ERK is a key signaling mediator controlling both proliferation and lineage specification during embryo development. How ERK choreographs differentiation and proliferation to achieve balanced developmental outcomes in lineages with variable ERK activities remains unclear. To investigate this, we established multiplex quantitative live-cell imaging to track human pluripotent stem cell differentiation into mesendoderm (ME), a lineage specified by gastrulation morphogens and dependent on high ERK activity. We found that distinct morphogen combinations generate varying ERK activity levels, which correlate with heterogeneous ME fate choices despite relatively uniform cell cycle dynamics. To dissect how heterogenous ERK levels directly modulate and coordinate ME differentiation and proliferation, we engineered a synthetic spectrum of titrated ERK activities. Our results showed that ERK fine-tunes ME differentiation potential and cell division speed under nonoverlapping activity ranges, enabling quantitative control of ME fate specification without major effect on cell cycle progression. Mechanistically, this uncoupling stems from differential transcriptional and translational sensitivities of ME-specifying genes versus cell cycle genes to ERK input. Together, our findings reveal how a single signaling pathway quantitatively balances differentiation and proliferation during lineage commitment and embryogenesis.

Multiplex live-imaging documented an elevated ERK activity accompanying pluripotency to mesendoderm specification

Both pluripotent stem cells (PSCs) and ME cells rely on activated FGF/ERK pathway to promote the respective cell fates [33–35]. PSCs can be induced to differentiate towards ME fate in the presence of WNT, Activin A, BMP4, and FGF2 (Figs 1A and S1A, S1B). To quantitatively dissect the real time dynamics of ERK during pluripotency to ME differentiation, as well as its functional relevance to the “proliferation-differentiation” balance, we established a multiplex quantitative live-cell imaging platform to automatically track ME differentiation (Figs 1A and S1C), ERK dynamics and cell cycle progression using two stable human pluripotent stem cells (hPSC) lines constructed (Fig 1A and 1B). One of the lines contained ERK-KTR fluorescence sensor [36] and knocked-in copGFP at the C terminus of ME specifier gene TBXT [37,38] (Figs 1B and S1D, S1E), while the other line contained both ERK-KTR and CDK2-KTR sensors [39] (Figs 1B and S1F). The integration of these sensors did not affect the differentiation potential of PSCs (S1B, S1G, and S1H Fig).

Live-cell tracking revealed 12 hours after ME induction, the fluorescent signal of ME marker protein TBXT started to emerge (Fig 1B–1D, and S1 Video), followed by a linear increment. The TBXT fluorescent signal reached a peak level and plateaued around 30 hours after induction (Fig 1D). The maximum expression of TBXT fluorescent protein corresponded with an upregulation of ME genes and down-regulation of pluripotency genes, marking a successful transition into the ME state (S1B, S1G, and S1H Fig). Over the course of ME differentiation, ERK-KTR sensor activity exhibited a prominent elevation relative to that in PSCs (Fig 1C–1E). We quantified the ERK activity in accordance with the TBXT expression dynamics, and found the elevation in ERK activity preceded the emergence of TBXT-copGFP fluorescent signal, and persisted throughout the differentiation process (Fig 1F). Consistently, we observed an increase in ERK phosphorylation level in ME differentiating cells compared to that in PSCs (S1I Fig).

Corresponding with an increase in ERK kinase activity, live-imaging of CDK2-KTR sensor during ME differentiation documented a mild shortening of cell cycle compared to that in PSCs (Fig 1G–1I; S2 Video). PSCs were known to possess high CDK2 activity and exhibited abbreviated cell cycles, which typically elongate following lineage specification [14,40]. In contrary to this notion, during ME differentiation we observed a swifter rebounding and a faster accumulation rate of CDK2 activity after mitotic division comparing to that in PSCs (Figs 1H, 1I and S1J). It is noted that the acceleration of CDK2 accumulation took place gradually with each subsequent cycle (Figs 1H, 1I and S1J) and was reflected by a low correlation of CDK2 activities between mother and daughter cells (S1K Fig). Collectively, employing a live-imaging platform for quantitative evaluation of ERK activity, ME differentiation and cell cycle dynamics in real time, we uncovered that ME differentiation is characterized by a prominent elevation in ERK activity comparing to pluripotency state, and this is followed by an increase in TBXT protein expression and a mild shortening of cell cycle.

Morphogens specifying ME fates engender elevated but heterogeneous ERK activity levels

Gastrulation morphogens, including WNT, Nodal, BMP, and FGF, collaboratively drive ME specification within the primitive streak [33,41–46], and induce heterogeneous cell potentials through their spatial and temporal variations [47–50]. To dissect how different gastrulating morphogens shape ERK activity profiles during ME differentiation, we employed the live-imaging system to track the dynamic states of cells after different morphogen treatments (S2A Fig). Our result revealed that individual morphogens supported varied levels of ERK activities (S2A and S2B Fig), but only with WNT pathway activation could the RNA and protein expression of TBXT emerge (S2C–S2E Fig), a result reflective of the essential role of WNT in TBXT transcription induction [51,52].

Thus, we performed analysis combining WNT activation with other morphogens (Fig 2A). WNT activation alone engendered a high level of ERK activity, although still lower than that observed in ME differentiating cells (Fig 2A and 2B). Combining WNT with BMP4 or FGF2 activation resulted in a further elevation in ERK activity; in contrast, co-treatment with Activin A dampened ERK activity (Fig 2A and 2B). This suggests that different gastrulation morphogen combinations can lead to varying ERK activity profile during ME differentiation.

Notably, co-activation of BMP4 or FGF2 with WNT also accelerated TBXT-copGFP induction and increased its peak expression level, whereas co-activation of Activin A did not (Figs 2C and S2F), consistent with previous reports of BMP4 and FGF2 promoting mesoderm gene expression [30]. Transcriptome profiling of different morphogen combinations confirmed this heterogeneity in cell states, showing that WNT combined with BMP4 or FGF2 promoted mesoderm-specific genes, while WNT with Activin A favored endoderm-specific gene expression (Figs 2D and S2G, S1 Table). Together, these results indicate that during gastrulation, varying morphogen treatments can support diversified ERK activity profiles, which correlate with heterogeneous TBXT expression and ME lineage potentials (Fig 2C and 2D). Notably, cell cycle length remained largely unaffected across treatments (S2H Fig).

