. Figure 4E,H: is the treatment done throughout the whole measurement?
Thank you for pointing this out. The treatments were applied to the cells 1-4 hours prior to mitotic entry, and the treatments were maintained throughout the whole measurement. We have now clarified this in the figure legends.
Figure 5C-E: Is this 1) a block to synchronize in G2, 2) a treatment with STLC, 3) a release, 4) growth measurement? (the text describing this in subsection “CDK1-driven mitotic protein synthesis supports daughter cell growth” could be made simpler). Again, schematics could be added to the figures to aid the reader.
We have now added a schematic of the workflow to the Figure 7—figure supplement 2A. We have also clarified our Results section “CDK1-driven mitotic protein synthesis supports daughter cell growth” to make this clearer.
2) Figure 2: In Figure 2D, the authors mention that MAR reaches 0 for all cell types. It looks like for S-HeLa and CD3+ T cells, MAR reaches negative values, which would mean that cells loose mass at M/A transition. Could the authors comment, and make sure that this is not a problem from measurement of adherent cells with the SMR?
The reviewer raises an important point. Most importantly, this negative MAR is not due to measurement bias, but reflects a genuine and common feature for many cells. More specifically, we show that:
The cell elongation in anaphase can induce a small bias in our mass measurements. We have now corrected this in all our data and figures using a correction method which we recently published (Kang et al., 2019). Examples comparing corrected and non-corrected data can be found in Figure 1—figure supplement 1E; Figure 2—figure supplement 1D. This correction does not remove the negative MAR observed in some cells, as our buoyant mass measurement is only affected by less than 1%.
If cytokinesis is block without inhibiting mitosis, and thus cell elongation does not take place, MAR can still become negative around the metaphase-to-anaphase transition (please see Figure 5—figure supplement 2B, where we inhibited cytokinesis using blebbistatin).
The negative MAR is more pronounced and observed in more cells if we shorten the window in which we analyze MAR (i.e. increase temporal resolution and sensitivity in MAR analysis). Please see Figure 5B, which displays that also L1210 cells can display negative MAR.
We have now pointed out to the reader in our Results section that we observe this negative MAR, and we have added a paragraph to the Discussion that explain how MAR reflects the flux of components (e.g. nutrients) in and out of the cell. Thus, this negative MAR may arise from, for example, cell blebbing and exocytosis during anaphase, but as we have evidence for this, we do not want to speculate on it.
Figure 2A is used to document a 15.8% increase in MAR in prophase vs. late G2 but it would be better to show a box plot with <MAR> for the 2 windows of time.
We have now added a violin plot to show this MAR increase. Please see our new Figure 2B.
3) Figure 4 is very interesting but requires clarification:
Interpretation of the effects of OTS treatment: It is hypothesized to act only via its off-target effect on CDK1, so why does it have a stronger effect on protein synthesis rate than direct inhibitors of CDK1?
CDK1 is one potential candidate for the mechanisms behind growth reduction upon OTS treatment, as CDK1 is a known OTS off-target (Klaeger et al., 2017, Science). Importantly, this does not exclude other (additional) mechanisms. We have now removed all suggestions that CDK1 is the mediator of OTS effects from our main text and we only point out in the supplementary figure legends that CDK1 is an OTS off-target. Even though we do not know the mechanisms, we wanted to include the OTS data in our manuscript. Previously, the MELK kinase, which is the designated target of the OTS drug, has been implicated in mitotic growth (Wang et al., 2016). Yet, our experiments suggest that MELK is not the mechanisms through which OTS affects mitotic growth since other MELK kinase inhibitors did not show mitosis specific effects on protein synthesis.
Interpretation of the authors for the negative MAR in Figure 4E and H: Does it suggest that the ratio of protein degradation rate/protein synthesis rate becomes positive in the absence of CDK1-driven translation?
We do not have direct evidence on how the protein degradation/synthesis ratio changes in the absence of CDK1-driven translation. However, we have further improved our Result and Discussion section to discuss the negative MAR in more detail (as pointed out above) and to clarify that MAR does not always reflect protein synthesis rates. Please see our updated Discussion for details.
Some of the statements in the text could be clearer: For example, in the third paragraph of subsection “CDK1 drives phosphorylation of 4E-BP1, cap-dependent protein synthesis and mass accumulation in mitosis”, it might be more precise to say: 'cap-dependent protein synthesis persists in mitosis and partially involves CDK1-mediated regulation of cap-dependent translation'. There remains a delta in protein synthesis rate between the inhibition of translation via the inhibition of CDK1 activity using RO-3306 and the inhibition of translation via a direct inhibition of 4EBP1 (Figure 4F), thus suggesting that part of the growth in mitosis involves cap-dependent translation but via another regulatory pathway than through CDK1.
The delta in protein synthesis rate in our old Figure 4F (Figure 6H in our updated manuscript) may be due to the fact that we cannot completely inhibit CDK1 as complete inhibition of CDK1 would prevent cells from entering mitosis. However, we acknowledge our initial statements may have been too strong and we have now amended the text to increase clarity and avoid any overstatements.
