Skip to main content
eLife logoLink to eLife
. 2022 Feb 1;11:e71356. doi: 10.7554/eLife.71356

Time-resolved single-cell sequencing identifies multiple waves of mRNA decay during the mitosis-to-G1 phase transition

Lenno Krenning 1,†,, Stijn Sonneveld 1,, Marvin E Tanenbaum 1,
Editors: Robert H Singer2, James L Manley3
PMCID: PMC8806192  PMID: 35103592

Abstract

Accurate control of the cell cycle is critical for development and tissue homeostasis, and requires precisely timed expression of many genes. Cell cycle gene expression is regulated through transcriptional and translational control, as well as through regulated protein degradation. Here, we show that widespread and temporally controlled mRNA decay acts as an additional mechanism for gene expression regulation during the cell cycle in human cells. We find that two waves of mRNA decay occur sequentially during the mitosis-to-G1 phase transition, and we identify the deadenylase CNOT1 as a factor that contributes to mRNA decay during this cell cycle transition. Collectively, our data show that, akin to protein degradation, scheduled mRNA decay helps to reshape cell cycle gene expression as cells move from mitosis into G1 phase.

Research organism: Human

Introduction

Cell division is essential for the development and homeostasis of multicellular organisms. Precise control over cell division is paramount, as errors may contribute to carcinogenesis (Hanahan and Weinberg, 2011; Malumbres and Barbacid, 2001). In order to divide, cells pass through a number of different phases, collectively referred to as the cell cycle. The cell cycle in somatic cells consists of four phases: (1) in G1 phase a cell grows and prepares for DNA replication; (2) in S phase the DNA is replicated; (3) in G2 phase a cell prepares for segregation of the replicated genome; (4) in M phase (or mitosis) the cell divides and can then either enter into G1 phase of the next cell cycle or it can (temporarily) exit the cell cycle and enter into G0 phase (i.e. quiescence). Progression through the cell cycle is accompanied by the periodic expression of many genes (referred to as cell cycle genes), whose protein products are likely required in a particular cell cycle phase (Bar-Joseph et al., 2008; Chaudhry et al., 2002; Cho et al., 2001; Cho et al., 1998; Grant et al., 2013; Whitfield et al., 2002). Deregulated expression of cell cycle genes can decrease the fidelity of cell division. For instance, reduced expression of G2 and M phase cell cycle genes impedes mitotic entry and affects the fidelity of chromosome segregation (Laoukili et al., 2005). Conversely, a failure to suppress expression of G2 and M phase genes as cells enter G1 phase can result in a shortened G1 phase and cause DNA replication errors (García-Higuera et al., 2008; Park et al., 2008; Sigl et al., 2009), and can even contribute to carcinogenesis (Bortner and Rosenberg, 1997; Coelho et al., 2015; Kalin et al., 2006; Kim et al., 2006; Vaidyanathan et al., 2016). These examples highlight the importance of tightly controlled gene expression for proper execution of the cell cycle.

To restrict cell cycle gene expression to the correct cell cycle phase, cells need to activate, but also repress the expression of cell cycle genes as they move from one phase to the next. Scheduled protein degradation plays an important role in repression of cell cycle gene expression by ensuring that protein expression is restricted to the appropriate cell cycle phase (Nakayama and Nakayama, 2006; Vodermaier, 2004). In addition, cells prevent de novo synthesis of proteins through inhibition of transcription to further restrict protein expression to the correct cell cycle phase (Bertoli et al., 2013; Sadasivam and DeCaprio, 2013). While inhibition of transcription will eventually lower mRNA levels and thus decrease protein synthesis rates, this process is relatively slow, as it requires turnover of the existing pool of mRNAs. To circumvent this, cells can shut down translation or degrade pre-existing transcripts when transitioning from one cell cycle phase to another. Indeed, control of mRNA translation also contributes to the regulation of gene expression during the cell cycle (Kronja and Orr-Weaver, 2011) and several hundreds of genes are subject to translational regulation at different phases of the cell cycle (Stumpf et al., 2013; Tanenbaum et al., 2015).

Regulation of mRNA stability during the cell cycle has been studied relatively little, but recent work suggests that this type of regulation also contributes to restriction of cell cycle gene expression; dynamic changes in mRNA stability during the cell cycle were observed in yeast using fluorescent in situ hybridization (FISH). Specifically, CLB2 and SWI5 mRNAs were shown to be degraded during mitosis (Trcek et al., 2011). Globally, mRNA synthesis and decay rates during the cell cycle of yeast were derived through metabolic mRNA labeling in synchronized populations, resulting in the identification of several hundred genes that show periodic changes in mRNA synthesis and degradation rates (Eser et al., 2014). Regulation of mRNA stability is also reported to occur during the human cell cycle. For instance, the transcription factor ERG was shown to control the degradation of a set of mRNAs during S phase (Rambout et al., 2016). Recently, global mRNA synthesis and degradation rates during the human cell cycle were determined (Battich et al., 2020). In this study, a newly developed method using a pulse-chase labeling approach simultaneously quantifies metabolically labeled and pre-existing unlabeled transcripts in individual cells. This method was used to determine synthesis and degradation rates of individual transcripts during the cell cycle. Together, these studies demonstrate that the stability of many mRNAs change during the cell cycle. However, due to the relatively long measurement time required for pulse-chase approaches (up to 6 hr), accurate dynamics and rapid changes, especially around the transition points in the cell cycle, are difficult to determine.

To obtain a highly quantitative view of transcriptome dynamics during cell cycle phase transitions, we established a method that combines singe-cell mRNA sequencing and live-cell imaging of cell cycle progression to map transcriptome-wide mRNA expression levels with high temporal resolution during the cell cycle. We focus specifically on the mitosis-to-G1 (M-G1) phase transition, during which cells divide and enter into a new cell cycle. This cell cycle phase transition was selected as gene expression needs to be ‘reset’ after cell division, requiring major changes to gene expression. The widespread protein degradation that occurs during the M-G1 phase transition is thought to contribute to this reset (Castro et al., 2005; Harper et al., 2002; Peters, 2002; Vodermaier, 2004). We hypothesized that, analogous to scheduled protein degradation, mRNA decay might play an important role in resetting cell cycle gene expression by limiting the carry-over of pre-existing G2/M-specific transcripts from one cell cycle into the next. Using our method, we identified two temporally distinct waves of mRNA decay: the first wave is initiated during mitosis and the second wave is initiated within the first hours of G1 phase. For several of these genes, we show that mRNA decay is stimulated by CNOT1, a subunit of the CCR4-NOT mRNA deadenylase complex that shortens the poly(A) tail of mRNAs, generally resulting in their decay (Garneau et al., 2007; Yamashita et al., 2005). Together, our findings demonstrate that, analogous to protein degradation, scheduled mRNA degradation occurs at the M-G1 phase transition. Scheduled mRNA degradation likely provides an important contribution to the reset of the transcriptome after cell division.

Results

Time-resolved transcriptome profiling during the cell cycle using the FUCCI system

To obtain a detailed view of mRNA levels as cells progress from M phase into G1 phase, we developed a method that connects live-cell microscopy with single-cell RNA sequencing (scRNA-seq), through fluorescence activated cell sorting (FACS). This method allows us to assign an accurate, ‘absolute’ cell cycle time (i.e. the time in minutes since the completion of metaphase) to individual sequenced cells, which we used to generate a high-resolution, time-resolved transcriptome profile of the M-G1 phase transition.

To assign an absolute cell cycle time for each cell, we expressed the fluorescent, ubiquitination-based cell cycle indicator (FUCCI) system in a human untransformed cell line, RPE-1 (RPE-FUCCI). In the FUCCI system an orange fluorescent protein (FUCCI-G1) is expressed in G1 and early S phase cells, as well as in G0 phase (quiescence), while a green fluorescent protein (FUCCI-G2) is expressed in late S, G2, and early M phase (Figure 1A–C, Figure 1—figure supplement 1A and Figure 1—video 1; Sakaue-Sawano et al., 2008). Importantly, the expression levels of both fluorescent markers change over time even within each cell cycle phase, potentially allowing precise pinpointing of the cell cycle time of individual cells based on the fluorescence intensity of the FUCCI reporter. We used live-cell microscopy to measure FUCCI-G1 fluorescence intensity in cells as they progressed through G1 phase, which revealed a monotonic increase during the first 6–8 hr after G1 entry (Figure 1—figure supplement 1B). To allow accurate calculation of a cell cycle time based on FUCCI-G1 fluorescence intensity, we fit the average FUCCI-G1 fluorescence intensity to a polynomial equation (Figure 1D and Supplementary file 1), which generates an accurate mathematical description of the data. Using this equation, a cell cycle time can be calculated for each cell based on its fluorescence intensity of the FUCCI-G1 marker. Cell cycle times of individual cells can be accurately determined during the first ~5 hr of G1/G0 phase, after which prediction accuracy decreases due to the increased cell-to-cell heterogeneity in FUCCI-G1 fluorescence at later time points in G1 phase (Figure 1—figure supplement 1B-C).

Figure 1. A method for time-resolved transcriptome analysis during the cell cycle.

(A) Schematic representation of the human cell cycle and the fluorescent, ubiquitination-based cell cycle indicator (FUCCI) system. (B) Representative images of RPE-FUCCI cells throughout the cell cycle (see Figure 1—video 1). RPE-FUCCI cells were incubated for 2 hr with SPY650-DNA to visualize the DNA. Cells were imaged every 15 min for a duration of 15 hr. Arrows indicate a single cell undergoing a complete cell cycle. (C) Fluorescence activated cell sorting (FACS) analysis of asynchronously growing RPE-FUCCI cells. Dashed boxes indicate the gating strategy used for the identification and isolation of cells in G1/G0, early S phase, mid/late S phase, and G2/M phases. (D) Modeling of FUCCI-G1 fluorescence intensities. Asynchronously growing RPE-FUCCI cells were analyzed by live-cell imaging (Figure 1—figure supplement 1A). Subsequently, FUCCI-G1 fluorescence intensities were measured and normalized to the average FUCCI-G1 fluorescence in early S phase cells (see Materials and methods). Red squares and shading represent the mean fluorescence and SEM, respectively, of three individual experiments. The mean FUCCI-G1 fluorescence was fit to a third-order polynomial (black line, equation above plot). The fit has no biological meaning, but serves to approximate the data to allow calculation of G1 phase cell cycle times based on FUCCI-G1 fluorescence intensity. (E) Differential gene expression analysis of RPE-FUCCI cells in G2/M phase versus G1 phase. Cells were ordered based on FUCCI fluorescence and differential gene expression analysis was performed using Monocle2 (see Materials and methods). (F) Venn diagram comparing differentially expressed genes (both up- and downregulated in G1 versus G2/M phase) identified after FUCCI- or Monocle2-based cell ordering. (G) Comparison of FUCCI- and Monocle2-based ordering of G1 phase cells. Dashed red line indicates identical order of cells. (H) Comparison of G1 phase cell cycle time from FUCCI-based ordering with G1 pseudo time based on trajectory inference by Monocle2. Both FUCCI-G1 phase cell cycle time and Monocle2 G1 pseudo time are normalized to values between 0 and 1 for comparison. Dashed line indicates identical timing of FUCCI and Monocle2.

Figure 1.

Figure 1—figure supplement 1. A method for time-resolved transcriptome analysis during the cell cycle.

Figure 1—figure supplement 1.

(A) Histogram of the DNA content of RPE-fluorescent, ubiquitination-based cell cycle indicator (FUCCI) cells. RPE-FUCCI cells were stained with Hoechst 33,342 for 30 min before fluorescence activated cell sorting (FACS) analysis. Four different gates were set based on FUCCI fluorescence, gating G1, early S, S, and G2/M phase cells. For each FUCCI-based gate we calculated the average Hoechst fluorescence intensity, relative to the intensity of G1 phase cells. Shown is one representative of three experiments. p-Value is based on a one-tailed unpaired Student’s t-test. p-Value is indicated as *** (p < 0.001). (B) Fluorescence microscopy time traces of RPE-1 cells expressing FUCCI-G1 and FUCCI-G2 markers. Asynchronously growing RPE-FUCCI cells were imaged every 5 min and the average nuclear intensities for both FUCCI markers were measured (red and green lines, respectively). Dark red and green lines represent the average of 30 cells, light red and green lines represent individual cells of one representative experiment of three experiments. For quantification of FUCCI-G1 fluorescence during G1 phase, cells that entered S phase within the 10 hr window post-mitosis were excluded, because the FUCCI-G1 reporter is rapidly degraded upon S phase entry. S phase entry is identified based on an increase in FUCCI-G2 fluorescence (Figure 1A–C and Figure 1—figure supplement 1A). See Supplementary file 2 for the number of cells included in each condition. (C) Measurement error in the G1 phase cell cycle time due to cell-to-cell variability in FUCCI-G1 fluorescence intensities. We calculated the standard deviation of the G1 phase cell cycle time (see Materials and methods). Since cell-to-cell variability in FUCCI-G1 fluorescence increases at later time points in G1 phase (Figure 1D and (B)), the measurement error of G1 phase cell cycle time increases at later time points in G1 phase. Red squares are individual data points. (D) Microscopy-based analysis of FUCCI fluorescence intensities in RPE-FUCCI cells. Early S phase cells were identified based on the presence of high FUCCI-G1 and low FUCCI-G2 fluorescence (yellow dots), and the FUCCI-G1 fluorescence intensity of early S phase cells was recorded (see Materials and methods). A representative experiment is shown (three experiments were performed with at least 700 cells quantified per experiment). (E) FACS plots of FUCCI fluorescence after different durations of Taxol treatment. To quantify the increase in FUCCI-G1 fluorescence over time, asynchronously growing RPE-FUCCI cells were treated with Taxol for the indicated durations to arrest cells in mitosis, thus preventing new cells from entering G1 phase. Over time, FUCCI-G1 fluorescence increases which results in the gradual loss of cells with low FUCCI-G1 fluorescence. The lowest FUCCI-G1 fluorescence intensity after a 1, 2, or 4 hr incubation with Taxol were identified (dotted lines), and used to calculate the FUCCI-G1 fluorescence at various times during G1 phase relative to early S phase (see Materials and methods). (F) Comparison of the relative FUCCI-G1 fluorescence intensities as determined by the polynomial equation (Figure 1D) and by Taxol treatment of cells followed by FACS (see panel S1E) at 1, 2, and 4 hr after mitosis. Bars and error bars indicate average ± SEM of three experiments. (G–I) Comparison of normalized read counts in G2/M from three sequencing experiments. Each dot represents the expression of one gene averaged over all cells in G2/M phase. Dotted red line indicates identical read counts in two experiments. (J) Histogram showing the position in the cell cycle of all cells subjected to single-cell RNA sequencing (scRNA-seq). The position in the cell cycle was determined based on the FUCCI-G1 fluorescence intensity as measured by FACS. (K) Gene Ontology term analysis of genes downregulated in G1 phase compared to G2/M phase. For this analysis, cells were grouped into either G2/M or G1 phase based on FUCCI fluorescence. (L) Pie chart showing the fraction of all genes with known cell cycle functions (derived from Cyclebase 3.0) that are up- or downregulated in G1 phase compared to G2/M phase, or that show no change in expression. (M) Single-cell trajectory of the mitosis-to-G1 (M-G1) phase transition constructed by Monocle2. Colors indicate the cell cycle position based on FUCCI-G1 fluorescence. (N) Differential transcriptome analysis of G1 phase versus G2/M phase RPE-FUCCI cells, aligned based on Monocle2 trajectory inference.
Figure 1—video 1. Example movie of RPE-1 cells expressing the G1 and G2 fluorescent, ubiquitination-based cell cycle indicator (FUCCI) reporters.
Download video file (968KB, mp4)
Video of asynchronously growing RPE-FUCCI cells imaged every 15 min. RPE-FUCCI cells were incubated with SPY650-DNA for 2 hr prior to imaging to visualize DNA. To facilitate visualization of the FUCCI reporters, SPY650-DNA is not shown in the video, but is included in the still images shown in Figure 1B.

In the scRNA-seq protocol the FUCCI fluorescence intensity of each sequenced cell is measured by FACS during FACS of cells into 384-well plates. To assess precise cell cycle times based on FUCCI fluorescence intensities measured by FACS, fluorescence intensities measured by FACS need to be compared with those obtained by microscopy. For this, we normalized FUCCI-G1 fluorescence intensities from both assays. Early S phase cells can be identified in both live-cell imaging experiments and by FACS analysis (Figure 1C, Figure 1-figure supplement 1A and D), and since the mean fluorescence intensity of the FUCCI-G1 marker in early S phase is constant, it can be used as a normalization factor to directly compare the FUCCI-G1 fluorescence intensity values obtained by imaging and FACS (Figure 1C, Figure 1-figure supplement 1D; see Materials and methods). Using this normalization factor and the fluorescence intensity of the FUCCI-G1 marker as assayed by FACS, it is possible to map individual G1 cells assayed by FACS onto time-lapse microscopy data, allowing us to pinpoint the precise cell cycle time of each cell.

To validate our method of converting FACS fluorescence intensities into absolute cell cycle times, we performed an alternative method to determine cell cycle times based on FUCCI fluorescence as measured by FACS; we blocked cells in mitosis using the microtubule stabilizing drug Taxol for various durations, preventing entry of cells in G1 phase. For cells already in G1 phase the FUCCI-G1 fluorescent signal continues to increase. As no new cells enter G1 phase, a gradual loss of cells with low FUCCI-G1 fluorescence is observed by FACS (Figure 1—figure supplement 1E). By mapping the population of cells that is lost after different durations of Taxol treatment we could calculate the FUCCI-G1 fluorescence intensity associated with cells that had spent various times in G1 phase. Comparison of both methods revealed very similar cell cycle times (Figure 1—figure supplement 1F). Thus, we conclude that we can accurately determine the time a cell has spent in G1 phase based on its FUCCI-G1 fluorescence as measured by FACS.

To identify changes to the transcriptome throughout the M-G1 phase transition, we FACS-isolated single G2, M, and G1 phase cells based on their FUCCI-G1 and FUCCI-G2 fluorescence (Figure 1C), and subjected them to scRNA-seq. In total, 1152 cells were sequenced in three replicate experiments, of which 841 cells passed quality checks (see Materials and methods) and were used to generate a high-resolution temporal transcriptome profile of the M-G1 phase transition. Since the FUCCI system does not discriminate between cells in G2 and M phase, and as there are few transcriptome changes between these two phases (Tanenbaum et al., 2015), we averaged the transcript levels of all cells in G2 and M phase (referred to as G2/M). The average G2/M expression levels of individual genes displayed a high correlation between different replicate experiments (ρ = 0.94–0.95) (Figure 1—figure supplement 1G-I), allowing us to pool the data from the different experiments. The final dataset consisted of 86 G2/M phase cells and 755 cells from various time points in G1 phase (up to 9 hr after the M-G1 phase transition) (Figure 1—figure supplement 1J and Supplementary file 1). After initial data processing, we performed differential transcriptome analysis comparing G2/M phase to G1 phase (see Materials and methods). This analysis identified 220 genes that were downregulated and 42 genes that were upregulated when cells from G2/M phase were compared with G1 phase cells (using a cutoff of >2-fold expression change, see Materials and methods) (Figure 1E and Supplementary file 1). Gene Ontology analysis revealed that these differentially expressed genes were strongly enriched for cell cycle functions, as expected (Figure 1—figure supplement 1K). Of all genes involved in the cell cycle (derived from Cyclebase 3.0; Santos et al., 2015), for which we could determine the expression, ∼53% is downregulated in early G1 phase in our dataset (>2-fold) (Figure 1—figure supplement 1L and Supplementary file 1).

To compare our method of cell cycle time determination with previous computational methods of (pseudo) time determination, we used Monocle2, an in silico trajectory inference method that orders cells based on their transcriptomes (Qiu et al., 2017a; Qiu et al., 2017b; Trapnell et al., 2014). We aligned cells using trajectory inference (Figure 1—figure supplement 1M, see Materials and methods), and subsequently performed differential transcriptome analysis, which identified 318 downregulated genes and 86 upregulated genes in early G1 phase compared to G2/M phase (Figure 1—figure supplement 1N and Supplementary file 1). There was a large overlap between the differentially expressed genes identified by Monocle2 and our FUCCI-based method (Figure 1F), and we found a good overall correlation between FUCCI-based ordering and Monocle2-based ordering of G1 phase cells (Figure 1G and Figure 1—figure supplement 1M). Monocle2 cannot assign absolute cell cycle times, instead it can compute a ‘pseudo time’ for each G1 phase cell assuming that transcriptome changes occur evenly over time. Comparing the pseudo time assigned by Monocle2 with the cell cycle time assigned by our FUCCI-based method revealed differences between both methods. In general, Monocle2 computed larger time intervals between cells early in G1 phase compared to our FUCCI-based method (Figure 1H). As Monocle2 computes the time intervals between cells based on the magnitude of transcriptome changes, a possible explanation for this observation is that transcriptome changes are larger in early G1 phase than at the end of G1 phase, and Monocle2 thus positions cells in early G1 phase too far apart in (pseudo) time. In conclusion, by using the FUCCI-based single-cell sequencing approach we could generate a high-resolution, time-resolved transcriptome profile of cells spanning the transition from M phase into G1 phase.

mRNA levels decline in multiple waves during the M-G1 phase transition

As discussed above, we found a large group of genes (220) for which mRNA levels decline at the M-G1 phase transition. To determine the precise moment when mRNA levels started to decline for each gene, we fit the data for individual mRNAs to a smoothing spline and determined the moment of maximum negative slope of the spline, which is the moment when the mRNA level declined most rapidly (referred to as the ‘spline analysis’; see Materials and methods). Strikingly, the decline in mRNA levels of various genes initiated at two distinct times in the cell cycle. The first ‘wave’ of mRNA decline occurred within 20 min of metaphase (Figure 2A and Supplementary file 1), which is around the time of mitotic exit, but before the start of G1 phase (Figure 1B). The second wave occurred at ~80 min after metaphase (Figure 2A and Supplementary file 1), which is in G1 phase (G1 phase starts between 15 and 30 min after metaphase; Figure 1B). To examine these two ‘waves’ of mRNA decline in more detail, we divided the 220 mRNAs into two groups: for the first group the maximum negative slope occurred during mitotic exit (immediate decrease) and for the second group the maximum negative slope occurred during early G1 phase (delayed decrease) (Supplementary file 1, see Materials and methods). Plotting the average slope over time for both groups revealed that the mRNAs in the immediate decrease group declined most rapidly during the M-G1 phase transition and continued to decline during the first 2–3 hr of G1 phase (Figure 2B, red lines), whereas the mRNAs in the delayed decrease group mostly declined between 1 and 4 hr after the start of G1 phase (Figure 2B, blue lines). For both groups, the slopes of individual mRNAs were mostly centered around zero at later times (>8 hr) in G1 phase, demonstrating that most mRNAs reached a new steady-state level at later time points in G1 phase.