ERK quantitatively fine-tunes ME fate specification at high activity range

Consistent with previous reports, our results confirm that higher ERK activity tended to associate with a higher TBXT expression [53–56] (Fig 2B and 2C). Yet, the elevated ERK did not dramatically increase the proliferation rate (S2H Fig). This raised the possibility that either additional factors may participate in balancing proliferation, or ERK itself may exert differential control on ME differentiation and cell cycle. To explore these possibilities, we employed a synthetic system to precisely titrate ERK activity to a series of distinct levels. Specifically, we employed inhibitors targeting ERK (Ulixertinib) and its upstream kinase MEK (PD0325901), which were frequently used to modulate ERK dynamics [57–59]. A prominent reduction of average ERK activity was observed upon addition of either inhibitor in our live-cell imaging experiment (S3A and S3B Fig). Upon high concentration of either inhibitor treatment, TBXT protein failed to emerge altogether during ME differentiation (S3C Fig).

To generate a spectrum of ERK activity, we adjusted the doses of either inhibitor to obtain a range of average ERK activity (spanning from 0.6–1.6 a.u) during ME differentiation (Figs 3A and S3D). With decreasing ERK activity, we observed a gradual dampening in both TBXT-copGFP induction kinetics (Fig 3B) and its average fluorescent intensity (Fig 3C), suggesting TBXT expression is finely tuned by ERK levels. When plotting the average ERK activity with TBXT expression level, distinct ERK response ranges can be identified. Notably, in an ERK activity range between 1–1.6 a.u., the corresponding TBXT expression exhibited a positive correlation with ERK; lower than this range, TBXT expression was ERK-insensitive and barely induced (Fig 3D). When binning single cells with the similar ERK activity range from different conditions (S3E Fig), a near sigmoidal-shape correlation curve can be fitted (Fig 3E). It is noted that the TBXT expression was irresponsive to low ERK level, but was up-regulated upon increasing ERK activity, and eventually plateaued (Fig 3E). Together, these results uncovered an input-output relationship between ERK activity and TBXT expression. To look for whether this relationship also applies in a physiological context, we analyzed single-cell transcriptomics from post-implantation human embryos [60]. Among the TBXT-expressing cells in primitive streak and nascent mesoderm, where the ME fate was specified (S3F Fig), those with higher TBXT level (S3G Fig) also exhibited increased ERK target gene expression (S3H Fig), suggesting elevated ERK activity also positively connected to higher TBXT expression during gastrulation.

In contrast to the effect of ERK on TBXT expression dynamics, we observed no alteration in endogenous OCT4-GFP expression (S3I-S3K Fig; S3 Video) upon inhibiting ERK by different dosages of Ulix (S3L Fig). Similarly, the induction of other lineage marker proteins, including SOX17 during ME differentiation or PAX6 during neurectoderm differentiation, was unaffected by ERK inhibition (S3M and S3N Fig). These data collectively demonstrate that changes in ERK activity can be specifically and quantitatively interpreted by TBXT, potentiating a heterogeneity in ME differentiation states.

ERK quantitatively regulates cell cycle remodeling at a lower response range

ERK activity is well-established as a quantitative regulator of cell proliferation in somatic cells and tissues [61–63]. However, during ME differentiation and gastrulation morphogen treatments, we did not observe prominent cell cycles remodeling with varying ERK activities (Figs 1G and S2H). To directly test ERK’s influence on proliferation in stem cells and developmental context, we either inhibited ERK by PD03 or Ulix during ME differentiation, or direct knocking down of ERK1/2 and MEK1/2 (S3O and S3P Fig). Both approaches resulted in clear remodeling of cell cycle progression (S3O–S3R Fig).

Thus, we further measured the cell cycle dynamics across a spectrum of ERK activity that fine-tunes ME fates (Fig 3A). In the ERK activity range where TBXT exhibited a positive response (1–1.6 a.u., Fig 3D), we observed a minimal alteration in cell cycle length (Figs 3F and S3S). However, with the ERK activity lower than 1 a.u., the cell cycle exhibited prominent elongation with decreasing ERK activity (S3S Fig). Altogether, the correlation between ERK level and cell division rate follows an “increase-plateau” pattern, resembling a Hill function (with n = 1.4) (Fig 3G). When single cells with similar ERK activity were binned across conditions, this trend became even more apparent (Fig 3H).

The differential ERK response profile for ME specification versus cell cycle regulation suggest that ME commitment requires higher ERK activity level. This observation aligns with our morphogen combination experiments (Fig 2A), where cells with varying TBXT expression (Fig 2C) maintained similar cell cycle durations (S2H Fig). To determine whether this coordination extends to human gastrulation, we analyzed single-cell RNA-seq data from embryos [60], focusing on TBXT-expressing cells (S3F Fig). Interestingly, TBXT-high and TBXT-low cells showed comparable expression of cell cycle-promoting genes (S3G and S3T Fig), indicating a similar ERK-dependent uncoupling of differentiation and proliferation in vivo. Together, our finding revealed a quantitative mechanism by which ERK permits lineage heterogeneity without significantly altering proliferation status during development.

Efficient ME protein translation requires high ERK activity

Next, we tried to identify the regulatory node that mediates the differential ERK sensitivity by ME differentiation and cell cycle progression. As data shown above, ERK inhibitor treatment from the onset of ME differentiation abolished TBXT expression (Fig 3B). To evaluate whether ERK activity is also required for the continuous accumulation of TBXT protein, we inhibited ERK and tracked the dynamics of TBXT fluorescent signal after the emergence of TBXT-copGFP expression. TBXT-copGFP was visible around 12 hours upon ME fate induction (Fig 1B and 1D), thus inhibitor was added at different time points after 12 hours (Fig 4A). An immediate dampening of ERK was observed (Fig 4B), and this coincided with the plateauing of TBXT fluorescent intensity followed subsequently by a decline (Fig 4B and 4C). Notably, despite these changes at the protein level, TBXT RNA expression remained largely unchanged (S4A Fig).

To investigate whether elevated ERK activity may elicit a post-transcriptional function in sustaining the accumulation of TBXT protein, we firstly assessed the protein stability upon ERK inhibition (Fig 4D). Notably, dampening of ERK did not dramatically alter the half-life of TBXT protein during ME differentiation (Figs 4E, 4F and S4B). To the contrary, ERK inhibition diminished the rate of Homopropargylglycine (HPG) incorporation (S4C and S4D Fig). Given that HPG is a methionine analog specifically incorporated into the newly synthesized proteins, this reduction suggests that ERK may enhance translation efficiency in ESCs.