4) Figure 5, testing the recovery of growth after the release of CDK1 inhibition during mitosis is an important result that would benefit from a bit more work.
5 c and d: why are the experiments treated separately? It is surprising that the control from exp2 in 5d is similar to the treated condition in exp1. A third experimental replicate for 5c and 5d is recommended to make sure that the results are reproducible. In addition, the number of events compared in the control and treated conditions should be comparable.
The two experiments were shown separately because of the difference in control cell growth rates. As requested by the reviewer, we have now added a third experimental repeat to show that the conclusion (partial CDK1 inhibition in mitosis results in degreased growth rate in daughter cells) is reproducible (Figure 7C, third panel; experiment #3). The number of cells we measure depends on multiple factors, not all of which we have control (For example, channel clogging in the middle of experiment). We have carried out the experiments by using the same experimental time (time after release from cell cycle arrest) to ensure that control and treated cells are comparable (despite the occasional difference in cell numbers).
The plot in 5f is supposed to show recovery of cell growth, but by examining confluency, proliferation, not growth is being measured. The size distribution of the different conditions at different time-points could prove this point (e.g. 1/2 cell cycle duration, 1 and 2 cell cycle durations after the treatment release). It would be even more interesting to add a few time points to the measurement on Figure 5E if feasible.
We agree with the reviewer that confluency is dependent on proliferation. However, in our model proliferation is not fully independent from growth, since proliferation (i.e. doubling time) is correlated with overall growth. To better address the reviewer’s comment, we have carried out two experiments:
As suggested by the reviewer, we have examined G1 cell protein synthesis rates at two time points after mitosis (Figure 7E). We do observe partial recovery of protein synthesis. However, as the cell cycle synchrony is quickly lost, and not all G2 arrested cells progress through mitosis (see our Figure 3A for an example), later time points cannot reliably separate the cells that have undergone mitosis in the presence of mitotic growth inhibition. Thus, the daughter cell growth may also recover more, but our measurements cannot resolve this.
As the reviewer also suggested, we have measured the size distribution of the different conditions at different time points. If we could achieve a perfect cell synchrony, this could reveal volume distribution shifting over time. Unfortunately, when we perturb mitotic growth (whether by temporary mitotic arrest or by partial CDK1 inhibition) we also interfere with the synchrony, as these treatments reduce the% of cells that eventually divide. Consequently, for example the sample that was released from temporary mitotic arrests will display larger average cell size than control sample at 6h timepoint after release because some of the treated cells are not dividing. Thus, we can only say that all conditions return back to normal size distribution 50 hours after release from G2 arrest (see Author response image 1). Because of the limited information that this experiment provides, we do not feel these results are suitable to be presented in our manuscript.

Cell size histograms of control (blue), RO-3306 (red) and STLC (yellow) treated L1210 cells 50 h after release from G2 (or mitotic for STLC) arrest.
Some single-cell experiments datasets were relatively small compared to the control condition. Moreover, the number of replicates in Figure 5A and B was unclear and whether measurements were made on the same day (with the same batch of cells) in the control and the treated conditions. If not, a justification should be provided (e.g. at least showing reproducible MAR measurement distributions on control populations in different experiments).
Thank you for pointing out this lack of clarity in our manuscript. All the replicates for all the drug treatments represent completely independent experiments (i.e. we only grow the cells through one cell division). For controls, where growth is not perturbed, we typically grow the cells through multiple cell division. Due to the low throughput of our measurements, the measurements typically represent data collected over multiple months (for drug treatments) or years (for controls). We have always carried out control experiments between the drug treatment experiments to ensure that the cell growth behavior is reproducible. As requested by the reviewer, we have now added a figure displaying the reproducibility of control experiments to Figure 2—figure supplement 2C. In addition, we have clarified the measurement details and measurement reproducibility in the Methods section.
5) Figure 6: is there only one biological replicate for Figure C-D?
No, the n referred to independent experiments for drug treated cells. But, as suggested by the editors, we have now removed the Figure 6 and all data relating to the asymmetric cell division and the localization of p-4E-BP1.
6) In the Discussion, the authors suggest that the growth during mitosis is comparable to that of prophase. This could be more directly tested with the data the authors already have in Figure 1. The comparison of the different cell lines is interesting but could be improved by comparing the relative mass accumulation to the relative duration of mitosis of each of these cell types. Perhaps by plotting DM_mitosis/DM_total cell cycle vs. DT_mitosis/DT_total cell cycle, or measuring <DM_mitosis/DT_mitosis> and comparing with <(DM_total_cell_cycle – DM_mitosis)/(DT_total_cell_cycle – DT_mitosis)>, or doing the same boxplot as Figure 1D but sorting the cells by increasing DT_mitosis/DT_total cell cycle. Mitosis duration has been shown to function as an insulated period of the cell cycle, with very constant duration across cell types, regardless of the variety of duration of the rest of the cell cycle (see Aurajo et al,. 2016). Perhaps the authors could use their data to see if the same homogeneous behavior holds true for growth. This would raise interesting hypotheses about whether general constraints on optimal energy resource allocation/protein synthesis exist during mitosis that lead to some reproducible growth behavior across cell types.