Figure 2. Reduction in mRNA levels occurs in multiple waves, during and after cell division.

(A) Time relative to metaphase of the highest rate of mRNA decrease for the 220 downregulated genes. For each gene a smoothing spline was fit to the data and the moment of maximum negative slope of the spline was determined (see Materials and methods). (B) Average slope of mRNA levels over time for genes that display immediate (thick red line) or delayed decrease (thick blue line). Thin red and blue lines show a random selection of 25 individual genes belonging to the immediate or delayed decrease group, respectively. (C) Validation of the two waves of mRNA decline. RPE-fluorescent, ubiquitination-based cell cycle indicator (FUCCI) cells at different stages of the cell cycle were isolated by fluorescence activated cell sorting (FACS) based on FUCCI fluorescence (see Figure 2—figure supplement 1A for gating strategy). mRNA expression levels of indicated genes was measured by RT-qPCR. Five genes from the immediate decrease group and five genes from the delayed decrease group were selected. Note that the moment of decrease as measured by RT-qPCR closely mirrors the moment of decrease determined by modeling of our single-cell sequencing data (see Supplementary file 1). Lines with error bars represent average ± SEM of three experiments. (D) Example images of TOP2A and CDK1 single molecule fluorescence in situ hybridization (smFISH) at the different stages of mitosis. Asynchronously growing RPE-1 cells were fixed and stained for DNA (DAPI), membranes (WGA), and TOP2A and CDK1 mRNA (using smFISH). Scale bar, 10 µm. (E–F) Quantification of TOP2A (E) and CDK1 (F) transcript number in different stages of mitosis. Each dot represents the average number of transcripts in a single experiment and lines with error bars represent average ± SEM of three experiments (at least 15 cells per experiment per condition analyzed, see Supplementary file 2 for the exact number of cells included). Single-cell TOP2A and CDK1 transcript counts are shown in Figure 2—figure supplement 1B-C. p-Values are based on a one-tailed unpaired Student’s t-test, and are indicated as * (p < 0.05), ** (p < 0.01), *** (p < 0.001), ns = not significant.

Figure 2.

Figure 2—figure supplement 1. Reduction in mRNA levels occurs in multiple waves, during and after cell division.

Figure 2—figure supplement 1.

(A) Gating strategy for the identification of G1, early S, and G2 phase cells in fluorescence activated cell sorting (FACS) analysis of asynchronously growing REP-fluorescent, ubiquitination-based cell cycle indicator (FUCCI) cells (top). The tables below the FACS plot display the FUCCI-G1 fluorescence intensity associated with cells at various times in G1 or S phase. The top table shows the mean FUCCI-G1 fluorescence intensity during early S phase obtained. In the lower table, the left column states the time a cell has spent in G1 phase. The middle column indicates the FUCCI-G1 fluorescence intensities relative to cells in early S phase (calculated using the polynomial equation). The right column shows the range in absolute FUCCI-G1 fluorescence intensities in each bin. These intensities were calculated by multiplying the mean FUCCI-G1 fluorescence intensity in early S phase (6517, see top table) by the fluorescence intensity relative to S phase (middle column, lower table). The resulting fluorescence intensity ranges were used to set the sorting gates for the isolation of cells at different times in G1 phase. These gates are indicated in the FACS plot above. (B–C) Quantification of the number of TOP2A (B) and CDK1 (C) transcripts in different phases of mitosis. Each dot represents the number of transcripts in a single cell and lines with error bars represent average ± SD (at least 15 cells per experiment per condition analyzed, see Supplementary file 2 for the exact number of cells included). One representative of three experiments is shown. (D) Fraction of mitotic cells in different isolated cell populations. Cells were either left untreated (asynchronous) or mitotic shake-off was performed to split the cells into two populations: Mitotic cells (collected cells from shake-off that are highly enriched for mitotic cells) and interphase cells (adherent cells remaining after shake-off that are depleted of mitotic cells). After isolation of different populations of cells, cells were fixed and stained with DAPI for DNA content measurements and for the mitosis-specific marker phosphorylated histone 3 ser 10, and the fraction of mitotic cells was determined by FACS for each cell population (see Materials and methods). Lines with error bars represent average ± SD of six experiments. (E–F) FACS strategies to enrich for G2 phase cells (E), early mitotic cells, or late mitotic cells (F). RPE-FUCCI cell populations depleted of mitotic cells (E) or enriched for mitotic cells (F) were isolated as in (D). The population of cells enriched for interphase cells (‘interphase’ in D) was used to isolate G2 phase cells. The population of cells enriched for mitotic cells (‘mitotic’ in D) was used to isolate early and late phase mitotic cells (prometaphase/metaphase and anaphase/telophase, respectively). Early and late mitotic cells were distinguished using FUCCI-G2 fluorescence levels: cells that express high levels of the FUCCI-G2 marker are early mitotic cells and cells that express low levels of the FUCCI-G1 marker are late mitotic cells. (G–H) Relative mRNA levels of indicated genes in G2, early and late mitosis, as measured by RT-qPCR. RPE-FUCCI cells in G2 phase, early and late mitosis were isolated by FACS (see E-F). mRNA expression levels of five genes from the immediate decrease group (G) and five genes from the delayed decrease group (H) were analyzed. Dots and error bars represent average ± SEM of three to five experiments. (I) Crystal violet staining of wild-type and p53 knock-out RPE-FUCCI cells that were treated with Nutlin-3a for 1 week. A representative experiment of two experiments is shown. (J) Validation of p55 knock-out cells. Wild-type and p53 knock-out RPE-FUCCI cells were irradiated using 10 Gy of ɣ-irradiation, or left unirradiated. Five hours later, cells were lysed and CDKN1a expression was analyzed by RT-qPCR. Lines with error bars represent average ± SEM of three experiments. (K) Analysis of G1 phase duration and the fraction of G0 (quiescent) cells in p53 knock-out cells. RPE-FUCCI wild-type or p53 knock-out cells were imaged for at least 21 hr. The duration of G1 phase in cells entering G1 during the first hour of imaging was analyzed. G0 (quiescent) cells are defined as cells that maintain FUCCI-G1 fluorescence for >20 hr. Pooled data of two experiments is shown. Per cell line 20 cells were included per experiment. (L) Accumulation of the FUCCI-G1 fluorescence in wild-type and p53 knock-out RPE-FUCCI cells. Wild-type and p53 knock-out RPE-FUCCI cells were imaged by time-lapse microscopy. FUCCI-G1 fluorescence was determined for cells at metaphase and 2 or 4 hr thereafter. Fluorescence intensities were normalized against the average fluorescence intensity of wild-type cells at 4 hr post-metaphase. Bots and error bars indicate average ± SEM of three experiments. At least 10 cells per condition per experiment were quantified (see Supplementary file 2). (M) Validation of the two waves of mRNA decline. RPE-FUCCI p53 knock-out cells at indicated stages of the cell cycle were isolated by FACS (see A for sorting strategy). mRNA expression levels of indicated genes were measured by RT-qPCR. Five genes from the immediate decrease group and five genes from the delayed decrease group were selected. Dots and error bars represent average ± SEM of three to five experiments. (N) Correlation plot comparing the relative expression of immediate decrease genes to the relative expression of delayed decrease genes in single cells. Cells were selected that have spent at least 4 hr in G1 phase (158 cells in total). For individual cells, we averaged the expression of all genes belonging to either the immediate or delayed decrease groups, thus creating two metagenes. We then normalized the expression of both metagenes to the average expression of either metagene in G2 phase cells. Red dashed line represents the linear model fit to the data. For all panels, p-values are based on an unpaired one-tailed Student’s t-test. p-Values are indicated as * (p < 0.05), ** (p < 0.01), *** (p < 0.001), ns = not significant. See Supplementary file 2 for the exact number of included cells in each condition.

To confirm that mRNA levels decline in two distinct temporal waves, cells were isolated by FACS (Figure 2—figure supplement 1A) and RT-qPCR was used to measure mRNA levels for five genes in the immediate decrease group (CDK1, TOP2A, UBE2C, FBXO5, and FZR1) and five genes in the delayed decrease group (ARL6IP1, CENPA, PSD3, UBALD2, and SRGAP1) in G2/M phase and at various time points in G1 phase. Consistent with the RNA sequencing data, we observed two distinct waves of mRNA decline by RT-qPCR (Figure 2C). Note that the minor increase in mRNA levels seen at the 1 hr time point in the delayed decrease group is likely an artifact caused by comparing a highly synchronized population of early G1 phase cells (1 hr time point, when cells have not yet initiated the decline of delayed genes and thus express the highest possible levels of these transcripts) to a somewhat more heterogeneous population of G2/M phase cells (0 hr time point).

To determine the moment of mRNA decline relative to cell division more precisely for the immediate decrease group, we assessed mRNA levels by single molecule fluorescence in situ hybridization (smFISH) and fluorescence microscopy during different stages of mitosis. We fixed asynchronously growing cultures of cells and stained them for two mRNAs from the immediate decrease group, TOP2A and CDK1, which were selected because of strong mRNA decline after metaphase (Figure 2C). To determine the mitotic stages and the outline of the individual cells, we stained the DNA with DAPI, and the membranes with fluorescent wheat germ agglutinin (WGA) (Figure 2D). Quantification of TOP2A and CDK1 mRNA levels at various stages of mitosis revealed a significant decrease in mRNA levels as early as anaphase, and a further decrease in telophase for both genes (Figure 2E–F and Figure 2—figure supplement 1B-C).

To further confirm the moment of mRNA decline of mRNAs belonging to the immediate decrease group, we assessed the moment of mRNA decline for additional mRNAs belonging to this group by RT-qPCR. We compared mRNA levels in G2 phase to mRNA levels in early mitosis (prometaphase/metaphase) and late mitosis (anaphase/telophase). G2 phase cells as well as cells in early and late mitosis were isolated by a combination of mitotic shake-off and FACS (Figure 2—figure supplement 1D-F, see Materials and methods). RT-qPCR analysis revealed that the levels of all five genes belonging to the immediate decrease group decline as cells move beyond metaphase (Figure 2—figure supplement 1G). In contrast, the levels of all five transcripts belonging to the delayed decrease group do not decline between G2 phase, early and late mitosis (Figure 2—figure supplement 1H). FBXO5 and UBE2C mRNA levels already show decline as cells move from G2 phase into early mitosis, while the levels of CDK1, TOP2A, and FZR1 mRNAs do not (their levels increase slightly between G2 phase and mitosis, which may reflect a cell synchronization artifact, as discussed before) (Figure 2—figure supplement 1G). These results suggest that genes belonging to the immediate decrease group decrease already in mitosis, while genes belonging to the delayed decrease group do not start to decrease until the onset of G1 phase.

Our previous analysis revealed two waves of mRNA decline. Since these waves of mRNA decline are determined based on cell populations, it is unclear if both waves occur in each individual cell, or whether these two waves rather occur in two different sub-populations of cells, for example in cells entering G1 or G0 phase (which cannot be distinguished based on the FUCCI-G1 reporter). To assess whether both waves of mRNA decline occur in cells entering G1 (rather than G0), we generated p53 knock-out cells (Figure 2—figure supplement 1I-J), which exclusively enter G1 after completion of mitosis (Figure 2—figure supplement 1K; Spencer et al., 2013; Yang et al., 2017). We confirmed that the synthesis rates of the FUCCI-G1 fluorescent reporter are unaltered in RPE-FUCCI Δp53 cells (Figure 2—figure supplement 1L), and assessed whether two waves of mRNA decrease occurred in G1 phase by RT-qPCR (see Figure 2—figure supplement 1A). We found that mRNA levels declined in two distinct waves (compare Figure 2—figure supplement 1M and Figure 2C), demonstrating that both waves of mRNA decline occur in G1 phase cells.

To further confirm that the two waves of mRNA decline occur in individual cells, we examined the expression of immediate decrease and delayed decrease genes in the same cells. If the two waves of mRNA decline occur in a distinct subset of cells, then cells showing the strongest decrease in the one set of genes (e.g. immediate decrease genes) should show little to no decrease in the other set of genes (e.g. delayed decrease genes). In contrast, if both waves of mRNA decline occur sequentially in the same cells, such anti-correlation is not expected. To investigate this, we averaged the expression of all immediate decrease and delayed decrease genes and compared the expression of both groups of genes in single cells. A time point of 4 hr after metaphase was selected for this analysis, as both types of mRNA decline (immediate and delayed) have been largely competed at this time point. We found that the expression levels of both groups of genes are highly correlated in individual cells (Figure 2—figure supplement 1N), demonstrating that both waves of mRNA decline occur in the same cells.

mRNA decay drives transcriptomic changes during the M-G1 phase transition

The decline in mRNA levels during early G1 phase may be caused by changes in the rate of mRNA synthesis (transcription) and/or degradation (mRNA stability). It is well established that transcription regulation controls the expression of cell cycle genes (Bertoli et al., 2013; Sadasivam and DeCaprio, 2013). Accordingly, comparison of transcription rates in G2 phase and G1 phase RPE-1 cells (data derived from Battich et al., 2020) shows that transcription is decreased for nearly all genes that are downregulated in G1 phase compared to G2/M phase (Figure 3A, see Materials and methods), regardless of whether mRNAs belong to the immediate decrease or delayed decrease group (Figure 3—figure supplement 1A). We wondered if transcription inhibition alone was sufficient to explain the rapid rate at which transcript levels decline in G1 phase, or whether an increase in mRNA degradation also contributes to the decreased expression in G1 phase for the genes that we find to be downregulated in G1 phase. To investigate whether the mRNA degradation rate is altered during the M-G1 phase transition, we calculated the degradation rate of individual mRNAs during the M-G1 phase transition using mathematical modeling (Figure 3B, see Materials and methods). Briefly, our model describes two phases for the mRNA levels over time: in the first phase, mRNA levels remain constant (at an initial level of m0), in the second phase mRNA levels decline to a new steady-state level. The onset of decline is described by tonset. The rate of decline is dependent on the mRNA degradation rate (γ), while the new steady-state mRNA level is dependent on the mRNA synthesis rate (µ) and on the mRNA degradation rate (γ). Using a quality of fit analysis (see Materials and methods), we identified the parameters (m0, tonset, µ, and γ) that resulted in the optimal fit with the data for each of the 220 downregulated genes. Visual inspection showed that the fits described the data well (Figure 3—figure supplement 1 and Supplementary file 1). Using this approach, we confirmed that the onset of decay for different genes occurred most strongly at two distinct times during the M-G1 phase transition; either during the M-G1 phase transition or during early G1 phase (Figure 3—figure supplement 1H), confirming the results from the spline analysis (Figure 2A).

Figure 3. mRNA decay occurs during a brief window of time as cells exit mitosis and enter G1 phase.

(A) Violin plot showing the ratio of transcription in G1 phase versus G2 phase for the 220 genes that we identified as downregulated in G1 phase. Data were retrieved from Battich et al., 2020. Battich et al. labeled new transcripts for 30 min using EU, and old and new transcripts were quantified using deep sequencing. We defined the relative rate of transcription as the number of labeled transcripts in G1 versus G2 phase. Dashed line indicates a ratio of 1, indicative of a similar transcription rate in G2 and G1 phase (a ratio <1 is indicative of reduced transcription in G1 phase). (B) Schematic of the mathematical model that was used to fit the decrease in mRNA levels as cells progress from mitosis into G1 phase. (C) Boxplot of mRNA half-lives for the genes that were identified as downregulated in G1 phase in our study. Half-lives at the mitosis-to-G1 (M-G1) transition are shown (this study), as well as the half-lives of the same genes determined in asynchronous cell populations in HeLa cells (Tani et al.), mouse embryonic stem cells (Herzog et al.) and mouse fibroblasts (Schwanhausser et al.). (D) Relative mRNA levels in mitosis after different times of transcription inhibition, as measured by RT-qPCR. RPE-1 cells were synchronized in G2 using the CDK1-inhibitor RO-3306. Subsequently, cells were released from RO-3306 into medium containing Taxol, to arrest cells in mitosis. Mitotic cells were collected by mitotic shake-off, and cultured for an additional 2 hr in the presence or absence of the transcription inhibitor Actinomycin D (blue and red lines, respectively). For comparison, mRNA levels during the M-G1 phase transition are shown (gray line). Note that mRNA of indicated genes is stable in mitosis, indicating that mRNA is degraded specifically during the M-G1 phase transition. Lines with error bars indicate average ± SEM of three experiments. (E) Relative mRNA levels in G2 and late G1 phase after different times of transcription inhibition, as measured by RT-qPCR. Asynchronously growing RPE-fluorescent, ubiquitination-based cell cycle indicator (FUCCI) cells were treated with Actinomycin D for indicated times. Cells were then fluorescence activated cell sorting (FACS)-sorted and G2 phase cells and late G1 phase cells (>4 hr into G1 phase) were isolated based on FUCCI reporter fluorescence. The mRNA levels of indicated genes were then measured by RT-qPCR. mRNA levels during the M-G1 phase transition are shown for comparison (gray lines). Note that mRNA levels are substantially less stable in cells during the M-G1 phase transition compared to G2 or late G1 phase cells. Lines with error bars indicate average ± SEM of three experiments. p-Values are based on a one-tailed unpaired Student’s t-test (C-E), and are indicated as * (p < 0.05), ** (p < 0.01), *** (p < 0.001), ns = not significant.

Figure 3.

Figure 3—figure supplement 1. mRNA decay occurs during a brief window of time as cells exit mitosis and enter G1 phase.

Figure 3—figure supplement 1.

(A) Violin plot showing the ratio of transcription in G1 phase versus G2 phase for the genes belonging to the immediate decrease or delayed decrease groups (Figure 2A and Supplementary file 1). Data were retrieved from Battich et al., 2020. New transcripts were labeled for 30 min using EU, and old and new transcripts were quantified using deep sequencing. The relative rate of transcription was defined as the number of labeled transcripts in G1 versus G2 phase. Dashed line indicates supplemnetary filea ratio of 1, indicative of a similar transcription rate in G2 and G1 phase (ratios < 1 are indicative of reduced transcription in G1 phase). (B–G) mRNA abundance over time for genes that undergo mRNA decay at the mitosis-to-G1 (M-G1) transition. Blue lines indicate the best fit obtained using the mathematical model described in Figure 3C. Representative genes of the immediate decrease group (CDK1, TOP2A, and UBE2C) and the delayed decrease group (CENPA, ALR6IP1, and UBALD2) are shown. (H) Histogram of the time (relative to metaphase) when mRNA levels start to decline is shown for genes that are downregulated during the M-G1 phase transition. mRNA levels over time were fit as in B–G and the onset time of mRNA decline was determined for each of the 220 downregulated genes. (I) Histogram of mRNA half-lives for the 220 genes that are downregulated during the M-G1 phase transition. To obtain mRNA half-lives, mRNA levels over time were fit as in B–G and the mRNA half-lives were calculated using the mathematical model described in Figure 3 (see Materials and methods). (J–L) Comparison of mRNA half-lives during the M-G1 phase transition with mRNA half-lives in asynchronous cells determined in previous studies (Herzog et al., 2017; Schwanhäusser et al., 2011; Tani and Akimitsu, 2012). Dashed lines indicate identical half-lives. Note that the half-lives of most genes are shorter during the M-G1 phase transition than in asynchronous growing cells. (M) Boxplot of mRNA half-lives of immediate and delayed decrease genes. For each gene, the half-live was determined from the moment mRNA levels start to decrease (see Supplementary file 1). (N) Analysis of transcription inhibition by Actinomycin D. Expression levels of the DNA damage-induced gene CDKN1a were measured by RT-qPCR in cells that were DNA damaged (exposed to 5 Gy ionizing radiation), in the presence or absence of Actinomycin D, relative to non-irradiated cells. Lines with error bars indicate average ± SEM of three experiments. (O) Mitotic index of RPE-1 cells treated with the transcription inhibitor Actinomycin D. RPE-1 cells were arrested in G2 phase using a CDK1 inhibitor (RO-3306). After 16 hr, the CDK1 inhibitor was removed and replaced by Taxol, thereby releasing cells from the G2 phase arrest and blocking cells in mitosis. Forty-five minutes later, mitotic cells were collected through mitotic shake-off, after which Actinomycin D was added for up to 2 hr. Cells were fixed and the fraction of mitotic cells was determined by fluorescence activated cell sorting (FACS) (by staining cells for DNA content and the mitosis-specific marker phosphorylated histone 3 at ser 10). Lines with error bars indicate average ± SEM of three experiments. p-Values are based on a one-tailed Student’s t-test, and indicated as * (p < 0.05), ** (p < 0.01), *** (p < 0.001), ns = not significant.

We used mRNA degradation rates extracted from the model to compute the half-lives of the 220 mRNAs that we found to be downregulated in G1 phase (Supplementary file 1). This revealed a median half-life of 61.5 min in the decay phase during the M-G1 phase transition (Figure 3—figure supplement 1I). We compared mRNA half-lives during the M-G1 phase transition with previously determined half-lives of the same mRNAs in asynchronously growing cells. For almost all mRNAs (98–100%) the M-G1 half-lives we computed are shorter than the reported half-lives of the same mRNAs in asynchronously growing cells (Figure 3C and Figure 3—figure supplement 1J-L; Herzog et al., 2017; Schwanhäusser et al., 2011; Tani et al., 2012). We observed no significant differences between the half-lives of mRNAs belonging to the immediate decrease group versus the delayed decrease group (Figure 3—figure supplement 1M). The comparatively short mRNA half-lives we find during the M-G1 phase transition indicate that these transcripts are subject to scheduled degradation at this stage of the cell cycle.

To confirm that mRNAs are subjected to scheduled degradation specifically during the M-G1 phase transition, we also examined their stability during mitosis, G2 phase and late G1 phase. To measure mRNA stability in mitosis, we synchronized and arrested RPE-1 cells in prometaphase of mitosis (see Materials and methods), followed by inhibition of transcription for 1 or 2 hr using Actinomycin D. Actinomycin D completely blocked de novo transcription (Figure 3—figure supplement 1N) and did not influence the arrest of cells in mitosis (Figure 3—figure supplement 1O). None of the 10 mRNAs tested (belonging to both the immediate and delayed decrease groups) showed an appreciable decrease in mRNA levels during the 2 hr time window of Actinomycin D treatment, indicating that these mRNAs are much more stable in mitosis than they are during the M-G1 phase transition (Figure 3D, compare red or blue lines to gray line).