Therefore, to assess whether ERK activity can directly and quantitatively regulate TBXT protein translation efficiency, we conducted polysome fractionation and sequencing analysis following treatment under graded ERK inhibition (S4E Fig). This approach separates mRNAs according to their ribosome occupancy, identifying the ‘heavy’ polysome fractions (F8-11) as the transcript pool undergoing active, high-efficiency translation. The remaining fractions, including free mRNA (F2-3), 40S/60S subunits (F4-5), and 80S monosomes (F6-7), contribute negligibly to translational output. Strikingly, the mRNA of many ME specific genes, such as TBXT and EOMES, showed a reduction in the polysome fraction, accompanied by an increase in free mRNA fraction when treated with 1.25 μM Ulix or higher (Fig 4G). In contrast, pluripotency genes, neurectoderm specifier genes and cell cycle genes maintained stable polysome loading across ERK inhibition conditions (Figs 4H and S4F). Together, these data highlight a specific role of ERK in supporting high translation rates for TBXT and other ME proteins. Notably, TBXT induction was already suppressed at 0.15 μM Ulix treatment, a concentration far below that required to inhibit its translation (1.25 μM). This suggests that additional ERK-dependent mechanisms, beyond translational control, contribute to TBXT expression regulation.

Gene-specific ERK sensitivity in transcriptional control underlies the observed response ranges

In addition to translational control, ERK pathway is also involved in transcription regulation [64]. Our live tracking of TBXT dynamics revealed that WNT-induced TBXT expression can be modulated by varying ERK activity. Specifically, TBXT-copGFP signal was abolished upon PD03 or Ulix treatment, while FGF2 addition enhanced TBXT induction kinetic and expression level (Figs 2C, 5A–5C and S5A). Notably, in this case, we observed a simultaneous loss of TBXT RNA expression alongside TBXT-copGFP reduction (S5B Fig), suggesting a direct role of ERK in modulating TBXT transcription.

Given the known involvement of ERK in regulating the transcription of cell cycle genes [64–66], we next sought to determine whether ERK activity differentially controls the transcription of cell cycle and ME lineage genes. To this end, we profiled the transcriptomic changes across critical ERK activity levels, corresponding to the disruption of ME differentiation or proliferation, respectively (Figs 5D and S5C). Genes down-regulated with decreasing ERK activity exhibited two major dynamics patterns: Cluster 1 showing reduced expression starting at 0.15 μM Ulix, where TBXT expression began to be suppressed; Cluster 2 which showed reduced expression from 1.25 μM Ulix or higher, where TBXT expression was completely abolished (Fig 3B and 3C). Interestingly, Cluster 1 was enriched in genes related to mesodermal lineages and WNT pathway, while Cluster 2 contained genes associated with MAPK pathway, metabolism and proliferation (Fig 5E). Particularly, the expression of ME-specific genes, including TBXT, MIXL1, and GATA6, continuously declined from 0 to 1.25 μM Ulix treatment (Fig 5F), mirroring the ERK response dynamics of TBXT-copGFP expression (Fig 3E). On the other hand, cell cycle genes such as CCND, CCNE, and CDK2, exhibited only a moderate down-regulation at higher ERK inhibitor levels (Fig 5G). Live-cell imaging confirmed a marked post-mitotic drop in CDK2 activity with higher ERK inhibitor concentration (S5D Fig), consistent with the reported function of ERK in regulating CDK2 dynamics [31].

To elucidate the molecular underpinnings for the gene-specific transcription sensitivity to ERK, we looked into the ERK binding patterns using the human ESC ChIP-seq profiling data [67]. We observed direct binding of ERK onto the promoters of cell cycle genes such as CCND, CCNA, and MYC (S5E Fig). In contrast, ERK did not bind directly to the promoters of ME marker genes, including TBXT and EOMES. Instead, we found that ELK, a transcription factor directly phosphorylated by ERK, bound to the promoter of TBXT and other ME-specific genes (Figs 5H and S5E). We also performed ERK ChIP-seq in cells upon differentiating into ME fate, and a similar enriched ERK binding on cell cycle genes but not ME genes can be seen (Figs 5I and S5F). These findings indicate that ERK-mediated transcriptional regulation differs between ME and cell cycle-specific genes, with ELK may potentially play an intermediary role in ME gene regulation. Collectively, these results show that the transcription of ME-specific genes is more sensitive to high ERK activity than that of cell cycle genes, providing a precise quantitative mechanism for regulating these two processes by ERK.

Materials availability

All biological materials are available upon request from the lead contact upon signing appropriate materials transfer agreements (MTAs). Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Hongqing Liang ([email protected]). The codes and associated data used in this study have been deposited in Mendeley Data under URL: https://data.mendeley.com/datasets/j8f8wxgfsb/1. RNA-seq data have been uploaded in GEO (https://www.ncbi.nlm.nih.gov/geo/) with series number, GSE273286. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Cell Culture

HEK293T cells were cultured in DMEM high glucose medium (HyClone, SH30243.01) supplemented with 10% fetal bovine serum (FBS, Gibco, 10270-106) on 0.12% gelatin pre-coated dishes. Primed human embryonic stem cell (hESC) line H1 was cultured in mTeSR1 medium [77] (50 µg/ml bFGF, 0.3 µg/ml TGFβ1, 0.13 mM pipecolic acid, 51.5 mM GABA, 21.2 mM LiCl, chemically defined lipid concentrates, L-glutamine-BME and NEAA; StemCell Technologies, 85851) on Matrigel-coated plates (Corning, 354277) under 5% CO₂ and normoxia condition. H1 cells were passaged at 1:5 ratio every 4–5 days upon 5 min treatment using Gentle Cell Dissociation Reagent (StemCell Technologies, 100-0485) at 37 °C.

Mesendoderm differentiation

H1 cells were dissociated into single cells by TrypLE (ThermoFisher, 12604021) treatment at 37 °C for 5 min. One million cells were seeded into each well of a six-well plate pre-coated plate with Matrigel (Corning, 354277), and 2 mL of mTESR1 medium supplemented with 1 μM Thiazovivin was added to each well. Media was changed 1 day after plating (on day 0) to RPMI CDM2 medium [78] supplemented with 30 ng/ml Activin A (STEMCELL, 78001.1), 10 ng/ml BMP4 (R&D, 314-BP-050), 20 ng/ml FGF2 (R&D, 233-FB-010), and 6 μM CHIR99021 (R&D, 4423-050). Media was refreshed daily throughout the differentiation process.

Electroporation

H1 cells were dissociated into single cells using TrypLE (ThermoFisher, 12604021) at 37 °C for 5 min. One million cells were centrifuged at 300 g for 5 min, and the pellet was resuspended in 100 µL of OPTI-MEM. Subsequently, 10 µg of plasmid DNA was mixed with the cell suspension. The mixture was subjected to electroporation using the DECAY+/− mode of the BEX instrument (BEX, CUY21EDIT2) under the voltage of 150 V, the pulse duration of 5 ms and the interval of 50 ms. During the electro-transfer phase, the voltage was set to 20 V, with pulse duration of 50 ms and interval of 50 ms.