Thank you for the interesting suggestion. Unfortunately, apart from S-HeLa cells, all the cell lines we analyzed have near identical duration of early mitosis (G2/M transition to M/A transition) and of the whole cell cycle. Thus, all the cell lines will have more or less the same relative duration of mitosis (this includes S-HeLa cells) and, indeed, all the cell lines display similar amount of growth in early mitosis (Figure 1D). We feel that the comparison suggested by the reviewer will only be useful if more cell lines with more variable mitotic or cell cycle durations are analyzed. Instead, we have now pointed out in our Figure 1D legend that the relative growth in early mitosis is very similar between all tested cell types.
7) The statement at the start of the second paragraph of the Discussion that growth is required for cell division has not been proven. The authors do not test directly the requirement for growth to complete cell division. Doing so would require inhibiting growth (independently of the cell cycle proteins) and assessing the rate of mitotic progression (and the rate of asymmetric divisions). The authors could perform this experiment (for example using 4EBP1-I) to make this point more rigorously.
As suggested by the editor, we have now removed the Figure 6 and all data relating to the asymmetric cell division and the localization of p-4E-BP1.
Reviewer #2
One of the main points raised by the reviewer at several occasions is that mitotic growth rate may depend purely on the time spent in mitosis. The reviewer suggests that this could explain why CDK1 inhibitions (which prolong mitosis) and mitotic arrests reduce growth rate. As stated in the major changes section at the beginning of our response letter: We have now studied if mitotic growth rate is purely dependent on the time cells have spent in mitosis or if additional growth reducing mechanism(s), that are specific to later mitotic stages (metaphase and anaphase), inhibit mitotic growth. In our new main figure (Figure 5C,D; Figure 5—figure supplement 2C) we study how growth rate depends on the time cells have spent in mitosis. By comparing the MAR of control cells and prometaphase arrested cells we show that in prometaphase both samples display similar MAR. However, as control cells proceed to metaphase and anaphase, their MAR decreases lower than what is seen in prometaphase arrested cells, although both samples have spent the same time in mitosis. Thus, time after mitotic entry alone does not explain the cell growth rate decrease observed during control cell mitosis. Therefore, additional mechanism(s) have to exist to reduce growth in metaphase and anaphase. Yet, the reviewer is correct that time after mitosis can, and probably will affect mitotic growth rate, as suggested by mitotic arrests. We have now clarified this in our manuscript writing. Please see subsection “Cells in metaphase and anaphase display mitotic stage specific inhibition of mass accumulation” of our Results section as well as our updated Discussion section for full details.
– Can the authors exclude that the morphological changes of the cell that occur during cell rounding in prometaphase affect their measurements, and perhaps cause the small bump in the graph describing cell mass in early prometaphase? Similar for the morphological changes in anaphase / telophase?
The reviewer raises an important concern here. All cell types presented in this work have spherical shapes throughout their whole cell cycle except cytokinesis, therefore making the cell shape changes in prophase minimal. However, cells do increase their volume (~10-20%) during early mitosis, where mitotic cell swelling takes place (Son et al., 2015). As stated in the major changes section at the beginning of our response letter:
We have now addressed how morphological changes (mitotic swelling, rounding and cytokinesis) affect our data. We have added a completely new figures (Figure 5; Figure 5—figure supplement 1,2) to our manuscript, in which we study mitotic growth dynamics when we inhibit mitotic swelling and cytokinetic cell elongation. Although the mitotic swelling is responsible for the increased MAR observed in early mitosis, it does not explain the rest of the MAR dynamics. Furthermore, mitotic protein synthesis dynamics did not display any change when swelling was inhibited. Inhibition of cytokinesis did not change MAR dynamics. Please see subsection “Cells in metaphase and anaphase display mitotic stage specific inhibition of mass accumulation” for full details.
In addition, we have identified that the cell elongation in anaphase can induce a small bias in our mass measurements. We have now corrected this in all our data and figures using a correction method which we recently published (Kang et al., 2019). Examples comparing corrected and non-corrected data can be found in Figure 1—figure supplement 1E; Figure 2—figure supplement 1D. This correction does not remove the negative MAR observed in some cells (our buoyant mass measurement is only affected by <1%).
– Is the reduction in cell growth in metaphase due to metaphase itself, of simply a consequence of being in mitosis for 20-40 min? When the authors delay metaphase onset, for example with a small molecule inhibitor of Eg5, growth progressively slows or halts after 20 min of entering prometaphase. Therefore, it appears that growth does slow down after cells enter prometaphase, but the slow down is somewhat slow and therefore not complete until metaphase.
The reviewer raises an important point, which we address in the beginning of this point-by-point response.
– If I'm correct, HeLa cells are the only non-lymphoid cells they use, and HeLa cells show a fairly strong reduction in growth during mitosis in their data (Figure 2D) (they don't distinguish between prometaphase and metaphase, but if looks like the growth is already reduced in PM). Their Abstract states: "Growth is only stopped in metaphase", but the HeLa cell data (As well as the DT40 cell data) appear to show a strong growth reduction already in prometaphase. Thus, there conclusions should only be limited to lymphoid cells, or the conclusion should better match the data, i.e. the observed slow down of growth in prometaphase.