To measure mRNA stabilities in G2 phase and late G1 phase, we inhibited transcription with Actinomycin D for 1 or 2 hr in asynchronously growing RPE-FUCCI cells. Subsequently, we FACS-sorted populations of G2 phase cells and late G1 phase cells (cells that had spent at least 4 hr in G1 phase) and determined mRNA levels of immediate and delayed decay genes with or without Actinomycin D treatment. For all genes tested, mRNA stability in G2 phase and late G1 phase substantially exceeded the mRNA stability calculated during the M-G1 phase transition (Figure 3E, compare red or blue lines to gray line). Collectively, these data demonstrate that for all genes tested, mRNAs are substantially more stable during G2 phase, mitosis (pre-anaphase), and late G1 phase compared to during the M-G1 phase transition and early G1 phase. Thus, these results uncover an active mRNA decay mechanism that specifically takes place during mitotic exit and early G1 phase.

CNOT1 stimulates mRNA decay during the M-G1 phase transition

Cytoplasmic mRNA degradation is often initiated by shortening of the poly(A) tail (Eisen et al., 2020), followed by degradation from either end of the mRNA (Garneau et al., 2007). Shortening of the poly(A) tail is frequently mediated by the CCR4-NOT complex (Yamashita et al., 2005). To test whether the CCR4-NOT complex is required for mRNA decay during the M-G1 phase transition, we depleted CNOT1, the scaffold subunit of the CCR4-NOT complex, using siRNA-mediated depletion in RPE-1 cells (Figure 4—figure supplement 1A). For initial experiments, we focused on TOP2A and CDK1 mRNAs, both of which are rapidly and robustly degraded at the M-G1 phase transition (Figure 2C). CNOT1 depletion resulted in substantially fewer mitotic cells and a strong enrichment of cells in G1 phase (Figure 4—figure supplement 1B), consistent with an important role for CNOT1 in cellular proliferation (Blomen et al., 2015; Hart et al., 2015; Wang et al., 2015). In the mitotic cells that could be identified, depletion of CNOT1 caused a 20–25% increase in the relative abundance of both TOP2A and CDK1 mRNAs in telophase (when decay of these mRNAs has normally occurred) compared to control cells (Figure 4A and Figure 4—figure supplement 1C-D). These results suggest that CNOT1-dependent mRNA deadenylation is involved in mRNA decay at the M-G1 phase transition.

Figure 4. CNOT1 promotes decay of mRNAs during mitotic exit and early G1 phase.

(A) Effect of CNOT1 depletion on TOP2A and CDK1 mRNA abundance in different states of mitosis. Cells were transfected with indicated siRNAs. Two days after transfection, cells were fixed and TOP2A and CDK1 mRNAs were visualized using single molecule fluorescence in situ hybridization (smFISH) (as in Figure 2D) and the number of mRNAs per cell was determined. To calculate the relative abundance of mRNAs during mitotic exit, we divided the average number of mRNAs present in telophase by the average number of mRNAs present in prophase, prometaphase, and metaphase mRNA abundance is similar during these phases of mitosis (Figure 2E–F). Relative abundance was used instead of absolute abundance, as the absolute number of detectable foci varied between experiments due to variations in labeling intensity of smFISH probes. Each dot represents a single experiment and lines with error bars indicate average ± SEM of three experiments. Per experiment, at least 10 cells during mitotic exit and 10 early mitotic cells were quantified (see Supplementary file 2 for the exact number of cells included). p-Values are based on a one-tailed Student’s t-test, and indicated as * (p < 0.05), ** (p < 0.01), *** (p < 0.001), ns = not significant. (B–C) Boxplot of poly(A) tail lengths in mitosis (B) and S phase (C) for immediate decrease genes, delayed decrease genes, or genes that are not subjected to mRNA decay (other genes). p-Values are based on a one-tailed Student’s t-test. (E) Expression levels of indicated mRNAs during the mitosis-to-G1 (M-G1) phase transition in control cells and cells depleted of CNOT1. RPE-fluorescent, ubiquitination-based cell cycle indicator (FUCCI) CRISPR interference (CRISPRi) cells infected with control- or one of three different CNOT1-targeting sgRNAs were sorted into populations of G2/M phase and G1 phase cells at 5 days post-sgRNA infection. mRNA levels of indicated genes were measured by RT-qPCR. Data from three CNOT1-targeting gRNAs were averaged. Lines and error bars indicate average ± SEM of three experiments. p-Values are based on a one-tailed Welch’s t-test. p-Values are indicated as * (p < 0.05), ** (p < 0.01), *** (p < 0.001), ns = not significant.

Figure 4.

Figure 4—figure supplement 1. CNOT1 promotes decay of mRNAs during mitotic exit and early G1 phase.

Figure 4—figure supplement 1.

(A) Quantification of CNOT1 expression levels following siRNA-mediated knockdown. RPE-1 cells were transfected with a CNOT1 siRNA, or a control siRNA-targeting luciferase. CNOT1 mRNA levels were measured by RT-qPCR at 48 hr post-siRNA transfection, and were normalized to CNOT1 expression levels in control siRNA treated cells. Lines with error bars indicate the average ± SEM of three experiments. (B) Cell cycle analysis of control- and CNOT1-depleted cells. RPE-fluorescent, ubiquitination-based cell cycle indicator (FUCCI) cells were transfected with indicated siRNAs. At 48 hr post-transfection, the cell cycle distribution was determined by fluorescence activated cell sorting (FACS) using FUCCI fluorescence. (C–D) Quantification of TOP2A (C) and CDK1 (D) transcript number in individual cells. Asynchronously growing RPE-1 cells were transfected with indicated siRNAs. Forty-eight hours later, cells were fixed and stained for DNA (DAPI), membranes (WGA), TOP2A, and CDK1 mRNA (using single molecule fluorescence in situ hybridization [smFISH]). A representative experiment of three experiments is shown. At least 10 cells per mitotic phase were included per experiment (see Supplementary file 2 for the exact number of cells included). (E) Quantification of CNOT1 expression levels following CRISPR interference (CRISPRi)-mediated knockdown. RPE-FUCCI CRISPRi cells were infected with the indicated sgRNAs, and selected for successful infection using puromycin. Five days post-infection, cells were sorted into the indicated cell cycle fractions based on FUCCI fluorescence (see Materials and methods). CNOT1 expression was measured by RT-qPCR. Bars and error bars indicate the average ± SEM of three experiments. (F) Cell cycle analysis of control- and CNOT1-depleted cells. RPE-FUCCI CRISPRi cells were infected with CNOT1 sgRNAs as in (E) and the cell cycle distribution of cells was determined as in (B). Bars and error bars represent average ± SEM of three experiments. (G) Accumulation of the FUCCI-G1 marker in control or CNOT1-depleted RPE-FUCCI CRISPRi cells. Control and CNOT1-depleted RPE-FUCCI CRISPRi cells were imaged using time-lapse microscopy. FUCCI-G1 fluorescence was determined for cells at metaphase and 2 or 4 hr thereafter. Fluorescence intensities were normalized to the average fluorescence intensity of wild-type cells at 4 hr post-metaphase. Dots and error bars indicate average ± SEM of three experiments. At least eight cells per condition per experiment were quantified. For all panels, p-values are based on an unpaired one-tailed Student’s t-test. p-Values are indicated as * (p < 0.05), ** (p < 0.01), *** (p < 0.001), ns = not significant. See Supplementary file 2 for the exact number of cells included.

A previous study found that the mRNAs of many cell cycle genes contain significantly shorter poly(A) tails in M phase compared to S phase (Park et al., 2016). Interestingly, re-analysis of this previous data revealed that the poly(A) tails of transcripts in the immediate decrease group were shorter than those of genes that did not show mRNA decay at the M-G1 phase transition (median value 58 versus 79, respectively) (Figure 4B). The delayed decrease group mRNAs also have significantly shorter poly(A) tails during mitosis than those in the control group, although the effect was modest (median value 75 versus 79, respectively) (Figure 4B). Importantly, the shortened poly(A) tails were specific to mitosis, as poly(A) tail lengths in S phase of immediate and delayed decrease groups were similar to those of control genes (Figure 4C). Collectively, these data show that CNOT1 aids the decay of TOP2A and CDK1 mRNAs during the M-G1 phase transition and suggest that CNOT1-dependent deadenylation in mitosis may contribute to the decay of many mRNAs at the M-G1 phase transition.

To determine whether CNOT1 is also involved in the second wave of mRNA decay, we depleted RPE-FUCCI cells of CNOT1 using CRISPR interference (CRISPRi) (Gilbert et al., 2014; Gilbert et al., 2013). Using CRISPRi, we could knock down CNOT1 in a large population of cells, allowing subsequent FACS-based isolation of sufficient numbers of early G1 phase cells. We used three independent single-guide RNAs (sgRNAs) to target CNOT1 by CRISPRi, which resulted in a modest (~50%) reduction of CNOT1 mRNA levels (Figure 4—figure supplement 1E). Nonetheless, the modest reduction of CNOT1 mRNA levels still caused a cell cycle arrest (Figure 4—figure supplement 1F), albeit to a lesser extent than the arrest caused by the more efficient siRNA-mediated depletion of CNOT1 (compare Figure 4—figure supplement 1B and F). CNOT1 depletion did not affect the synthesis rates of the FUCCI-G1 fluorescent reporter (i.e. accumulation of fluorescence over time) during G1 phase (Figure 4—figure supplement 1G), allowing us to isolate control and CNOT1-depleted cells at similar times in G1 phase based on FUCCI fluorescence through FACS, as before (see Figure 2—figure supplement 1A). G2/M phase and early G1 phase cell populations were isolated by FACS and mRNAs of the immediate decrease and delayed decrease groups were measured by RT-qPCR. Even though depletion of CNOT1 was modest in these experiments, a small but reproducible decrease in decay was observed for eight of the ten genes tested (Figure 4D). Taken together, these data demonstrate that both waves of post-mitotic mRNA decay are stimulated by CNOT1.

Discussion

Assigning a precise cell cycle time to individual, sequenced cells

Understanding of gene expression control has flourished due to the development of single-cell sequencing techniques. To investigate transcriptome changes over time, trajectory inference methods have been developed that allow in silico ordering of cells, based on (dis)similarities in transcriptomes (Saelens et al., 2019). This creates single-cell trajectories of a biological process of interest, such as differentiation or the cell cycle (Figure 1—figure supplement 1M; Haghverdi et al., 2016; Trapnell et al., 2014), and is useful to study dynamics in gene expression. However, due to clustering based on transcriptome (dis)similarities, pseudo time may under/overestimate true cellular state durations (Tian et al., 2019). In addition, trajectories lack real temporal information and are therefore not ideal to determine absolute mRNA synthesis and degradation rates. To circumvent these issues, we have developed a method that combines live-cell microscopy with scRNA-seq to generate a high-resolution, time-resolved transcriptome profile in human cells. Using the FUCCI system, we have applied our method to map gene expression dynamics during the M-G1 phase transition of the cell cycle. Even though the FUCCI system has previously been used to order single-cell transcriptomes along the cell cycle (Battich et al., 2020; Hsiao et al., 2020; Mahdessian et al., 2021), a unique feature of our method is that it uses precisely calibrated FUCCI reporter fluorescence intensities for accurate assignment of cell cycle times of individual, sequenced cells. We use these calibrated fluorescence intensities to align cells along the cell cycle according to their cell cycle ‘age’. While cell-to-cell heterogeneity in the FUCCI fluorescence introduces some noise in the cell cycle timing of individual cells (Figure 1—figure supplement 1B-C and Figure 1—video 1), measurements on many single cells allows averaging of the timing of individual cells, resulting in a high-resolution, time-resolved transcriptome profile. Using our transcriptome profile of the M-G1 phase transition, we identify hundreds of mRNAs that show sharp transitions in their expression levels as cells progress from mitosis to G1 phase (Figure 1E). The availability of temporal information allowed us to quantitatively determine mRNA degradation rates for these transcripts (Figure 3B–C and Supplementary file 1), which is not possible using trajectory inference methods. Importantly, our quantitative measurements of mRNA stability allowed us to differentiate mRNA decline at the M-G1 phase transition by combined transcription inhibition and mRNA degradation from transcription inhibition alone. Similar approaches are likely possible for other biological events for which fluorescence reporters are available, making this approach a broadly applicable method.

mRNA decay contributes to a reset of the transcriptome at the M-G1 phase transition

It is evident that the expression of G2/M-specific genes is reduced following the completion of cell division, both through scheduled protein degradation and transcriptional inactivation (Figure 3A; Bar-Joseph et al., 2008; Castro et al., 2005; Chaudhry et al., 2002; Cho et al., 2001; Cho et al., 1998; Grant et al., 2013; Harper et al., 2002; Nakayama and Nakayama, 2006; Peters, 2002; Vodermaier, 2004; Whitfield et al., 2002). Here, we identify widespread scheduled mRNA decay during mitotic exit and early G1 phase as an additional mechanism acting to reset gene expression following cell division. Why would mRNA transcripts be actively degraded when the clearance of transcripts will eventually be achieved by transcription shut-down alone? All the mRNAs we tested are quite stable during mitosis and late G1 phase (Figure 3D–E). Therefore, transcription inhibition by itself would lead to significant carry-over of transcripts into the next cell cycle. Considering the half-live of these mRNAs in G2 and M phase (Figure 3C), they would persist in G1 phase for many hours. Therefore, decay-mediated clearance of mRNAs as cells exit mitosis could aid to limit expression of the encoded proteins in G1 phase, especially since most mRNAs that are degraded during the M-G1 phase transition are not translationally repressed in G1 phase (Tanenbaum et al., 2015). As these genes include many genes that encode for proteins with important functions in cell cycle control (Figure 1—figure supplement 1K), we speculate that their continued expression in G1 phase may perturb normal cell cycle progression, and potentially could even contribute to cellular transformation (García-Higuera et al., 2008; Park et al., 2008; Sigl et al., 2009). Thus, scheduled mRNA decay during the cell cycle may be important to restrict gene expression of many cell cycle genes to their appropriate cell cycle phases.

CNOT1 promotes two waves of mRNA decay during the M-G1 phase transition

We have identified two consecutive waves of mRNA decay as cells progress through mitosis and into G1 phase (Figure 2A and C and Figure 3—figure supplement 1H). The fact that mRNA decay occurs in two waves may indicate the existence of two distinct mechanisms that act consecutively to degrade transcripts. Interestingly, these two waves of mRNA degradation during the M-G1 phase transition are highly reminiscent of the two consecutive waves of protein degradation that occur at the M-G1 transition (Alfieri et al., 2017; Sivakumar and Gorbsky, 2015).

The regulation of mRNA decay often occurs through (sequence) specific interactions between mRNAs and RNA binding proteins (RBPs). Through direct interactions with mRNAs, RBPs can recruit components of the RNA decay machinery, such as the CCR4-NOT complex, to the mRNA. Recruitment of CCR4-NOT, a key regulator of gene expression, will then induce deadenylation of the target transcript, generally followed by degradation (Garneau et al., 2007). Interestingly, a previous report found that the poly(A) tail lengths of the genes we identified as immediate decay are already shortened in early mitosis (Figure 4B; Park et al., 2016). The observation that poly(A) tails of transcripts decayed during the M-G1 phase transition are shorter in mitosis could suggest that CCR4-NOT-dependent deadenylation during early mitosis marks these transcripts for decay during mitosis and early G1 phase. Indeed, we identified CNOT1, an essential member of the CCR4-NOT complex, as a regulator of post-mitotic mRNA decay (Figure 4A and D). We note that the effects of CCR4-NOT depletion on mRNA decay are modest in our experiments, but the magnitude of the effect is likely caused, at least in part, by the inability to effectively deplete CNOT1, while maintaining cells in a cycling state. Nonetheless, it is possible that additional mechanisms contribute to mRNA decay during the M-G1 phase transition. Such mechanisms could involve PARN-dependent deadenylation, or may be independent of mitotic deadenylation and instead rely on mRNA decapping or endonucleolytic cleavage of the mRNA. Rapid inducible degradation systems (Banaszynski et al., 2006; Nishimura et al., 2009; Yesbolatova et al., 2020) could be applied to quickly and potently inhibit CNOT1 and other mRNA decay factors such as DCP2 and XRN1 to shed more light on this question in the future.

Our data shows that both waves of mRNA decay during the M-G1 phase transition are stimulated by CNOT1 (Figure 4D). Nonetheless, these waves of mRNA decay may be regulated independently, involving distinct RBPs. Binding of distinct RBPs could ensure the timely decay of specific sets of mRNA during either mitotic exit or early G1 phase. It will be interesting to investigate which RBPs are involved in recognizing different subsets of mRNAs that need to be degraded during particular times in the cell cycle. Identification of such RBPs will allow a better understanding of the function and mechanisms of scheduled mRNA degradation during the cell cycle.

Materials and methods

Key resources table.

Reagent type (species) or resource Designation Source or reference Identifiers Additional information
Cell line (Homo sapiens) hTERT RPE-1 ATCC CRL-4000 Cell line maintained in DMEM/F12
Cell line (Homo sapiens) HEK293T ATCC CRL-3216 Cell line maintained in DMEM
Antibody Anti-Histone three phospho-serine 10 (pH3 Ser10) (Rabbit polyclonal) Upstate 06–570 FACS (1:500)
Recombinant DNA reagent pMD2.G Addgene #12,259 Lentiviral packaging plasmid
Recombinant DNA reagent psPAX2 Addgene #12,260 Lentiviral packaging plasmid
Recombinant DNA reagent mkO2-hCdt1(30/120) Sakaue-Sawano et al., 2008 FUCCI-G1 marker
Recombinant DNA reagent mAG-hGem(1/110) Sakaue-Sawano et al., 2008 FUCCI-G2 marker
Recombinant DNA reagent dCas9-BFP-KRAB Jost et al., 2017 CRISPRi construct
Recombinant DNA reagent CRISPRia-v2 with sgRNA non-targeting control Addgene plasmid #84832 and this paper sgRNA GCTGCGCTCCGAGCAACCAC
Recombinant DNA reagent pCRISPRia-v2 with sgRNA CNOT1 #1 Addgene plasmid #84832 and this paper sgRNA GCTCCGGGAAACGCTTCCAG
Recombinant DNA reagent CRISPRia-v2 with sgRNA CNOT1 #2 Addgene plasmid #84832 and this paper sgRNA GCGGAGCTCTAGGGAGTGAG
Recombinant DNA reagent CRISPRia-v2 with sgRNA CNOT1 #3 Addgene plasmid #84832 and this paper sgRNA GCGGAGCTCTAGGGAGTGAG
Sequence-based reagent siRNA luciferase This paper sgRNA CGUACGCGGAAUACUUCGAUU
Sequence-based reagent CDK1 qPCR For This paper qPCR primers CTATCCCTCCTGGTCAGTACATGG
Sequence-based reagent CDK1 qPCR Rev This paper qPCR primers CTCTGGCAAGGCCAAAATCAGCCAG
Sequence-based reagent TOP2A qPCR For This paper qPCR primers GTCTCTCAAAAGCCTGATCCTGCC
Sequence-based reagent TOP2A qPCR Rev This paper qPCR primers GTCATCACTCTCCCCCTTGGATTTC
Sequence-based reagent UBE2C qPCR For This paper qPCR primers GATGTCAGGACCATTCTGCTCTCC
Sequence-based reagent UBE2C qPCR Rev This paper qPCR primers GCTCCTGGCTGGTGACCTGC
Sequence-based reagent FBXO5 qPCR For This paper qPCR primers GATCCTAGAAGATGATAAGGGGG
Sequence-based reagent FBXO5 qPCR Rev This paper qPCR primers CACCTTGATTGGATAACTTGGTT
Sequence-based reagent FZR1 qPCR For This paper qPCR primers GCACGCCAACGAGCTGGTGAGC
Sequence-based reagent FZR1 qPCR Rev This paper qPCR primers CAGACACAGACTCCCACTTTACC
Sequence-based reagent ARL6IP1 qPCR For This paper qPCR primers CTACCTTGTTCCCATTCTAGCGCC
Sequence-based reagent ARL6IP1 qPCR Rev This paper qPCR primers GGCGTTTCCACCAACCCACAGC
Sequence-based reagent CENPA qPCR For This paper qPCR primers GCCCTATTGGCCCTACAAGAGGC
Sequence-based reagent CENPA qPCR Rev This paper qPCR primers GGCTCTGGAGAGTCCCCGG
Sequence-based reagent PSD3 qPCR For This paper qPCR primers CTTAAAACTGCCGACTGGAGGGTC
Sequence-based reagent PSD3 qPCR Rev This paper qPCR primers CTTCAGTTGCTCCTCCTGAGACAG
Sequence-based reagent UBALD2 qPCR For This paper qPCR primers CGGCCGACCAGGCGAAGCAG
Sequence-based reagent UBALD2 qPCR Rev This paper qPCR primers CAGCGCATCGGGGAAGTTGGG
Sequence-based reagent SRGAP1 qPCR For This paper qPCR primers GGCAGCCTGACCAACATCAGCCG
Sequence-based reagent SRGAP1 qPCR Rev This paper qPCR primers GGGGCATGCTTTGCTGTGCTCTG
Sequence-based reagent CNOT1 qPCR For This paper qPCR primers GTAGTGCCCTTTGTTGCCAAAG
Sequence-based reagent CNOT1 qPCR Rev This paper qPCR primers GGAGGTTTCCAGGTTTTAGCTC
Sequence-based reagent CDKN1A qPCR For This paper qPCR primers GCACCTCACCTGCTCTGCTGC
Sequence-based reagent CDKN1A qPCR Rev This paper qPCR primers CCTCTTGGAGAAGATCAGCCGG
Sequence-based reagent Alt-R crRNA targeting p53 This paper; ordered from Integrated DNA Technologies (IDT) crRNA UCGACGCUAGGAUCUGACUG
Sequence-based reagent Alt-R tracrRNA Integrated DNA Technologies (IDT) tracrRNA
Sequence-based reagent sgRNA non-targeting control This paper sgRNA GCTGCGCTCCGAGCAACCAC
Sequence-based reagent sgRNA CNOT1 #1 This paper sgRNA GCTCCGGGAAACGCTTCCAG
Sequence-based reagent sgRNA CNOT1 #2 This paper sgRNA GCGGAGCTCTAGGGAGTGAG
Sequence-based reagent sgRNA CNOT1 #3 This paper sgRNA GCGGAGCTCTAGGGAGTGAG
Sequence-based reagent siRNA luciferase This paper sgRNA CGUACGCGGAAUACUUCGAUU
Sequence-based reagent siRNA CNOT1 Dharmacon ON-TARGET plus siRNA pool of 4
Sequence-based reagent smFISH probe for TOP2A ThermoFisher VA1-14609 Fluorophore: Alexa Fluor 546
Sequence:
Proprietary
Sequence-based reagent smFISH probe for CDK1 ThermoFisher VA6-17545 Fluorophore: Alexa Fluor 647
Sequence:
Proprietary
Chemical compound, drug Propidium Iodide Sigma-Aldrich P4170
Chemical compound, drug Taxol (Paclitaxel) Sigma-Aldrich T1912
Chemical compound, drug RO-3306 Calibochem 217699
Chemical compound, drug Actinoymcin D Sigma-Aldrich A9415
Chemical compound, drug Hoechst 33,342 ThermoFisher H3570
Chemical compound, drug TriSure Bioline BIO-380032
Commercial assay or kit Bioscript Reverse Transcriptase Kit Bioline BIO-27036
Commercial assay or kit SYBR-Green Supermix Bio-Rad #1708880
Commercial assay or kit viewRNA smFISH kit ThermoFisher QVC0001
Software, algorithm Matlab Mathworks
Software, algorithm R R
Other DAPI stain ThermoFisher D1306 1 µg/ml
Other Wheat Germ Agglutinin ThermoFisher W11261 1 µg/ml

Cell culture

HEK293T cells were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 5% fetal bovine serum (FBS, Sigma-Aldrich) and 1% penicillin/streptomycin (Gibco). RPE-1 cells and derivatives were maintained in DMEM/Nutrient Mixture F-12 (DMEM/F12, Gibco) supplemented with 10% FBS and 1% penicillin/streptomycin. RPE-1 cells were obtained from ATCC, and are not part of the commonly misidentified cell lines. The RPE-1 cells used are all free of mycoplasm.