Construction of knock-in fluorescent reporter cell line

Fluorescent reporter knock-in cell lines (TBXT-copGFP and OCT4-GFP knock-in cell line) were constructed following protocol described by He and colleagues [79]. Briefly, the HR180PA-1 plasmid was modified (System Biosciences, HR180PA-1, PBHR100A-1) to clone the target gene fragment and used as the donor plasmid. The guide RNA (sgRNA) targeting the desired gene (TBXT or POU5F1) was designed using the website https://www.zlab.bio/CRISPOR, and cloned into the PX458 plasmid containing the Cas9 gene. Both plasmids were simultaneously electroporated into H1 cells (S1C Fig). Following electroporation, cells were cultured for 48 hours before subjected to selection using 1 μg/ml of puromycin. After 2 days of selection, single clones were isolated from the remaining cells and subjected to PCR and western blot (WB) verification. The clones were validated by PCR, western blotting and examined under fluorescence microscope. Clones with the correct reporter and gene expression, karyotype and differentiation potential were used for subsequent experiments.

Lentivirus production

HEK293T cells were cultured to ~70% confluency. The transfer plasmid, psPAX2 (Addgene #12260), and pCMV-VSVG (Addgene #8454) were mixed at a ratio of 4:4:1 and incubated with 1 mg/mL of Polyethylene Glycol (PEI) (Polysciences, 24765-1) diluted in Opti-MEM for 25 min. The mixture was then added dropwise to the HEK293T cells and gently mixed. After 48 and 72 hours, the supernatant was harvested and filtered. The virus containing supernatant was concentrated using Virus Concentration Solution (Origene, TR30026) by centrifugation at 3,500 g. The virus pellet was resuspended in PBS, aliquoted, and stored at −80 °C.

Lentiviral transduction and isolation of stable clones

For making CDK2-KTR and ERK-KTR cell lines, half a million cells were seeded in 1 ml mTESR1 supplemented with 1 μM Thiazovivin, along with 10 μL lentiviruses and polybrene for integrating the reporters of ERK-KTR, CDK2-KTR, and H2B-iRFP (S1E Fig). After 1 day of viral treatment, the medium was replaced with mTeSR1 and continuously cultured for three more days. To allow stable integration of fluorescent reporters. H1 cells were then dissociated into single cells using TrypLE (ThermoFisher, 12604021) and 5,000 cells were seeded to allow the outgrowth of single cells. Upon emergence of single clones, they were gently scraped out using a 10 µL pipette tip under microscope and transferred to Matrigel pre-coated 48-well plates in mTeSR1 medium. Upon the clonal cells grown out, the expression of the fluorescent sensors was examined under fluorescence microscope. Clones bearing the correct reporter expression, karyotype, differentiation potential were used for subsequent experiments.

Live-cell Imaging

Around 0.6 million cells were seeded into Matrigel pre-coated IBIDI confocal wells (IBIDI, 81156) with 2 ml of culture medium. Fluorescent reporter bearing cell lines were mixed with wild-type cells at a ratio of 1:100 for better capturing and segmentation of fluorescent cells. Twenty-four hours after seeding, live-cell imaging was started using a Nikon Ti2 microscope equipped with a Cage incubator H201 (OKOlab) set with constant temperature (37 °C) and air composition (normaxia 5% CO₂). CY5, TRITC, and FITC channels were utilized for imaging with the fixed laser intensities for different channels. Live-cell imaging was performed at 10 min’ interval, and the culture medium was refreshed every 24 hours during imaging. At the end of imaging, cell samples were collected for RNA extraction and qPCR analysis to assess the consistency of differentiation status. Batches of samples that showed consistent differentiation were selected for subsequent image quantification.

Tracking single cells from live-cell imaging data

We developed an automated cell tracking pipeline (S1C Fig) to index each cell in a live-cell image series in four steps: (1) We performed nuclei instance segmentation to locate cells on each image frame with an open-source platform, Cellpose [80], and assigned a unique lineage index to each cell in the first frame. (2) Between neighboring frames, we matched each cell in the first frame to a distinct cell in the second frame via a global minimization of the total cell movement. To lower computational complexity, we only considered, for each cell in the first frame, the best two matches from single-cell proximity analysis as candidates for global optimization. A cell was determined to disappear with its tracking ended if it was >10 pixels away from all cells in the second frame. (3) To track cells moving in and outside the image area, we defined a hypothetical boundary 10 pixels away from the real edge. Cells moving outside this boundary were determined to disappear with their tracking ended. Cells entering the boundary were assigned new lineage indices. (4) To determine if the remaining unmatched cells in the second frame resulted from cell division, we tried to match each of them to a distinct cell in the first frame as in step (2). A cell was considered a new lineage if it was >10 pixels away from all cells in the first frame. All lineage information was recorded.

Quantification of live-cell imaging data

To quantify the fluorescence signal in each cell, we obtained the nuclear region from nuclei segmentation and defined the cytoplasmic region from a four-pixel dilation of the nuclear region. To exclude any possible extracellular signal, we only considered the brightest 50% pixels in the defined cytoplasmic region. Raw images were de-noised with a top-hat filter [81]. Background fluorescence was estimated from pixels outside a 40-pixel dilation of all nuclei. For each fluorescent channel, the average nuclear and cytoplasmic signals of a cell were calculated from the corresponding regions with background subtraction (S1C Fig). For CDK2 and ERK KTR sensors, a ratio between the cytoplasmic and nuclear signals was computed. Due to the loss of cytoplasm and nuclei locality information during mitosis, four frames from nuclear membrane breaking down till mitotic division were excluded from quantification.

Cell cycle duration was measured between two mitotic events. Based on live-cell imaging of H2B-labeled chromatin, mitosis onset was identified by the initial splitting of the bright, condensed nuclear H2B signal, and cycle completion was marked by the recurrence of this splitting in the daughter cells. Analysis was restricted to cells that could be clearly tracked through two full mitoses.

We applied a linear mixed-effects model (LME) to test the significance of changes between different treatments and meanwhile account for variation between culture dishes. In this model, different treatment conditions were modeled as a fixed effect, and wells (data from the same treatment) were included as a random effect to account for the nonindependence of cells measured within the same culture dish. Pairwise comparisons between a reference group and a treatment group were extracted from the model’s fixed effects table.

Correlation analysis between ERK and TBXT expression or cell division

To analyze the relationship between ERK activity and TBXT intensity, ERK activity data from all treatment conditions were pooled together. The maximum and minimum values were determined, and the values in between were divided into 30 equal bins. ERK activities within bins were averaged to get the mean ERK activity for every bin. Similarly, TBXT intensity value corresponding to bin was calculated. A sigmoid function, f (x) = a./ (1 + exp(-b*(x-c))) + d, was applied to fit the ERK activity and TBXT expression level. The ERK activity-TBXT binned data were used for curve fitting, and the four parameters (a, b, c, and d) were iteratively optimized to achieve the best fit and the final fitting function.