Thank you for pointing this out. Our original idea with this sentence was that growth being stopped means that MAR becomes (near) zero, but we now realize this may be misleading to the readers. Since we don’t optically monitor the chromosome condensation and alignment of each cell, we cannot pinpoint the exact prometaphase-to-metaphase transition. Also, we don’t have information on the relative duration of each mitotic phase in S-HeLa or DT40 cells, so we cannot conclude if some of the growth reduction would take place already late prometaphase. We have therefore amended our Abstract by stating that growth rates decrease as cells approach metaphase-to-anaphase transition (to avoid claims which we cannot fully validate). Then, in our Results section we specify that this reduction of growth may start in late prometaphase. We have also specified in our results and Discussion sections that our results reflect the growth behavior of suspension grown animal cells.
– The authors are not always very accurate in their description of the different mitotic phases. For example, cells treated with STLC are in prometaphase, not in metaphase. In contrast, MG132 and proTAME to arrest cells in metaphase. Since the main focus of this study is the precise timing of cell growth during different stages of mitosis, such inaccuracies are fairly confusing.
Thank you for clarifying this. We have now amended our text by stating that STLC arrests the cells in a prometaphase state.
– Based on the STLC and Vinblastine etc experiments, I belief it should be concluded that the growth inhibition does not occur specifically in metaphase, but rather occurs after entering mitosis, with some delay. Consistent with this, HeLa cells are slower in reaching anaphase, and show a greater reduction in growth before anaphase.
The reviewer raises an important point, which we address in the beginning of this point-by-point response.
– Throughout the manuscript the authors claim that CDK1 stimulates translation. Yet, in late prometaphase and early metaphase there is plenty of CDK1 activity, but low translation rates. Similarly, in G2 there is low CDK1 activity, but high translation rates. So there is a fairly poor correlation between CDK1 activity and translation rates. This is poorly discussed throughout the manuscript.
Thank you for pointing this out. In G2, for example, signaling components like mTOR are well-known to promote growth through 4E-BP1 and cap-dependent translation. Our data is consistent with this, as inhibition of mTOR reduces protein synthesis and 4E-BP1 phosphorylation in G2. During mitosis CDK1 has been shown to substitute for the role of mTOR (Shuda et al., 2015), which is consistent with our data or CDK1 being required for the phosphorylation of 4E-BP1 and protein synthesis in mitosis. The reviewer is correct that around late prometaphase and metaphase growth slows down although CDK1 remains active. We have now clarified our writing that additional mechanism(s) are needed to explain this decrease in growth rates (please see our updated Discussion). Furthermore, we have provided experimental evidence that CDK1 is responsible for promoting growth in mitosis. As stated in the major changes section at the beginning of our response letter:
4E-BP1 is a negative regulator of cap-dependent translation, and the phosphorylation of 4E-BP1 by mTOR or CDK1 inactivates 4E-BP1. We have now knocked down 4E-BP1 and monitored how this affects mitotic protein synthesis rates. Our new data shows that in the absence of 4E-BP1 CDK1 inhibition no longer reduces mitotic protein synthesis as radically. Thus, this rescue experiment validates that at least part of the CDK1 driven mitotic growth is mediated by 4E-BP1. These new data can be found in our new Figure 6G, Figure 6—figure supplement 4, and in subsection “CDK1 drives phosphorylation of 4E-BP1, protein synthesis and mass accumulation in mitosis”. We have also clarified in our Discussion that other CDK1 dependent and independent growth regulations may also exist, as our data does not rule this out.
– "However, treatment with both 4EGI-1 and RO-3306 did not significantly change mitotic protein synthesis rates when compared to treatment with 4EGI-1alone. Thus, cap-dependent protein synthesis persists in mitosis, and CDK1 regulates growth at least partly through cap-dependent translation."
I don't understand how the authors make this conclusion (and similar ones elsewhere in the manuscript) based on these experiments. Is the reasoning that 4EGI-1 inhibits cap-dependent translation, and in the absence of cap-dependent translation CDK1 inhibition does not reduce protein synthesis, thus CDK1 must be acting through cap-dependent translation? If so, I don't belief that the current data are solid enough to support this conclusion. 4EGI-1 treatment only reduces protein synthesis rates by 50%, both in G2 and M, so clearly there is still substantial cap-dependent translation under 4EGI-1 inhibitor treatment.
The reviewer is correct about the logic of our work (when cap-dependent translation is inhibited, then partial inhibition of CDK1 no longer influences mitotic protein synthesis, as would be expected if CDK1 promotes growth through cap-dependent translation). The fact that approximately 50% of the protein synthesis rates persist in the presence of 4EGI-1 does not make this experiment invalid. However, we do now point out to the reader that 4EGI-1 reduced translation rates approximately 50%. In addition, our new experimental evidence detailed in the major changes section at the beginning of our response letter (and in the previous answer) shows that CDK1 is at least partly promoting translation through 4E-BP1, a controller of cap-dependent translation. While we believe that this data validates our claims that “CDK1 regulates growth at least partly through cap-dependent translation”, we have also changed our manuscript text to more clearly point out that we cannot rule out other CDK1 dependent and independent mechanisms (please see our updated Discussion section).