Transfections

Plasmid transfections were performed using FuGENE HD (Promega) according to the manufacturer’s protocol. Cas9 RNPs were transfected using RNAiMAX (Invitrogen). In short Cas9 loaded with duplexed tracrRNA and crRNAs targeting TP53 (all from Integrated DNA Technologies [IDT]) were transfected according to the manufacturer’s protocol. siRNAs were transfected at a final concentration of 10 nM using RNAiMAX according to the manufacturer’s protocol. For microscopy, cells were grown and transfected in 96-well microscopy plates (Matriplate, Brooks). For RT-qPCR, cells were grown and transfected in 96-well culture in late (Greiner Bio-One). For FACS analysis of the cell cycle, cells were grown and transfected in six-well culturing plate (Greiner Bio-One). Two days post-transfection, cells were either fixed for smFISH, the RNA was harvested for RT-qPCR analysis, or the cells were dissociated and resuspended in ice-cold PBS for FACS analysis. For knockdown of CNOT1 we used ON-TARGET plus siRNAs from Dharmacon. As a control, we used a custom siRNA targeting luciferase (5’- CGUACGCGGAAUACUUCGAUU-3’) from Dharmacon.

Lentivirus production

Lentivirus was produced by transfecting HEK293T cells with packaging plasmids (pMD2.G and psPAX2; Addgene #12,259 and #12260, respectively) and lentiviral plasmids carrying the transgene of interest. Two days post-transfection, virus was harvested by collecting the culture medium, pelleting cell debris by centrifugation, and collecting the supernatant.

Generation of cell lines

To generate RPE-FUCCI cells, RPE-1 cells were transduced with lentivirus expressing mkO2-hCdt1(30/120) (FUCCI-G1) and lentivirus expressing mAG-hGem(1/110) (FUCCI-G2) (Sakaue-Sawano et al., 2008). Single clones were isolated by FACS on a BD FACSFUSION system. One clone was selected that showed cyclic expression of both reporter constructs. To generate RPE-FUCCI CRISPRi cells, RPE-FUCCI cells were transduced with lentivirus carrying dCas9-BFP-KRAB (Jost et al., 2017), and the 15% highest BFP-positive cells were isolated by FACS. RPE-FUCCI Δp53 cells were generated by transfecting Cas9 protein loaded with tracrRNA and crRNAs targeting p53 into RPE-FUCCI cells. Knock-out cells were selected by treatment with Nutlin-3a (Cayman chemical) for 7 days.

Irradiation

For irradiation, cells were placed into the irradiation chamber of a Gammacell Exactor (Best Theratronics) equipped with a 137Cs source, and irradiated with the indicated doses of ɣ-irradiation.

Synchronization of cells in mitosis

In order to synchronize cells in mitosis, we first arrested cells in G2 by treating cells with the CDK1-inhibtor RO-3306 (10 μM, Calbiochem) for 16 hr. Subsequently, RO-3306 was removed and the cells were washed twice with PBS before applying fresh medium supplemented with Taxol, which blocks cells in mitosis. Finally, 45 min after Taxol addition, mitotic cells were separated from the interphase cells by shaking of the culture dish (shake-off). This specifically detaches mitotic cells, which were then harvested by collecting the culture medium. For some experiments mitotic cells were isolated through shake-off from asynchronously growing populations without pre-treatment with RO-3306 and Taxol, which is indicated in the corresponding figure legends.

Transcription inhibition

To inhibit transcription, we treated cells with 1 µg/ml Actinomycin D (Sigma-Aldrich) for the indicated durations.

FACS analysis

To visualize the cell cycle distribution of RPE-FUCCI cells, cells were dissociated, resuspended in ice-cold PBS and analyzed by FACS. To visualize and quantify the DNA content of RPE-FUCCI cells, cells were incubated with 2 μM of Hoechst 33,342 (ThermoFisher) for 30 min. Cells were then dissociated, resuspended in ice-cold PBS and analyzed by FACS. Cell cycle phases were gated based on FUCCI fluorescence (Figure 1C). FlowJo software was used to quantify the mean Hoechsts 33,342 fluorescence intensity of cells within each FUCCI gate. In order to identify mitotic cells, cells were fixed in 80% ethanol (–20°C). Thereafter, cells were stained using an antibody targeting the mitosis-specific marker phosphorylated histone 3 (4N pH3)-ser10 (Upstate, 06–570) and propidium iodide to label DNA content. The mitotic fraction was determined as the fraction of 4N pH3-ser10-positive cells.

CRISPRi

For CRISPRi, RPE-FUCCI CRISPRi cells were infected with lentivirus particles expressing a non-targeting sgRNA, or an sgRNA targeting CNOT1 (Horlbeck et al., 2016) and a puromycin resistance cassette followed by BFP. Two days post-infection, infected cells were selected with puromycin (10 µg/ml) for 3 days to eliminate uninfected cells. Sequences of sgRNAs used in this study can be found in Supplementary file 3.

FACS-isolation of cells in different stages of the cell cycle for qPCR

RPE-FUCCI cells were collected by trypsinization and subsequently resuspended as single cells in PBS supplemented with 0.5% FBS. Cells were sorted using a BD FACSFUSION system. To isolate cells at different moments during G1, we measured the average FUCCI-G1 expression in early S phase cells (for gating strategy, see Figure 2—figure supplement 1A and Supplementary file 1). Using the FUCCI-G1 fluorescence relative to early S phase, we calculated the upper and lower bounds for the FACS-gating strategy to isolate cells at different times throughout G1 phase. To isolate cells in G2/M, we isolated FUCCI-G2-positive cells from asynchronous populations (Figure 2—figure supplement 1A). To specifically isolate G2 cells, mitotic cells were first removed from the population by shake-off (Figure 2—figure supplement 1D), and FUCCI-G2 cells were isolated (Figure 2—figure supplement 1E). To isolate early versus late mitotic cells, mitotic cells were first isolated by shake-off (Figure 2—figure supplement 1D). Subsequently, early and late mitotic cells were separated by sorting mitotic cells expressing high versus low levels of the FUCCI-G2 reporter, respectively (Figure 2—figure supplement 1F). For each cell cycle fraction, at least 2500 cells were isolated for downstream RT-qPCR analysis. Following FACS isolation the cells were pelleted and resuspended in TriSure lysis buffer, and subsequently stored at –20°C or processed for RNA isolation (see below).

RT-qPCR

For RT-qPCR analysis, cells were lysed in TriSure (Bioline) and RNA was extracted according to the manufacturer’s protocol. First strand synthesis was performed using Bioscript (Bioline). mRNA expression levels were quantified using SYBR-Green Supermix (Bio-Rad) on a Bio-Rad Real-time PCR machine (CFX Connect Real-Time PCR Detection System). Relative mRNA expression levels were calculated using the ΔΔCt method. GAPDH and RPN1 were selected as reference genes for normalization, based on their reported high mRNA stability (Schwanhäusser et al., 2011). Importantly, the use of reference genes for the calculation of gene expression using the ΔΔCt method will correct for changes in mRNA abundance after cell division (i.e. the number of transcripts will decrease by 2-fold upon cell division), as the reference genes are also subject to this effect. RT-qPCR primers were designed using Primer3, for sequences see Supplementary file 3.

smFISH

smFISH was performed using viewRNA probes targeting TOP2A (probe# VA1-14609) and CDK1 (probe# VA6-17545) (ThermoFisher). Staining was done according to the manufacturer’s protocol. In brief, cells were grown in 96-well microscopy plates (Matriplate, Brooks) and fixed for 30 min using 4% formaldehyde. Then, cells were permeabilized with detergent solution for 5 min at room temperature (RT), and subsequently treated with protease solution for 10 min at RT. To label the RNAs, cells were incubated with probes targeting TOP2A and CDK1 for 3 hr at 40°C. Subsequent probes (preAmplifier, Amplifier, and Label Probe) were incubated for 1 hr at 40°C. Between probe incubations, cells were washed with wash buffer for 3 × 1 min. After the final incubation (with Label Probe), cells were washed and incubated with DAPI (ThermoFisher, D1306) and WGA, conjugated to Alexa Fluor 488 (ThermoFisher, W11261) to label DNA and membranes, respectively.

Microscopy

For live-cell microscopy, RPE-FUCCI cells were grown on microscopy plates and imaged using a Nikon Ti-E with PFS, equipped with an Andor Zyla 4.2Mpx sCMOS camera, CFI S Plan Fluor ELWD 20× air objective (0.45 NA), a Lumencor SpectraX light source and Chroma-ET filter sets (89401 and 24002). Temperature and CO2 control were provided by an OKO-lab Boldline microscope cage and CO2 controller. Where indicated, DNA was visualized by incubation of cells with SPY650-DNA (Spirochrome) for 2 hr prior to imaging. SPY650-DNA was resuspended in DMSO according to manufacurer’s protocol and used at a 4000-fold dilution. Image analysis was performed using ImageJ software.

For imaging of smFISH stained samples, we used a Nikon TI2 inverted microscope with a perfect focus system, equipped with a Yokagawa CSU-X1 spinning disc, a 100× oil objective (1.49 NA),a Prime 95B sCMOS camera (Photometrics), and a Chroma filter set (ZET405/488/565/640).

Crystal violet staining

1500 RPE-FUCCI cells (wild-type or p53 knock-out) were seeded per well in six-well dishes. Four hours after seeding 5 µM Nutlin-3a (Cayman chemical) or DMSO were added to indicated wells. After 7 days, cells were washed with PBS and fixed with 100% MeOH for 10 min. Cells were then washed with PBS and incubated with 1.5% crystal violet overnight. The next morning, crystal violet was removed and the plates were washed three times with water and air-dried.

Single Cell RNA-Seq of FACS-Sorted Cells (SORT-seq)

scRNA-seq of FACS-sorted cells (SORT-seq) was performed as described previously (Muraro et al., 2016). Briefly, we sorted in total 104 G2/M phase cells (FUCCI-G1-negative and FUCCI-G2-positive cells, Figure 1C) and 893 G1 phase cells (FUCCI-G1-positive and FUCCI-G2-negative cells, Figure 1C) as single cells in three 384-well plates. After sorting, these cells were subjected to scRNA-seq based on the CEL-Seq2 protocol (Hashimshony et al., 2016). scRNA-seq was performed by Single Cell Discoveries (https://www.scdiscoveries.com). After sequencing, we continued with cells (841 in total) that passed quality tests (we removed cells with less than 5900 UMIs or more than 111.000 UMIs to lose low-quality cells and doublets, respectively). Finally, we normalized for differences in mRNA recovery per cell using Monocle2 (R package). Normalization of mRNA recovery corrects for the 2-fold decrease of mRNA content that occurs as a consequence of cell division. Each 384-well plate contained G1 phase cells from 0 to 4 hr after the start of G1 phase, but only one plate contained G1 phase cells from 4 to 9 hr after the start of G1 phase. Therefore, we only used cells from 0 to 4 hr after the start of G1 phase to identify differentially expressed genes. In subsequent analyses (i.e. the spline analysis and the modeling) we did use all G1 phase cells.

Cell cycle timing using the FUCCI system

To obtain a temporal transcriptome profile as cells progress from mitosis into G1 phase, we wanted to compute a cell cycle time for each sorted G1 phase cell (i.e. how much time a cell has spent in G1 phase at the moment of sorting). Since FUCCI-G1 levels positively correlate with the amount of time a cell has spent in G1 phase, we reasoned that we could use the measured FUCCI-G1 levels to infer a cell cycle time for a G1 phase cell. To characterize precisely how FUCCI-G1 levels increase during G1 phase, we imaged RPE-FUCCI cells under the microscope with a time interval of 5 min and selected cells that progressed through mitosis into G1 phase. Next, we measured the mean fluorescence intensities of both FUCCI sensors in a region of interest (ROI) in the nucleus using ImageJ and subtracted background signal measured in an extracellular ROI. In each experiment we quantified the fluorescence intensities of the FUCCI sensors in 30 cells.

To compute the average FUCCI-G1 time trace during G1 phase, we aligned the time traces of individual cells at the metaphase-to-anaphase transition, which is defined by a sudden decrease in FUCCI-G2 fluorescence (Sakaue-Sawano et al., 2008). Next, since the total amount of time a cell spends in G1 phase differs for each cell, we clipped individual time traces at the end of G1 phase, which ends shortly after the FUCCI-G2 levels start to increase (Grant et al., 2018). To determine the moment the FUCCI-G2 levels start to increase, we first corrected the FUCCI-G2 traces for fluorescence crosstalk from the FUCCI-G1 marker, which is also excited by the 488 nm laser used for imaging of the FUCCI-G2 marker. To correct the FUCCI-G2 time traces for crosstalk from the FUCCI-G1 marker, we subtracted at each time point 31% of the FUCCI-G1 fluorescence intensity from the FUCCI-G2 fluorescence intensity. Next, we determined the time point when the mean FUCCI-G2 fluorescence intensity reached a threshold value, which was set by visual inspection, and clipped all FUCCI-G1 time traces at the time point of FUCCI-G2 increase. Finally, we computed for each experiment the average FUCCI-G1 levels from the moment of metaphase and fit the average of three experiments to a third-order polynomial (Figure 1D).

To directly compare the FUCCI-G1 levels that we measured on the microscope to the FUCCI-G1 levels that we measured on the FACS, we normalized both microscopy- and FACS-measured FUCCI-G1 levels to the average FUCCI-G1 level of early S phase cells. To quantify the mean FUCCI-G1 fluorescence intensity in early S phase cells on the microscope, we analyzed the mean nuclear intensities of at least 700 cells per experiment (Figure 1—figure supplement 1D). As above, we compensated for fluorescence crosstalk from the FUCCI-G1 marker into the FUCCI-G2 channel by subtracting 31% of the FUCCI-G1 fluorescence intensity from the FUCCI-G2 fluorescence intensity. Next, we determined the range of fluorescence intensities for both the FUCCI-G1 and FUCCI-G2 markers (by subtracting the lowest fluorescence intensity from the highest fluorescence intensity), and defined the early S phase population as those cells with FUCCI-G2 intensities between 2.5% and 10% of the range of FUCCI-G2 intensities and FUCCI-G1 intensities higher than 2.5% of the range of FUCCI-G1 intensities (Figure 1—figure supplement 1D, yellow dots). We computed the average FUCCI-G1 level of the early S phase cells, and normalized the third-order polynomial fit against the average FUCCI-G1 level of the early S phase cells.

To quantify the mean FUCCI-G1 fluorescence intensity in early S phase cells on FACS, we analyzed the FUCCI sensors on FACS (Figure 1C). We define the early S phase population as those cells that have high FUCCI-G1 levels and have started to increase the expression of the FUCCI-G2 marker (Figure 1C), and computed the average FUCCI-G1 level in early S phase cells. To obtain a cell cycle time for each G1 phase cell that was sequenced, we determined the FUCCI-G1 fluorescence intensity level that was obtained by FACS and normalized the FUCCI-G1 level to the average early S phase FUCCI-G1 value. Finally, we used the third-order polynomial fit to infer the cell cycle time of each G1 phase cell from its normalized FUCCI-G1 level.

Because of cell-to-cell variability in FUCCI-G1 fluorescence during G1 phase (Figure 1—figure supplement 1B), the cell cycle times computed for individual G1 phase cells based on their FUCCI-G1 fluorescence intensity is an approximate cell cycle time, which is based on the average FUCCI-G1 fluorescence intensity of many cells. To estimate the error in the cell cycle time caused by this cell-to-cell heterogeneity in FUCCI fluorescence intensities, we first determined at every time point the mean and standard deviation (SD) of all 90 FUCCI-G1 fluorescence intensity traces. Next, for each time point we calculated the mean intensity +1 or –1 SD. Using the intensity values of +1 and –1 SD, we calculated the cell cycle time using our polynomial model (Figure 1D). Next, we computed the difference in cell cycle time between the calculated cell cycle times for +1 and –1 SD intensity values and the cell cycle time for the mean intensity value. Finally, we determined the average time difference of the +1 and –1 SD and refer to this time as the SD of FUCCI-G1 timing.

Cell cycle timing using Monocle2

To rank cells using Monocle2 (R package), we used all G2/M phase cells and G1 phase cells that were from the first 4 hr of G1 phase (based on FUCCI cell cycle timing; see section SORT-seq). Next, we performed an initial differential transcriptome analysis comparing G2/M phase (FUCCI-G1 marker negative and FUCCI-G2 marker positive) and G1 phase cells (FUCCI-G1 marker positive and FUCCI-G2 marker negative) to select differentially expressed genes that Monocle2 can use in subsequent steps to reconstruct the single-cell trajectory. Monocle2 selected a total of 430 genes that were used to reconstruct the single-cell trajectory, and both the cell rank- and the Monocle2-assigned pseudo times were compared to FUCCI-based ranking and cell cycle timing.

Differential transcriptome analysis

Differential transcriptome analysis was performed with Monocle2 (R package), either using FUCCI-based or Monocle2-based cell cycle time. For the differential transcriptome analysis (both for the FUCCI-based and the Monocle2-based analysis) we used all G2/M phase cells and only G1 phase cells from the first 4 hr of G1 phase (see section SORT-seq). To increase the confidence of our differential transcriptome analysis, we only selected genes for analysis that were clearly detected in all three 384-well plates. To select detected genes, we computed for each gene in each single 384-well plate its average expression in G2/M phase cells (as we didn’t want to bias against genes that were downregulated in G1 phase), and only selected genes which had an average expression of at least two reads in each single 384-well plate. This resulted in a dataset of 3985 genes. Finally, after differential transcriptome analysis, genes that showed at least a 2-fold increase or decrease in expression and had at least a p-value of 1.2547E–5 (based on a Bonferroni correction from a p-value of 0.05) were selected as upregulated or downregulated genes, respectively.

Spline analysis

For the spline analysis (performed in Matlab R2018b), we used the full set of 841 cells (see section SORT-seq). We selected the 220 genes that were identified in the differential transcriptome analysis as downregulated in G1 phase, and fit each gene profile with a smoothing spline. Next, we computed the derivative of the splines at each time point and determined the time when the derivative was minimal for each gene (i.e. the moment mRNA levels decreased most). To compare different genes to each other, we normalized the derivative of each gene to its minimum value (i.e. setting the minimum value to –1). Finally, we determined for each gene the first time point during which the normalized derivative was at least –0.95 (where –1.0 is the minimum slope after normalization), and divided genes into two groups; one group in which the minimum slope was reached at the first time point (i.e. during mitosis) and one group in which the minimum slope was reached during G1.

Calculation of transcription rates

To assess transcription rates in different cell cycle phases of the set of genes that is downregulated in G1 phase, we made use of a previously published dataset (Battich et al., 2020) (accession number GSE128365, RPE-1 labeled and spliced dataset). In the experiment used to create this dataset, RPE-FUCCI cells were incubated for varying amount of times with EU, an analog of uridine. Transcripts containing EU were biotinylated and separated from transcripts without EU, using streptavidin magnetic beads. Using a sc-seq pipeline (scEU-seq, Battich et al., 2020), the amount of labeled and unlabeled transcripts was determined in single cells. For our analysis, we used data of cells that were incubated for 30 min in EU (i.e. those cells whose ‘Condition_Id’ is listed as ‘Pulse_30’ in the metadata). We identified cells in G1 phase (i.e those cells whose ‘Cell_cycle_relativePos’ is between 0.00 and 0.33, see Battich et al., 2020) and cells in G2 phase (i.e. those cells whose ‘Cell_cycle_relativePos’ is between 0.83 and 1.00, see Battich et al., 2020). For each of the 220 downregulated genes, we computed the average number of labeled transcripts in G1 phase cells and in G2 phase cells and we computed the ratio of labeled transcripts in G1 phase compared to G2 phase (see Figure 3A and Figure 3—figure supplement 1A). We excluded genes for which the average number of labeled transcripts in G2 phase is zero.

Modeling mRNA decrease mRNA levels (m) depend on the synthesis (µ) and degradation (γ) rate, and the change in mRNA levels over time can be described as follows:

dmdt=μγm (1)

To describe the mRNA levels as cells progress from mitosis into G1 phase, we assumed a simple model in which the observed decrease of mRNA levels is explained by a decrease in the synthesis rate and/or an increase in the degradation rate at a specific time point during M or early G1 phase (referred to as the onset time or tonset). When mRNA levels start at a given value (m0), the solution of Equation 1 results in the following expression for the mRNA levels over time.

m(t)=μγ+(m0μγ)eγt (2)

Furthermore, we assumed that mRNA levels remain constant before the onset time, resulting in the following pair of equations to describe the mRNA levels as cells progress from mitosis into G1 phase.

m(t)=m0t<tonsetm(t)=μγ+(m0μγ)eγ(ttonset)t<tonset (3)

For each gene, we optimized tonset (performed in Matlab R2018B) using an iterative search (between 0 and 370 min after metaphase in steps of 10 min), in which we optimized m0, µ, and γ using least square fitting for each tonset. Finally, we computed a sum of squared errors (SSE) between the data (using the full dataset of 841 cells) and model for each tonset and selected the tonset with the minimal SSE.