To analyze the relationship between ERK activity and cell division rate, the same binning strategy as described for TBXT intensity was applied. The Hill function, defined as f(x)=(Vmax * x.^ n)./(Kd ^ n + x.^ n), was used to fit the data.

The ERK activity range was divided into 30 equally spaced intervals (bins). Cells within the same ERK activity bin were grouped, and the mean TBXT-copGFP fluorescence intensity was calculated for each bin to generate a data point with average level and the standard error of the mean (SEM). These bin-averaged TBXT-copGFP values were plotted against the corresponding ERK activity bins to visualize the population-level dose–response relationship. The data points were then fitted with a sigmoidal function to model the nonlinear dependence of TBXT expression on ERK activity (S3E Fig).

Flow Cytometry

H1 cells were dissociated into single cells using TrypLE (ThermoFisher, 12604021), followed by centrifugation at 300 g for 5 min to collect the cell pellet. The cells were washed in PBS and resuspended in pre-cooled 70% ethanol for fixation. Cells were then washed with PBS and then resuspended in FxCycle PI/RNase Staining Solution (ThermoFisher, F10797) for 30 min to staining for DNA. Cells were filtered through a 40 µm strainer and analyzed using a BD LSRFortessa Cell Analyzer. The flow cytometry data were analyzed using FlowJo X V10 software.

RNA extraction and RT-qPCR

Total RNA was extracted using RNAiso Plus (Takara, 9109). For cDNA synthesis, 500 ng of total RNA was utilized with the HiScript III 1st Strand cDNA Synthesis Kit + gDNA wiper (Vazyme Biotech, R312-01). Quantitative PCR (qPCR) was performed using Fast SYBR Green Master Mix (Thermo Fisher Scientific, 4385612) on the Roche LightCycler 480 (LC480) system.

Immunofluorescence staining

Cells cultured on plastic coverslips (ThermoFisher, 174969) were fixed with 4% paraformaldehyde (PFA) for 30 min at room temperature (RT), permeabilized with 0.5% Triton X-100/PBS for 30 min, and blocked in 2% bovine serum albumin (BSA) in 0.1% Triton X-100/PBS for 30 min at room temperature. Cells were then incubated with primary antibodies overnight at 4 °C, followed by incubation with fluorophore-conjugated secondary antibodies (Thermo Fisher Scientific, A-21201 and R37118) and 5 μg/mL DAPI (Solarbio, C0065) for 2 hours at room temperature. Antibodies were diluted in 2% BSA with 0.1% Triton X-100/PBS. Coverslips were mounted onto microscope slides, and images were acquired using Nikon Ti2 microscope.

Western blotting

Cell lysates in RIPA buffer with protease inhibitor cocktail or co-immunoprecipitated protein (Co-IP) samples were mixed with sample loading buffer (Bosterbio, AR1112-10) and boiled for 5 min at 95 °C to denature the proteins. It was then kept on ice before resolving using 10%–12.5% SDS-PAGE and transferred onto a polyvinylidene difluoride (PVDF) membrane (Sigma-Aldrich, 3010040001) using a wet tank transfer system. The PVDF membrane was blocked in 5% nonfat milk in PBST for one hour and then incubated with primary antibodies diluted in blocking buffer at 4 °C overnight. Subsequently, the membrane was incubated with secondary antibodies conjugated with horseradish peroxidase (HRP) (GenScript, A00098). Finally, the membrane was visualized using the Azure C300 system.

Nascent peptide labeling and detection

To detect and quantify nascent protein production, we utilized the BeyoClick HPG Protein Synthesis Kit with Alexa Fluor 555 (BeyoClick). Cells were differentiated into the ME lineage for 18 hours, and subsequently treated with varying concentrations of Ulix for 1 hour. Upon washing with PBS, cells were incubated in DMEM/F12 medium (methionine-free) containing the same concentrations of Ulix but supplemented with homopropargylglycine for 10 min. Cells were then fixed with 4% paraformaldehyde, and the Click-iT reaction using Alexa Fluor 555 detection reagents was performed in accordance with the manufacturer’s protocol. As positive control, 100 μg/mL of cycloheximide (MCE) was added 1 hour prior to the incubation with HPG. Cells were subjected to immunofluorescence (IF) staining and quantification of fluorescence signal under confocal imaging.

Polysome profiling and RNA-seq

Fifteen million H1 cells grown in 15 cm culture dishes were seeded in mTeSR1 medium supplemented with 1 μM Thiazovivin. Twenty-four hours later, 6 µM CHIR, 6 µM CHIR + 1 µM PD03, and 6 µM CHIR + 10 µM Ulix were added, respectively, into the dishes. Cell were cultured for another 30 hours before collected for polysome profiling. Polysome profiling was conducted according to the protocol reported by Xue and colleagues [82]. Briefly, the cell lysate was loaded on sucrose gradients prepared by Gradient Master (Biocomp) and using SW-41 Ti rotor at 36,000 rpm at 4 °C for 2.5 hours to separate the fraction. The fraction was analyzed by Piston Gradient Fractionator (Biocomp) according to the manufacturer’s instructions to collect fractions. Total RNA was separated into 10 fractions based on ribosome size gradients. Adjacent fractions were combined to get free RNA, 40S/60S subunits, 80S monosomes, and polysomes portions, respectively. In addition, total RNAs from each treatment group were included for RNA-sequencing. To calculate the RNA fraction for every gene, the TPM values from all four fractions for one gene were added up to obtain the total TPM. The proportion for different fractions of one gene was calculated by the TPM of individual fraction divided by the total TPM.

RNA-sequencing

Fifty thousand H1 cells were seeded onto each 96-well plate wells pre-coated with Matrigel. Twenty-four hours after seeding, the medium was replaced by 50% DMEM:F12 (Nacalai Tesque, 08460-p5) and 50% Neurobasal (Gibco, 21103049) supplemented with 1% N2 (Gibco, 17502048), 2% B27 (Gibco, 17504044), 1% L-glutamine (Gibco, 25030081), 1% NEAA (Gibco, 11140050), 0.18% 2-Mercaptoethanol (Gibco, 21985023), 50 μg/mL BSA (Biosharp, BS114-100) and different combinations of growth factors and small molecules were added. The experimental design was as shown in S2G Fig. Each treatment was conducted in triplicates, and cells were harvested at 48 hours for RNA extraction and purification. High-throughput barcode transcriptomic analysis was performed using the AccuraCode HTP OneStep RNAseq Kit (1071065) to carry out adaptor ligation and prepare RNA-seq libraries. The high-throughput sequencing results were processed on a Linux system to generate an analysis matrix, and further data analysis was conducted using the R package.