– Did the authors examine the time from mitotic entry to the metaphase-to-anaphase transition in control cells vs cells treated with a low dose of CDK1 inhibitor? It is certainly possible that chromosome alignment is perturbed under these conditions, so the time from NEB to the M-A transition is longer under partial CDK1 inhibition. All the data in this study shows that translation rates steadily decrease either from the moment of mitotic entry, or with a ~20 min delay (apparently depending on the cell line), including STLC treatment.
– Therefore, a delay in prometaphase induced by the CDK1 inhibitor would be sufficient to explain the observed decrease in the protein synthesis rates. This is especially true, because the authors align their traces at the M-A transition; if CDK1 inhibition slows down chromosome alignment, then cells that are, say 10 min before the M-A transition in control cells have been in mitosis shorter than cells 10 min before the M-A transition in CDK1 inhibited cells, thus potentially explaining why CDK1 inhibitor treatment apparently reduced translation.
Again, the reviewer raises a good point. We have examined the time from mitotic entry to metaphase-anaphase transition in cells treated with RO3306, and the reviewer is correct that the duration of early mitosis is increased under partial CDK1 inhibition, especially in the L1210 cells. However, this alone does not explain the reduced growth in early mitosis, because partial CDK1 inhibition also removes the growth rate increase seen in control cells after mitotic entry (note that mitotic arrests, like STLC treatment, still display an increased growth during early mitosis). In addition, as discussed above, we have now knocked down 4E-BP1 and show that this partly rescues RO3306 mediated mitotic growth inhibition, indicating that CDK1 regulates growth at least partly through 4E-BP1 (and not only by lengthening mitosis). Importantly, this does not rule out the option that reviewer suggests (i.e. time after mitotic entry could, and probably will affect growth rate). We have now clarified to the reader that partial CDK1 inhibition also increased the duration of early mitosis (subsection “CDK1 drives phosphorylation of 4E-BP1, protein synthesis and mass accumulation in mitosis” of our Results section) and we point out the mitotic entry in the MAR traces where CDK1 is inhibited (please see arrows in Figures 6D,F).
– Subsection “CDK1-driven mitotic protein synthesis supports daughter cell growth”: "…the growth reduction in daughter cells must reflect a consequence of mitotic growth inhibition".
This is an overstatement, CDK1 inhibition does all kinds of things besides inhibition of growth. For example, it may delay mitosis, which activates p53, among other things and has consequences in G1 (Uetake and Sluder, 2010, Curr Biol), it could induce chromosome missegregation, or change the levels or localization of different proteins. Thus, attributing the effects of growth in daughter cells specifically to the role of CDK1 in stimulating mitotic translation or growth seems premature.
Thank you for pointing this out. We realize that our original statement was too bold and thus we have now softened our claims. However, we have also provided additional data to strengthen the relationship between CDK1 driven mitotic growth and daughter cell growth. As stated in the major changes section at the beginning of our response letter:
Using the 4E-BP1 knockdown to rescue mitotic growth effects of CDK1 inhibition, we now show that daughter cell protein synthesis rates are less impaired in 4E-BP1 knockdown cells than in control knockdown cells. Thus, rescuing the mitotic growth inhibition also rescues the daughter cell growth inhibition. However, these rescues were not complete, and we have now clarified our text by pointing out that other mechanisms may also be involved. Please see our new Figure 7E, subsection “CDK1-driven mitotic protein synthesis supports daughter cell growth” and also the discussion of our manuscript.
– Figure 6. Just because 4EBP1 localizes a bit more to the cleavage plane, does not prove that translation is preferentially occurring there. An OPP assay could be performed to support this claim.
As suggested by the editor, we have now removed the Figure 6 and all data relating to the asymmetric cell division and the localization of p-4E-BP1.
– "Control cell divisions were mostly symmetrical (Figure 6C). However, when we analyzed cells treated with RO-3306, OTSSP167 or cycloheximide, we observed that all these treatments resulted in more asymmetric cell divisions"
CDK1 phosphorylates hundreds of proteins during mitosis, including many proteins that regulate microtubule and spindle function. Thus, the spindle is probably also not functioning completely normally after CDK1 inhibition, which is far more likely to be the cause of the asymmetric cell division. Therefore, I don't think that linking the observed affects on cell growth after CDK1 inhibition to the asymmetric divisions is warranted.
As suggested by the editor, we have now removed the Figure 6 and all data relating to the asymmetric cell division and the localization of p-4E-BP1.
– "Our single-cell measurements show that mass accumulation behavior during mitosis is dynamic, dependent on mitotic stage"
I disagree that the data proves that mass accumulation depends on the mitotic stage per se. The rate of mass accumulation appears to depend mainly on the time since entering mitosis, with a gradual shutdown of protein synthesis and growth occurring after mitotic entry. This conclusion is supported by the strong growth inhibition before the M-A transition in HeLa cells, as well as the strong growth inhibition in STLC treated cells, which never reach metaphase.