Calculating half-lives

We computed mRNA half-lives from the degradation rates (γ) (that we obtained from the modeling) using Equation 4.

Half-life =ln2γ (4)

Statistics

Statistical comparisons were made using an unpaired one-tailed Student’s t-test (Figures 2E, F4A, B and C, Figure 2—figure supplement 1K-L, Figure 3—figure supplement 1A and M, and Figure 4—figure supplement 1G), a paired one-tailed Student’s t-test (Figure 3B) or a one-tailed Welch’s t-test (Figure 4D).

Acknowledgements

We thank members of the Tanenbaum group for helpful discussions, and Xiaowei Yan for help during the initial stages of the project. We thank the Hubrecht Institute flow cytometry facility and single-cell sequencing facility (now Single Cell Discoveries) for their technical support. This work was financially supported by the European Research Council (ERC) through an ERC starting grant (ERCSTG 677936-RNAREG) to MET; MET is also supported by the Oncode Institute that is partially funded by the Dutch Cancer Society (KWF).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Marvin E Tanenbaum, Email: m.tanenbaum@hubrecht.eu.

Robert H Singer, Albert Einstein College of Medicine, United States.

James L Manley, Columbia University, United States.

Funding Information

This paper was supported by the following grant:

  • European Research Council ERCSTG 677936-RNAREG to Marvin E Tanenbaum.

Additional information

Competing interests

No competing interests declared.

No competing interests declared.

Author contributions

Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft.

Conceptualization, Data curation, Formal analysis, Investigation, Software, Validation, Writing – original draft.

Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – review and editing.

Additional files

Supplementary file 1. Analyses used in this study.

This file contains analyses that were used throughout the study, in Figure 1D–E, Figure 1—figure supplement 1J,L,N, Figures 2A–B3C and Figure 3—figure supplement 1B-G.

elife-71356-supp1.xlsx (7.1MB, xlsx)
Supplementary file 2. Sample size indication.

This file contains sample sizes for all experiments that involved single-cell analyses.

elife-71356-supp2.xlsx (16.5KB, xlsx)
Supplementary file 3. Nucleotide sequences.

This file contains nucleotide sequences for reagents that were used in this study; RT-qPCR primer sequences and sgRNA sequences.

elife-71356-supp3.xlsx (10.8KB, xlsx)
Source data 1. Single-cell transcript counts plate 1.
elife-71356-data1.csv (13.6MB, csv)
Source data 2. Single-cell transcript counts plate 2.
elife-71356-data2.csv (13.5MB, csv)
Source data 3. Single-cell transcript counts plate 3.
elife-71356-data3.csv (13.2MB, csv)
Source data 4. Single-cell sequencing metadata.
elife-71356-data4.xlsx (41.5KB, xlsx)
Transparent reporting form

Data availability

Source data containing single cell transcript counts that were used in this study are provided as supplementary data.

References

  1. Alfieri C, Zhang S, Barford D. Visualizing the complex functions and mechanisms of the anaphase promoting complex/cyclosome (APC/C) Open Biology. 2017;7:170204. doi: 10.1098/rsob.170204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Banaszynski LA, Chen L-C, Maynard-Smith LA, Ooi AGL, Wandless TJ. A rapid, reversible, and tunable method to regulate protein function in living cells using synthetic small molecules. Cell. 2006;126:995–1004. doi: 10.1016/j.cell.2006.07.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bar-Joseph Z, Siegfried Z, Brandeis M, Brors B, Lu Y, Eils R, Dynlacht BD, Simon I. Genome-wide transcriptional analysis of the human cell cycle identifies genes differentially regulated in normal and cancer cells. PNAS. 2008;105:955–960. doi: 10.1073/pnas.0704723105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Battich N, Beumer J, de Barbanson B, Krenning L, Baron CS, Tanenbaum ME, Clevers H, van Oudenaarden A. Sequencing metabolically labeled transcripts in single cells reveals mRNA turnover strategies. Science. 2020;367:1151–1156. doi: 10.1126/science.aax3072. [DOI] [PubMed] [Google Scholar]
  5. Bertoli C, Skotheim JM, de Bruin RAM. Control of cell cycle transcription during G1 and S phases. Nature Reviews. Molecular Cell Biology. 2013;14:518–528. doi: 10.1038/nrm3629. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Blomen VA, Májek P, Jae LT, Bigenzahn JW, Nieuwenhuis J, Staring J, Sacco R, van Diemen FR, Olk N, Stukalov A, Marceau C, Janssen H, Carette JE, Bennett KL, Colinge J, Superti-Furga G, Brummelkamp TR. Gene essentiality and synthetic lethality in haploid human cells. Science. 2015;350:1092–1096. doi: 10.1126/science.aac7557. [DOI] [PubMed] [Google Scholar]
  7. Bortner DM, Rosenberg MP. Induction of mammary gland hyperplasia and carcinomas in transgenic mice expressing human cyclin E. Molecular and Cellular Biology. 1997;17:453–459. doi: 10.1128/MCB.17.1.453. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Castro A, Bernis C, Vigneron S, Labbé JC, Lorca T. The anaphase-promoting complex: a key factor in the regulation of cell cycle. Oncogene. 2005;24:314–325. doi: 10.1038/sj.onc.1207973. [DOI] [PubMed] [Google Scholar]
  9. Chaudhry MA, Chodosh LA, McKenna WG, Muschel RJ. Gene expression profiling of HeLa cells in G1 or G2 phases. Oncogene. 2002;21:1934–1942. doi: 10.1038/sj.onc.1205264. [DOI] [PubMed] [Google Scholar]
  10. Cho RJ, Campbell MJ, Winzeler EA, Steinmetz L, Conway A, Wodicka L, Wolfsberg TG, Gabrielian AE, Landsman D, Lockhart DJ, Davis RW. A Genome-Wide Transcriptional Analysis of the Mitotic Cell Cycle including control of mRNA transcription. Molecular Cell. 1998;2:65–73. doi: 10.1016/S1097-2765(00)80114-8. [DOI] [PubMed] [Google Scholar]
  11. Cho RJ, Huang M, Campbell MJ, Dong H, Steinmetz L, Sapinoso L, Hampton G, Elledge SJ, Davis RW, Lockhart DJ. Transcriptional regulation and function during the human cell cycle. Nature Genetics. 2001;27:48–54. doi: 10.1038/83751. [DOI] [PubMed] [Google Scholar]
  12. Coelho PA, Bury L, Shahbazi MN, Liakath-Ali K, Tate PH, Wormald S, Hindley CJ, Huch M, Archer J, Skarnes WC, Zernicka-Goetz M, Glover DM. Over-expression of Plk4 induces centrosome amplification, loss of primary cilia and associated tissue hyperplasia in the mouse. Open Biology. 2015;5:150209. doi: 10.1098/rsob.150209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Eisen TJ, Eichhorn SW, Subtelny AO, Lin KS, McGeary SE, Gupta S, Bartel DP. The Dynamics of Cytoplasmic mRNA Metabolism. Molecular Cell. 2020;77:786–799. doi: 10.1016/j.molcel.2019.12.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Eser P, Demel C, Maier KC, Schwalb B, Pirkl N, Martin DE, Cramer P, Tresch A. Periodic mRNA synthesis and degradation co‐operate during cell cycle gene expression. Molecular Systems Biology. 2014;10:717. doi: 10.1002/msb.134886. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. García-Higuera I, Manchado E, Dubus P, Cañamero M, Méndez J, Moreno S, Malumbres M. Genomic stability and tumour suppression by the APC/C cofactor Cdh1. Nature Cell Biology. 2008;10:802–811. doi: 10.1038/ncb1742. [DOI] [PubMed] [Google Scholar]
  16. Garneau NL, Wilusz J, Wilusz CJ. The highways and byways of mRNA decay. Nature Reviews. Molecular Cell Biology. 2007;8:113–126. doi: 10.1038/nrm2104. [DOI] [PubMed] [Google Scholar]
  17. Gilbert LA, Larson MH, Morsut L, Liu Z, Brar GA, Torres SE, Stern-Ginossar N, Brandman O, Whitehead EH, Doudna JA, Lim WA, Weissman JS, Qi LS. CRISPR-mediated modular RNA-guided regulation of transcription in eukaryotes. Cell. 2013;154:442–451. doi: 10.1016/j.cell.2013.06.044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Gilbert LA, Horlbeck MA, Adamson B, Villalta JE, Chen Y, Whitehead EH, Guimaraes C, Panning B, Ploegh HL, Bassik MC, Qi LS, Kampmann M, Weissman JS. Genome-Scale CRISPR-Mediated Control of Gene Repression and Activation. Cell. 2014;159:647–661. doi: 10.1016/j.cell.2014.09.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Grant GD, Brooks L, Zhang X, Mahoney JM, Martyanov V, Wood TA, Sherlock G, Cheng C, Whitfield ML. Identification of cell cycle-regulated genes periodically expressed in U2OS cells and their regulation by FOXM1 and E2F transcription factors. Molecular Biology of the Cell. 2013;24:3634–3650. doi: 10.1091/mbc.E13-05-0264. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Grant GD, Kedziora KM, Limas JC, Cook JG, Purvis JE. Accurate delineation of cell cycle phase transitions in living cells with PIP-FUCCI. Cell Cycle. 2018;17:2496–2516. doi: 10.1080/15384101.2018.1547001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Haghverdi L, Büttner M, Wolf FA, Buettner F, Theis FJ. Diffusion pseudotime robustly reconstructs lineage branching. Nature Methods. 2016;13:845–848. doi: 10.1038/nmeth.3971. [DOI] [PubMed] [Google Scholar]
  22. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144:646–674. doi: 10.1016/j.cell.2011.02.013. [DOI] [PubMed] [Google Scholar]
  23. Harper JW, Burton JL, Solomon MJ. The anaphase-promoting complex: it’s not just for mitosis any more. Genes & Development. 2002;16:2179–2206. doi: 10.1101/gad.1013102. [DOI] [PubMed] [Google Scholar]
  24. Hart T, Chandrashekhar M, Aregger M, Steinhart Z, Brown KR, MacLeod G, Mis M, Zimmermann M, Fradet-Turcotte A, Sun S, Mero P, Dirks P, Sidhu S, Roth FP, Rissland OS, Durocher D, Angers S, Moffat J. High-Resolution CRISPR Screens Reveal Fitness Genes and Genotype-Specific Cancer Liabilities. Cell. 2015;163:1515–1526. doi: 10.1016/j.cell.2015.11.015. [DOI] [PubMed] [Google Scholar]
  25. Hashimshony T, Senderovich N, Avital G, Klochendler A, De LY, Anavy L, Gennert D, Li S, Livak KJ, Rozenblatt-rosen O, Dor Y, Regev A. CEL-Seq2: sensitive highly-multiplexed. Genome Biology. 2016;10:1–7. doi: 10.1186/s13059-016-0938-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Herzog VA, Reichholf B, Neumann T, Rescheneder P, Bhat P, Burkard TR, Wlotzka W, von Haeseler A, Zuber J, Ameres SL. Thiol-linked alkylation of RNA to assess expression dynamics. Nature Methods. 2017;14:1198–1204. doi: 10.1038/nmeth.4435. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Horlbeck MA, Gilbert LA, Villalta JE, Adamson B, Pak RA, Chen Y, Fields AP, Park CY, Corn JE, Kampmann M, Weissman JS. Compact and highly active next-generation libraries for CRISPR-mediated gene repression and activation. eLife. 2016;5:e19760. doi: 10.7554/eLife.19760. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Hsiao CJ, Tung PY, Blischak JD, Burnett JE, Barr KA, Dey KK, Stephens M, Gilad Y. Characterizing and inferring quantitative cell cycle phase in single-cell RNA-seq data analysis. Genome Research. 2020;30:611–621. doi: 10.1101/gr.247759.118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Jost M, Chen Y, Gilbert LA, Horlbeck MA, Krenning L, Menchon G, Rai A, Cho MY, Stern JJ, Prota AE, Kampmann M, Akhmanova A, Steinmetz MO, Tanenbaum ME, Weissman JS. Combined CRISPRi/a-Based Chemical Genetic Screens Reveal that Rigosertib Is a Microtubule-Destabilizing Agent. Molecular Cell. 2017;68:210–223. doi: 10.1016/j.molcel.2017.09.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kalin TV, Wang I-C, Ackerson TJ, Major ML, Detrisac CJ, Kalinichenko VV, Lyubimov A, Costa RH. Increased levels of the FoxM1 transcription factor accelerate development and progression of prostate carcinomas in both TRAMP and LADY transgenic mice. Cancer Research. 2006;66:1712–1720. doi: 10.1158/0008-5472.CAN-05-3138. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kim I-M, Ackerson T, Ramakrishna S, Tretiakova M, Wang I-C, Kalin TV, Major ML, Gusarova GA, Yoder HM, Costa RH, Kalinichenko VV. The Forkhead Box m1 transcription factor stimulates the proliferation of tumor cells during development of lung cancer. Cancer Research. 2006;66:2153–2161. doi: 10.1158/0008-5472.CAN-05-3003. [DOI] [PubMed] [Google Scholar]
  32. Kronja I, Orr-Weaver TL. Translational regulation of the cell cycle: when, where, how and why? Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences. 2011;366:3638–3652. doi: 10.1098/rstb.2011.0084. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Laoukili J, Kooistra MRH, Brás A, Kauw J, Kerkhoven RM, Morrison A, Clevers H, Medema RH. FoxM1 is required for execution of the mitotic programme and chromosome stability. Nature Cell Biology. 2005;7:126–136. doi: 10.1038/ncb1217. [DOI] [PubMed] [Google Scholar]
  34. Mahdessian D, Cesnik AJ, Gnann C, Danielsson F, Stenström L, Arif M, Zhang C, Le T, Johansson F, Shutten R, Bäckström A, Axelsson U, Thul P, Cho NH, Carja O, Uhlén M, Mardinoglu A, Stadler C, Lindskog C, Ayoglu B, Leonetti MD, Pontén F, Sullivan DP, Lundberg E. Spatiotemporal dissection of the cell cycle with single-cell proteogenomics. Nature. 2021;590:649–654. doi: 10.1038/s41586-021-03232-9. [DOI] [PubMed] [Google Scholar]
  35. Malumbres M, Barbacid M. To cycle or not to cycle: a critical decision in cancer. Nature Reviews. Cancer. 2001;1:222–231. doi: 10.1038/35106065. [DOI] [PubMed] [Google Scholar]
  36. Muraro MJ, Dharmadhikari G, Grün D, Groen N, Dielen T, Jansen E, van Gurp L, Engelse MA, Carlotti F, de Koning EJP, van Oudenaarden A. A Single-Cell Transcriptome Atlas of the Human Pancreas. Cell Systems. 2016;3:385–394. doi: 10.1016/j.cels.2016.09.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Nakayama KI, Nakayama K. Ubiquitin ligases: cell-cycle control and cancer. Nature Reviews. Cancer. 2006;6:369–381. doi: 10.1038/nrc1881. [DOI] [PubMed] [Google Scholar]
  38. Nishimura K, Fukagawa T, Takisawa H, Kakimoto T, Kanemaki M. An auxin-based degron system for the rapid depletion of proteins in nonplant cells. Nature Methods. 2009;6:917–922. doi: 10.1038/nmeth.1401. [DOI] [PubMed] [Google Scholar]
  39. Park HJ, Costa RH, Lau LF, Tyner AL, Raychaudhuri P. Anaphase-promoting complex/cyclosome-CDH1-mediated proteolysis of the forkhead box M1 transcription factor is critical for regulated entry into S phase. Molecular and Cellular Biology. 2008;28:5162–5171. doi: 10.1128/MCB.00387-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Park JE, Yi H, Kim Y, Chang H, Kim VN. Regulation of Poly(A) Tail and Translation during the Somatic Cell Cycle. Molecular Cell. 2016;62:462–471. doi: 10.1016/j.molcel.2016.04.007. [DOI] [PubMed] [Google Scholar]
  41. Peters JM. The Anaphase-Promoting Complex. Molecular Cell. 2002;9:931–943. doi: 10.1016/S1097-2765(02)00540-3. [DOI] [PubMed] [Google Scholar]
  42. Qiu X, Hill A, Packer J, Lin D, Ma YA, Trapnell C. Single-cell mRNA quantification and differential analysis with Census. Nature Methods. 2017a;14:309–315. doi: 10.1038/nmeth.4150. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Qiu X, Mao Q, Tang Y, Wang L, Chawla R, Pliner HA, Trapnell C. Reversed graph embedding resolves complex single-cell trajectories. Nature Methods. 2017b;14:979–982. doi: 10.1038/nmeth.4402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Rambout X, Detiffe C, Bruyr J, Mariavelle E, Cherkaoui M, Brohée S, Demoitié P, Lebrun M, Soin R, Lesage B, Guedri K, Beullens M, Bollen M, Farazi TA, Kettmann R, Struman I, Hill DE, Vidal M, Kruys V, Simonis N, Twizere J-C, Dequiedt F. The transcription factor ERG recruits CCR4-NOT to control mRNA decay and mitotic progression. Nature Structural & Molecular Biology. 2016;23:663–672. doi: 10.1038/nsmb.3243. [DOI] [PubMed] [Google Scholar]
  45. Sadasivam S, DeCaprio JA. The DREAM complex: master coordinator of cell cycle-dependent gene expression. Nature Reviews. Cancer. 2013;13:585–595. doi: 10.1038/nrc3556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Saelens W, Cannoodt R, Todorov H, Saeys Y. A comparison of single-cell trajectory inference methods. Nature Biotechnology. 2019;37:547–554. doi: 10.1038/s41587-019-0071-9. [DOI] [PubMed] [Google Scholar]
  47. Sakaue-Sawano A, Kurokawa H, Morimura T, Hanyu A, Hama H, Osawa H, Kashiwagi S, Fukami K, Miyata T, Miyoshi H, Imamura T, Ogawa M, Masai H, Miyawaki A. Visualizing spatiotemporal dynamics of multicellular cell-cycle progression. Cell. 2008;132:487–498. doi: 10.1016/j.cell.2007.12.033. [DOI] [PubMed] [Google Scholar]
  48. Santos A, Wernersson R, Jensen LJ. Cyclebase 3.0: a multi-organism database on cell-cycle regulation and phenotypes. Nucleic Acids Research. 2015;43:D1140–D1144. doi: 10.1093/nar/gku1092. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Schwanhäusser B, Busse D, Li N, Dittmar G, Schuchhardt J, Wolf J, Chen W, Selbach M. Global quantification of mammalian gene expression control. Nature. 2011;473:337–342. doi: 10.1038/nature10098. [DOI] [PubMed] [Google Scholar]
  50. Sigl R, Wandke C, Rauch V, Kirk J, Hunt T, Geley S. Loss of the mammalian APC/C activator FZR1 shortens G1 and lengthens S phase but has little effect on exit from mitosis. Journal of Cell Science. 2009;122:4208–4217. doi: 10.1242/jcs.054197. [DOI] [PubMed] [Google Scholar]
  51. Sivakumar S, Gorbsky GJ. Spatiotemporal regulation of the anaphase-promoting complex in mitosis. Nature Reviews. Molecular Cell Biology. 2015;16:82–94. doi: 10.1038/nrm3934. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Spencer SL, Cappell SD, Tsai F-C, Overton KW, Wang CL, Meyer T. The proliferation-quiescence decision is controlled by a bifurcation in CDK2 activity at mitotic exit. Cell. 2013;155:369–383. doi: 10.1016/j.cell.2013.08.062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Stumpf CR, Moreno MV, Olshen AB, Taylor BS, Ruggero D. The translational landscape of the mammalian cell cycle. Molecular Cell. 2013;52:574–582. doi: 10.1016/j.molcel.2013.09.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Tanenbaum ME, Stern-Ginossar N, Weissman JS, Vale RD. Regulation of mRNA translation during mitosis. eLife. 2015;4:e07957. doi: 10.7554/eLife.07957. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Tani H, Akimitsu N. Genome-wide technology for determining RNA stability in mammalian cells: Historical perspective and recent advantages based on modified nucleotide labeling. RNA Biology. 2012;9:1233–1238. doi: 10.4161/rna.22036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Tani H, Mizutani R, Salam KA, Tano K, Ijiri K, Wakamatsu A, Isogai T, Suzuki Y, Akimitsu N. Genome-wide determination of RNA stability reveals hundreds of short-lived noncoding transcripts in mammals. Genome Research. 2012;22:947–956. doi: 10.1101/gr.130559.111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Tian L, Dong X, Freytag S, Lê Cao KA, Su S, JalalAbadi A, Amann-Zalcenstein D, Weber TS, Seidi A, Jabbari JS, Naik SH, Ritchie ME. Benchmarking single cell RNA-sequencing analysis pipelines using mixture control experiments. Nature Methods. 2019;16:479–487. doi: 10.1038/s41592-019-0425-8. [DOI] [PubMed] [Google Scholar]
  58. Trapnell C, Cacchiarelli D, Grimsby J, Pokharel P, Li S, Morse M, Lennon NJ, Livak KJ, Mikkelsen TS, Rinn JL. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nature Biotechnology. 2014;32:381–386. doi: 10.1038/nbt.2859. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Trcek T, Larson DR, Moldón A, Query CC, Singer RH. Single-molecule mRNA decay measurements reveal promoter- regulated mRNA stability in yeast. Cell. 2011;147:1484–1497. doi: 10.1016/j.cell.2011.11.051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Vaidyanathan S, Cato K, Tang L, Pavey S, Haass NK, Gabrielli BG, Duijf PHG. In vivo overexpression of Emi1 promotes chromosome instability and tumorigenesis. Oncogene. 2016;35:5446–5455. doi: 10.1038/onc.2016.94. [DOI] [PubMed] [Google Scholar]
  61. Vodermaier HC. APC/C and SCF: controlling each other and the cell cycle. Current Biology. 2004;14:R787–R796. doi: 10.1016/j.cub.2004.09.020. [DOI] [PubMed] [Google Scholar]
  62. Wang T, Birsoy K, Hughes NW, Krupczak KM, Post Y, Wei JJ, Lander ES, Sabatini DM. Identification and characterization of essential genes in the human genome. Science. 2015;350:1096–1101. doi: 10.1126/science.aac7041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Whitfield ML, Sherlock G, Saldanha AJ, Murray JI, Ball CA, Alexander KE, Matese JC, Perou CM, Hurt MM, Brown PO, Botstein D. Identification of genes periodically expressed in the human cell cycle and their expression in tumors. Molecular Biology of the Cell. 2002;13:1977–2000. doi: 10.1091/mbc.02-02-0030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Yamashita A, Chang TC, Yamashita Y, Zhu W, Zhong Z, Chen CYA, Shyu AB. Concerted action of poly(A) nucleases and decapping enzyme in mammalian mRNA turnover. Nature Structural & Molecular Biology. 2005;12:1054–1063. doi: 10.1038/nsmb1016. [DOI] [PubMed] [Google Scholar]
  65. Yang HW, Chung M, Kudo T, Meyer T. Competing memories of mitogen and p53 signalling control cell-cycle entry. Nature. 2017;549:404–408. doi: 10.1038/nature23880. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Yesbolatova A, Saito Y, Kitamoto N, Makino-Itou H, Ajima R, Nakano R, Nakaoka H, Fukui K, Gamo K, Tominari Y, Takeuchi H, Saga Y, Hayashi K-I, Kanemaki MT. The auxin-inducible degron 2 technology provides sharp degradation control in yeast, mammalian cells, and mice. Nature Communications. 2020;11:5701. doi: 10.1038/s41467-020-19532-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

Editor's evaluation

Robert H Singer 1

This work is a very significant contribution on how mRNA stability regulation takes place across the cell cycle, particularly after mitosis. It also describes an original method for time-resolved transcriptome analysis during the cell cycle, which is potentially very interesting for the whole molecular and cell biology community.