RNA-sequencing data analysis

RNA-sequencing reads were aligned to the GRCh38 reference genome using AccuraCode pipeline. We generated count data with FeatureCounts and identified differentially expressed genes (DEGs) with DESeq2. The control group for comparison was N2B27 basal medium condition, and every growth factor treatment condition was individually compared with control to identify pair-wise DEGs which had expression fold change ≥ 2.0 or ≤ −2 and an adjusted p-value cutoff less than 0.05. The pair-wise DEGs from individual conditions were combined to create an overall DEG list for subsequent analysis. Gene expression levels were then calculated in TPM for subsequent analysis. Using Mfuzz package, we clustered the DEGs based on similar pattern of expression under different condition. Next, we obtained lineage-specific gene lists from single-cell RNA-seq data of post-implantation human embryos [60] and overlapped with differentially expressed genes to identify the lineage preferences of DEGs.

ChIP-seq and data analysis

Cross-linking was performed with 1% formaldehyde (Sigma-Aldrich, 47,608–250ML-F) for 10 min at RT and quenched with 0.14 M glycine for 10 min. Following lysis (10 mM Tris-HCl pH 8.0, 0.25% Triton X-100, 10 mM EDTA, 100 mM NaCl, protease inhibitor cocktail), chromatin was sonicated to an average fragment size of 500 bp. Samples were incubated using anti-ERK2 antibodies overnight at 4 °C (Cat: sc-1647), followed by 30 µL Protein G Dynabeads for 2 hours at RT. DNA was eluted by 50 mM Tris-HCl pH 8.0, 1 mM EDTA, 1% SDS, digested with proteinase K at 60 °C overnight, and purified using the FastPure DNA Extraction Mini Kit (Vazyme Biotech, DC301) for qPCR or Illumina sequencing.

A community-curated bioinformatics pipelines “nf-core/chipseq” (v1.1.0) was used to analyze ChIP-seq data. Raw fastq files were filtered by FastQC (v0.11.8). Trim Galore (v0.5.0, https://www.bioinformatics.babraham.ac.uk/projects/trimgalore/) was used for trimming adapters. The remaining high-quality reads were mapped to human reference genome from Ensembl (GRCh38.99) by BWA (v0.7.17). PCR duplicates were identified by PICARD (v2.19.0, http://broadinstitute.github.io/picard/) and removed by SAMtools (v1.9). MACS2 (v2.1.2) was used for calling peaks. The narrow peaks with q-value (minimum FDR) < 0.05 were kept as final peaks. Bigwig tracks were generated using Deeptools (v3.4.3) by normalizing to RPKM using binsize of 10 bp. ChIP-seq signals over genomic regions were plotted by Deeptools (v3.4.3).

S1 Fig. Establishment and characterization of pluripotent stem cell reporter lines.

(A) The immunofluorescence staining pictures showing the antibody staining of TBXT protein at selective time points during ME differentiation. The violin plot on the right showing the quantification of TBXT protein fluorescence intensity that is normalized to DAPI staining and the median and quartiles of the data are shown. Scale bar: 100 μm. (B) Bar chart showing the log₂ RNA expression fold change of pluripotency and germ layer specific genes at different time points during ME differentiation in wild-type (WT) PSC lines. Data presented as mean ± SD (n = 3). (C) Schematic illustration of the automated cell tracking and quantification pipeline. The workflow is demonstrated for nonmitotic cells (top row) and mitotic cells (bottom row) tracking. Firstly, nuclei were segmented (white dashed lines) on each frame using Cellpose. For matching and tracking, cells in the current frame were matched to their locations in the previous frame (yellow dashed outlines) by minimizing total cell movement. Pink arrows indicate the lineage assignment, including the splitting of a mother cell into two daughter cells during mitosis. For signal quantification, the data collection area for the cytoplasm (blue dashed lines) is defined by a 4-pixel dilation of the nuclear region (white dashed lines). Valid signal was extracted by considering the brightest 50% of pixels within the cytoplasmic region to calculate the final Cytoplasmic-to-Nuclear (C/N) ratio. (D) Cartoon illustrating the structure of the donor construct and Cas9-sgRNA plasmid for constructing the TBXT-copGFP knock-in cell line. (E) Western blot images showing the copGFP tagged TBXT, endogenous TBXT and GAPDH protein expression in TBXT-copGFP knock-in PSC line in pluripotent state (ME differentiation -) or upon ME differentiation (ME differentiation +). (F) Cartoon illustrating the construct map for making the stable PSC lines integrated with ERK-KTR sensor, CDK2-KTR sensor and H2B-iRFP. (G and H) Bar chart showing the log₂ RNA expression fold change of pluripotency and germ layer specific genes at different time points during ME differentiation in the PSC clones stably integrated with ERK_TBXT-copGFP (G) or ERK_CDK2 sensors (H). Data presented as mean ± SD (n = 3). (I) Western blot showing ERK1/2 phosphorylation and GAPDH protein expression level in PSCs and ME differentiating cells or upon Ulix and PD03 treatment. (J) Violin plot illustrating the average slope of CDK2 accumulation in pluripotent cell cycles, or in the first and second cell cycles of ME differentiating cells. The median and quartiles of the data were shown on the violin plot. t test was performed, ****p < 0.0001. (K) Contour plots showing the CDK2 activity correlation between mother and daughter cells measured from live-cell imaging in PSCs and ME differentiating cells. Contour color representing cell density. The yellow line represented the fitted trend line with R² shown on top of the graph. The data underlying this figure are provided in “supplement figure/SF 1” folder from Mendeley Data repository: https://data.mendeley.com/datasets/j8f8wxgfsb/1.

https://doi.org/10.1371/journal.pbio.3003711.s001

(PDF)

S2 Fig. The effect of morphogen combinations on ERK activity.