The reviewer raises an important point, which we address in the beginning of this point-by-point response.
– "Indeed, we show that growth is only stopped during anaphase"
Not correct, in all cell lines, growth is already stopped in metaphase, in many cell lines, it is already strongly reduced in prometaphase.
Thank you for pointing out the different interpretation of the data and text. As stated above, since we don’t optically monitor the chromosome condensation and alignment of each cell, we cannot pinpoint the exact prometaphase-to-metaphase transition. Also, we don’t have information on the relative duration of each mitotic phase in S-HeLa or DT40 cells, so we cannot conclude if some of the growth reduction would take place already late prometaphase. We have therefore amended our Abstract by stating that growth rates decrease as cells approach metaphase-to-anaphase transition (to avoid claims which we cannot fully validate). Then, in our Results section we specify that this reduction of growth may start in late prometaphase.
– "Mechanistically, we found that CDK1 drives M-phase growth through 4E-BP1 and cap-dependent protein synthesis"
As stated above, I don't think this is shown. In G2 there is very low CDK1 activity, yet strong growth and translation. Similarly, in late prometaphase/metaphase there is also strong CDK1 activity, yet growth gets downregulated.
As stated above: In G2 signaling components like mTOR are well-known to promote growth through 4E-BP1 and cap-dependent translation. Our data is consistent with this, as inhibition of mTOR reduces protein synthesis and 4E-BP1 phosphorylation in G2. During mitosis CDK1 has been shown to substitute for the role of mTOR (Shuda et al., 2015), which is consistent with our data or CDK1 being required for the phosphorylation of 4E-BP1 and protein synthesis in mitosis. Now, the reviewer is correct that around late prometaphase and metaphase growth slows down although CDK1 remains active. We have now clarified our writing that additional mechanism(s) are needed to explain this decrease in growth rates. Furthermore, we have provided experimental evidence that CDK1 is responsible for promoting growth in mitosis. As stated in the major changes section at the beginning of our response letter:
4E-BP1 is a negative regulator of cap-dependent translation, and the phosphorylation of 4E-BP1 by mTOR or CDK1 inactivates 4E-BP1. We have now knocked down 4E-BP1 and monitored how this affects mitotic protein synthesis rates. Our new data shows that in the absence of 4E-BP1 CDK1 inhibition no longer reduces mitotic protein synthesis as radically. Thus, this rescue experiment validates that at least part of the CDK1 driven mitotic growth is mediated by 4E-BP1. Please see our new Figure 6G, Figure 6—figure supplement 4, and subsection “CDK1 drives phosphorylation of 4E-BP1, protein synthesis and mass accumulation in mitosis”.
We have clarified in our Discussion that other CDK1 dependent and independent growth regulations may also exist, as our data does not rule this out.
– The authors should be careful not to be selective in their choice of the different studies examining translation in mitosis to find support for their conclusions, for example:
"This is consistent with the observations that transcription and translation of
growth related genes are prioritized during mitosis". They do not discuss Stumpf et al. and Tanenbaum et al. here, who performed genome-wide analysis of translation in mitosis, and which are discussed in other parts of the manuscript.
Thank you for pointing out this shortcoming of our writing. We have now added a note to our Discussion stating that “it should be noted that the active mitotic translation of growth promoting components remains controversial (Stumpf et al., 2013; Tanenbaum et al., 2015)”.
– Title: I'm not sure that the title is sufficiently supported by the data. While the authors provide one experiment that translation is needed for cell division symmetry; cells treated with CHX (As discussed above, I don't think their inhibition of CDK1 experiments prove that cell growth in M is needed for division symmetry), they do not show that "cell growth" is needed for cell division symmetry. The authors themselves state the translation and cell growth are not the same: "Importantly, protein and RNA synthesis rates are only proxies of overall growth (biomass increase), which is determined by the balance between synthesis (anabolic) and degradation (catabolic) rates"
We understand that concerns raised by the reviewer and, as suggested by the editors, we have now removed our last main figure and all data relating to the asymmetric cell division and the localization of p-4E-BP1. Consequently, we have changed our title to “Mammalian cell growth dynamics in mitosis”.
Reviewer #3
1) The main concern is the specificity of the effect of CDK1 inhibition on early mitotic growth. The authors show by multiple approaches that CDK1 activity is required for the maintenance of growth during early mitosis – this is fine. They then propose that it is because CDK1 phosphorylates 4E-BP1. Nevertheless, they do not formally prove this point, and CDK1 could act via other pathways. The authors should amend their text to make it clear that they did not prove that the early mitosis growth is due to CDK1 acting via 4E-BP1.