Decision letter

Editor: Robert H Singer1

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

[Editors' note: this paper was reviewed by Review Commons.]

Decision letter after peer review:

Thank you for submitting your article "Time-resolved single-cell sequencing identifies multiple waves of mRNA decay during mitotic exit" for consideration by eLife. Your article has been reviewed by 3 peer reviewers at Review Commons, and the evaluation at eLife has been overseen by a Reviewing Editor and James Manley as the Senior Editor. We have also sought a new reviewer with specific expertise and discussed the reviews with one another. We would like to invite a revised submission, with these comments in mind:

Reviewer #1:

The manuscript describes interesting and relevant insights into cell-cycle controlled mRNA decay and in my opinion is suitable for publication in eLife after revision. Specifically, I recommend the following adjustments:

1) Include in the Discussion section the possible involvement of other mRNA decay factors in cell-cycle dependent RNA decay, as already indicated in the response to reviewer 1. Given the lack of data demonstrating the biological relevance of a controlled RNA decay after mitosis, the discussion should in my opinion toned down appropriately.

2) The proposed experiments addressing a possible relationship between G0/G1 bifurcation and the described two-wave mRNA decay model will certainly strengthen the manuscript and I highly recommend including such data in a revised version.

3) As acknowledged by the authors in response to reviewer 3, further analyses on concomitant transcription levels of immediate decay and delayed decay genes would certainly help to clarify the overall impact of cell-cycle controlled mRNA decay.

Reviewer #2:

I believe the work is potentially suitable for publication in eLife but additional data is needed to support the major claims.

This paper essentially makes three major points:

Description of a new methodology of combining FUCCI sensor levels with sc-RNA-seq to get time-resolved RNA dynamics.

• For this main part of the manuscript, I think the authors need to do a few more experiments to fully validate this method, to determine the resolution and level of accuracy it provides and as Reviewer #4 states "to improve reproducibility" by other labs trying to replicate their results.

• In their rebuttal, the authors have already included some additional analysis that helps differentiate G0 and G1 cells. They also propose to perform new experiments in cells lacking p53 which should be incapable of entering G0. These would be nice additions to the manuscript and would address some of my comments.

• Further characterization of the accuracy and time resolution is also needed in order to correctly interpret their results. There is variability in the expression levels of the FUCCI sensors independent of time. Some sort of quantification of this is needed to understand the error in their time estimates.

• These additions should be relatively straight forward and would improve the manuscript.

mRNA is decayed in two waves after mitosis

• The authors have mostly addressed my comments on this point and have proposed several experiments to further address these comments including additional validation of FBXO5 and to look at mRNA levels at a higher time resolution within mitosis (telophase, metaphase, as well as anaphase).

• I also agree with Reviewer #4 that the author's comments about the importance of these waves of mRNA decay are not supported by any experiments. The authors should either edit the text to be more conservative or perform additional experiments to support this claim.

CNOT1 contributes to the observed mRNA decay after mitosis

• The data supporting this conclusion are relatively weak. The change in mRNA decay upon CNOT1 knockdown are all relatively minor. While the changes seem to be statistically significant, it’s not clear if these changes are large enough to be biologically relevant. They also don't appear robustly repeatable as CDK1 mRNA is rescued in Figure 4A but not in 4D.

• I think the authors should edit the text here to reflect that CNOT1 may contribute somewhat to the decay but that it is not required, as requested by Reviewer 4, and to discuss additional factors that may be involved as requested by Reviewer 3.

In summary, I think the authors have either already addressed or have proposed experiments/analysis to address most of the reviewers comments about the methodology as well as the two waves of mRNA decay. However, the authors do not appear to have a clear plan to try and address the mechanism of the mRNA decay and whether or not CNOT1 is a major regulator of this decay. I think the relatively weak phenotypes with CNOT1 depletion are a major weakness of the manuscript but I recognize that the experiments needed to support their claims may not be feasible in a reasonable time.

Reviewer #3:

I think that the paper by Tanenbaum and colleagues is suitable for publication in eLife as long as they:

1) Perform the new experiments that they described in the rebuttal letter. Particularly important is to perform measurements of mRNA sysnthesis by addressing nascent transcription rates. The authors propose to include an analysis addressing the relative transcription levels of 'immediate decay' and 'delayed decay' genes. I agree with the authors that these results could clarify most of the concerns in this respect.

2) Carefully revise the text to avoid too strong conclusions not directly related to the results, as pointed out by the fourth referee.

3) Modify the Discussion section in order to consider the possibility that other mRNA decay factors may act in parallel to CNOT1 or in concert with it. This is particularly convenient if we considered the information supplied by the authors in response to Reviewer 1, describing the results of experiments addressing the possible involvement of DCP2 in post-mitotic mRNA decay (Figure 1A,B of the rebuttal letter).

In conclusion, I consider that this manuscript, with the new experiments and changes proposed by the authors, is a significant contribution on how mRNA stability regulation takes place accross the cell cycle, particularly after mitosis. This results should be of interest for a broad audience in the molecular and cell biology community, which would appreciate a paper like this in eLife.

Reviewer #4:

In this work, Krenning and colleagues aim to investigate the mechanisms controlling periodic cell cycle mRNAs expression in mammalian cells. While previous studies extensively characterized the control of periodic mRNA synthesis, as well as cell cycle regulated protein destruction, here the authors investigate the control of mRNA degradation, which they suggest may contribute to sharp transitions between cell cycle phase and unidirectional cycles. To this end, they combine already existing tools – single-cell sequencing and cell cycle monitoring by using the fluorescent reporter FUCCI (Sakaue-Sawano et al. 2008 PMID: 18267078) – to precisely correlate the cell cycle phase of a cell and its transcriptome.

Here lies the novelty of this work. Previous reports (by the authors and others) described a similar approach, coupling FUCCI, single cell sequencing and metabolic labelling, to measure mRNA synthesis and degradation rates in mammalian cells (e.g. Battich et al. 2020, PMID: 32139547). Using this approach however, the RNA labelling times were long (up to 6 hours for pulse-chase experiments), precluding to capture rapid changes in mRNA expression that may occur during cell cycle phase transitions. To overcome this limitation, Krenning and colleagues set to precisely calibrate the FUCCI reporter fluorescence during the cell cycle of HEK293T cells, using a combination of live imaging and FACS analysis. This fluorescence-based calibration is used to obtain a cell cycle progression (pseudo)time for each cell by FACS. Then single cells are sorted to perform single-cell sequencing, thus providing the significant advantage that cells do not be synchronized before gene expression measurements. Using this tool and a combination of cell cycle synchronization, transcription inhibition and mathematical modelling, the authors conclude that two-waves of mRNA degradation occur for a sub-class of cell cycle mRNAs during the M/G1 phase transition. Furthermore, they show that depletion – even if incomplete – of the decay factor CNOT1 promotes partial stabilization of the mRNAs degraded during the M/G1 phase transition.

While I think that the method is original and it has the potential to address important biological questions, I found two major weakness in this study. First, the accuracy of the FUCCI reporter in determining cell cycle progression (pseudo)time in single cells is not definitively demonstrated. A key aspect that remains unclear is how variable is the expression of the FUCCI reporter and how the calibration suffers from cell-to-cell heterogeneity in the reporter expression and cell cycle duration. This could affect the results of experiments where the reporter is used to sort cells based on the cell cycle stage to measure mRNA expression on pools of cells (e.g. Figure 2C, Figure 3-figure supplement 1A-F). Second, the authors conclude that "mRNA degradation occurs at the M/G1 phase transition, and provides an important contribution to the reset of the transcriptome after cell division" and that "mRNA decay requires CNOT1". I find that these claims are very strong and they are not fully supported by the data presented in this manuscript. For instance no experiment is performed to determine the importance of this decay mechanism and its consequences on cell cycle. Furthermore, the CNOT1 downregulation experiment shows that CNOT1 provides a small, even if significant, contribution to decay, not that it is required.

Overall, the impact of this work on the cell cycle field is two-fold. First, it describes technology that, provided that an accurate calibration is done, can be applied to numerous model organisms since the FUCCI system is available, for instance, for mice, flies and zebrafish (see review by Zielke and Edgar, 2015, PMID: 25827130). Second, this study provides novel insights on the regulation of mRNA degradation during the mammalian cell cycle.

Major concerns:

The authors claim that their method can accurately measure the cell cycle stage and the "absolute" time an individual cell spent in the cell cycle with a resolution in the scale of minutes (p. 5). I have few concerns here, that affect also the strength of the biological conclusions. Here my specific points:

a) The time resolution estimate is based on an average cell cycle length determined by fluorescence microscopy (Figure 1A). The cell-to-cell variability in cell cycle length should be computed and used to estimate the error of the time resolution.

b) The authors combined microscopy and FACS to calibrate the fluorescence measured with the two methods. In particular they base their calibration on the identification of cells in early S phase to normalize the two fluorescence datasets. Thus, accurate identification of early S phase cells seems very important to compare the fluorescence detected with the two methods. It is unclear from the text and methods section whether they used also independent approaches to determine the cell cycle phase. For instance in the FUCCI paper (Sakaue-Sawano et al. 2008 PMID: 18267078) Hoechst staining is used to independently determine the cell cycle. Furthermore, by microscopy, how did the authors determine when cells were undergoing early S-phase? Did they use FUCCI alone? To clarify this point, they could add a fluorescence microscopy video where they indicate how they classify cell cycle stages. This will be very important to ensure reproducibility of the method.

c) The FUCCI reporter expression heterogeneity is not clearly discussed in the text and it is unclear whether it may affect the accuracy of the method. The authors should provide a sample microscopy video (same as in (b) to show how heterogeneous (or not) is the expression of the reporter across many synchronized cells.

d) The biological discovery of this paper is that mRNA levels decline in multiple waves during mitotic exit. I find the text at times confusing. For instance at page 7, the paragraph title "mRNA levels decline in multiple waves after cell division", suggests that both degradation waves occur after completion of cell division. But this is not what is shown in Figure 2D. The "immediate" degraded mRNAs are degraded before the end of mitosis, according to the FISH quantifications. I highly recommend to clarify the text.

e) The method presented in this manuscript provides the ability to precisely estimate the duration of the different cell cycle phases of HEK293 cells. One result that I would recommend adding is the duration of the different cell cycle phase of an average HEK293 cell, based on your FUCCI reporter estimates. This will be important to interpret many of your plots where the X axis is presented as "Time after metaphase". I make this point because I think that some of the mRNA levels declines reported in your study may be explained by a dilution effect occurring upon cell division. I think that this is particularly important for the "delayed" mRNA pool. Could the authors please comment this point in the text? How did they consider dividing cells while performing single cell imaging?

f) FISH is used as in independent mean to measure mRNA reduction in single cells at different stages of the cell cycle. For these experiments, two mRNAs belonging to the "immediate" degraded group are measured, TOP2A and CDK1 and both nicely show a decrease before the end of the cell cycle. Could you add the single cell data to the FISH quantifications shown in figures 2E, 2F, and 4A, to report the cell-to-cell variability observed for the expression of these mRNA?

g) As a follow-up question, why didn't you perform FISH also on two mRNAs from the "delayed decrease" group, to show changes in mRNA expression? This seems an important control if the aim is to demonstrated trough impendent methods that two waves of degradation occur during the M/G1 phase transition.

h) The authors conclude that "mRNA degradation occurs at the M/G1 phase transition, and provides an important contribution to the reset of the transcriptome after cell division". However, no experiment is performed to determine the importance of this decay mechanism and its consequences on cell cycle. For instance it could be tested by selectively downregulating the expression of CNOT during the M/G1 phase transition by using an inducible degron system. I understand that this may be a complicated experiment to do, but if a demonstration is not provided, then the text should remain more conservative in describing the results.

i) The authors conclude that "mRNA decay requires CNOT1". The CNOT1 downregulation experiments show that CNOT1 contributes to decay, not that it is required. This should be changed in the text.

eLife. 2022 Feb 1;11:e71356. doi: 10.7554/eLife.71356.sa2

Author response


Reviewer #1:

The manuscript describes interesting and relevant insights into cell-cycle controlled mRNA decay and in my opinion is suitable for publication in eLife after revision. Specifically, I recommend the following adjustments:

1) Include in the Discussion section the possible involvement of other mRNA decay factors in cell-cycle dependent RNA decay, as already indicated in the response to reviewer 1.

In the Discussion section we now mention several other potential mRNA decay mechanisms that may contribute to cell cycle dependent RNA decay. See page 15, lines 5 to 8:

“Nonetheless, it is possible that additional mechanisms contribute to mRNA decay during the M-G1 phase transition. Such mechanisms could involve PARN-dependent deadenylation, or may be independent of mitotic deadenylation and instead rely on mRNA decapping or endonucleolytic cleavage of the mRNA.”

Given the lack of data demonstrating the biological relevance of a controlled RNA decay after mitosis, the discussion should in my opinion toned down appropriately.

We have toned down our conclusions regarding the biological relevance of RNA decay after mitosis. We have done this at the following locations in the text:

Introduction: Page 4, line 36 to page 5, line 2

“Together, our findings demonstrate that, analogous to protein degradation, mRNA degradation occurs at the M-G1 phase transition, and provides an important contribution to the reset of the transcriptome after cell division” changed to “Together, our findings demonstrate that, analogous to protein degradation, scheduled mRNA degradation occurs at the M-G1 transition. Scheduled mRNA degradation likely provides an important contribution to the reset of the transcriptome after cell division”.

Discussion: Page 14, lines 10-11

“Therefore, decay-mediated clearance of mRNAs as cells exit mitosis will aid to limit expression of the encoded proteins in G1 phase” changed to “Therefore, decay-mediated clearance of mRNAs as cells exit mitosis could aid to limit expression of the encoded proteins in G1 phase”.

Discussion: Page 14, lines 13-16

“As these genes include many genes that encode for proteins with important functions in cell cycle control, their continued expression in G1 phase may perturb normal cell cycle progression…” changed to “As these genes include many genes that encode for proteins with important functions in cell cycle control, we speculate that their continued expression in G1 phase may perturb normal cell cycle progression…”.

2) The proposed experiments addressing a possible relationship between G0/G1 bifurcation and the described two-wave mRNA decay model will certainly strengthen the manuscript and I highly recommend including such data in a revised version.

We have included experiments performed in p53 knock-out cells (which prevents cells from entering G0 phase). These experiments demonstrate that mRNA decay still occurs in two independent waves, demonstrating that these two waves of mRNA decay do not represent distinct waves occurring in G0 and G1 phase cells (see new Figure 2—figure supplement 1I-M). We discuss these results in the main text on page 9, lines 12 to 23.

We have analyzed the expression of genes belonging to the immediate and delayed decay groups in individual cells. This analysis shows that both sets of genes are downregulated in the same set of cells (new Figure 2—figure supplement 1N), indicating that both waves of mRNA decay occur in individual cells, irrespective of G1/G0 status. This result is discussed in the main text on page 9, lines 24 to 36.

3) As acknowledged by the authors in response to reviewer 3, further analyses on concomitant transcription levels of immediate decay and delayed decay genes would certainly help to clarify the overall impact of cell-cycle controlled mRNA decay.

We have now included analysis of mRNA synthesis rates for immediate and delayed decay genes (see Figure 3A and Figure 3—figure supplement 1A). These results indicate that G1 phase transcription shutdown occurs for the large majority of genes that undergo scheduled mRNA decay, demonstrating a dual mechanism of gene silencing for these genes. These data are discussed in the main text on page 10, lines 2 to 9.

Reviewer #2:

I believe the work is potentially suitable for publication in eLife but additional data is needed to support the major claims.

This paper essentially makes three major points:

Description of a new methodology of combining FUCCI sensor levels with sc-RNA-seq to get time-resolved RNA dynamics.

• For this main part of the manuscript, I think the authors need to do a few more experiments to fully validate this method, to determine the resolution and level of accuracy it provides and as Reviewer #4 states "to improve reproducibility" by other labs trying to replicate their results.

To address this point, we have now included multiple experiments that test the accuracy of our method.

First, we have quantified the cell-to-cell heterogeneity in FUCCI-G1 fluorescence and the corresponding heterogeneity (i.e. error) in the calculated G1 timing of single cells (see new Figure 1—figure supplement 1C and Methods). This analysis precisely quantifies the accuracy of our timing method at the single cell level. It is important to point out that mRNA decay rates which we report (e.g. figures 3B and 3C, Supplementary file 1) are calculated based on many cells, thus averaging out much of the cell-to-cell heterogeneity in FUCCI fluorescence. Therefore, the calculated mRNA decay rates are expected to be very accurate. We have inserted this new analysis in the manuscript (Figure 1—figure supplement 1C) and we discuss it in the main text (page 5, lines 28 to 31) and in the discussion (page 13, lines 21 to 24).

Second, we have included Hoechst-based analysis of DNA content (new Figure 1—figure supplement 1A), which independently confirms our gating strategy for cells in different phases of the cell cycle. This provides an extra quality check that helps to improve the reproducibility of our FACS-sorting based isolation of cells in different positions along the cell cycle.

Finally, we have included example images of RPE-FUCCI cells as they progress through mitosis and G1/G0 phase (new Figure 1B). We included these images in the section of the text describing the FUCCI system (page 5, lines 15-30). These images provide clear examples of how cells in the different cell cycle phases are identified by microscopy, and as such will also improve the reproducibility of this work by others.

• In their rebuttal, the authors have already included some additional analysis that helps differentiate G0 and G1 cells. They also propose to perform new experiments in cells lacking p53 which should be incapable of entering G0. These would be nice additions to the manuscript and would address some of my comments.

We have included experiments performed in p53 knock-out cells (which prevents cells from entering G0 phase). These experiments demonstrate that mRNA decay still occurs in two independent waves, demonstrating that these two waves of mRNA decay do not represent distinct waves occurring in G0 and G1 phase cells (see new Figure 2—figure supplement 1I-M). We discuss these results in the main text on page 9, lines 12 to 23.

We have analyzed the expression of genes belonging to the immediate and delayed decay groups in individual cells. This analysis shows that both sets of genes are downregulated in the same set of cells (new Figure 2—figure supplement 1N), indicating that both waves of mRNA decay occur in individual cells, irrespective of G1/G0 status. This result is discussed in the main text on page 9, lines 24 to 36.

• Further characterization of the accuracy and time resolution is also needed in order to correctly interpret their results. There is variability in the expression levels of the FUCCI sensors independent of time. Some sort of quantification of this is needed to understand the error in their time estimates.

We agree that this in an important point. As mentioned above, we have quantified the cell-to-cell heterogeneity in FUCCI-G1 fluorescence and the corresponding heterogeneity (i.e. error) in the calculated G1 timing of single cells (see new Figure 1—figure supplement 1C and Methods). This analysis precisely quantifies the accuracy of our timing method at the single cell level. It is important to point out that mRNA decay rates which we report (e.g. figures 3B and 3C, Supplementary file 1) are calculated based on many cells, thus averaging out much of the cell-to-cell heterogeneity. Therefore, the calculated mRNA decay rates are expected to be very accurate. We have inserted this new analysis in the manuscript (Figure 1—figure supplement 1C) and we discuss it in the main text (page 5, lines 28 to 31) and in the discussion (page 13, lines 21 to 24).

• These additions should be relatively straight forward and would improve the manuscript.

mRNA is decayed in two waves after mitosis

• The authors have mostly addressed my comments on this point and have proposed several experiments to further address these comments including additional validation of FBXO5 and to look at mRNA levels at a higher time resolution within mitosis (telophase, metaphase, as well as anaphase).

We have now included additional analysis of the expression levels of 5 ‘immediate’ and 5 ‘delayed’ decay genes in G2 phase compared to early / late mitosis using qPCR. These data are included in the manuscript (Figure 2—figure supplement 1G-H), and demonstrate that the levels of ‘immediate decay’ genes decrease as cells move from early to late mitosis, while ‘delayed’ decay gene expression does not decrease until G1 phase. We discuss these results in the main text at lines page 8, line 33 to page 9, line 11.

• I also agree with Reviewer #4 that the author's comments about the importance of these waves of mRNA decay are not supported by any experiments. The authors should either edit the text to be more conservative or perform additional experiments to support this claim.

We have toned down our conclusions regarding the biological relevance of RNA decay after mitosis. We have done this at the following locations in the text:

Introduction: Page 4, line 36 to page 5, line 2

“Together, our findings demonstrate that, analogous to protein degradation, mRNA degradation occurs at the M-G1 phase transition, and provides an important contribution to the reset of the transcriptome after cell division” changed to “Together, our findings demonstrate that, analogous to protein degradation, scheduled mRNA degradation occurs at the M-G1 transition. Scheduled mRNA degradation likely provides an important contribution to the reset of the transcriptome after cell division”.

Discussion: Page 14, lines 10-11

“Therefore, decay-mediated clearance of mRNAs as cells exit mitosis will aid to limit expression of the encoded proteins in G1 phase” changed to “Therefore, decay-mediated clearance of mRNAs as cells exit mitosis could aid to limit expression of the encoded proteins in G1 phase”.

Discussion: Page 14, lines 13-16

“As these genes include many genes that encode for proteins with important functions in cell cycle control, their continued expression in G1 phase may perturb normal cell cycle progression…” changed to “As these genes include many genes that encode for proteins with important functions in cell cycle control, we speculate that their continued expression in G1 phase may perturb normal cell cycle progression…”

CNOT1 contributes to the observed mRNA decay after mitosis

• The data supporting this conclusion are relatively weak. The change in mRNA decay upon CNOT1 knockdown are all relatively minor. While the changes seem to be statistically significant, it’s not clear if these changes are large enough to be biologically relevant. They also don't appear robustly repeatable as CDK1 mRNA is rescued in Figure 4A but not in 4D.