(A) Heat maps illustrating the dynamics of ERK-KTR sensor activity and TBXT-copGFP fluorescence intensity over 36 hour following treatment with 6 μM CHIR, 30 ng/mL Activin A, 10 ng/mL BMP4, 20 ng/mL FGF2 or basal medium (CDM2) to activate corresponding morphogen pathways. Cells were imaged immediately upon replacing with the imaging medium. More than 600 single cells were quantified from all morphogen conditions. (B) Violin plot displaying the distribution of average ERK activity over 36 hour in ME differentiating medium or upon individual treatment with 6 μM CHIR, 30 ng/mL Activin A, 10 ng/mL BMP4, 20 ng/mL FGF2 or in CDM2 basal medium. (C) Bar plots showing the log₂ RNA expression fold change of TBXT at 36 hour upon 6 μM CHIR, 30 ng/mL Activin A, 10 ng/mL BMP4, or 20 ng/mL FGF2 treatment compared with PSC and ME control. t test was performed, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (D) Graph showing the population trajectories of TBXT-copGFP fluorescent intensity over 36 hours in ME differentiating medium or upon individual treatment with 6 μM CHIR, 30 ng/mL Activin A, 10 ng/mL BMP4, 20 ng/mL FGF2 or in CDM2 basal medium. The population means were shown in the solid line and SEMs were shown in shades of the corresponding color. (E) Violin plot displaying the distribution of average TBXT-copGFP expression level over 36 hour in ME differentiating medium or upon individual treatment with 6 μM CHIR, 30 ng/mL Activin A, 10 ng/mL BMP4, 20 ng/mL FGF2 or in CDM2 basal medium. (F) Violin plot displaying the distribution of average TBXT-copGFP fluorescent intensity over 36 hour upon individual treatment of 6 μM CHIR, 6 μM CHIR +30 ng/mL Activin A, 6 μM CHIR +10 ng/mL BMP4, 6 μM CHIR + 20 ng/mL FGF2 or in CDM2 basal medium. (G) The schematics showing the experimental design for transcriptome profiling of morphogen treatment. Cells were treated with CHIR as well as in combination with 100 ng/ml Activin A, 50 ng/ml BMP4, 50 ng/ml FGF2. Morphogens were applied at indicated concentrations and every condition was done in triplicates. Treated cells under different conditions were collected at 48 hours for transcriptome profiling. (H) Violin plot displaying the distribution of cell cycle length upon 6 μM CHIR, 6 μM CHIR + 30 ng/mL Activin A, 6 μM CHIR + 10 ng/mL BMP4, 6 μM CHIR +20 ng/mL FGF2 treatment or in CDM2 basal medium. For (B), (E), (F), and (H), the median and quartiles of the data were shown on the violin plot. A linear mixed-effects model was applied and pairwise comparisons were performed against the ME or CHIR control group, ns: not significant (p > 0.05), ***p < 0.001, ****p < 0.0001. The data underlying this figure are provided in “supplement figure/SF 2” folder from Mendeley Data repository: https://data.mendeley.com/datasets/j8f8wxgfsb/1.

https://doi.org/10.1371/journal.pbio.3003711.s002

(PDF)

S3 Fig. ERK activity in relationship with TBXT expression during ME differentiation.

(A) Violin plots showing the distribution of average ERK activity in PSCs and PSCs treated with 1 μM PD03 or 10 μM Ulix. Cells were imaged immediately upon replacing with the treatment medium. (B) Florescent images showing the ERK-KTR sensor activity with or without 1 μM PD03 or 10 μM Ulix treatment in ME differentiating cells. Scale bar: 20 μm. (C) Western blot showing TBXT-copGFP and GAPDH protein expression level in ME differentiating cells and ME differentiating cells treated with 1 μM PD03 or 10 μM Ulix for 24 hours. (D) Heat maps illustrating the dynamics of ERK-KTR sensor activity and TBXT-copGFP fluorescence intensity over 36 hour following treatment with different dosages of PD03 or Ulix during ME differentiation. Ulix was applied at 0.05, 0.15, 1.25, 2.5, 5, and 10 μM and PD03 was applied at 0.01, 0.1, 0.2, and 0.8 μM, respectively, at the beginning of ME differentiation. More than 1,700 single cells were quantified across different conditions. (E) Analysis of TBXT expression as a function of ERK activity: (Left) Single-cell scatter plot of TBXT-copGFP intensity versus ERK activity (C/N ratio), and cells were grouped into 30 bins based on ERK activity; (Middle) Red dotted boxes showing representative intervals based on ERK activity; (Right) Bin-averaged TBXT values were calculated for each interval and plotted against the corresponding ERK activity, revealing a dose–response relationship. (F) Violin plot showing the distribution of TBXT gene expression TPM in single cells from different lineages of the gastrulating human embryos [60]. (G) Bar plot showing the expression TPM of TBXT genes in all the identified TBXT-expressing single cells from the gastrulating human embryos. The single cells were ranked by the TBXT expression level and the top or bottom 50% of TBXT-expressing cells were denoted as TBXT-high and TBXT-low, respectively. (H) Bar plot showing the average expression level of ERK pathway target genes (DUSP5, DUSP6, MAPK1, MAP2K1, and YWHAZ) in TBXT-high and TBXT-low cells from the gastrulating human embryos. (I) Confocal images showing the fluorescent signals of OCT4-GFP and H2B-iRFP in the stable PSC line in pluripotent state or after ME differentiation. Scale bar: 100 µM. (J) Graph showing the population trajectories of OCT4-GFP fluorescent intensity over 36 h in PSCs and ME differentiating cells treated with 1.25, 2.5, 5, and 10 μM Ulix. Cells were imaged immediately upon replacing with the imaging medium. The population means were shown in the solid lines and SDs were shown in shades of the corresponding color. More than 610 single cells were quantified were across different conditions. (K and L) Violin plots illustrating the average OCT4-GFP fluorescent intensity (K) and average ERK activity (L) in PSCs and ME differentiating cells treated with different doses of Ulix. (M) Confocal images showing the expression of SOX17 protein at different time points during ME differentiation with or without 10 μM Ulix treatment (left). Scale bar: 100 μm. Violin plot on the right quantifying SOX17 protein fluorescence intensity normalized against DAPI signals across single cells (right). (N) Confocal images showing the expression of PAX6 protein at different time points during NE differentiation with or without 10 μM Ulix treatment (left). Scale bar: 100 μm. Violin plot on the right quantifying PAX6 protein fluorescence intensity normalized against DAPI signals across single cells (right). (O) Bar plots showing the flow cytometry analysis of cell cycle dynamics (G1, S, G2/M) in MEK/ERK double-knockdown (KD) and control KD (NC) cells 36 hour post ME differentiation. (P) Bar plots showing the flow cytometry analysis of cell cycle dynamics (G1, S, G2/M) in ME differentiating cells without or with 1 μM PD03 and 10 μM ULIX treatment. (Q and R) Violin plots showing the cell cycle length (Q) and cell division rate (R) in PSCs and PSCs treated with 1 μM PD03 or 10 μM Ulix for 24 hours. The median and quartiles of the data were shown on the violin plot. t test was performed. ns, ****p < 0.0001. (S) Correlation plots showing the cell cycle length plotted against average ERK activity under different concentrations of PD03 and Ulix treatment in ME differentiating cells or in PSCs for 24 hours. ME-1 and ME-2 refer to the first and second full cell cycle measured during ME differentiation. Every dot represents the average values in one treatment condition, with error bars representing the SEM. (T) Bar plot showing the average expression level of cell cycle-promoting genes (CCNL1, ORC1, ORC2, PCNA, MCM2, MCM7, CDK1, SMC1A, CCNB1, CCND2, CCND3, and CCNE2) in TBXT-high and TBXT-low cells from the gastrulating human embryos. The cell cycle-promoting genes that were expressed in more than 10% of single cells from the human gastrulating embryo scRNA-seq data were selected. For data in (A), (K), (L), (Q), and (R), the median and quartiles of the data were shown on the violin plot. A linear mixed-effects model was performed and pairwise comparisons were made against the respective control group, ns: not significant (p > 0.05), ***p < 0.001, ****p < 0.0001. For data in (M) and (N), the median and quartiles of the data were shown on the violin plot. One way ANOVA test was performed, ns, not significant (p > 0.05), ****p < 0.0001. The data underlying this figure are provided in “supplement figure/SF 3” folder from Mendeley Data repository: https://data.mendeley.com/datasets/j8f8wxgfsb/1.