The reviewer raises an important concern. We have now addressed this concern both experimentally and by rewriting our manuscript, as stated in the major changes section at the beginning of our response letter:
4E-BP1 is a negative regulator of cap-dependent translation, and the phosphorylation of 4E-BP1 by mTOR or CDK1 inactivates 4E-BP1. We have now knocked down 4E-BP1 and monitored how this affects mitotic protein synthesis rates. Our new data shows that in the absence of 4E-BP1 CDK1 inhibition no longer reduces mitotic protein synthesis as radically. Thus, this rescue experiment validates that at least part of the CDK1 driven mitotic growth is mediated by 4E-BP1. Please see our new Figure 6G, Figure 6—figure supplement 4, and subsection “CDK1 drives phosphorylation of 4E-BP1, protein synthesis and mass accumulation in mitosis”.
We have clarified in our Discussion that other CDK1 dependent and independent growth regulations may also exist, as our data does not rule this out.
They should also discuss the potential role of mitotic swelling, which they showed is also CDK1 dependent, and could affect early mitotic growth rate – when cells enter mitosis, they increase volume in a CDK1 dependent manner. One would thus expect that inhibiting CDK1 activity, which would prevent this swelling, could have an effect on overall cell density and affect the growth rate in this early mitotic phase.
The reviewer raises another important concern. As stated in the major changes section at the beginning of our response letter:
We have now addressed how morphological changes (mitotic swelling, rounding and cytokinesis) affect our data. We have added completely new figures (Figure 5; Figure 5—figure supplement 1,2) to our manuscript, in which we study mitotic growth dynamics when we inhibit mitotic swelling. These experiments show that mitotic swelling is responsible for the increased MAR observed in early mitosis, but not for the rest of the MAR dynamics. Furthermore, mitotic protein synthesis dynamics did not display any change when swelling was inhibited. Please see subsection “Cells in metaphase and anaphase display mitotic stage specific inhibition of mass accumulation” for full details.
Notably, as we show in our previous work (Son et al., 2015), partial CDK1 inhibition with 1μM RO-3306 (same concentration used in this work) does not significantly change the magnitude of mitotic swelling in L1210 cells. Despite this, the partial CDK1 inhibition removes the increased growth observed in control cells after mitotic entry, suggesting that CDK1 inhibition results in reduction of growth that is independent of mitotic cell swelling.
2) One consequence of the previous concern is that the last piece of data provided is not really easy to interpret. It is in fact unlikely that the effect of CDK1 inhibition on the asymmetry of division is due to its effect on mitotic growth. This effect of CDK1 inhibition on mitotic spindle positioning, or in fact of almost any drug inducing a mitotic delay, as already been reported and rather depends on the importance of CDK activity for other processes that are needed to orient the spindle properly (see for example Beamish JBC 2009). I would remove this part, which does not bring much to the article and is unlikely to be correct – it is probably not because of a lack of early mitotic growth that the division ends up being asymmetrical upon CDK1 inhibition. In my opinion, the conclusion drawn “Together, these results show that CDK1 and the maintenance of active translation in M-phase are required for the fidelity of cytokinesis” is not supported by the data. But it is really not important for the rest of the article.
3) Similarly, I would remove the very short chapter on the localization of 4E-BP1. It is not clear how it helps the main message of the article.
Thank you for pointing out these concerns. As suggested by the editors, we have now removed our last main figure and all data relating to the asymmetric cell division and the localization of p-4E-BP1.
4) To help the reader estimate the global effect of inhibiting early mitotic growth, the authors should provide, just like in Figure 1, the effective growth in mitosis (in% and not just the MAR), cytokinesis and early G1 comparing control and cells in which early mitotic growth has been inhibited.
Thank you for the suggestion. However, our experimental approach does not reveal where G1 ends. Thus, we cannot reveal the total growth during G1, only the growth rate at the beginning (first 30min) of G1. In order to do what the reviewer is suggesting, we would have to fix the duration analyzed for G1. As also the duration of early mitosis is relatively constant (Araujo et al., 2016), the early mitosis and G1 phases would display results comparable to our current MAR analysis (please see Author response image 2). We believe this would not provide additional information to the reader and therefore we do not present it in our manuscript.

Comparison of cell growth in early mitosis, cytokinesis and early G1 after indicated drug treatments.
Normalized mass accumulation rate is displayed on top, while total mass accumulated is displayed on the bottom. Both formats of data presentation lead to the same conclusion.
[Editors' note: further revisions were requested prior to acceptance, as described below.]
Essential revisions:
The 4E-BP1 knockdown experiment does not really prove that translation in mitosis is needed for growth in G1. One could still envision a scenario in which CDK1 inhibition causes some sort of stress in G1 phase, either through an increased mitotic duration or for example through induction of chromosome missegregation (or DNA damage associated with chromosome missegregation). Such stress could trigger a translational downregulation in G1, which is (partially) restored by depletion of 4E-BP1. In this scenario it is not the role of CDK1 in mitotic translation / growth that is responsible for slower G1, but another function of CDK1. Also, the depletion of 4E-BP1 does not increase G1 translation through its activity in M phase, but through its activity in G1 phase in this scenario.
The authors should discuss this alternative possibility if they cannot fully exclude it.