We agree with the reviewer that he change in mRNA decay upon CNOT1 knockdown is modest. In accordance with this, we have softened our statements regarding the involvement of CNOT1 in mRNA decay (see below).

We do note that CDK1 levels in Figure 4D are lower upon knockdown of CNOT1, as is the case in Figure 4A (note that Figure 4D is plotted on a log scale, while Figure 4A is plotted on a linear scale). It is true that the effect of CNOT1 depletion is slightly stronger in Figure 4A than in Figure 4D (causing the effects in Figure 4D to be not statistically significant), but this is likely caused by the higher level of knockdown of CNOT1 by siRNA (Figure 4A) compared to CRISPRi (Figure 4D).

• I think the authors should edit the text here to reflect that CNOT1 may contribute somewhat to the decay but that it is not required, as requested by Reviewer 4, and to discuss additional factors that may be involved as requested by Reviewer 3.

As suggested by the reviewer we have weakened our statements regarding the involvement of CNOT1 in mRNA decay during the M-G1 phase transition, at several locations in the manuscript.

Introduction: page 4, line 33

“For several of these genes, we show that mRNA decay requires CNOT1” changed to “For several of these genes, we show that mRNA decay is stimulated by CNOT1”.

Main text, page 12, lines 3 to 4

“These results suggest that CNOT1-dependent deadenylation is important for mRNA decay at the M-G1 phase transition” changed to “These results suggest that CNOT1-dependent deadenylation is involved in mRNA decay at the M-G1 phase transition”.

Main text, page 12, line 14 to 15

“Collectively, these data show that CNOT1 is important for the decay of TOP2A and CDK1 mRNAs during the M-G1 phase transition […]” changed to ““Collectively, these data show that CNOT1 aids the decay of TOP2A and CDK1 mRNAs during the M-G1 phase transition […]”.

Also, we now discuss additional factors that could be involved in mRNA decay during the M-G1 phase transition in the discussion. On page 15, lines 5 to 8, we write:

“Nonetheless, it is possible that additional mechanisms contribute to mRNA decay during the M-G1 phase transition. Such mechanisms could involve PARN-dependent deadenylation, or may be independent of mitotic deadenylation and instead rely on mRNA decapping or endonucleolytic cleavage of the mRNA.”

In summary, I think the authors have either already addressed or have proposed experiments/analysis to address most of the reviewers comments about the methodology as well as the two waves of mRNA decay. However, the authors do not appear to have a clear plan to try and address the mechanism of the mRNA decay and whether or not CNOT1 is a major regulator of this decay. I think the relatively weak phenotypes with CNOT1 depletion are a major weakness of the manuscript but I recognize that the experiments needed to support their claims may not be feasible in a reasonable time.

We agree with the reviewer that it will be very interesting to perform additional work to provide a more detailed mechanism of mRNA decay during the M-G1 transition. However, as the reviewer indicates, this would require a significant amount of additional work, so we feel it is beyond the scope of the current manuscript.

Reviewer #3:

I think that the paper by Tanenbaum and colleagues is suitable for publication in eLife as long as they:

1) Perform the new experiments that they described in the rebuttal letter. Particularly important is to perform measurements of mRNA synthesis by addressing nascent transcription rates. The authors propose to include an analysis addressing the relative transcription levels of 'immediate decay' and 'delayed decay' genes. I agree with the authors that these results could clarify most of the concerns in this respect.

We have now included analysis of mRNA synthesis rates for immediate and delayed decay genes (see Figure 3A and Figure 3—figure supplement 1A). These results indicate that G1 phase transcription shutdown occurs for the large majority of genes that undergo scheduled mRNA decay, demonstrating a dual mechanism of gene silencing for these genes. These data are discussed in the main text on page 10, lines 2 to 9.

2) Carefully revise the text to avoid too strong conclusions not directly related to the results, as pointed out by the fourth referee.

As suggested by the reviewer we have weakened our statements regarding the involvement of CNOT1 in mRNA decay during the M-G1 phase transition, at several locations in the manuscript.

Introduction: page 4, line 33

“For several of these genes, we show that mRNA decay requires CNOT1” changed to “For several of these genes, we show that mRNA decay is stimulated by CNOT1”.

Main text, page 12, lines 3 to 4

“These results suggest that CNOT1-dependent deadenylation is important for mRNA decay at the M-G1 phase transition” changed to “These results suggest that CNOT1-dependent deadenylation is involved in mRNA decay at the M-G1 phase transition”.

Main text, page 12, line 14 to 15

“Collectively, these data show that CNOT1 is important for the decay of TOP2A and CDK1 mRNAs during the M-G1 phase transition […]” changed to ““Collectively, these data show that CNOT1 aids the decay of TOP2A and CDK1 mRNAs during the M-G1 phase transition […]”.

In addition, we have toned down our conclusions regarding the biological relevance of RNA decay after mitosis. We have done this at the following locations in the text:

Introduction: Page 4, line 36 to page 5, line 2

“Together, our findings demonstrate that, analogous to protein degradation, mRNA degradation occurs at the M-G1 phase transition, and provides an important contribution to the reset of the transcriptome after cell division” changed to “Together, our findings demonstrate that, analogous to protein degradation, scheduled mRNA degradation occurs at the M-G1 transition. Scheduled mRNA degradation likely provides an important contribution to the reset of the transcriptome after cell division”.

Discussion: Page 14, lines 10-11

Therefore, decay-mediated clearance of mRNAs as cells exit mitosis will aid to limit expression of the encoded proteins in G1 phase” changed to “Therefore, decay-mediated clearance of mRNAs as cells exit mitosis could aid to limit expression of the encoded proteins in G1 phase”.

Discussion: Page 14, lines 13-16

“As these genes include many genes that encode for proteins with important functions in cell cycle control, their continued expression in G1 phase may perturb normal cell cycle progression…” changed to “As these genes include many genes that encode for proteins with important functions in cell cycle control, we speculate that their continued expression in G1 phase may perturb normal cell cycle progression…”.

3) Modify the Discussion section in order to consider the possibility that other mRNA decay factors may act in parallel to CNOT1 or in concert with it. This is particularly convenient if we considered the information supplied by the authors in response to Reviewer 1, describing the results of experiments addressing the possible involvement of DCP2 in post-mitotic mRNA decay (Figure 1A,B of the rebuttal letter).

In the Discussion section we now mention several other potential mRNA decay mechanisms that may contribute to cell cycle dependent RNA decay. On page 15, lines 5 to 8, we write:

“Nonetheless, it is possible that additional mechanisms contribute to mRNA decay during the M-G1 phase transition. Such mechanisms could involve PARN-dependent deadenylation, or may be independent of mitotic deadenylation and instead rely on mRNA decapping or endonucleolytic cleavage of the mRNA.”

In conclusion, I consider that this manuscript, with the new experiments and changes proposed by the authors, is a significant contribution on how mRNA stability regulation takes place accross the cell cycle, particularly after mitosis. This results should be of interest for a broad audience in the molecular and cell biology community, which would appreciate a paper like this in eLife.

Reviewer #4:

[…] The authors claim that their method can accurately measure the cell cycle stage and the "absolute" time an individual cell spent in the cell cycle with a resolution in the scale of minutes (p. 5). I have few concerns here, that affect also the strength of the biological conclusions. Here my specific points:

a) The time resolution estimate is based on an average cell cycle length determined by fluorescence microscopy (Figure 1A). The cell-to-cell variability in cell cycle length should be computed and used to estimate the error of the time resolution.

The reviewer is right that cell-to-cell heterogeneity is important to consider in calculating an accurate cell cycle time. We note though that our cell cycle time values are not based on (average) cell cycle length, but rather on the FUCCI-G1 fluorescence intensities, which are dependent on the time since the previous cell division, rather than on the overall cell cycle duration. The FUCCI-G1 fluorescence intensity continues to increase the longer a cell spends in G1 phase, at least for the first 7 hours of G1 phase, and reports directly on the time a cell has spent in G1 phase, irrespective of that cell’s overall cell cycle duration. Times beyond 7 hours spent in G1 phase are more challenging, as some cells will enter S phase after 7 hrs of G1 phase, at which time the FUCCI-G1 reporter no longer accurately reports on the time in G1 phase. However, our analysis was limited to the first 7 hours of G1 phase, thus circumventing this issue.

b) The authors combined microscopy and FACS to calibrate the fluorescence measured with the two methods. In particular they base their calibration on the identification of cells in early S phase to normalize the two fluorescence datasets. Thus, accurate identification of early S phase cells seems very important to compare the fluorescence detected with the two methods. It is unclear from the text and methods section whether they used also independent approaches to determine the cell cycle phase. For instance in the FUCCI paper (Sakaue-Sawano et al. 2008 PMID: 18267078) Hoechst staining is used to independently determine the cell cycle. Furthermore, by microscopy, how did the authors determine when cells were undergoing early S-phase? Did they use FUCCI alone? To clarify this point, they could add a fluorescence microscopy video where they indicate how they classify cell cycle stages. This will be very important to ensure reproducibility of the method.

Per reviewers’ suggestion, and to improve reproducibility of our method, we have now included a video (Video 1), as well as still images of RPE-FUCCI cells to indicate the cell cycle classification used in this study (new Figure 1B). These have been added as figure references in the section describing the FUCCI system in the main text (page 5, lines 13-18).

Also, we have now included FACS-based analysis of DNA content of RPE-FUCCI. This analysis confirmed the gating strategy we used to identify G1 phase, early S, rest of S and G2/M phase cells (new Figure 1—figure supplement 1A) and is incorporated in our introduction of the FUCCI system in the main text (page 5, lines 15 to 30).

Finally, a detailed description for the criteria used to identify early S phase cells based on microscopy data can be found in the Methods section entitled “Cell cycle timing using the FUCCI system”, page 32, lines 9 to 23.

c) The FUCCI reporter expression heterogeneity is not clearly discussed in the text and it is unclear whether it may affect the accuracy of the method. The authors should provide a sample microscopy video (same as in (b) to show how heterogeneous (or not) is the expression of the reporter across many synchronized cells.

We agree with the reviewer that this in an important point. We have now included additional analysis of the heterogeneity of FUCCI-G1 fluorescence and have determined the corresponding error in the G1 timing of single cells (see new Figure 1—figure supplement 1C), which we discuss in the main text (page 5 lines 28 to 31) and in the discussion (page 13, lines 21 to 24). In addition, we have added a video of many cells expressing the FUCCI reporters (Video 1), as suggested by the reviewer.

d) The biological discovery of this paper is that mRNA levels decline in multiple waves during mitotic exit. I find the text at times confusing. For instance at page 7, the paragraph title "mRNA levels decline in multiple waves after cell division", suggests that both degradation waves occur after completion of cell division. But this is not what is shown in Figure 2D. The "immediate" degraded mRNAs are degraded before the end of mitosis, according to the FISH quantifications. I highly recommend to clarify the text.

The reviewer is correct that this description is correct. We apologize for this inaccuracy and we have modified the title of our manuscript and the title of the relevant paragraph to more accurately reflect the data presented.

Old title: “Time-resolved single-cell sequencing identifies multiple waves of mRNA decay during mitotic exit” changed to “Time-resolved single-cell sequencing identifies multiple waves of mRNA decay during the mitosis-to-G1 phase transition”.

Paragraph title (page7): “mRNA levels decline in multiple waves after cell division” changed to “mRNA levels decline in multiple waves during the M-G1 transition”.

e) The method presented in this manuscript provides the ability to precisely estimate the duration of the different cell cycle phases of HEK293 cells. One result that I would recommend adding is the duration of the different cell cycle phase of an average HEK293 cell, based on your FUCCI reporter estimates. This will be important to interpret many of your plots where the X axis is presented as "Time after metaphase". I make this point because I think that some of the mRNA levels declines reported in your study may be explained by a dilution effect occurring upon cell division. I think that this is particularly important for the "delayed" mRNA pool. Could the authors please comment this point in the text? How did they consider dividing cells while performing single cell imaging?

We thank the reviewer for this suggestion. We have now included analysis of G1 phase length, based on FUCCI-G1 fluorescence, in both RPE-FUCCI wild-type as well as RPE-FUCCI p53-knock-out cells (Figure 2—figure supplement 1K). These data show that the average G1 phase length is ~12 hours, and the shortest G1-phase durations are ˜6-7 hours. The entire cell cycle duration in RPE cells is ~24 hrs. So within the time-window of our analysis (~7 hours post metaphase), almost no cells with have entered the next S-phase and no cells will have passed through another mitosis. Therefore, the reported mRNA decline is not affected by the next cell division. Additionally, it is important to note that using our FACS-based isolation of cells, we only include cells that are still in G1 phase (see sorting strategy in Figure 1C), ensuring that progression into the next cell cycle phase(s) (e.g. S-phase) does not complicate our analysis.

Regarding the dilution effect occurring upon cell division (i.e. mRNAs are divided over the two daughter cells), specifically when comparing G2/M sequenced cells to early G1 phase cells; in this case there is indeed a “dilution effect”, which we have corrected in all our experimental approaches. In our qPCR-based analysis we always normalize gene expression to two independent control genes which were previously reported to have long half-lives (Ribophorin and GAPDH), using the ∆∆Ct-method. Since these control mRNAs will also be diluted as a consequence of cell division, normalization of gene expression to these two control genes will compensate for any dilution of mRNAs caused by division. As such, we believe that the reported decrease in mRNA expression is due to turn over rather than due to dilution. In the sc-RNA sequencing analysis, we have normalized all cells for the number of UMIs retrieved (using an algorithm from Monocle2). In this way we compensate for the dilution effect and we only look at the effect of mRNA decay. We have updated the methods section to more clearly explain this important point (page 29, line 32 to page 30 line 2, and page 31, lines 12 to 13).

Finally, for our FISH data (Figures 2E, 2F and 4A) where we imaged mRNA content in single cells, we always included both daughter cells for our analysis of mRNA content, as this prevents a dilution effect that is caused by cell division itself. With regards to FUCCI-G1 fluorescence quantification, we analyzed FUCCI-G1 fluorescence levels in both daughter cells individually, as this is also how FUCCI-G1 fluorescence will be analyzed using FACS.

f) FISH is used as in independent mean to measure mRNA reduction in single cells at different stages of the cell cycle. For these experiments, two mRNAs belonging to the "immediate" degraded group are measured, TOP2A and CDK1 and both nicely show a decrease before the end of the cell cycle. Could you add the single cell data to the FISH quantifications shown in figures 2E, 2F, and 4A, to report the cell-to-cell variability observed for the expression of these mRNA?

We have now included the single cell data of a representative experiment corresponding to the data shown in figures 2E and 2F (new Figure 2—figure supplement 1B-C), and corresponding to figure 4A (new Figure 4—figure supplement 1C-D).

g) As a follow-up question, why didn't you perform FISH also on two mRNAs from the "delayed decrease" group, to show changes in mRNA expression? This seems an important control if the aim is to demonstrated trough impendent methods that two waves of degradation occur during the M/G1 phase transition.

We did not include any genes belonging to the delayed decay group in our smFISH experiments as we had already validated the existence of two independent waves of mRNA decay using RT-qPCR (Figure 2C.) (and due to the high price of smFISH probes). The smFISH experiment was performed to more precisely determine the moment of mRNA decay for genes belonging to the immediate decay group. We have now included additional experiments investigating the mRNA levels of 5 immediate decay genes and 5 delayed decay genes in cells FACS-sorted in G2, early and late mitosis. These results confirm that immediate decay genes start to decline in mitosis, while delayed decay genes remain stable during mitosis (new Figure 2—figure supplement 1E-H).

h) The authors conclude that "mRNA degradation occurs at the M/G1 phase transition, and provides an important contribution to the reset of the transcriptome after cell division". However, no experiment is performed to determine the importance of this decay mechanism and its consequences on cell cycle. For instance it could be tested by selectively downregulating the expression of CNOT during the M/G1 phase transition by using an inducible degron system. I understand that this may be a complicated experiment to do, but if a demonstration is not provided, then the text should remain more conservative in describing the results.

As suggested, we have toned down our conclusions regarding the biological relevance of RNA decay after mitosis. We have done this at the following locations in the text:

Introduction: Page 4, line 36 to page 5, line 2

“Together, our findings demonstrate that, analogous to protein degradation, mRNA degradation occurs at the M-G1 phase transition, and provides an important contribution to the reset of the transcriptome after cell division” changed to “Together, our findings demonstrate that, analogous to protein degradation, scheduled mRNA degradation occurs at the M-G1 transition. Scheduled mRNA degradation likely provides an important contribution to the reset of the transcriptome after cell division”.

Discussion: Page 14, lines 10-11

“Therefore, decay-mediated clearance of mRNAs as cells exit mitosis will aid to limit expression of the encoded proteins in G1 phase” changed to “Therefore, decay-mediated clearance of mRNAs as cells exit mitosis could aid to limit expression of the encoded proteins in G1 phase”.

Discussion: Page 14, lines 13-16

“As these genes include many genes that encode for proteins with important functions in cell cycle control, their continued expression in G1 phase may perturb normal cell cycle progression…” changed to “As these genes include many genes that encode for proteins with important functions in cell cycle control, we speculate that their continued expression in G1 phase may perturb normal cell cycle progression…”.

We do note that we only state that mRNA decay may affect the transcriptome (i.e. mRNA levels), not that mRNA decay has functional consequences for the cell cycle.

i) The authors conclude that "mRNA decay requires CNOT1". The CNOT1 downregulation experiments show that CNOT1 contributes to decay, not that it is required. This should be changed in the text.

As suggested by the reviewer we have weakened our statements regarding the involvement of CNOT1 in mRNA decay during the M-G1 phase transition, at several locations in the manuscript.

Introduction: page 4, line 33

“For several of these genes, we show that mRNA decay requires CNOT1” changed to “For several of these genes, we show that mRNA decay is stimulated by CNOT1”.

Main text, page 12, lines 3 to 4

“These results suggest that CNOT1-dependent deadenylation is important for mRNA decay at the M-G1 phase transition” changed to “These results suggest that CNOT1-dependent deadenylation is involved in mRNA decay at the M-G1 phase transition”.

Main text, page 12, line 14 to 15

“Collectively, these data show that CNOT1 is important for the decay of TOP2A and CDK1 mRNAs during the M-G1 phase transition […]” changed to ““Collectively, these data show that CNOT1 aids the decay of TOP2A and CDK1 mRNAs during the M-G1 phase transition […]”.

[Editors' note: we include below the reviews that the authors received from Review Commons, along with the authors’ responses.]

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

In the manuscript "Time-resolved single-cell sequencing identifies multiple waves of mRNA decay during mitotic exit", Krenning et al. describe a method that connects live-cell microscopy with single-cell RNA sequencing in order to monitor global changes in mammalian mRNA gene expression in a cell-cycle-dependent manner. To this end they employ a fluorescent, ubiquitination-based cell cycle indicator (FUCCI system) in human untransformed RPE-1 cells coupled to SORT-Seq in order to generate a high-resolution, time-resolved transcriptome profile of cells spanning the transition form M phase into G1 phase. By comparing FACS-based sampling with an in silico trajectory inference method the authors provide convincing evidence that this system allows for high-resolution, time-resolved transcriptome profiling of cells in the transition from M-phase into G1 phase. Subsequent analysis of changes in steady-state gene expression revealed a set of >200 transcripts that undergo rapid decay in two distinct waves, first around the time of mitotic exit and second upon G1 entry. Those results were independently validated by single molecule FISH of select transcripts, revealing that the first wave of RNA decay initiates at the start of anaphase and the second wave starts during early G1 phase. Using mathematical modeling followed by selective validation (using cell-cycle staging combined with global inhibition of RNA decay), the authors derive and confirm precise mRNA decay rates for cell cycle regulated mRNAs that were overall lower than what has been observed in previously described population-based half-life measurements. Finally, the authors establish a role of deadenylation by the CCR4-NOT complex in selective mRNA decay during M-phase/G1 transition by revealing a partial stabilization upon NOT1-depletion by RNAi.

I have only a minor comment to help improve the mechanistic insights: While the authors attribute the partial stabilization of cell-cycle regulated mRNAs to an incomplete depletion of NOT1 (which is possible), an alternative hypothesis would be that decapping and 5´-to-3´ decay further contribute to mRNA turnover. Notably, the authors describe the use of siRNAs targeting Dcp2 in the methods section (without referring to it in the results part), indicating that they may have data that could clarify this. They authors may consider adding this data to complete an otherwise nicely executed and well-written manuscript.

We agree with the reviewer that decapping and 5’-to-3’ decay could contribute to mRNA turnover, either following initial CNOT1-dependent deadenylation or in the absence of deadenylation. Indeed, we had perform studies addressing the involvement of DCP2 in the decay of CDK1 and TOP2a mRNAs during mitosis, but we decided to omit these results, as they are largely inconclusive. We found a small but significant reduction in decay of TOP2A, and a small, but not significant reduction in the decay of CDK1. As we depleted DCP2 by siRNA, it is difficult to interpret these results, as the lack of effect could be explained by insufficient knockdown. We have included these experiments as Author response image 1A,B.

Author response image 1. Figure 1.

Author response image 1.

(A) Cells were transfected with the indicated siRNAs and fixed 48 hours later. mRNAs, DNA and membranes were labeled with smFISH probes, DAPI and fluorescent wheat germ agglutinin, respectively. mRNA numbers of cells in late mitosis were counted and divided by the number of mRNAs present in cells in early mitosis. (B) Cells were transfected with the indicated siRNAs. 48 hours later cells were lysed and the RNA was harvested. Gene-expression levels were analyzed by qPCR, relative to cells treated with siRNAs targeting Luciferase.

Due to the inconclusive nature of these experiments, we propose to omit these data from the manuscript. Instead, we will include additional discussion on the possible involvement of other decay factors, either acting in parallel to CNOT1 or in concert with CNOT1, in the Discussion section. We will make the following changes:

– Remove the DCP2 data from Figure S4A.

– Change the figure legends for Figure S4A accordingly.

– Remove the siRNA from the materials section.

– Remove the DCP2 data in Supplemental Table 2.

– Include discussion on the possible involvement of other decay factors.