https://doi.org/10.1371/journal.pbio.3003711.s003

(PDF)

S4 Fig. High ERK activity regulates TBXT protein translation during ME differentiation.

(A) Bar chart showing the log₂ RNA expression fold change of TBXT gene at different time points after inhibiting ERK at 16 hour of ME differentiation. Data were presented as mean ± SD, n = 3. (B) A replicate of western blot showing TBXT-copGFP and GAPDH protein level as described in Fig 4D for ME differentiating cells without or with 10 μM Ulix treatment. (C) Confocal images showing the incorporation of the RFP labeled HGP into nascent peptides in ME differentiating cells without or with 0.15, 1.25, or 5 μM Ulix treatment. 50 μg/ml cycloheximide (CHX) was used as a control. Blue channel indicates nuclear staining by Hoechst. Scale bar: 100 μm. (D) Box plot showing the quantification of HGP-RFP signals in single cells under different conditions in (C). t test was performed, ****p < 0.0001. (E) Gradient centrifugation fractionation plot of polysomes in 6 μM CHIR treated cells upon different dosages of Ulix addition (0, 0.15, 1.25, and 5 μM). The X-axis represents the fractions obtained by centrifugation through a 10%−45% sucrose gradient. Fractions 1–10 denotes the ribosome fractions obtained through 10 consecutive separations. Fractions representing free RNAs (F1-2), ribosomal subunits 40S/60S (F3-4), monosomes 80S (F5-6) and polysomes (F7-10) are indicated by different colors and collected respectively. The Y-axis represents the absorbance at 260 nm. (F) Plots showing the proportions of mRNA bound to different ribosome fractions upon different dosages of Ulix addition (0, 0.15, 1.25, and 5 μM) for 24 hours in 6 μM CHIR treated cells. The RNA expression was detected by RNA-seq for house-keeping genes (ACTIN, GAPDH), pluripotent genes (SOX2, POU5F1), endoderm genes (FOXA1, FOXA2) and ectoderm genes (MAP2, PAX6). The X-axis represents different ribosome fractions, while the Y-axis represents the percentage of total RNA in every fraction. The data underlying this figure are provided in “supplement figure/SF 4” folder from Mendeley Data repository: https://data.mendeley.com/datasets/j8f8wxgfsb/1.

https://doi.org/10.1371/journal.pbio.3003711.s004

(PDF)

S5 Fig. ERK activity differentially regulates ME gene and cell cycle gene transcription.

(A) Violin plot illustrating the average TBXT-copGFP fluorescent intensity over 36 hour in 6 µM CHIR, 6 µM CHIR+1 µM PD03, 6 µM CHIR+10 µM Ulix conditions. The median and quartiles of the data were shown on the violin plot. A linear mixed-effects model was performed, ****p < 0.0001. (B) Bar chart illustrating the RNA expression fold change of TBXT gene at different time points upon treatment with CHIR, or CHIR with 1 µM PD03 or 10 µM Ulix. (C) Heat maps on the left showing the Z-score of RNA expression for up-regulated genes in CHIR treated cells upon different dosages of Ulix treatment (0, 0.15, 1.25, and 5 μM). Genes were clustered based on their different response dynamics to different dosages of Ulix. Plots on the right showing the enriched ontology terms for the up-regulated gene clusters. (D) Graph showing the population trajectories of CDK2-KTR sensor dynamics in ME differentiating cells or upon 1.25 μM or 5 μM Ulix treatment. The population means were shown as solid lines and standard deviation were shown in shades of the corresponding color. Individual cells in the population were synchronized in silico by the time of mitosis. Cells were imaged immediately upon replacing with the ME medium. (E) Track views showing the ERK2 (blue), ELK1 (orange) ChIP-seq signals [66] and input (gray) signal in regions nearby ME specific genes (TBXT, EOMES, GATA6, and DKK1) and cell cycle genes (CCND1, MYC, CCNA2, and JUN), respectively. (F) Track views showing the ERK2 (blue) ChIP-seq signals and input (gray) signal in regions nearby ME specific genes (TBXT, EOMES, GATA6, and DKK1) and cell cycle genes (CCND1, MYC, CCNA2, and JUN), respectively. The data underlying this figure are provided in “supplement figure/SF 5” folder from Mendeley Data repository: https://data.mendeley.com/datasets/j8f8wxgfsb/1.

https://doi.org/10.1371/journal.pbio.3003711.s005

(PDF)

S1 Video. Live-cell imaging of ERK-TBXT cell line in pluripotent state and ME differentiation.

Pluripotent stem cells expressing the ERK-RFP sensor (red), TBXT-copGFP fusion protein (green), and H2B-iRFP fusion protein (blue) were imaged in PSC and ME differentiation medium for 36 hours with 10-minute intervals using a 20x objective lens.

https://doi.org/10.1371/journal.pbio.3003711.s007

(MP4)

S2 Video. Live-cell imaging of ERK-CDK2 sensor cell line in pluripotent state and ME differentiation.

Pluripotent stem cells expressing the CDK2-venus sensor (green), ERK-RFP sensor (red), and H2B-iRFP fusion protein (blue) were imaged in PSC and ME differentiation medium for 36 hours with 10-min intervals using a 20x objective lens.

https://doi.org/10.1371/journal.pbio.3003711.s008

(MP4)

S3 Video. Live cell imaging of ERK-OCT4 cell line in pluripotent state and ME differentiation.

Pluripotent stem cells expressing the ERK-RFP sensor (red), GFP-OCT4 fusion protein (green), and H2B-iRFP fusion protein (blue) imaged in PSC and ME differentiation medium for 36 hours with 10-min intervals using a 20x objective lens.

https://doi.org/10.1371/journal.pbio.3003711.s009

(MP4)