We would like to thank the reviewers and editors for pointing out that alternative explanations exist that might explain how CDK1 inhibition affects daughter cell growth. We agree with the criticism and we have revised the results and Discussion section accordingly. First, in the Results section, we now state that “Thus, 4E-BP1 mediates mitotic protein synthesis of CDK1 (Figure 6), and this may also affect daughter cell protein synthesis. However, we cannot fully exclude the possibility that daughter cell growth is affected by some other effects of partial CDK1 inhibition, such as chromosome missegregation, which could consequently reduce daughter cell growth independently of mitotic growth.” In addition to this, we have also amended our Discussion section so that “CDK1 activity”, not necessarily “CDK1-driven mitotic growth”, affected daughter cell growth. We also state that “CDK1 inhibition may reduce daughter cell growth independently of mitotic growth”.
https://doi.org/10.7554/eLife.44700.028Article and author information
Author details
Funding
Wellcome (110275/Z/15/Z)
- Teemu P Miettinen
National Cancer Institute (CA217377)
- Scott R Manalis
Koch Institute Frontier Research Program (P30-CA14051)
- Scott R Manalis
Samsung Scholarship
- Joon Ho Kang
The authors declare that the funders had no involvement in study design, data collection, interpretation or presentation.
Acknowledgements
We thank from Dr Samejima and Dr Earnshaw for providing the DT40 CDK1as cells and Dr Elias for providing the S-Hela cells. We thank E Vasile for technical support with microscopy, L Atta for assistance in isolating primary T-cells, and M Björklund, P Winter and L Mu for useful comments on the manuscript. This work was supported by Koch Institute Frontier Research Program, Koch Institute Support (core) Grant P30-CA14051 and Cancer Systems Biology Consortium U54 CA217377 from the National Cancer Institute. TPM is supported by the Wellcome Trust Sir Henry Postdoctoral Fellowship grant 110275/Z/15/Z. JHK acknowledges support from Samsung scholarship.
Senior Editor
- Jonathan A Cooper, Fred Hutchinson Cancer Research Center, United States
Reviewing Editor
- Jon Pines, Institute of Cancer Research Research, United Kingdom
Reviewers
- Marvin Tanenbaum, Hubrecht Institute,, Netherlands
- Matthieu Piel, Institut Curie, France
Version history
- Received: December 24, 2018
- Accepted: May 5, 2019
- Accepted Manuscript published: May 7, 2019 (version 1)
- Version of Record published: May 24, 2019 (version 2)
Copyright
© 2019, Miettinen et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 6,489
- Page views
-
- 708
- Downloads
-
- 31
- Citations
Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.
Download links
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Further reading
-
- Cell Biology
- Developmental Biology
Cylicins are testis-specific proteins, which are exclusively expressed during spermiogenesis. In mice and humans, two Cylicins, the gonosomal X-linked Cylicin 1 (Cylc1/CYLC1) and the autosomal Cylicin 2 (Cylc2/CYLC2) genes, have been identified. Cylicins are cytoskeletal proteins with an overall positive charge due to lysine-rich repeats. While Cylicins have been localized in the acrosomal region of round spermatids, they resemble a major component of the calyx within the perinuclear theca at the posterior part of mature sperm nuclei. However, the role of Cylicins during spermiogenesis has not yet been investigated. Here, we applied CRISPR/Cas9-mediated gene editing in zygotes to establish Cylc1- and Cylc2-deficient mouse lines as a model to study the function of these proteins. Cylc1 deficiency resulted in male subfertility, whereas Cylc2-/-, Cylc1-/yCylc2+/-, and Cylc1-/yCylc2-/- males were infertile. Phenotypical characterization revealed that loss of Cylicins prevents proper calyx assembly during spermiogenesis. This results in decreased epididymal sperm counts, impaired shedding of excess cytoplasm, and severe structural malformations, ultimately resulting in impaired sperm motility. Furthermore, exome sequencing identified an infertile man with a hemizygous variant in CYLC1 and a heterozygous variant in CYLC2, displaying morphological abnormalities of the sperm including the absence of the acrosome. Thus, our study highlights the relevance and importance of Cylicins for spermiogenic remodeling and male fertility in human and mouse, and provides the basis for further studies on unraveling the complex molecular interactions between perinuclear theca proteins required during spermiogenesis.
-
- Cell Biology
- Structural Biology and Molecular Biophysics
Previously we showed that 2D template matching (2DTM) can be used to localize macromolecular complexes in images recorded by cryogenic electron microscopy (cryo-EM) with high precision, even in the presence of noise and cellular background (Lucas et al., 2021; Lucas et al., 2022). Here, we show that once localized, these particles may be averaged together to generate high-resolution 3D reconstructions. However, regions included in the template may suffer from template bias, leading to inflated resolution estimates and making the interpretation of high-resolution features unreliable. We evaluate conditions that minimize template bias while retaining the benefits of high-precision localization, and we show that molecular features not present in the template can be reconstructed at high resolution from targets found by 2DTM, extending prior work at low-resolution. Moreover, we present a quantitative metric for template bias to aid the interpretation of 3D reconstructions calculated with particles localized using high-resolution templates and fine angular sampling.