Reviewer #1 (Significance (Required)):

Overall, this is an impressive and well-controlled body of work providing novel insights into cell-cycle controlled mRNA decay. It reveals novel insights into the targets and molecular mechanisms of mRNA decay in the transition of mitosis to G1, thereby adding significantly to our understanding of the scope of tightly controlled gene expression for proper execution of the cell cycle.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

Summary:

In this study Krenning, Sonneveld, and Tanenbaum have investigated the temporal control of mRNA decay after mitosis. Previous work has demonstrated that following mitosis, regulated protein degradation serves an important role in erasing any lingering memory of the previous cell cycle. Left unanswered however, is whether mRNA decay is similarly regulated after mitosis, and if so, what role does it play in the cell cycle. The authors use time-lapse imaging and single-cell RNAseq to measure decay rates of mRNAs after mitosis. They use the FUCCI sensors to identify the precise age of each individual cell prior to performing single-cell RNAseq. Using this pseudo-timelapse approach, they identify several genes that are actively degraded after mitosis. Interestingly, they find that some mRNAs are decayed rapidly and immediately after mitosis while other mRNAs are decayed only after a delay of about 1.5 hours after mitosis. The authors find that, at least for several of the genes, the protein CNOT1 mediates the decay. The authors conclude that regulated mRNA decay occurs after mitosis and helps reset the transcriptome at the onset of a new cell cycle.

The question being asked in this manuscript is interesting and potentially very important to the cell cycle field. However, I believe this study suffers from several technical issues highlighted below as well as relatively modest effect sizes, particularly when it comes to ascribing CNOT1 as the key regulator mediating mRNA decay upon mitotic exit.

Major Comments:

1) The authors repeatedly claim that they can accurately measure the time a cell has spent in G1 phase, due to using live-cell imaging to calibrate their FACS data. However, following mitosis, cells bifurcate into either G0/Quiescence or into G1 phase. Importantly, the FUCCI-G1 sensor is not capable of distinguishing between these two cell cycle phases. Thus, while two cells may have spent the same amount of time since mitosis, one cell may be in Quiescence while the other cell is in G1 phase and they would both have the same level of FUCCI-G1. Even by live-cell imaging it’s not possible to distinguish between these two cell cycle phases with the markers used in this study. I believe you can see evidence for this in the Flow cytometry data in Figure 1C where there is a large population of cells with very high FUCCI-G1 levels (even higher than most of the S phase cells) but very low FUCCI-G2 levels. These cells likely represent cells that entered G0 after mitosis, remained there for several hours, and thus accumulated very high levels of the FUCCI-G1 sensor. This technical limitation has several implications. First, it means what is likely going into the single-cell mRNA seq workflow is a mixture of G0 and G1 cells. Since the authors observe two distinct "waves" of mRNA decay, it makes me wonder if one wave might be occurring in G1 cells while the other wave is occurring in G0 cells?

Second, the authors state several times they can accurately determine the time a cell has spent in G1 phase. The more accurate statement however is that the authors can accurately determine the time that has elapsed since mitosis, since they cannot distinguish between G0 and G1 phase. This is admittedly a nuanced distinction but an important one given that quiescence cells are in a very different cellular state than cells in G1 phase. The authors actually use the correct x-axis label of "Time after metaphase (min)" in several of their figures, but they do not use the same language in the main text or to describe their conclusions. To overcome the technical challenge of distinguishing between G0/G1 cells and to address these two points, the authors could use an additional FUCCI sensor, mVenus-p27K(-), which accumulates in G0 cells but not in G1 cells (PMID: 24500246).

We thank the reviewer for pointing out the G0/G1 bifurcation that occurs after metaphase, and agree that based on the FUCCI-G1 reporter we cannot distinguish between cells entering G0 or G1 phase. The reviewer is also correct that, in principle, it is possible that one wave of mRNA decay occurs specifically in cells entering G1 phase, while the other occurs in cells entering G0 phase. To address this point we have performed additional analyses and we propose to perform an additional experiment as well. In addition, we will make textual changes to the manuscript, addressing the G0/G1 bifurcation.

As the reviewer suggests, it could be possible that one wave of mRNA decay occurs in G1 cells, while the other wave occurs in G0 cells. We have performed new analysis comparing genes that are differentially expressed (DE) between G0 and G1 phases (genes were selected based on the study mentioned by the reviewer, PMID 24500246) to genes that are decayed following cell division. We found that many genes that are subject to post-mitotic decay are also differentially expressed between G0/G1 phases. Of the genes belonging to the ‘immediate decay’ group, 57% is also differentially expressed between G1 and G0. Of the genes belonging to the ‘delayed decay’ group, 73% is differentially expressed between G1 and G0. This implies that genes that are differentially expressed between G0 and G1 phases can be decayed during both waves of mRNA decay. There results argue against one wave of mRNA decay occurring in G1 cells and the other in G0 cells.

Additionally, if the two decay waves would occur in two distinct sub-populations of cells (i.e. G1 and G0 cells), it is expected that cells that show the strongest decay in one set of genes (e.g. immediate decay genes) would show no, or substantially less, decay of the other set of genes (e.g. delayed decay genes). In other words, the levels of immediate and delayed decay genes should anti-correlate. In contrast, if both waves of mRNA decay occur sequentially in the same cells, irrespective of whether cells are/become G1 or G0 cells, such anti-correlation in the levels of immediate and delayed decay genes is not expected. We have now performed this analysis and found that cells with low levels of immediate decay genes, also tend to have low levels of delayed decay genes (Author response image 2) . The expected anti-correlation is not observed, rather our results show that both waves of mRNA decay occur in the same cells, indicating that these two waves of decay do not reflect decay events that occur selectively either in G0 and G1 phase.

Author response image 2. Comparison of immediate decay versus delayed decay in single cells.

Author response image 2.

We generated two metagenes, one including all genes belonging to the ‘immediate decay’ wave, and one including all genes belonging to the ‘delayed decay’ wave. Metagenes were created by summing up all reads for the genes that belonging to the same wave of decay in a single cell, thus creating one expression value per metagene per cell. Then we analyzed metagene expression at 240 minutes post metaphase normalized to its expression at G2/M.

While we found a good correlation in the decay of genes belonging to both decay waves, we wanted to further investigate if we could detect evidence for a bifurcation of G1 and G0 cells on the whole transcriptome level. For this, we looked back at our Monocle2 analysis, which generated a transcriptome-based trajectory. In case G1 and G0 cells would differ on the transcriptome level, we would expect to see a bifurcation in the trajectory at some point during G1 phase. We did not detect a bifurcation in cells during the first 4 hours after cell division (Figure 1-figure supplement 1M). These observations show that, even though the cells are a mix of cells in G1 and G0, or a mix of cells that will enter G0 or G1, the transcriptomes of all cells (and thus the mRNA decay) does not yet show signs of the G0-G1 bifurcation during the first 4 hours following cell division (the time-window used to identify the two waves of mRNA decay).

Lastly, to further explore this point experimentally, we propose the following experiment; we will generate p53 knockout cells, as the quiescence/proliferation decision is dependent on p53 and its transcriptional target p21 (Spencer et al., 2013; Yang et al., 2017). We will perform new qPCR-based analysis, directly comparing the mRNA decay of populations of wild-type cells (containing both G1 and G0 cells) to p53 knockout cells (containing only G1 cells). This will experimentally validate whether the lack of G0 cells prevents the mRNA decay for either group of genes.

2) One of the main novelty claims of the paper is that mRNA decays following mitotic exit in two waves. In order to support this claim, the authors plot the time since metaphase vs normalized transcripts for many single cells. For several genes including CDK1, TOP2A, and UBE2C, there is an immediate drop from time 0 to the next most densely populated timepoint of about 20 minutes. This results in a bimodal histogram as seen in Figure 2A, where there is one single bin representing genes that are maximally decayed at time 0, while another population is more normally distributed with a median decay time of ~80minutes. I have two concerns about this data. First, there are very few if not zero cells analyzed that were between 0 and 20 minutes old at the time of collection. Therefore the authors do not know if the mRNA decayed immediately after mitosis or with a 20minute delay after mitosis. Thus, rather than some genes decaying immediately after mitotic exit resulting in that one bin at time=0 and some genes decaying with a delay, there could actually be just one wave of mRNA decay that is broadly and normally distributed from 20-80minutes after mitosis (see histogram Figure 2 and Figure S3G; if you ignore anything less than 20minutes due to lack of data during this time window, then there is just a single distribution). While likely technically challenging due to cytokinesis, sampling cells that are between 0-20minutes old may be important to accurately measuring the decay rates of mRNAs upon mitotic exit.

We apologize for the confusion caused by this data. While it is true that only few data points exist between 0 and 20 min post-metaphase, the data nonetheless shows that the first wave of decay initiates before the start of G1. We will modify the text to explain this point more clearly (see below for a detailed explanation) and we will perform an additional experiment to further strengthen this point.

To determine the moment of initiation of decay, we performed a data fitting approach in which we fit the mRNA levels over time of each gene with an exponential decay distribution with a variable delay time (i.e. an initial plateau phase followed by the exponential decay). We then searched for the delay time (in intervals of 10 minutes) and the decay rate that best fit the data. This analysis revealed that for many genes a delay time of 0 min (relative to metaphase) generated the best fit. It is perhaps counterintuitive that a delay time of 0 minutes could create the best fit for a dataset that includes very little data points in the first 20 min of the analysis. Yet, this is possible because one can fit the later timepoints of an exponential decay distribution and based on the fit of the later datapoints extrapolate the earlier datapoints. This approach allowed us to model the decay for each gene, determining the best starting point of the exponential decay distribution. The modeling approach could have determined decay to initiate at 0, 10, 20, 30, 40 (etc) minutes post metaphase as the best fit. However, for 75% of the ‘immediate decay genes’ 0 minutes post metaphase was identified as the best fit, indicating that decay of the immediate decay genes is initiated quite synchronously during mitotic exit.

Importantly, we experimentally validated that decay initiates during mitotic exit for two genes that were predicted to initiate decay at 0 minutes post metaphase, CDK1 and TOP2A, using smFISH (see manuscript Figure 2D-F). These data show that declining mRNA levels of CDK1 and TOP2a are detectable as early as anaphase, which independently validates (the end of) mitosis as the moment of mRNA decay initiation.

To further corroborate these findings, we propose to include smFISH data for an additional gene that is predicted to initiate decay at 0 minutes post metaphase FBXO5. In addition, we propose to separate cells in telophase from prometaphase/metaphase cells and also from cells in G2 (based on mitotic shake-off and FACS sorting) and assess the mRNA levels of 5 immediate decay genes using qPCR in these different phases. These experiments will hopefully confirm that decay indeed already initiates in anaphase/telophase for additional genes.

Second, it’s not entirely clear from the methods section, but I am assuming that for time 0 of the plots in Figure S3A-F, the authors are using G2/M sorted cells as defined by FUCCI-G2 high/FUCCI-G1 low status. However, depending on what precise FACS gates were used to do this sort, these cells could be anywhere in early G2, late G2, prometaphase, metaphase, etc and they are all averaged together. Thus, when the authors claim that the mRNAs are decaying immediately after mitosis, that's based on comparing the mRNA levels in this mixed G2/M population with cells that are 20minutes or more after mitosis. For genes like CDK1 where the mRNA is almost completely gone at 20 minutes after mitosis when they first have any cells to measure, its entirely possible these mRNAs were already decayed either before or within mitosis, rather than upon mitotic exit. What we really need to know is what are the mRNA levels precisely at metaphase and plot that value as time=0 (for example in figure S3A-F). Perhaps the authors could use CDK1i plus Taxol treatment to accumulate cells in mitosis like they describe in the methods for the transcription inhibition experiments. (If this is what was done for the Sort-Seq experiments as well as the data in Figure S3A-F, my apologies for not completely understanding, but the methods section should be updated to make this more clear)

The reviewer is correct that the cells represented at t=0 are a mix of G2 cells and early mitotic cells. While the concern that mRNA decay already initiates during G2 or early mitosis is understandable, there are several experiments that demonstrate that mRNA decay does not initiate until post-metaphase.

1. Our previous mRNA sequencing experiments has directly compared mRNA levels in G2 vs early mitosis, which revealed that there are little to no changes in mRNA expression between G2 and mitosis for the genes that undergo mRNA decay post-metaphase (Tanenbaum et al., eLife, 2015).

2. Using smFISH, we could show that the levels of two genes identified in this study, CDK1 and TOP2a, are stable during mitosis, until cells enter anaphase.

To further support the conclusion that mRNA decay initiates post-metaphase, and mRNA levels between G2 and early mitosis remain largely unchanged we will perform the experiment proposed above (separating G2 cells from early mitotic cells and telophase cells) and examine mRNA levels at these different time-points. This will allow us to experimentally determine the putative changes in gene expression between G2, early mitosis and post metaphase.

3) As the authors point out in the discussion, the effect size of CNOT1 depletion on mRNA decay are rather small. Notably the genes that appear to be statistically significant are all the genes that are hardly decayed after mitosis anyways (see Figure 4D, ARLGIP1, PSD3, etc). These genes appear to be only down about 2-fold (-1 on a log2 scale axis) in the control group, which would be expected during mitosis when the mother cell splits in half into two daughter cells. Depending on the method of normalization, this change in mRNA levels may simply be due to this. I understand the technical limitations in knocking down an essential gene, but given that one of the major novelty claims of this study is that CNOT1 mediates the mRNA decay upon mitotic exit, more robust proof is warranted. The authors could employ the inducible protein degradation system they referenced in the discussion. Additionally they could perhaps show data that CNOT1 is somehow differentially regulated during mitotic exit that would account for this sudden change in mRNA stability. If the authors were to include additional corroboratory data demonstrating CNOT1 is most active during mitotic exit for example, it would help to overcome that low effect size upon mild CNOT1 depletion.

The statement that depletion of CNOT1 only significantly prevents mRNA decay for genes that are hardly decayed after mitosis is not accurate. CNOT1 knockdown significantly inhibited the decay of CDK1 and TOP2a mRNAs (manuscript Figure 4A), and of UBE2C and FZR1 (manuscript Figure 4D). These are all genes that show a very strong mRNA post metaphase. In fact, for all the genes tested, CNOT1 depletion reduced post-metaphase decay for 9 out of 10 genes.

The reviewer correctly points out that, depending on the normalization method used for qPCRs, one may expect a 2-fold reduction in the absolute mRNA levels due to cell division itself. In our experiments, we compare gene expression of decayed genes to two independent control genes that are not degraded post mitosis. This comparison will correct for any effects on mRNA levels caused by cell division. We will update the methods section to clarify this important point.

Finally, the reviewer suggests several new lines of experimentation to further study the involvement of CNOT1 in mRNA decay during mitotic exit. While we agree that using degron-tagging of the endogenous CNOT1 locus and developing methods for activity profiling of CNOT1 would be very interesting and exciting, these approaches represent a significant investment in time and we feel that this work goes beyond the scope of the current study.

Minor comments:

1) The authors state that they can "accurately determine the time a cell has spent in G1 phase based on its FUCCI-G1 fluorescence as measured by FACS" (Page 6, last line of the first paragraph). It would be nice to know how accurately? What is the level of uncertainty in your measurements? +/- how many minutes?

The reviewer raises an important point. The inference of timing of single cells is indeed subject to variability in FUCCI-G1 fluorescence. We will measure the experimental error in this inference and include it in the manuscript. It is important to note that mRNA decay rates are calculated based on many cells, averaging out any measurement error.

2) The authors only focused on those genes whose mRNA’s were decayed upon mitotic exit, but of equal interest would be those cell cycle genes that were not observed to decay upon mitotic exit. The authors may want to highlight these genes, because it might shed light on what aspects of the cell cycle cells want to reset, and what aspects of the cell cycle the cells may wish to “remember” from the previous cell cycle. For example, what is the status of the cyclins? CIPs? Origin licensing genes? There is growing evidence that the status of the mother cell can influence the cell “ycle of ”he daughter cell and mRNA levels of key cell cycle genes have been implicated in this (PMIDs: 28514656, 28317845, 28869970).

This is an interesting point. We have provided an easily-searchable excel sheet with an overview of all genes and their expression levels over time during the M-G1 transition (Supplementary file 1), so that anyone can examine the expression of their favorite gene.

Reviewer #2 (Significance (Required)):

This paper offers a potential conceptual advance to our understanding of how cells reset after the cell cycle is completed. By demonstrating that cells regulate protein as well as mRNA degradation upon cell cycle exit, the authors show cells utilize multiple mechanisms to reset the biochemical state of the cell at the onset of a fresh cell cycle. This work would be of potential interest to the cell cycle field, particularly those interested in G1 regulation. It would also be of interest to people interested in mRNA regulation as it would demonstrate a clear window during which mRNA decay was specifically upregulated.

While my expertise lies in the cell cycle and the use of FUCCI sensors to identify cell cycle stages, I have less expertise in single-cell RNAseq and methods of mRNA quantification.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

In this paper, Dr. Tanenbaum and colleagues present a new method to determine the exact point of the cell cycle in which a cell is. This new method has the potential to be applied to any cell-cycle phase and even to other processes. Focusing on M-G1 transition, they do sequencing of single human cells. They identify two groups of cell cycle-related genes, expressing mRNAs which are degraded either immediately after exit from mitosis or later on during G1. One of the factors involved in this degradation is identified as CNOT1.

In my opinion, the new method is well stablished and has the potential to be very useful to future work. In general, I think that the main conclusions are based on more than one approaches and are convincing. In relation to these, I have two major concerns:

1. Although it is clear that a scheduled mRNA decay exits, this does not exclude the possibility of a concomitant effect on mRNA synthesis. A measurement of nascent transcription is needed.

The reviewer raises an important point, and we believe that concomitant transcription shut down may indeed play a role. Therefore, we propose to include an analysis addressing the relative transcription levels of ‘immediate decay’ and ‘delayed decay’ genes.

2. As mentioned in the discussion, it is possible that the limited effect of depleting CNOT1 is due to the partial knockdown. However, it is also possible that a different pathway of mRNA degradation is involved. This should be addressed by targeting other decay factors (for example Xrn1 and/or an exosome component).

We agree with the reviewer, and we had attempted to test the involvement of other mRNA decay factors, including the decapping factor DCP2 in post-mitotic mRNA decay. We decided to omit these results, as they are largely inconclusive unfortunately. We found a small but significant reduction in decay of TOP2A, but no significant reduction in the decay of CDK1. As we depleted DCP2 by siRNA, it is difficult to interpret these results, as the lack of effect could be explained by insufficient knockdown. We have included these experiments here (Figure 1A,B, see response to reviewer 1). As these data are inconclusive, we propose to leave them out of the manuscript. Instead, we will include additional discussion on the possible involvement of other decay factors, either acting in parallel to CNOT1 or in concert with CNOT1, in the Discussion section.

Minor comments:

1. Cell cycle control is not absolutely universal. The authors should mention that their results and conclusions correspond to human cells; if not in the title, at least in the abstract.

We will include this statement in the abstract.

2. Is SORT-seq (mentioned exclusively in the methods section) the same as scRNA-seq?

SORT-seq is a combination of FACS-sorting and scRNA-seq. We will explain this more carefully in the results and methods sections of the manuscript.

3. Figure 2: there is no correspondence between the TOP2A images and their quantification. This experiment also needs an unrelated mRNA FISH as a negative control.

We apologize for this mistake and will update the figure with a more representative image.

4. Figure 3D: if I correctly understood this experiment, “Time in Actinomycin D” is a better title for the X axis. “Time after mitotic shake-off” is misleading because it suggests that the cells were released from the mitotic blockage.

We thank the reviewer for pointing this out. We will update the figure accordingly.

5. Figure S3L: the blockage of transcription by Actinomicin D should be demonstrated by a more general method, such as EU (5-ethynyl uridine) incorporation.

Actinoymcin D is a very well-established inhibitor of RNA polymerases. We show through expression of a control gene (CDKN1a) that the drug is active in our experiments, so we feel that additional analysis is not essential in this case.

6. Figure 4A: what is the cell cycle phenotype of CNOT1 siRNA? If a reduction with CRISPRi of 50 % causes a phenotype (Figure S4C), then a reduction of 90 % by the siRNA (Figure S4A) should cause a stronger phenotype.

We agree with the reviewer that a higher knockdown efficiency is expected to more profoundly affect cell cycle progression. We propose to include the analysis of cell cycle progression upon siRNA mediated CNOT1 depletion in the revised manuscript.

7. Do the two waves of protein degradation (mentioned in the second paragraph of page 12) affect the same genes than the two waves of mRNA degradation?

This is an interesting point that we will address. We will provide a supplemental table containing an analysis comparing molecular targets of the APC/C during mitotic exit with the genes classified as ‘immediate decay’ or ‘delayed decay’.

Reviewer #3 (Significance (Required)):

The new method described in this paper could be useful to address many questions in the field of cell-cycle control. Moreover, with modifications, it could be applied to any time-dependent biological problem. In addition, the scheduled degradation of mRNAs in the M-G1 transition is a discovery of biological significance. In my opinion, the paper is technically interesting to a broad range of investigators, and biologically relevant for those studying the cell cycle.

Keywords of my field of expertise: transcription, chromatin, gene expression.

References:

Spencer, S. L., Cappell, S. D., Tsai, F., Overton, K. W., Wang, C. L., and Meyer, T. (2013). The Proliferation-Quiescence Decision Is Controlled by a Bifurcation in CDK2 Activity at Mitotic Exit. Cell, 155(2), 369–383. https://doi.org/10.1016/j.cell.2013.08.062

Yang, H. W., Chung, M., Kudo, T., and Meyer, T. (2017). Competing memories of mitogen and p53 signalling control cell-cycle entry. Nature, 549(7672), 404–408. https://doi.org/10.1038/nature23880

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Supplementary file 1. Analyses used in this study.

    This file contains analyses that were used throughout the study, in Figure 1D–E, Figure 1—figure supplement 1J,L,N, Figures 2A–B3C and Figure 3—figure supplement 1B-G.

    elife-71356-supp1.xlsx (7.1MB, xlsx)
    Supplementary file 2. Sample size indication.

    This file contains sample sizes for all experiments that involved single-cell analyses.

    elife-71356-supp2.xlsx (16.5KB, xlsx)
    Supplementary file 3. Nucleotide sequences.

    This file contains nucleotide sequences for reagents that were used in this study; RT-qPCR primer sequences and sgRNA sequences.

    elife-71356-supp3.xlsx (10.8KB, xlsx)
    Source data 1. Single-cell transcript counts plate 1.
    elife-71356-data1.csv (13.6MB, csv)
    Source data 2. Single-cell transcript counts plate 2.
    elife-71356-data2.csv (13.5MB, csv)
    Source data 3. Single-cell transcript counts plate 3.
    elife-71356-data3.csv (13.2MB, csv)
    Source data 4. Single-cell sequencing metadata.
    elife-71356-data4.xlsx (41.5KB, xlsx)
    Transparent reporting form

    Data Availability Statement

    Source data containing single cell transcript counts that were used in this study are provided as supplementary data.


    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES